The Role of Microenvironment in Development of Skin Cancer and Metastasis

Home , Balloon cell nevus, Benign melanocytic nevus, Nevus anemicus

Arash Chitsazan BSc MSc

A thesis submitted for the degree of Doctor of Philosophy at The University of Queensland in 2018 Faculty of Medicine Abstract

The overarching aim of this thesis is to study the mechanisms by which giant congenital nevi form and to study if such mechanisms are targetable. Virtually nothing is known about genes conferring susceptibility to the development of giant congenital nevi, largely untreatable lesions which increase the risk of patients for development of neurocutaneous melanocytoma and malignant melanoma. To discover such genes, we utilized a nevus- prone transgenic mouse model together with the Collaborative Cross (CC), a genetic resource for discovery of genes for complex diseases. It not only allows gene discovery, but also a systems genetics approach that enabled us to determine how the gene functions in vivo to modify the nevogenesis. We also show that the identity of somatic oncogenic mutation (e.g. in NRAS) does not always define the histopathological subtype of nevus, as is generally thought. In sum, we have performed an unbiased genetic screen and discovered a novel gene (Cdon), a dependence receptor of sonic hedgehog (Shh) as the modifier gene for nevus exacerbation. We then used three different mouse models to validate it as the causal gene and determine its mechanism of action. We also show that Shh pathway components are active in the epidermis adjacent to human congenital nevi (Chapter 2 and 3). In summary this is the first time a gene has been successfully mapped and functionally validated using CC. In Chapter 4 we discuss the result from an unbiased genetic screen to map the modifier gene for development of sub-cutaneous (hypodermis) metastasis using 45 CC strains. Gja1 which encodes a gap junction subunit connexion 43 (Cx43) is the best candidate. Our gene expression and protein expression studies suggest that Gja1 could be a strong candidate for further study of sub-cutaneous metastasis.

In Chapter 5 we went on to study the gene expression of normal skin, melanocytoma and malignant melanoma samples from our NRAS-Cdk4 mouse model to study the transformation and progression of melanocytic lesions. Dusp6 a negative regulator of MAPK pathway was highly upregulated with transformation and progression. Our protein expression study validated this result on murine and human nevi versus malignant melanoma samples.

Declaration by author

This thesis is composed of my original work, and contains no material previously published or written by another person except where due reference has been made in the text. I have clearly stated the contribution by others to jointly-authored works that I have included in my thesis.

I have clearly stated the contribution of others to my thesis as a whole, including statistical assistance, survey design, data analysis, significant technical procedures, professional editorial advice, financial support and any other original research work used or reported in my thesis. The content of my thesis is the result of work I have carried out since the commencement of my higher degree by research candidature and does not include a substantial part of work that has been submitted to qualify for the award of any other degree or diploma in any university or other tertiary institution. I have clearly stated which parts of my thesis, if any, have been submitted to qualify for another award.

I acknowledge that an electronic copy of my thesis must be lodged with the University Library and, subject to the policy and procedures of The University of Queensland, the thesis be made available for research and study in accordance with the Copyright Act 1968 unless a period of embargo has been approved by the Dean of the Graduate School.

I acknowledge that copyright of all material contained in my thesis resides with the copyright holder(s) of that material. Where appropriate I have obtained copyright permission from the copyright holder to reproduce material in this thesis and have sought permission from co-authors for any jointly authored works included in the thesis.

Publications during candidature

Peer-reviewed papers:

Chitsazan, A., Ferguson, B.,Villani, R., Handoko, HY., Mukhopadhyay, P., Gabrielli, B., Mooi, WJ., Soyer, HP., Lambie, D., Khosrotehrani, K., Morahan, G., and Walker, GJ. Keratinocyte Sonic Hedgehog Up-regulation Drives the Development of Giant Congenital Nevi via Paracrine Endothelin-1 Secretion. J Investigative Derma. Accepted for publication.

Chitsazan, A., Ferguson, B., Ram, R., Mukhopadhyay, P., Handoko, H.Y., Gabrielli, B., Soyer, P.H., Morahan, G., and Walker, G.J. A mutation in the Cdon gene potentiates congenital nevus development mediated by NRAS(Q61K). Pigment Cell Melanoma Res. 2016;29(4):459-64

Young, A., Ngiow, S.F., Madore, J., Reinhardt, J., Landsberg, J., Chitsazan, A., Rautela, J., Bald, T., Barkauskas, D.S., Ahern, E., Huntington, ND., Schadendorf, D., Long, GV., Boyle, GM., Hölzel, M., Scolyer, RA., Smyth, MJ. Targeting Adenosine in BRAF-Mutant Melanoma Reduces Tumor Growth and Metastasis. Cancer Res. 2017;77, 4684-4696.

Publications included in this thesis

Conception and design (85%) Chitsazan A Analysis and interpretation (82.5%) Drafting and production (82.5%)

Ferguson B Conception and design (2.5%)

Ramesh R Analysis and interpretation (2.5%)

Mukhopadhyay P Analysis and interpretation (2.5%)

Handoko HY Conception and design (2.5%)

Drafting and production Gabrielli B (2.5%)

Soyer HP Analysis and interpretation (2.5%)

Morahan G Drafting and production (5%)

Walker GJ Conception and design (10%) Drafting and production (10%) Analysis and interpretation (10%)

Conception and design (85%) Chitsazan A Analysis and interpretation (83.5%) Drafting and production (80.5%)

Ferguson B Conception and design (2.5%)

Villani R Conception and design (2.5%)

Handoko HY Conception and design (2.5%)

Mukhopadhyay P Analysis and interpretation (2.5%)

Gabrielli B Drafting and production (2%)

Mooi WJ Drafting and production (2%)

Soyer HP Analysis and interpretation (2%)

Lambie D Analysis and interpretation (2%)

Khosrotehrani K Drafting and production (2%)

Morahan G Drafting and production (5%)

Conception and Design (7.5%) Walker GJ Drafting and production (8.5%) Analysis and interpretation (10%)

Contributions by others to the thesis

No contributions by others.

Statement of parts of the thesis submitted to qualify for the award of another degree

None. Research Involving Human or Animal Subjects

Research involving human subjects: Approval number: 2011001201/HREC/11/QPAH/363 Approving committee: Human research ethics committee A & B

Research involving animal subjects: Project number: P240 Approval number: A98004M Approving committee: QIMR Berghofer Medical Research Institute Animal Ethics Committee

Acknowledgements

I owe a debt of gratitude to all people who have helped me during my candidacy. A huge thanks to Graeme for his meticulous mentorship and heart warming friendship. To Brian for his support, guidance and friendship. To Peter for his help and enthusiasm with diagnosis of murine melanocytic lesions. Big shut out to Walker lab, G floor staff, QIMRB histology and microscopy departments. Thanks to the UQDI and UQ for their financial support. Finally, a huge thanks to my family and friends for their love and support.

Financial support

This project was funded by Melanoma Research Alliance, Washington DC., USA.

Keywords collaborative cross, ednothelin-1, giant congenital nevus, melanocytoma, melanoma, cdon, shh, lde225, bosentan

Australian and New Zealand Standard Research Classifications (ANZSRC)

ANZSRC code: 111299, Oncology and Carcinogenesis not elsewhere classified, 80%

ANZSRC code: 111202, Cancer Diagnosis, 10%

ANZSRC code: 111205, Chemotherapy, 10%

Fields of Research (FoR) Classification

FoR code: 1112, Oncology and Carcinogenesis, 100%

List of Abbreviations used in the thesis

α-Melanocyte-stimulating hormone α-MSH adrenocorticotropic hormone ADH basal cell carcinoma BCC collaborative cross CC congenital melanocytic nevi CMN dysplastic nevus syndrome DNS epithelial to mesenchymal transition EMT ednothelin-1 EDN1 endothelin receptor B EDNRB human keratin 1 HK1 hepatocyte growth factor HGF leukemia inhibitory factor LIF malignant melanoma MM non-melanoma skin cancers NMSC sonic hedgehog SHH stem cell factor SCF senescence-associated β-galactosidase SAβ-gal ultraviolet radiation UVR

Table of Contents

1.1 Skin ……………………………………………………………………………………...... ….19

Sonic hedgehog signaling in skin homeostasis ……………………………………….……..20

1.2 The melanocyte lineage in development ………………………………………………….22

1.3 Melanocyte growth in culture ……………………………………………………….………24

1.4 Non-cell autonomous effect of keratinocytes on melanocyte proliferation and differentiation ………………………………….…………………………………………..…..…24

1.5 Keratinocyte cytokines which are melanocyte mitogens ………………………....…….25

Endothelins ………………………………….…………………………..……….………………25

Stem cell factor ………………………………….………………………………………………..26

Leukemia inhibitory factor ……………………………….…………………………………..….26

Hepatocyte growth factor ………………………………….……………………………………26

1.6 Melanocytic nevi ………………………………….…………………………………………26

Congenital nevi ………………………………….………………………………………………28

Acquired nevi ………………………………….…………………………………………………28

1.7 Dermoscopic patterns of melanocytic nevi ………………………………….…………....30

1.8 Skin type related nevus pattern ………………………………….……………………..….31

1.9 Nevogenesis ………………………………….…………………….………………………..31

1.10 Nevi as a risk factor for melanoma …………………….…………….………….……….33

1.11 Transgenic model used for current research …………………………….……………..35

Cdk4R24C/R24C::Tyr-NRASQ61K mouse model ………………..……………….…………..36

1.12 Natural genetic factors that regulate nevus development and density……………………………………………………………………………………....……..39

1.13 Collaborative Cross ……………………………………………………………….….……39

1.14 Successful precedent for the “transgene modifier of cancer” …………………..……41

1.15 Forward and reverse genetic study ………………………………………………..….….42

2.1 Manuscript 1 …………………………….………….………………………………....…….45

2.2 Contribution ……………………………….……………………………………………....….45

3.1 Manuscript 2 ……………………………….………………………….……………………..90

3.2 Contribution……………………………….……………………….………………….………90

4.1 Introduction ………………………………………..………………….…………………….124

4.2 Hypothesis ………………………………………..……………………….………………..125

4.3 Materials and Methods ………………………………………..………….…………….…127

Mice and phenotyping ………………………………………..…………….………………....127

Genetic analyses ………………………………………..………………….……………….…127

Gene expression study of genes differentially expressed during anagen and telogen ………………………………………..………………………………………….……...…..……127

Assessment of RNA quantity and quality from frozen sections ……….……………..…..127 cRNA Synthesis ……………………………………..…………………………………………127

Analysis of cRNA size ……………………………………..…………………….…....……….128

Expression array analysis ……………………………………..…….………………….……..128

Haematoxylin and eosin staining protocol ……………………………………..……………128

Multiplexed fluorescent immunohistochemistry with Tyramide signal amplification ………………………………………………………………………………..……….….……….128

4.4 Results ………………………………………..……………………………….……….……130

Histopathology of murine MS ……………………………………..…………..……..…….….130

Genetic background affects MS count ………………………………………………..……..132

Gene expression analysis of candidate genes ………………………………………….…..135

In vivo study of Connexin 43 expression …………………………………………………..137

4.5 Discussion ……………………………………..………………………………………..…..140

5.1 Introduction ……………………………………..………………………………………..…148

5.2 Hypothesis ……………………………………..……………………………………….…..149

5.3 Methods ……………………………………..…………………………………………..…..150

Mouse melanoma model ……………………………………..…………………….………….150

Histopathological classification of melanocytic lesion …………………………………….150

Non contact laser capture microdissection ……………………………………..….………..151

Preparation of slides ……………………………………..………………………….…………151

Sample collection ……………………………………………………………………………….152

Formalin Fixed Parafﬁn Embedded (FFPE) and PAXgene Fixed Paraffin Embedded (PFPE) Sections ……………………………………….……………………………………….152

Flash frozen sections ………………………………………..………………………………..152

Assessment of RNA quantity and quality from FFPE and PFPE sections ………………………………………..…………………………………………………………..152

Assessment of RNA quantity and quality from frozen sections ………………………………………..………………………………………………….……….153 cRNA Synthesis ………………………………………..………………………………………153

Analysis of cRNA size ………………………………………..…………………………..……153

Expression array analysis ………………………………………..…………………………...153

Immunohistochemistry and multiplexed fluorescent immunohistochemistry with Tyramide signal amplification ………………………………………..…………………………………...154

5.4 Results ………………………………………..…………………………………………….156

Whole-genome gene expression study of murine melanoma progression ……………………………………………………………………………………………………156

Laser capture microdissection …………………………….…………..……………………..158

In vivo validation of Dusp6 expression changes during melanocytic lesion progression ………………………………………..…………………………………………………………...160

DUSP6, ETS1, and BRN2 expression in human nevi and thin melanoma ………………161

5.6 Discussion ………………………………………..…………………………….………..….166

5.7 Future aims ………………………………………..…………………………….………….169

6.1 Summary of key findings and discussion ……………………………………..………..173

Melanocytes reach the dermis by breaking hair follicles ………………………………….173

Somatic NRAS mutation is not the sole driver of nevogenesis, microenvironment can also play a role ……………………………………….……………………………….…………..….173

The Cdon-Shh-Edn1 axis activation exacerbates nevogenesis ………………………….174

A region on murine chromosome 10 is linked to the development of melanocytic microsatellite ………………………………………..……………………………………….….176

The dual role of Dusp6 in development of melanocytic lesions ………………………….179

6.2 Challenges and shortcomings - Chapter 2 ………………………………………….….180

Scoring nevus density of pigmented CC strains …………………………………….……..180

6.3 Challenges and shortcomings - Chapter 3 ……………………………………..………180

Orthogonal approaches measuring gene and protein expression in skin ………………………………………………………………………………………………..…..180

Lack of GWAS study on human large and giant congenital nevi ……………………………………..……………………………………………………………..180

6.4 Challenges and shortcomings - Chapter 4 …………………………………………….182

6.5 Challenges and shortcomings - Chapter 5 …………………………………………….182

6.6 Future research direction ……………………………………..………………………….182

Future studies stem from chapter 3 ……………………………………..……….…….….…182

Future studies stem from chapter 4 ………………………….…………..………..…………183

Future studies stem from chapter 5 …………………………….………….……...…………183

6.7 Conclusions ……………………………………..…………………………………………184

References……………………………………………………..………………………….……186

List of Figures and Tables

Figure 1.1 ǀ The skin structure …………………………………………………………………19

Figure 1.2 ǀ Proposed model of Shh signaling pathway regulation ……………………….21

Figure 1.3 ǀ An overview of melanocyte development ………………………………………23

Figure 1.4 ǀ Schematic image showing the action of keratinocyte-derived cytokines on the proliferation and differentiation of mammalian melanocytes …………………………….…25

Figure 1.5 ǀ Color of melnocytic nevi is dependent on location in skin layers ………………………………………………………………………………………………….….27

Figure 1.6 ǀ Classification of congenital and acquired melanocytic nevi ………………………………………………………………………………………………….….29

Figure 1.7 ǀ Dermascopic pattern of nevi ……………………………………………….…….30

Figure 1.8 ǀ Skin type dictates the color of nevi ……………………………………………...31

Figure 1.9 ǀ Schema of melanoma progression from a nevus ……………………………..33

Figure 1.10 ǀ Clinical image of a patient with high mole count who is at high risk of developing melanoma …………………………………………………………………………..34

Figure 1.11 ǀ H&E stain of the band like nevus specific to the Cdk4::Tyr-NRAS mice ……………………………………………………………………………………………………..37

Figure 1.12 ǀ Schematic of nevi in Cdk4::NRAS mouse skin looking from above ………………………………………………………………………………………………….….38

Figure 1.13 ǀ Staging schema for murine melanoma ……………………………………….38

Figure 1.14 ǀ Collaborative cross is a typical eight-way RI strain …………………….…….40

Figure 1.15 ǀ Multiple diverse phenotypic assays using collaborative cross ………………………………………………………………………………………………..……41

Figure 1.16 ǀ Genome wide scan for the mouse albino locus ………………………….…..42

Figure 4.1 ǀ Microsatellite showing angiotropism ………………………..……………….…126

Figure 4.2 ǀ MS usually appear in the vicinity of vessels …………………………………..129

Figure 4.3 ǀ MSs creates a special niche in the hypodermis ……………………………..131

Figure 4.4 ǀ nevogenesis and melanomagenesis in Tyr::CreERT2; BrafCA; Ptenlox/lox (TBP) mice ………………………………………………………………………………………132

Figure 4.5 ǀ Nodal metastasis appear early in mice with melanocytoma ………………………………………………………………………………………………..…..134

Figure 4.6 ǀ Genetic background dictates MS count of NRAS::Cdk4 mice ……………………………………………………………………………………………….……136

Figure 4.7 ǀ Mapping the modifier gene for development of MS ……………………….…139

Figure 4.8 ǀ Gja1 expression is highly correlated with known metastatic genes ……………………………………………………………………………………………….…....141

Figure 4.9 ǀ Connexin 43 is expressed in the murine deep dermis and hypodermis ………………………………………………………………………………………………….…143

Figure 4.10 ǀ CX43 expression in human melanocytic lesions ……………………………145

Figure 5.1 ǀ Regulation of MAPK pathway via ERK dephosphorylation ……………………………………………………………………………………………….……149

Figure 5.2 ǀ Whole-genome gene expression study of stage 3 MM vs. normal skin ……………………………………………………………………………………………………155

Figure 5.3 ǀ Whole-genome gene expression study of melanoma progression ……………………………………………………………………………………………………157

Figure 5.4 ǀ Whole-genome gene expression study of microdissected nevi and melanoma samples using LCM ……………………………………………………………………………159

Figure 5.5 ǀ Qsep 100 analysis of total RNA extracted from flash frozen blocks RNA isolated post-LCM and RNA ladder versus negative control ………………………………………………………………………………………….…..…….160

Figure 5.6 ǀ Dusp6 is up-regulated with melanoma progression ………………….……. 161

Figure 5.7 ǀ Immunofluoresence with tyramide signal amplification (IF-TSA) of human congenital nevi and benign nevus with a panel of DUSP6, SOX10, ETS1 and BRN2 antibodies ……………………………………………………………………………………..…163 17

Figure 5.8 ǀ Immunofluoresence with tyramide signal amplification (IF-TSA) of thin melanoma with a panel of DUSP6, SOX10, ETS1 and BRN2 antibodies ……………………………………………………………………………………………………164

Figure 5.9 ǀ Quantitation of DUSP6, BRN2, and ETS1 in human melanocytic lesions ………………………………………………………………………………………………..…..165

Figure 6.1 ǀ Cx43 mediates brain metastasis …………………………………………..…...178

Figure 6.2 ǀ Convergence of signaling pathways under the activity of MAPK could exacerbates the effect of somatic NRAS or BRAF mutations…………………………………………………………………………………..…….181

Appendix 1: List of genes positively (red) and negatively (yellow) correlated with Gja1 expression ………………………………………………………………………………....……221

Appendix 2: Top 10 genes correlated with candidate genes from angiotropism linkage…………………………………………….………………………………………….…225

Appendix 3: List of gene differentaly expressed between S3 and normal tail skin …………………………………………..……………………………………………..………...231

Appendix 4: Gene expression profiling S3 vs. S1 melanoma ……………………………………………………………………………………………………235

Appendix 5: List of genes differentially expressed between human metastasis and primary melanoma …………………….…………………………………………………………………245

Appendix 6: Approval for research involving human subjects…………………………………………………………………………………….……246

Appendix 7: Approval for research involving animal subjects…………………………………………………………………………………………247

Literature Review

1.1 Skin

Diverse cell types, including keratinocytes, melanocytes, and immune cells, form a complex tissue called skin as a protective barrier in mammals. The epidermis is the outermost layer of skin. It contains different stem cell populations located in a complex structure called hair follicle. Hair follicles have multiple tasks including hair generation, anchoring of sensory neurons, arrector pili muscles, and blood vessels (Figure 1.1). The majority of skin stem cells and their niches are located in hair follicles, making it an interesting model for studying the interaction between different stem cell populations and their niches.

Hsu 2014

Figure 1.1 ǀ The skin structure. Skin is composed of epidermis, dermis and hypodermis (fat layer). Hair follicles are composed of three areas, infundibulum, junctional zone, and bulge. Melanocyte stem cells reside in the bulge area (from Hsu 2014).

Also, disruption of such interactions has huge influence in development of hair follicle related cancers e.g. basal cell carcinoma. Dermis and hypodermis make up the middle and inner layer of skin. Melanocytes reside in the epidermis and hair follicle in human. During neoplastic hyperproliferation, melanocytes could invade the dermis or hypodermis and occupy one or multiple layers of skin.

Murine skin has a higher density of hair follicles while human skin has longer interfollicular epidermis. Also, human skin is composed of thicker epidermis and dermis while the presence of rete ridges is only seen in human epidermis. Murine epidermis can also be differentiated from human epidermis due to presence of Vγ5+ dendritic T cells which are absent in human epidermis (Pasparakis 2014).

Sonic hedgehog signaling in skin homeostasis

The sonic hedgehog (SHH) pathway is a critical factor in the regulation of skin homeostasis. SHH signaling is mediated through five membrane bound receptors; CDON, BOC, GAS1, PTCH1 and SMO (Mehlen 2014). SMO is a positive regulators of SHH while PTCH1 and CDON (Delloye-Bourgeois 2013) negatively regulates the pathway via inhibition of SMO. Activation of SMO in keratinocytes results in the upregulation of transcription factors such as GLI1 and secretion of factors such as ET-1, TGFB2, WIF1 and Sema3E (Youssef 2012). This activation also has an impact on the hair follicle stem cell niche which is evident by upregulation of stemness markers such as SOX9, FOXC1 and TCF3 (Youssef 2012).

Upregulation of the SHH pathway is a critical feature of the development of basal cell carcinoma (BCC). The American Cancer society estimates that in 2012, 5.4 million cases of non-melanoma skin cancers (NMSCs) were diagnosed in 3.3 million people, of which approximately 8 in 10 cases would have been BCC. Pigmentation of BCCs is dependent on ethnic background since 50-80% of Asian BCCs are pigmented but only 3% of Caucasians develops pigmented BCC (Moore 2012, Chow 2011). Asians develop BCC later in life and develop fewer BCCs over their lifetime than Caucasians, despite similar estimated lifetime daily sun exposure (Moore 2012).

Grachtchouk et al. studied the role of GLI2 overexpression in keratinocytes using albino and pigmented K5-Gli2 transgenic mice. K5 promoter is active in keratinocytes of the epidermal basal layer and outer root sheath of hair follicles. By three months of age, K5– 21

Gli2 trangenics developed multiple skin tumours. Tumours on the albino founder exhibited features characteristic of human BCCs. Many tumours on the agouti and black founders resembled human pigmented BCC (Grachtchouk 2000).

Bosnac2009

Figure 1.2 ǀ Proposed model of Shh signaling pathway regulation. Cdon acts as decoy receptor for Shh. In the presence of high level of Shh ligand, Cdon and Ptch1 bind to Shh and Smo is activated which in turn signals the transcription of Shh target genes (from Bosnac 2009).

Further studies using constitutive Shh transgenic mouse model under control of human keratin 1 (HK1) promoter emphasized the governing role of Shh on proliferation of melanocytes. The HK1-Shh transgenic mice did not develop BCC at all but showed extensive dermal hyperplasia of pigmented melanocytes. Similar results were reported by Grachtchouk et al., who used truncated keratin 5 (∆K5) promoter to assess the potential role of the human M2SMO mutant in BCC development in adult transgenic mice. ΔK5- M2SMO mice developed pigmented nevi and showed signs of hair loss but did not develop BCC. 22

Interestingly, most of the aforementioned models develop inguinal lesions which nonetheless express a similar profile of marker proteins as that of human BCC. The competition between keratinocytes and melanocytes over secretory cues released by transgenic keratinocytes could play a vital role in development of melanocytic nevus over BCC. Another explanation would be that the secreted cues are simply not adequate for development of BCC but sufficient for more sensitive and sympathetic melanocytes to undergo hyperplasia.

In 2012, Eberl et al. studied the effect of Shh activation in the context of SMO activation using K5creER;SmoM2;EGFR+/+ mice which resulted in development of pigmented BCC. This phenotype was most prominent in the ears. Loss of EGFR due to tamoxifen induction inhibited the development of BCC and nevogenesis in dermis (Ebrel 2012). Convergence of nevogenesis and BCC development under control of EGFR is very interesting which ushers in new targeted therapy possibilities.

In mammals, the stem cell niche within hair follicles is home to epithelial hair follicle stem cells and melanocyte stem cells, which sustain cyclical bouts of hair regeneration and pigmentation (Blanpain 2009, Cotsarelis 2006, Tumbar 2004, and Nishimura 2002). Dissecting the inter-stem-cell crosstalk governing this intricate coordination has been difficult, because mutations affecting one lineage often affect the other. The perturbation of hair follicle structure as a result of melanocyte proliferation or as an independent factor could be the determining factor in nevogenesis.

1.2 The melanocyte lineage in development

The neural tube is the origin of neural crest, from which melanocytes are derived. Neural crest cells are derived from the dorsal part of neural tube and go through epithelial to mesenchymal transition (EMT) and become migratory.

These cells migrate to different parts of the embryo and generate a multitude of specialized structures and tissues (Mayor and Theveneau, 2013). Melanocytes, glia, and neurons are derived from the neural crest in the trunk. Cells that leave neural crest early and follow a ventral migration through the anterior scleretome differentiate into glia and neurons. Cells that leave neural crest later and follow the dorsolateral path between the ectoderm and through the early dermis differentiate into melanocytes (Erickson and Goins, 1995, Henion and Weston, 1997). Fate determination process is very complex and needs 23 more study. Other studies suggest a dual neural crest origin of melanocytes encompassing; 1) early differentiation from NC (dorsolateral migration). 2) late differentiation from melanoblast or Schwann cell progenitors (Henion and Weston, 1997, Dupin et al., 2003). Melanoblasts then migrate large distances in order to populate the epidermis and early hair follicles. There is a huge increase in melanoblast populations in mouse skin between mouse embryonice day 10.5 (E10.5) and E15.5 from <100 to around 20,000 (Luciani 2011).

Figure 1.3 ǀ An overview of melanocyte development (from Mort 2015).

The migration from dermis to epidermis and colonisation of the epidermis by melanoblasts is complete by E15.5 in mice (Mann 1962). Melanoblasts then colonise the developing hair follicle (E15.5 onward) and start to produce pigment by E16.5 (Hirobe 1984). The appearance of white patches in mouse mutants and humans is the result of unsuccessful proliferation or migration of melanoblasts (Lamoreux et al., 2010, Fleischman et al., 1991).

Another interesting aspect of melanocyte migration is changes in the expression of cell adhesion molecules since pre-migratory neural crest populations express N-cadherin which is down regulated during migration (due to EMT changes induced by up-regulation of Snail) (Nieto et al., 1994; Jiang et al., 1998). Melanoblast that reach dermis by E11.5 do not express E-cadherin and P-cadherin but majority of melanoblast that migrate from dermis to epidermis show 200 fold upregulation of cadherins and become E-cadherinHigh and P-cadherinLow ((Nishimura et al., 1999). Melanoblasts are sensitive to keratinocyte 24 cadherin expression as they localize in the hair follicles which is concomitant with switch from E-cadherinhigh/P-cadherinlow to E-cadherin-/P-cadherinhigh phenotype (Nishimura et al., 1999).

The switch in cadherin expression of melanoblasts in the presence of keratinocytes suggests a non-cell autonomous effect on melanoblasts migration.

1.3 Melanocyte growth in culture

Melanocytes generally can only be propagated in culture via the inclusion in the growth media containing cholera toxin and phorbol ester. However co-culture with keratinocytes is also successful (Scott and Haake, 1991). In mouse 2D serum free culture of epidermal cell suspensions from neonatal mice keratinocytes proliferate first and melanoblast and melanocytes divide in the periphery. Keratinocytes are the first cells to die after 8-9 days which lead to melanoblast or melanocyte pure cultures after 14 days (Hirobe 1992a). In neonatal mice, keratinocytes also have important role in differentiation of melanocytes in the presence of agents that increase cAMP production (α-MSH, ACTH, DBcAMP) (Hirobe 1992c, Hirobe and Abe 2000). The concentration of Ca2+ in human keratinocyte- melanocyte co-culture has a huge impact on the release of mitogenic secretory cues released from keratinocytes. Low Ca2+ medium or undifferentiated keratinocytes co- cultured with melanocytes have higher impact on the proliferation and melanogenesis of the latter while high Ca2+ medium or co-culture of differentiated keratinocytes with melanocytes only result in melanogenesis. (Hennings et al., 1980, Valyi-Nagy et al., 1993, Abdel-Naser, 1999, Valyi-Nagy et al., 1993).

1.4 Non-cell autonomous effect of keratinocytes on melanocyte proliferation and differentiation

Melanocyte behaviour, including proliferation, melanogenesis, and dendritogenesis, is mainly controlled by keratinocytes in normal or UV induced skin (Archambault et al., 1995; Todd et al., 1993, Toyoda et al., 1999, Seiberg et al., 2000). Figure 1.4 shows multitude of keratinocyte derived cytokines, which in turn stimulate melanocyte proliferation by binding to their receptors on melanocyte.

Figure 1.4 ǀ Schematic image showing the action of keratinocyte-derived cytokines on the proliferation and differentiation of mammalian melanocytes (from Hirobe 2004).

1.5 Keratinocyte cytokines which are melanocyte mitogens

Endothelins

Ednothelin-1 (EDN1/ET-1), EDN2 (ET-2), and EDN3 (ET-3) exert their mitogenic effect by binding to their receptor on melanoblasts/melanocytes, Endothelin receptor B (EdnrB) (Sakurai 1990). Their mitogenic effect in newborn mice melanocytes may be in combination with factors that elevate cAMP or in cooperation with bFgf (Hirobe 2001). These cytokines increase the number of mitotic melanoblasts/melanocytes and also push the cell cycle to S phase and expedite the proliferation of these cells (Hirobe 2004). Endothelins also trigger the DNA synthesis in human epidermal melanocyte cultures in cooperation with cAMP elevating agents and serum (Imokawa et al., 1992). UV exposure increases EDN1 secretion from keratinocytes both in vitro and in vivo. The increases in EDN1 secretion will result in down-regulation of E-Cadherin in melanocytes and melanoma cells. Blockade of EDN1 receptor (EDNRB) inhibits melanoma progression and metastasis (Jamal 2002).

Stem cell factor

Stem cell factor (SCF/Kitl) exerts its mitogenic and melanogenic effects on mouse melanoblasts/melanocytes by binding to melanocytes. Kit receptor then activates the melanocyte MAPK pathway (Grichnik et al., 1998, Geissler et al., 1988) induces transcription of MITF.

Leukemia inhibitory factor

Leukemia inhibitory factor (LIF) has mitogenic and melanogenic effects on murine epidermal melanocytes in the presence of cAMP elevating agents (Hirobe 2002) by expediting G0/G1S and G2M transitions (Hirobe 2004). Mechanistically this is done by binding of LIF to its receptor on melanocytes, gp130LIFRa and activation of JAK/STAT and MAPK pathways (Coughlin et al., 1988, Auernhammer and Melmed, 2000).

Hepatocyte growth factor

Hepatocyte growth factor (HGF) has mitogenic and melanogenic effects on murine melanocytes in vitro (Hirobe 2004) in the presence of cAMP elevating agents through activation of c-Met receptor of melanocytes. Activated c-Met in melanoblasts and melanocytes expedite S phase transition and increases the number of melanoblast and melanocytes in the mitotic state (Hirobe 2004a). HGF also stimulates the proliferation of non-cutaneous melanocytes (eye, ear, dermal melanocytes) in vivo and has less effect in growth of cutaneous melanocytes (Aoki 2009). HGF also stimulates the hyperproliferation of human epidermal melanocytes without the need of cAMP elevating agents (Bottaro et al., 1991, Halaban et al., 1993; Matsumoto et al., 1991).

1.6 Melanocytic nevi

Melanocytic nevi have distinct colors and patterns which allow the physician to draw partial conclusions about the localization of pigmented cells within the skin. Black and brown colors (due to melanocytes or pigmented parakeratosis) indicate pigmentation in the epidermis, while gray and blue (due to melanocytes or pigment-laden melanophages) correspond to pigmented cells within the superficial and deep dermis, respectively 27

(Zalaudek 2009) (Figure 1.5a). Nevi can occupy the epidermis or dermis which are classified as epidermal or dermal nevi (Fig 1.5b,c). They also could reside in both layers hence compound nevi (Fig 1.5d).

b a

Blue Black Brown Gray reticular superficial dermoepidermal papillary dermis epidermis junction dermis

Figure 1.5 ǀ Color of melanocytic nevi is dependent on location in skin layers. a, if they reside in epidermis the color will be black while superficial lesions have a gray color. Nevi forming in dermatoepidermal junction look brown. Nevi residing deep in dermis have a blue color. Scale bar = 500 μm. b, c, and d, H&E images of epidermal, dermal and compound nevi respectively. Red arrows point to location of nevus melanocytes (nevocytes) in the epidermis and yellow arrows point to the location of dermal nevi. Scale bars = 200 μm.

Congenital nevi

Congenital nevi are classified based on their size and their location in skin (Figure 1.6a). NRAS and BRAF mutations are the most common oncogenic changes (55% and 39.3%) in congenital melanocytic nevi (CMN) while BRAF mutation (79%) is the most frequent mutation in acquired nevi. Large and medium size CMN tend to have NRAS mutations (66%) but smaller CMN have BRAF mutations (79%) (Raeve 2006, Ichii-Nakato 2006, Bauer 2007, Wu 2007 and Dessars 2009). Ultraviolet radiation (UVR) has no role since the mutation occurs in utero. The preference for NRAS mutations has been suggested could be explained by the fact that they may favor upregulation of CRAF. CRAF shares many functional aspects with BRAF but also has a unique function, the inhibition of apoptosis (Dumaz 2006, Wojnowski 2000 and Smalley 2009). Majority of CMNs show angiotropism which is defined by clear and recognizable presence of nevus cells in the proximity of the external surfaces of vessles or lymphatic channels. These aggregates show no sign of internalization to the vessel or lymphatic channel (Barnhill 2010). In melanoma, angiotropism is indicative of micrometastasis, often associated with regional lymph node metastasis (Day 1981, Day 1982, Harrist 1984), and is considered a poor prognosis factor (Wilmott 2012). Since angiotropism is common in CMN and metastatic melanoma, this extravascular migration may be an important mechanism in metastasis which could be linked to perturbation of developmental processes during embryonic development.

Acquired nevi

In acquired nevi (mainly epidermal and junctional) BRAF and NRAS mutations occur in almost 79% (Pollock 2003, Uribe 2006, Venesio 2008 and Ross 2011) and 4.6% (Saldanha 2004, Gill 2004, Poynter 2006 and Venesio 2008) of cases of acquired nevi respectively. 65% of dysplastic/atypical nevi carry BRAF mutations (Pollock 2003 and Uribe 2006). The origin of the most common BRAFV600E mutation is still obscure (Figure 1.6b). There is no compelling evidence that UVR directly induces this mutation: 1) BRAFV600E occur more frequently in malignant melanomas (MM) from intermittently sun- exposured skin (Poynter 2006 and Maldonado 2003); 2) BRAFV600E does not occur as a result of a UV signature (C>T transition) (Davies 2002); 3) CMNs harboring BRAF mutations that occur in utero (Wu 2007), 4) BRAFV600E mutation commonly occurs in internal cancer like colorectal carcinoma.

a Congenital melanocytic nevi

Small congenital nevus Large congenital nevus Giant congenital nevus Mongolian spot

Lee 2017 Lee 2017 Kinsler 2018 Ma 2015

Hairy Congenital agminated Speckled café-au-lait congenital nevus nevus lentiginous nevus macules

Cengiz 2017 Rocha 2012 Schaffer 2001 Wintermeier 2014

b Acquired melanocytic nevi

Acquired agminated nevus Blue nevus Spitz nevus

Shin 2012 Sand 2008 Sharma 2015

Naevus of Reed Halo nevus Acral nevus Unna nevus

Webber 2011 Patrizi 2005 Palleschi 2008 Witt 2010

Figure 1.6 ǀ Classification of congenital and acquired melanocytic nevi. a, Congenital nevi are classified as small, large, giant, Mongolian spots, hairy, agminated, speckled lentiginous and café-au-lait macules. b, Acquired nevi are classified as agminated, blue, Spitz, Reed, Halo, acral and Unna.

1.7 Dermoscopic patterns of melanocytic nevi

The presence of uniform and regularly distributed globular, reticular, starburst, blue, and complex patterns (Kreusch 1990, Yadav 1993, Soyer 2000, Ferrara 2005, Schärer 2007, Arevalo 2008, Zalaudek 2009, Shiga 2012) identify a given lesion as a melanocytic nevus (ie, per definition absence of melanoma-specific patterns) (Argenziano 2003). Each pattern corresponds to a specific underlying histopathologic correlate (Figure 1.7). Dermascopic patterns of melanocytic nevi

Blue Reticular Globular

Shiga 2012 Marchetti 2014 Zalaudek 2009

Complex Starburst

Zalaudek 2009 Marchell 2005

Figure 1.7 ǀ Dermascopic pattern of nevi. Pigmentation patterns of melanocytic nevi give them specific dermascopic patterns. Blue nevi are uniform structureless lesions with blue color, reticluar nevi are meshwork of brown lines, globular nevi show aggregated oval structures, complex nevi are composed of two different patterns and starburst nevi show radial lines around the periphery of lesion.

1.8 Skin type related nevus pattern

Individuals with skin type I exhibit a predominant nevus type characterized by light brown color and central hypopigmentation. In contrast, nevi in patients with skin type IV tend to be dark brown with a central hyperpigmentation (so-called black or hypermelanotic nevi). Nevi of skin types II and III are prone to be light to dark brown with multifocal pigmentation (Figure 1.8) (Giorgi 2006, Zalaudek 2007).

Skin type II Skin type III Skin type I multifocal multifocal Skin type IV central hyperpigmentation hyperpigmentation central hypopigmentation or hypopigmentation or hypopigmentation hyperpigmentation

Figure 1.8 ǀ Skin type dictates the color of nevi. People with lighter skin pigmentation have hypopigmented nevi while people with darker skin colors have hyperpigmented nevi.

1.9 Nevogenesis

Nevogenesis is a multifaceted process involving complex interplay of environmental and genetic factors. Although the nature of this process is not well understood. Activation of certain molecular pathways is crucial to drive nevogenesis (Ross 2011). The life cycle of benign melanocytic lesions can consist of four stages: initiation, promotion, senescence, and occasionally involution. In general, initiation is the result of a mutation in nevus progenitor cell that permits future growth. Activation and proliferation of mutated cell begins the promotion phase, which may involve a change in the relationship between the nevocyte and its local environmental that favors proliferation. After a period of growth, nevi stop proliferating through the activation of senescence pathways. Senescence keeps nevi stable for an extended period of time and, occasionally, the cycle comes to end with involution (Ross 2011).

Although some models of nevogenesis (Robinson 1998, Hui 2001, Grichnik 2008) propose the single cell origin of the melanocytic lesion, the location of the putative progenitor cell is still unclear. This would depend upon the location of the nevi (i.e. epidermal, dermal, or junctional), hence the progenitor cell could arise from epidermis, dermis, or both. It is possible that an immature melanocytic stem cell serves as the progenitor. UV light exposure and other mutagenic processes could cause such secondary changes. These genetic alterations would result in abnormal proliferation of the progenitor cell in response to normal environmental signals which activate this cell to produce melanocytes.

Another model for nevogenesis predicts that the nevus derives from a diﬀerentiated melanocyte that drives the nevogenesis. The oncogenic mutation enables the cell to regain proliferative capacity, dedifferentiate, and develop invasive properties which allows the cell to migrate into greater depths of the dermis (Ross 2011). It is more difficult to explain the second model with the clinical behavior of nevi since the initial nevogenic mutation induces nevus growth immediately.

Nnevocytes are senescent, hence express several markers of senescence, namely, tumor p16INK4a (p16), macroH2A, 53BP1, and senescence-associated β-galactosidase (SAβ-gal) (Michaloglou 2005, Gray-Schopfer 2006 and Suram 2012). Inactivation of tumor suppressor genes e.g. PTEN and CDKN2A (p16), induction of embryonic epithelial- mesenchymal transition (EMT), and activation of Wnt signaling are just some of mechanisms thought to contribute to nevus to melanoma progression via enabling escape from senescence (Dankort 2009, Vredeveld 2012, Delmas 2007 and Dhomen 2009).

The location, differentiation state and gene expression profile of the cell of origin for benign moles remains to be discovered. Further comprehensive research is necessary to fully understand nevogenesis, and in particular, why a small proportion of nevi progress to melanoma (Figure 1.9).

Figure 1.9 ǀ Schema of melanoma progression from a nevus (from Vultur and Herlyn 2013).

1.10 Nevi as a risk factor for melanoma

The survival rates for metastatic melanoma are directly proportional to the thickness of the primary melanoma. Surgical excision of primary melanomas of <1 mm thickness results in a 5 year survival rate of >90%, whereas for excision of lesions of >4 mm average 5 year survival is 40% (Balch 2001). Thus the case for early detection of suspicious nevi for prevention of melanoma is very strong. The morphology and type of premalignant nevi are indicator of melanoma risk. However, total body surface nevus count is very important, with individuals harbouring >100 nevi having a 7-fold elevated risk of developing melanoma (Chang 2009).

Nevi are melanoma precursor lesions, and 35-50% melanomas arise from pre- existing moles (Skender-Kalnenas 1995), although the overall risk of any individual nevus developing into a melanoma is small, estimated at ~1/30,000 nevi (Tsao 2003). Thus, although nevi represent a reservoir of premalignant cells, the transformation rate is extremely low, making surveillance a time consuming task.

This situation is exacerbated in dysplastic nevus syndrome (DNS) (Figure 1.10). DNS is classified differently by different groups but generally refers to people with >50, 2 mm melanocytic lesions, at least one with atypical features, and one or more close relative with melanoma (Duffy 2012). The presence of one dysplastic nevus is associated with a 2 fold increased risk of developing melanoma, and 10 or more nevi with 12-fold increased risk (Tucker 1997).

There is a significant inherited component to nevogenesis, with members of melanoma- prone families often having large numbers of atypical nevi (Olsen, 2010), and persons with 2 or more family members with melanoma having a >85-fold increased risk (Friedman 2009). The presence of morphological atypia increases the risk of melanoma >4 fold (Arumi-Uria 2003).

Figure 1.10 ǀ Clinical image of a patient with high mole count who is at high risk of developing melanoma. Extensive scarring from previous excision of suspicious nevi is evident (UQ Dermatology Research Centre).

Nevi are very common premalignant lesions in high mole count individuals who are at high risk of developing melanoma in whom they may more frequently develop into melanoma. Clinically, these high risk patients tend to have large numbers of suspicious lesions surgically excised often unnecessarily with the associated cost and morbidity (Figure 1.10). Thus, there is a clear clinical need in the assessment of DNS patients, to develop rapid, non-invasive methods to accurately assess the transformation potential of 35 individual nevi, providing the treating clinician with a readout that can be readily applied to determining the treatment of each nevus. This is done by morphological features, but they are often not very predictive. Hence there are no clear molecular markers to distinguish dysplastic and non-dysplastic nevi. BRAF and CDKN2A and TP53 mutations are reported in DN, and clinical and histopathological markers are suggested to classify these lesions, but there is still a deal of debate as to what constitutes a dysplastic nevus (Friedman 2009, Duffy 2012).

Knowing the particular molecular changes occurring during the transition of nevus to melanoma would greatly enhance our understanding of why some nevi progress. The shortcoming of the correlative approaches possible using patient derived samples is that when the lesion is excised for analysis the future of that lesion can only be conjectured. Any markers identified will be correlated based on statistical associations. While it is not ethically possible to continuously sample a suspicious lesion without treating it, this type of longitudinal study is possible using an animal model of the disease. For example using a mouse model it should be possible to interfere with the expression or function of genes identified in patient lesion based studies, to determine whether a particular gene is involved in melanomagenesis. This project will use a mouse model of nevus progressing to melanoma, which will provide the opportunity to perform a variety of analyses of lesions at various stages of nevogenesis and melanomagenesis.

1.11 Transgenic model used for current research

The contribution of one or multiple genes to nevogenesis, melanomagenesis, and metastasis can be studied using transgenic animal models. Since there are many types and subtypes of human nevi, the ideal model of nevus to melanoma progression does not exist. An ideal model should show identical: 1) Histopathological features; 2) Progression patterns and stages; 3) Driving molecular pathways; 4) Response to treatment; with human nevi and melanoma. However human melanomas are so diverse that it is impossible to precisely model all of these aspects except for the case of a “generalised” human melanoma. Nevertheless, transgenic mouse models are the best mammalian

36 model systems to study nevogenesis and melanomagenesis. They have a similar genome, comparable organs and physiology.

Also, their genome is well described, and a large panel of mutants already exist, the introduction of new genetic modifications is quite simple, and environmental and therapeutic studies can readily be tested in a whole organism (Gremel 2009). NRAS, KRAS, HRAS, BRAF, CDK4, CDKN2A, PTEN, TRP53, CTNNB1 and STK11are just some of the genes that have been studied using transgenic mouse models throughout the years (Ferguson 2010, Wurm 2012, and Vultur and Herlyn 2013).

Cdk4R24C/R24C::Tyr-NRASQ61K mouse model This project will use Cdk4R24C/R24C::Tyr-NRASQ61K (or Cdk4::NRAS) mouse model for nevi progressing to melanoma (Ferguson 2010, Wurm et al., 2012). In these mice nevus like-lesions were observed in the papillary dermis, immediately beneath the epidermis in a band-like distribution (Figure 1.11). In terms of classical morphology the nevi mimic a superficial congenital nevus pattern. This phenotype was observed in both UVR-treated and control mice. All nevi show extremely low levels of proliferation by Ki-67 staining and little or no histological features of malignancy. The groupings of nevus cells have been studied by three dimensional modelling based on serial sectioning of the skin (Figure 1.12). Over the course of about 6 months, very slow growing flat lesions will develop out of these nevi, and a small portion of which will develop into a fast growing melanoma.

Figure 1.11 ǀ H&E stain of the band like nevus specific to the Cdk4::Tyr-NRAS mice. In terms of classical morphology these cells are reminiscent of a superficial congenital nevus pattern.

A staging system (Wurm 2012, Figure 1.13) was developed for the murine lesions, based on Breslow thickness, a prognostic indicator in human melanoma. The small plaques histopathologically mimic human congenital nevi or dysplastic nevi. The dysplastic pattern is characterised by melanocytes singly and in nests throughout the superficial dermis and focally in the epidermis. In some cases we could histopathologically confirm that a melanoma was emanating from one of the nevi. Because we observe multiple nevi of different morphologies, and melanoma, in the same animal, it was speculated that this mouse model mimics the critical aspects of the familial atypical multiple mole melanoma (FAMMM) syndrome (Lynch 1980), or DNS on the FVB strain background.

Lines of Group of

hair follicles nevocytes Figure 1.12 ǀ Schematic of nevi in Cdk4::NRAS mouse skin looking from above. This was modelled by serial sectioning. We calculate that there are about 350 nevus nests on the dorsal surface of each mouse.

Notably, dermal nevi are sometimes seen in these patients (Tucker 2002). Careful histopathological assessment of many murine lesions was performed by an experienced dermatopathologist (Professor H Peter Soyer), who characterized the nevi and assessed how they relate morphologically to human lesions.

Figure. 5. Nevus Histopathological

staging system for Cdk4::Tyr-NRAS melanocytic lesions.

Images of lesions representative of each

stage of the progression model are presented at various magnifications.

Figure 1.13 ǀ Staging schema for murine melanoma. Staging is based on dermascopy of MMs and H&E staining (from Wurm 2012).

1.12 Natural genetic factors that regulate nevus development and density

Risk factors for invasive MM are (approximate relative risk is shown in brackets) (Gandini 2005a, Gandini 2005b, Mackie 2009): 1) Three or more clinically atypical (dysplastic) nevi (12); 2) Previous personal primary invasive melanoma (8.5); 3) Multiple banal melanocytic nevi: 100 vs. <15 nevi (6.9); 4) High solar exposure in early childhood (4.3); 5) Red hair (2.4). Therefore, genetic factors are arguably more important to melanoma risk than sun exposure. Genome Wide Association Studies (GWAS) have shown that SNPs within or near many genes (MC1R, ASIP, TYR, SLC45A2, TYRP, PLA2G6, ATM, CASP8, MX2, TERT, and NID1) are in linkage disequilibrium with melanoma-predisposing variants (Gerstenblith, 2010, Bishop 2009, Barrett 2011, Nan 2011, Gerstenblith 2010, Kim and Chanock 2012).

Loci associated with nevus count include CDKN2A/MTAP (Falchi 2009) and IRF4 (Duffy 2010), with several other genes also having very small effects. The nevus genes discovered to date account for less than 10% of the variance in nevus count, surprisingly low given that it is highly heritable (70-80%, Law 2013). Many of these genes are of unknown function, and where this is known, it has not been functionally determined how and where they act during melanoma progression.

1.13 Collaborative Cross

The Collaborative cross (CC) is a panel of recombinant inbred (RI) mice, derived from a genetically diverse set of founder strains and designed specifically for complex trait analysis (Figure 1.14). The genomes of eight genetically diverse founder strains: 1) 129S1/SvImJ; 2) A/J; 3) C57BL/6J; 4) CAST/EiJ; 5) NOD/LtJ; 6) NZO/HlLtJ; 7) PWK/PhJ; and 8) WSB/EiJ are combined (Figure 1.14) to capture ~90% of the known variation present in laboratory mice while other RI panels e.g. AXB/BXA and BXD only capture ~13% of known variations (Churchill 2004).

Figure 1.14 ǀ Collaborative cross is a typical eight-way RI strain. The color scheme indicates the parental origin of genomic segments.

The CC is ideal for the discovery of genes for complex diseases (i.e. those where many genes are involved) such as cancer. Each of the 100’s of CC strains is inbred, hence infinitely available, and the genome sequence is known. Therefore the CC provides not only gene mapping power, but also integration, such that any other characteristics of nevus susceptible strains (i.e. immune system, or tissue gene expression, in fact any variable that can be measured) is available (Figure 1.15). By breeding Cdk4R24C::Tyr- NRAS males (FVB strain background) with females of many CC strains, it has been noted that there is a tremendous variation in the density of dermal nevi in these mice when crossed against up to 70 CC strains (Chitsazan 2016). This provides an opportunity to use the CC to discover genetic modifiers of nevus density.

Figure 1.15 ǀ Multiple, diverse phenotypic assays using collaborative cross (Churchill 2004).

1.14 Successful precedent for the “transgene modifier of cancer” approach Genes which do not predispose to a disease, but which can influence its course or severity, are “modifier genes”. They are very difficult to identify, especially in humans (Nadeau, 2001). An excellent example of the power of the transgene modifier approach comes from the work of Hunter et al. (Lifsted, 1998). To find modifier genes of breast cancer metastasis, they crossed a relevant transgenic mouse with each of 27 strains. Recombinant inbred mouse lines are generally derived from just two parental strains. Despite the smaller size and lower heterogeneity of their panel compared to the CC that are using, they discovered a set of interacting genes in a pathway that regulates metastasis; they named the pathway “Diaspora”, alleles of which predict survival outcomes for patient genotype groups (Crawford, 2008). A similar strategy has been used to identify a modulator of breast cancer metastasis in mouse and humans by a non-cell-autonomous mechanism (Faraji, 2012). These studies used a small number of closely related common inbred strains. Instead, we can use the CC to harness the available genetic repertoire of the mouse. The mapping power of the CC has recently been demonstrated using coat colour as a known Mendelian trait (Figure 1.16, Ram, 2012).

Figure 1.16 ǀ Genome wide scan for the mouse albino locus. Figure shows mapping the mouse albino locus on murine chromosome 7 (Tyr) using 64 collaborative cross strains (Ram 2012).

1.15 Forward and reverse genetic study Understanding the biological underpinnings of complex diseases such as autoimmunity, cancer and infection requires study of appropriate tissues, spatio-temporal progression, genetic and gene expression. Next step is studying the observed changes in animal models such as mouse. Mouse models are used to phenocopy the human disease to a degree. This model also provides an opportunity to study the desired phenotype in human by removing many factors that could have huge impact on the result of study e.g. diet, stress, exercise, sun exposure etc. Hence mouse models are used to simplify the study by removing factors that can’t be removed in human studies.

Mouse models are currently used for forward and reverse genetic studies which complement each others. Forward genetic studies (FGSs) use innate genetic variation between different mouse strains with known genotype (e.g. Collaborative Cross mouse strains) to conduct unbiased genetic screens. In order to perform such analysis, the genetic variation should first be quantitated e.g. nevus density in chapter 2 and microsatellite count in chapter 4. The unbiased genetic screen will provide a region that can be fine mapped using available recombinations in CC mouse strains. Finally, the candidate gene or genes can be studies using reverse genetic studies (RGSs) such as

43 development of knock outs, knock ins or transgenic mice for candidate genes or mutations (e.g. Cdon+/- mouse in chapter 3).

Studies using mouse models have provided many seminal contributions to biomedical research e.g. Noble prizes in physiology and medicine in 1980, 1984, and 2011 for discovery of major histocompatibility complex, development of monoclonal antibody, and activation of innate immunity respectively. FGSs also discovered many important genes involved in metabolic diseases (Ob and Lepr), circadian rythms (Bmal1, Npas2, Cry1, Cry2), and central nervous system (Hnrpul1).

Cdk4R24C/R24C::Tyr-NRASQ61K (or NRAS-Cdk4) mouse model (Ferguson 2010; Wurm 2012) carry the NRASQ61K mutation common in malignant melanoma, and a germline Cdk4R24C mutation, mimicking families who carry such mutations or those in the gene encoding its binding partner, p16 (Zuo 1996). They develop malignant melanomas characterized by downward invasion into the dermis and hypodermis, cellular atypia, and other histological signs of malignancy. Because we are using an animal model, we can obtain very accurate and precise data on these phenotypes for gene mapping.

My thesis is comprised of three projects where I use NRAS-Cdk4 mouse model (Ferguson 2010; Wurm 2012) to conduct following studies: 1) FGS and RGS of nevus exacerbation, 2) FGS of microsatellite count, and 3) Detecting cell autonomous changes involved in melanoma progression and metastasis.

Mapping the Modifier Gene for Nevus Exacerbation

Chapter 2

This chapter is comprised of a publication emanating from my thesis project. This publication is the preliminary result of a congenital nevus modifier gene mapping using collaborative cross mouse model.

2.1 Publication

2.3 Contribution

My contribution to the publication above included;

 Mouse breeding  Acquisition of mouse lesions  Design of scoring approach for scoring of nevus density in mouse using H&E stained slides  Design of immunohistochemical staining protocol to examine CDON and SOX10 expression  Acquisition of patient congenital melanocytic nevi samples from collaborators  Processing and immunohistochemical staining of lesion set  Analysis of staining patterns of lesion set  Preparation of the manuscript

Contribution by others:

 B.F. provided technical support, animal breeding, sample collection. 46

 R.R. provided bioinformatic support and analysed data.  P.M. provided bioinformatic support and analysed data.  H.Y.H. provided technical support.  B.G. contributed to experimental design and supplied pathological specimens.  H.P.S. contributed to Pathological diagnosis.  G.M. provided the CC mouse strains, and genetic analysis software.  G.J.W. designed the project, analysed data and prepared manuscript.

A Mutation in the Cdon Gene Potentiates Congenital Nevus Development Mediated by NRASQ61K

Arash Chitsazan1,2, Blake Ferguson1, Ramesh Ram3, Pamela Mukhopadhyay1, Herlina Y. Handoko1,Brian Gabrielli2, Peter H Soyer4, Grant Morahan3,* and Graeme J. Walker1,*

1 QIMR Berghofer Medical Research Institute, Herston, QLD, Australia

2 The University of Queensland Diamantina Institute, Translational Research Institute, The University of Queensland (UQ), Brisbane, QLD,Australia

3 Centre for Diabetes Research, Harry Perkins Institute of Medical Research, Perth, WA, Australia

4 Dermatology Research Centre, UQ School of Medicine, Translational Research Institute, Brisbane, QLD ,Australia

Keywords

Giant Congenital Nevus, Melanoma, QTL

Corresponding author: Dr Graeme Walker, QIMR Berghofer Medical Research Institute, 300 Herston Rd, Herston, QLD 4006, Australia. Tel: +61738453715, Fax: +61738453508 Email: [email protected]

Summary

Congenital nevi develop before birth and sometimes cover large areas of the body. They are presumed to arise from the acquisition of a gene mutation in an embryonic melanocyte that becomes trapped in the dermis during development. Mice bearing the Cdk4 R24C::Tyr-NRASQ61K transgenes develop congenital nevus-like lesions by post-natal day 10, from melanocytes escaping the confines of hair follicles. We interbred these mice with the collaborative cross (CC), a resource that enables identification of modifier genes for complex diseases (those where multiple genes are involved). We examined variation in nevus cell density in 66 CC strains and mapped a large-effect quantitative trait locus (QTL) controlling nevus cell density to murine chromosome 9. The best candidate for a gene that exacerbates congenital nevus development in the context of an NRAS mutation is Cdon, a positive regulator of sonic hedgehog (Shh) that is expressed mainly in keratinocytes.

Signiﬁcance

Giant congenital nevi can cause a variety of clinical problems and can convert to melanoma. They are not easily removed as they can be located deep within the dermis. No genome-wide association studies have been performed to help us understand the susceptibility to the development of congenital nevi. For the first time, we utilized a systems genetics approach to define innate variation that dramatically influences the density of congenital nevi. Understanding genes that control the development of these lesions will be a first step in the development of new treatments and control measures for these lesions.

Congenital nevi can manifest as small commonplace “birthmarks” anywhere on the body, large lesions on the head and neck, or giant lesions covering substantial areas of the torso (Krengel et al., 2013). Large congenital nevi are often associated with increased skin fragility, dryness and itching, and overheating due to less subcutaneous fat. The risk of otherwise extremely rare childhood melanoma is significantly increased for those with giant congenital nevi (Krengel et al. 2006). In an early prospective study of large congenital nevus patients, Hale et al. (2005) reported pediatric melanoma in 2.3%, while Vourc’h-Jourdain et al. (2013) found that 51 of 2578 similar patients developed melanoma in fourteen selection-based studies, about a 2% absolute risk.

Congenital nevi develop before or occasionally soon after birth. Located in the superficial and sometimes deep dermis and hypodermis, they are thought to arise from an oncogenic mutation in an embryonic melanocyte precursor. NRASQ61K mutation is the most common somatic mutation, present in ~95% of giant congenital melanocytic nevi (Kinsler et al. 2013; Charbel et al, 2014). However, there is little information about innate genetic differences between individuals, whereby variation in epigenetic or melanocyte- microenvironment interactions may also be important in nevus development. Many transgenic mice carrying melanocyte-specific oncogenic mutations develop dermal melanocytic proliferations reminiscent of human congenital or blue nevi (e.g. Dohmen et al. 2009; Shakhova et al. 2012; Pawlikowski et al. 2015). The development and size of these lesions can be dramatically altered with genetic engineering that results in changes in expression of stem cell or hepatocyte growth factor from keratinocytes (Kunisada et al. 1998, 2000), or increased Wnt signaling within melanocytes (Pawlikowski et al. 2013).

To investigate innate genetic variation that regulates congenital nevus development, we used Cdk4R24C/R24C::Tyr-NRASQ61K mice (Ferguson et al., 2010, 2015) in conjunction with a cohort of genetically-defined mouse strains, the Collaborative Cross (CC) (Morahan et al. 2008; Welsh et al. 2012). The CC is derived from eight diverse founder strains and captures ~90% of genetic variation present in the species (Churchill et al., 2004, Collaborative Cross Consortium 2012), making it ideal for the discovery of genes for complex traits (those where many genes are involved). The genome sequence at any locus for each strain is easily inferred based upon knowledge of the haplotype blocks inherited from the founder(s), for which the genome sequence is known. In a proof of 50 concept, the CC could rapidly map the variants responsible for mouse coat colour (Ram et al. 2014).

By breeding Cdk4R24C/R24C::Tyr-NRASQ61K/+ males with females from many CC strains we generated progeny heterozygous for both the Cdk4 and NRAS mutations. As we previously reported (Chai et al. 2014), at between days 0 and 5 after birth (P0-P5) melanocytes are migrating downwards to the hair follicle (unlike in humans, mouse interfollicular melanocytes do not normally persist at the epidermal-dermal junction). At P10 there are more melanocytes in the epidermis, and by P21 these have developed into nests of cells (nevi) within the superficial dermis, around hair follicles (Figure 1A). Hence the nevi begin to develop at P10, apparently from melanocytes that break away from the hair follicle (Figure 1B). In littermates not carrying NRASQ61K, melanocyte numbers were much lower at all timepoints, and between P10 and p42 there was no sign of any congregation of melanocytes in the form of nevi (Figure S1).

We collected skin from >3 mice of each cross, and the density of nevus cells in dermis was scored between 1 (least dense) and 10 (most dense). Cdk4R24C/R24C::Tyr-NRASQ61K/+ mice develop many thousands of nevi over their dorsal surface (Chai et al. 2014). Note that nevus density is not a measure of the number of nevi in each mouse skin, instead, an arbitrary score for the density of nevus cells right across the skin section analyzed. For albino strains we estimated the relative density of nevus cell nuclei across the section. Note that this, being a biological variable, can vary greatly across a given section, so an estimate has to be made. Figures S2 and S3 show examples of each of the scores (1-10). For lightly pigmented low nevus count strains we assessed the number of lightly pigmented (brown) dendritic nevus cells. For heavily pigmented strains (i.e. dark pigmentation covering a large portion of the dermis), we also try to assess dendritic “nevus cells”, but essentially we scored based on the area covered by dense pigmentation as a marker for the density score.

The median score per strain is shown in Figure S4. Normal skin was collected when each mouse was sacrificed (generally had developed a melanoma of ~10mm in diameter). We determined whether nevus count was dependent upon whether or not a mouse had received neonatal UVR. In the 12 strains for which we had data for both spontaneous and

51 neonatal UVR-induced melanoma we compared nevus count between the two groups and found no correlation (Figure 1C). Hence nevus cell density (Figure 1D,E) is genetically determined. We previously reported the presence of these dermal nevi, and early stage 2 or 3 melanomas whose location in the superficial dermis suggests that they have emanated from these nevi (Wurm et al. 2010). We compared the age of onset of stage 2/3 melanomas with the nevus cell density score for all mice in our study (Figure 1F), but we found a negative correlation (correlation co-efficient = -0.18, p= 0.00048, Spearman Rank Correlation test with an in-house R script).

We tested 66 CC strains to discover quantitative trait loci (QTLs) that modify median nevus density per strain in Cdk4R24C::Tyr-NRASQ61K mice. The construction of the CC founder haplotypes is described in Ram et al. (2014). For mapping we used a logistic regression matrix model over the reconstructed haplotypes matrix to produce genome-wide distribution of P values (ANOVA chi-squared). We used a false discovery rate of p=0.0001 to define significant genome wide linkage. We identified a major effect QTL on mouse chromosome 9 (-1 -Log10(P) interval=28.2-36.2 megabases (Mb), a region containing 23 genes) (Figure 2A). We also detected linkage at lower levels of significance to chromosome 11 (Figure S5, (-1 -Log10(P) interval=12.3-17.4 Mb, containing 13 genes), and chromosome 19 (-1 -Log10(P) interval=38.5-42.8 Mb, containing about 70 genes).

We began to ascertain which genes within the intervals are the best candidates by cataloguing DNA variants that vary between susceptible or resistant strains (see Figure

2B for the coefficients together with the corresponding ANOVA test –log10(P) values). For instance, examination of the founder haplotype coefficients through the linked interval (Figure 2C) on chromosome 9 shows that the causal variant is derived from the NOD/LtJ founder (hereafter termed NOD). Mining the Sanger Mouse Genomes database for NOD- specific variants revealed just 2 genes (Cdon and Pus3) carrying non-synonymous mutations, while 11 had NOD-specific single nucleotide polymorphisms (SNPs) in their 5’ or 3’ UTR, or introns (Figure 2D). The other gene in the chromosome 9 interval carrying a NOD-specific missense mutation is Pus3 (encoding Pseudouridylate Synthase 3), whose protein product is involved in tRNA metabolism but not skin or melanocyte biology. Other less likely candidates are listed in Figure 1d. Most of these have no known role in skin or melanocyte biology. 52

The interval on chromosome 11 carries 13 genes (Figure S5A,B). Here also the nevus “susceptibility” allele is inherited from NOD. On chromosome 11 only one candidate carries a NOD-specific missense variant, Cobl (Cordon-Bleu). It encodes a protein that binds actin and regulates neuronal behavior (Hou et al., 2015). On chromosome 19 the NOD haplotype again carries the causal allele, and of the 70 genes in the 4 Mb interval, only one, Tctn3 (Tectonic Family Member 3) carries a NOD-specific missense variant. Neither Cobl nor Tctn3 are reported to regulate melanocyte behavior. Many strains that carry the susceptibility haplotype on chromosome 9 also carry the susceptibility haplotype on chromosomes 11 and 19, (Figure S5C), suggesting some cooperation between the causal variants. Susceptibility to common diseases generally involves cooperation between multiple low risk genetic variants.

This may be similar for congenital nevus, however the linkage to chromosome 9 is by far the strongest hit in the genome, suggesting a large effect locus modifying congenital nevus formation is located here. Because our method of scoring nevus cell density is in some ways different for albino and pigmented strains, we ran a genome wide linkage analysis on albino and pigmented strains separately (Figure S6). Despite the relatively smaller samples sizes than with all strains analyzed altogether, the chromosome 9 locus was under the most significant peak in both cases.

The best candidate under the Chromosome 9 peak is Cdon, the NOD allele of which carries a missense mutation; none of the other founder strains have this change (Figure 2D). Cdon has five Immunoglobulin (Ig) and three fibronectin type III domains (Sanchez- Arrones et al. 2012). The CdonP456S mutation is in the Ig domain and could modify the glycosylation, phosphorylation and interactions of the mutant protein. Cdon is a neural cell adhesion molecule that is an activator of sonic hedgehog (SHH) (Allen et al. 2011). SHH- GLI signaling is crucial for melanoma progression via cross-talk between the GLI and RAS-MEK/AKT pathways. SHH-GLI1 signaling in human hair follicle matrix controls normal proliferation of human melanocytes (Stecca et al. 2007). The Functional Networks of Tissues in Mouse (http://fntm.princeton.edu) database provides a means to assess tissue-specific co-expression and functional interactions between genes. The most

53 significant genes interacting with Cdon were in the hair follicle, particularly Shh (sonic hedgehog). We stratified the hair follicle gene network looking for genes interacting with Tyrosinase (Tyr) to identify melanocyte-specific networks, and Keratin 14 (Krt14) to identify keratinocyte-specific networks. In the Krt14 and Cdon (but not the Tyr) networks, Shh was the strongest associated gene, suggesting that the activity of Cdon via Shh may be mostly within follicular keratinocytes. Using a commercial anti-CDON antibody we were not able to obtain acceptable staining of mouse skin, but in human skin CDON is mostly expressed in the follicular inner root sheath (Figure 2E, F). If Cdon deregulation facilitates escape of melanocytes from the hair follicle in the nevus-prone mice, it may be via disrupting cell-cell adhesion, since in zebrafish Cdon promotes neural crest migration by regulating N- cadherin localization, and Cdon knockout suppresses melanocyte migration to the ventral surface (Powell et al. 2015). Further, transgenic mice overexpressing Shh under a keratin gene promoter have hyperpigmented skin due to increased melanocytes near the epidermis, and develop discrete pigmented lesions on hind-limbs, made up of masses of highly pigmented melanocytes near the basal epidermis (Ellis et al. 2003). This non cell- autonomous influence of the Shh on pigmented lesion formation has to our knowledge not been followed up.

In a mouse model carrying the human NRASQ61K cDNA, lesions very reminiscent of congenital nevi formed at ~post-natal day 10, probably from melanocytes escaping the hair follicle. While these mice do not perfectly phenocopy human congenital nevus, they are one of the best characterized mouse models of dermal nevi. Mice are the only mammals in which we could have take the CC approach. Nevus cell density varied greatly with genetic background, with CC progeny in some strains hardly developing the lesions. Nevus cell density was not influenced by neonatal UVR exposure. Further, the gene we have localized seems to only influence the formation of the nevi, not whether they proceed to melanoma. Genetic linkage analysis of nevus density revealed a highly significant locus on chromosome 9, with Cdon a strong candidate modifier driving high nevus density. Other minor, but significant loci were located on chromosomes 11 and 19. Fine mapping of the genomic intervals can now be performed by testing further CC strains with the Cdk4R24C::Tyr-NRASQ61K mice and/or using the diversity outbred (DO) system (Churchill et al. 2012). DO mice carry many more recombinations than the CC strains, allowing more detailed mapping within the intervals.

These novel findings may provide the first clues as to how innate genetic variation may control the growth and development of congenital nevi.

Conflict of Interest

The authors report no conflict of interest.

Acknowledgements

This work was supported by Melanoma Research Alliance, Washington DC, and the Cancer Council of Queensland.

References

Allen, B.L., et al. (2011). Overlapping roles and collective requirement for the coreceptors GAS1, CDO, and BOC in SHH pathway function. Dev. Cell 20, 775–787.

Chai E, Ferguson B, Prow T, Soyer P, Walker G. (2014). Three-dimensional modelling for estimation of nevus count and probability of nevus-melanoma progression in a murine model. Pigment Cell Melanoma Res. 27, 317-9.

Charbel, C., et al. (2014). NRAS mutation is the sole recurrent somatic mutation in large congenital melanocytic nevi. J. Invest. Dermatol. 134, 1067-74.

Churchill GA, Gatti DM, Munger SC, Svenson KL. (2012). The Diversity Outbred mouse population. Mamm. Genome 23, 713-8.

Collaborative Cross Consortium. (2012). The genome architecture of the Collaborative Cross mouse genetic reference population. Genetics 190, 389-401.

Dhomen, N., et al. (2009). Oncogenic Braf induces melanocyte senescence and melanoma in mice. Cancer Cell 15, 294-303.

Ellis T, Smyth I, Riley E, Bowles J, Adolphe C, Rothnagel JA, Wicking C, Wainwright BJ. (2003).

Overexpression of Sonic Hedgehog suppresses embryonic hair follicle morphogenesis. Dev Biol. 263, 203-15.

Ferguson, B., Konrad Muller, H., Handoko, H.Y., et al. (2010). Differential roles of the pRb and Arf/p53 pathways in murine naevus and melanoma genesis. Pigment Cell Melanoma Res. 23, 771-80

Ferguson, B., et al. (2015). Melanoma susceptibility as a complex trait: genetic variation controls all stages of tumor progression. Oncogene 34, 2879-86.

Hale, E.K., et al. (2005). Association of melanoma and neurocutaneous melanocytosis with large congenital melanocytic naevi--results from the NYU-LCMN registry. Br J Dermatol. 152, 512-7.

Hou, W., Izadi, M., Nemitz, S., Haag, N., Kessels, M.M., and Qualmann, B. (2015). The Actin Nucleator Cobl Is Controlled by Calcium and Calmodulin. PLoS. Biol. 13, e1002233.

Jalili, A., et al. (2013). NVP-LDE225, a potent and selective SMOOTHENED antagonist reduces melanoma growth in vitro and in vivo. PLoS One. 8, e69064.

Kinsler, V.A., et al. (2013). Multiple congenital melanocytic nevi and neurocutaneous melanosis are caused by postzygotic mutations in codon 61 of NRAS. J Invest Dermatol. 133, 2229-36.

Krengel, S., Hauschild, A., Schäfer, T. (2006). Melanoma risk in congenital melanocytic naevi: a systematic review. Br J Dermatol. 155, 1-8.

Krengel, S., Scope, A., Dusza, S.W., Vonthein, R., Marghoob, AA. (2013). New recommendations for the categorization of cutaneous features of congenital melanocytic nevi. J Am Acad Dermatol. 68, 441-51.

Kunisada T, et al. (1998). Transgene expression of steel factor in the basal layer of epidermis promotes survival, proliferation, differentiation and migration of melanocyte precursors. Development 125, 2915-23.

Kunisada T, et al. (2000). Keratinocyte expression of transgenic hepatocyte growth factor affects melanocyte development, leading to dermal melanocytosis. Mech Dev. 94, 67-78.

Morahan, G., Balmer, L., Monley, D. (2008). Establishment of "The Gene Mine": a resource for rapid identification of complex trait genes. Mamm. Genome 19, 390-3.

Powell, D.R., Williams, J.S., Hernandez-Lagunas, L., Salcedo, E., O'Brien, J.H., Artinger, K.B. (2015). Cdon promotes neural crest migration by regulating N-cadherin localization. Dev Biol. S0012-1606, 30097-X.

Pawlikowski, J.S., et al. (2013). Wnt signaling potentiates nevogenesis. Proc. Natl. Acad. Sci. USA. 110, 16009-14.

Pawlikowski, J.S., et al. (2015). Acute Inhibition of MEK Suppresses Congenital Melanocytic Nevus Syndrome in a Murine Model Driven by Activated NRAS and Wnt Signaling. J Invest. Dermatol. 135, 2093-101.

Welsh, C.E., et al. (2012). Status and access to the Collaborative Cross population. Mamm. Genome 23, 706-12.

Ram, R., Mehta, M., Balmer, L., Gatti, D.M., Morahan, G. (2014). Rapid identification of major-effect genes using the collaborative cross. Genetics 198, 75-86.

Sanchez-Arrones, L., Cardozo, M., Nieto-Lopez, F., and Bovolenta, P. (2012). Cdon and Boc: Two transmembrane proteins implicated in cell-cell communication. Int. J. Biochem Cell Biol. 44, 698-702.

Shakhova, O., et al. (2012). Sox10 promotes the formation and maintenance of giant congenital naevi and melanoma. Nat. Cell Biol. 14, 882-90.

Stecca, B., Mas, C., Clement, V., Zbinden, M., Correa, R., Piguet, V., Beermann, F., Ruiz, I. (2007). Melanomas require HEDGEHOG-GLI signaling regulated by interactions between GLI1 and the RAS-MEK/AKT pathways. Proc. Natl. Acad. Sci. USA. 104, 5895- 5900.

Vourc'h-Jourdain, M., Martin, L., Barbarot, S; aRED. (2013). Large congenital melanocytic nevi: therapeutic management and melanoma risk: a systematic review. J. Am. Acad. Dermatol. 68, 493-8.

Wurm, E.M., Lin, L.L., Ferguson, B., Lambie, D., Prow, T.W., Walker, G.J., Soyer, H.P. (2012). A blueprint for staging of murine melanocytic lesions based on the Cdk4 (R24C/R24C)::Tyr- NRAS(Q61K) model. Exp. Dermatol. 21, 676-81.

Figure legends

Figure 1. (A) Sox10-stained skin taken from Cdk4R24C/R24C::Tyr-NRASQ61K mice at various time-points after birth. Yellow arrows denote nevi. See immunohistochemistry staining methods in supplementary methods file. (b). Yellow arrows show Sox10-positive melanocytes escaping (left panel), or escaped (right panel), the hair bulb. (C) Average nevus score per strains for mice with and without previous exposure to neonatal UVR. Two people (AC and GW) independently scored them blind, and a third (BF) agreed on the score. Experiments were undertaken with institute animal ethics approval (A98004M). 3- day-old mice were exposed to a single UVB exposure as described (Ferguson et al., 2010) (D) H&E skin sections from albino strains. Upper panel: a strain with very high nevus cell density, lower panel, a strain with very low density. (E) A pigmented strain. Upper panel: a strain with very high, lower panel, one with very low density. (F) Heat map showing the lack of correlation between congenital nevus cell density and neonatal UVR-induced and spontaneous melanoma. . Red denotes high nevus cell density and early melanoma age of onset, yellow low score and late onset. Black denotes data not yet available. All scale bars = 100μM

Figure 2. (A) Genome-wide scan based on nevus cell density score in 66 CC strains. Genotyping, construction of CC strain haplotypes, and linkage analysis was performed as described in (Ram et al., 2014). Nevus score was treated as a multinomial analysis with values ranging from one to ten. The x-axis shows the chromosomal position and the y-axis shows the 2log10(P) values; the P-values were derived from the linkage haplotype data. (B) Plot of LOD scores along chromosome 9. (C) Founder allele coefficients: plot of the calculated log-odds ratio of eight founder alleles over the chromosome where the founders are colour-coded. (D) Genes within the -1 -Log10(P) interval. Red asterisk in blue box denoted missense mutation, black asterisk in yellow box denotes an intronic, 5’or 3’ UTR variant. (D) Table showing genes in the chromosome 9 interval. Columns at right show genes carrying variants specific to the NOD haplotype. Candidates carrying missense variants are marked in red with a blue box. Genes carrying synonymous, 5 or 3’ UTR, or intronic variants, highlighted in yellow. (E) Staining for CDON in the human anagen (E) and telogen (F) hair follicle, with positivity (red) in the inner root sheath indicated by yellow arrows. There is some much weaker positivity in the suprabasal epidermis (not shown). 59

Supplementary Figure legends.

Figure S1. Graph shows spread of nevus scores for 66 CC strains.

Figure S2. (A) Summary of linkage peak on murine chromosome 11 (see legend to Figure 2). Founder strain coefficients across chromosome 11 are shown below. (B) Genes within the -1 -Log10(P) interval. Red asterisk in blue box denotes missense mutation, black asterisk in yellow box denotes an intronic variant. (C) Yellow boxes denote strains that carry the NOD founder haplotype.

Figure S3. (A) Sox10-stained skin taken from Cdk4R24C/+::Tyr-NRASQ61K mice at various time-points after birth. (B). (B) Sox10-stained skin taken from Cdk4R24C/+ mice at various time-points after birth. See immunohistochemistry staining methods in supplementary methods file.

Figure S4. Examples of Cdk4R24C/+::Tyr-NRASQ61K mouse skin for each or our arbitrary nevus cell density scores 1-10. Nevus cell deposits in the subcutis fat layer were not scored.

Figure S5. Same images as in Figure S4, except at the original size for ease of closer examination of specific sections.

Figure S6. Genome-wide scan based on nevus cell density score in (A), 26 albino CC strains, and (B), 40 pigmented CC strains. Red arrows denote the major linkage peak on chromosome 9 containing Cdon. The x-axis shows the chromosomal position and the y- axis shows the 2log10(P) values; the P-values were derived from the linkage haplotype data. Panels below contain plot of LOD scores along chromosome 9, and colour coded founder allele coefficients.

Chitsazan_Fig1

Chitsazan_Fig2

Figure S1

Figure S2

Figure S3

Figure S4

Figure S5. Same images as in Figure S4, except at the original size for ease of closer examination of specific sections.

Figure S6. Genomewide scan based on nevus cell density score in (A), 26 albino CC strains, and (B), 40 pigmented CC strains. Red arrows denote the major linkage peak on chromosome 9 containing Cdon. The x-axis shows the chromosomal position and the y- axis shows the 2log10(P) values; the P-values were derived from the linkage haplotype data. Panels below contain plot of LOD scores along chromosome 9, and color-coded founder allele coefficients.

Functional Validation of Cdon as the Modifier Gene for Development of Giant Congenital Nevus

Chapter 3

This chapter is comprised of a publication emanating from my thesis project.

2.1 Publication

2.3 Contribution

My contribution to the publication above included;

 Mouse breeding  Acquisition of mouse lesions  Design of scoring approach for scoring of nevus density in mouse using H&E stained slides  Design of 5-Plex immunohistochemical staining protocol  Acquisition of patient congenital melanocytic nevi samples from collaborators  Processing and immunohistochemical staining of lesion set  Analysis of staining patterns of lesion set  Drug treatment  Gene expression  Preparation of the manuscript

Contribution by others:

 B.F. provided technical support, animal breeding, sample collection.  R.V. gene expression study of Ptch1 KO mice  P.M. provided bioinformatic support and analysed data.  H.Y.H. provided technical support.  B.G. contributed to experimental design and supplied pathological specimens.

 H.P.S. contributed to Pathological diagnosis.  D.L. contributed to Pathological diagnosis.  W.J.M. contributed to Pathological diagnosis.  K.K. gene expression study of Ptch1 KO mice  G.M. provided the CC mouse strains, and genetic analysis software.  G.J.W. designed the project, analysed data and prepared manuscript.

Keratinocyte Sonic Hedgehog Up-regulation Drives the Development of Giant Congenital Nevi via Paracrine Endothelin-1 Secretion

Arash Chitsazan1,2, Blake Ferguson1, Rehan Villani2, Herlina Y. Handoko1, Pamela Mukhopadhyay1, Brian Gabrielli3, Wolter J. Mooi4, H Peter Soyer2, Duncan Lambie2, Kiarash Khosrotehrani2, Grant Morahan5, Graeme J. Walker1.

1Drug Discovery Group, QIMR Berghofer Medical Research Institute, Herston, Queensland, Australia. 2The University of Queensland Diamantina Institute, Translational Research Institute, Brisbane, Queensland, Australia. 3Mater Medical Research Institute, The University of Queensland, Translational Research Institute, Brisbane, Queensland, Australia. 4Department of Pathology, Vrije University Medical Center, Amsterdam, Netherlands. 5Centre for Diabetes Research, Harry Perkins Institute of Medical Research, Perth, Western Australia, Australia.

Keywords

Cdon, Collaborative Cross, Endothelin-1, Giant Congenital Nevus, Melanocytoma, Melanoma

Corresponding author: Dr Graeme Walker, QIMR Berghofer Medical Research Institute, 300 Herston Rd, Herston, QLD 4006, Australia. Tel: +61738453715, Fax: +61738453508 Email: [email protected]

Abstract

Giant congenital nevi are associated with serious clinical complications such as neurocutaneous melanosis and melanoma. Virtually nothing is known about why some individuals develop these lesions. We previously identified the sonic hedgehog (Shh) pathway regulator Cdon as a candidate nevus modifier gene. Here we validate this by studying Cdon knockout mice, and go on to establishment the mechanism by which Shh exacerbates nevogenesis. Cdon knockout mice develop blue nevi without the need for somatic melanocyte oncogenic mutation. In a mouse model carrying melanocyte NRASQ61K we found that strain backgrounds that carry genetic variants that cause increased keratinocyte Shh pathway activity, as measured by Gli1 and Gli2 expression, develop giant congenital nevi. Shh components are also active adjacent to human congenital nevi. Mechanistically, this exacerbation of nevogenesis is driven via the release of the melanocyte mitogen endothelin-1 from keratinocytes. We then suppressed nevus development in mice using Shh and endothelin antagonists. Our work suggests a novel aspect of nevus development whereby keratinocyte cytokines such as endothelin-1 can exacerbate nevogenesis, and provides potential therapeutic approaches for giant congenital nevi. Further, it highlights the notion that germline genetic variation, in addition to somatic melanocyte mutation, can strongly influence the histopathological features of melanocytic nevi.

Introduction

The presence of multiple nevi is the strongest risk factor for development of melanoma (Gandini et al., 2005). Whereas common acquired nevi emanate from the dermo-epithelial junction of the skin, many nevi are dermal or have a dermal component. Congenital nevi are mostly dermal, and observed at birth or in the very young (Sowa et al., 2008; Price et al., 2010). They may be small birthmarks anywhere on the body, large lesions mostly on the head and neck, or giant lesions sometimes covering large areas of body. The latter have an increased risk for transformation melanoma (Zaal et al., 2004). Some patients develop neural melanosis, which carries a poor prognosis, with no effective treatment. One way to approach understanding the formation of these lesions is to look for natural gene variation that confers susceptibility to their development. But in terms of germline gene variation there are no GWAS or genetic association studies conducted on human giant congenital nevi patients, nor familial studies that might reveal rare mutations.

Congenital nevi probably derive from melanocytes that have escaped the developmental route from the neural crest to the skin, and become trapped in the dermis, where they proliferate. Many transgenic mice carrying melanocyte-specific oncogenic mutations develop dermal melanocytic proliferations reminiscent of giant congenital nevi (Shakhova et al., 2012; Pawlikowski et al., 2015). Transgenic Cdk4R24C::Tyr-NRASQ61K (hereafter termed Cdk4::NRAS) mice begin to develop dermal nevi during the second week after birth (Chai et al., 2014; Chitsazan et al., 2016). Since the corresponding developmental period in which embryonic melanocytes have mostly migrated to the hair bulb is pre-natal in humans, beginning at around 18 weeks (Gleason et al., 2008), and examination of aborted foetuses at this stage have indicated the presence of deep dermal nevi at this stage (Cramer et al., 2012). one might expect that the mechanisms promoting escape of melanocytes from the hair follicle to the dermis should be similar in both species. Hence in order to discover genes affecting congenital nevogenesis, we utilised the Collaborative Cross (CC) (Morahan et al., 2008; Welsh et al., 2014), a resource of characterised mouse strains, in conjunction with the Cdk4:;NRAS model, to identify a quantitative trait locus (QTL) on mouse chromosome 9 (Chitsazan et al., 2016) influencing nevocyte density. Examination of the founder haplotype coefficients through the linked interval on chromosome 9 showed that the nevus-susceptible strains carried the causal variant derived from the NOD/LtJ founder (hereafter termed NOD). One of the best candidates in this QTL was Cdon, a sonic hedgehog (Shh) pathway regulator. Shh signalling is controlled by dependence receptors Cdon and Ptch1 (Delloye-Bourgeois et

95 al., 2013). Their inactivation by Shh removes the inhibition on smoothened (Smo), a positive regulator of Gli transcription factors (Ruizi Altaba et al., 1997; Taipale et al., 2002; Ouspenskaia et al., 2016). Deregulation of the SHH pathway drives the formation of the keratinocyte-derived skin cancer, basal cell carcinoma (BCC), which are sometimes pigmented. Here we define a cell extrinsic role for the Shh-endothelin1 axis in the development of giant congenital nevi.

Results

Genetic Background Dictates the Histopathology of Melanocytic Hyperplasia Cdk4::NRAS mice develop dermal nevi by postnatal day 10, from melanocytes escaping the hair follicles (Chitsazan et al., 2016). Cdk4R24C/R24C::Tyr-NRASQ61K/+ males were bred with females from many CC strains to generate progeny heterozygous for Cdk4 and NRAS transgenes. Dorsal skin was collected from >3 progeny of each CC cross, and the density of dermal nevocytes scored from 1 (least dense) to 10 (most dense). The method of scoring each strain for nevus cell density in the dermis is described in Chitsazan et al., (2016). Here we went on to diagnose lesions at each density level according to histopathological criteria for diagnosing human melanocytic hyperplasia. Those with a score of 1-4 were best defined as superficial or deep blue nevi (Figure 1a; Figure S1), a score of 5-6 as deep penetrating nevi, 7 as epitheloid blue nevus, and 8-10 as melanocytoma. The nevus density score was closely related to depth and histopathological diagnosis (Figure 1b). In general a particular histopathological parameter may be shared by more than one lesion subtype, and we observed a gradation of nevocyte density scores closely, but not perfectly, reflecting nevus subtype (Figures 1c). Despite all mice carrying a melanocyte NRASQ61K mutation making them susceptible to the development of nevi, germ-line genetic variation dramatically alters nevus histopathological presentation.

Validation of a candidate Modifier Gene for Nevus Exacerbation Previously we identified Cdon as candidate nevus modifier gene by studying Cdk4::NRAS mice crossed onto many CC strain backgrounds (Chitsazan et al., 2016). We used heterozygous loss of Cdon, since homozygous loss is embryonic lethal. Cdon+/- mice (n=4), and not littermate wild type controls (n=3), developed blue nevi in old age. Dermoscopy of the lesions showed a blue hue suggesting nevocyte location in the deep dermis (Figure 2a). Dendritic and epithelioid melanocytes were present. Whereas in the normal skin of these mice hair follicles were in telogen (the resting phase of the hair cycle), surrounding the blue nevi there were more hair follicles and they were in in anagen (the 96 growth phase of the hair cycle) (Figures 2b and Figure S4a and b). Also, these aged Cdon+/- mice had fewer hair follicles than controls (Figure S4). The differences in hair follicle number and the inappropriate stage of cycle are unsurprising given the pivotal role of Shh in hair follicle homeostasis (Sato et al., 1999). Cdon+/- mice develop dermal nevi without the need for somatic BRAF or NRAS mutation in melanocytes.

We then assessed the visual appearance of Cdk4::NRAS mice carrying the CdonNOD allele (Figure 2c), versus controls carrying the non-NOD allele (by genotyping the CdonP456S mutation). Both cohorts were hyperpigmented, but susceptible mice carrying CdonNOD developed leathery thickenings reminiscent of giant congenital nevi. Dermoscopy of the tail and dorsal skin of resistant mice showed a scattered blue hue on a white background with some small superficial and deep nodules, whereas in the susceptible mice carrying CdonNOD both the deeper nevi and nodules were much larger (Figure 2c). Histologically, the former were blue nevi, while the latter consisted of strings interconnected melanocytomas producing a macroscopic appearance of giant congenital nevus. The formation of leathery lesions can be observed in patients with either medium- sized (Figure 2d) or giant congenital nevi (Figure 2e). The human and murine giant congenital nevi lesions showed other similarities including generalised hyperpigmentation, some large plaques, and ectopic hair growth (hypertrichosis). lnactivation of the Shh pathway in mouse keratinocytes, in conjunction with melanocyte NRASQ61K mutation, produced large leathery melanocytic plaques reminiscent of human giant congenital nevi. Thus, while we do not know that status of the SHH pathway in such individuals, upregulated keratinocyte SHH and EDN1 might be one potential mechanism.

In vivo Activation of Shh Signaling in Cdk4::NRAS Mice with Giant Congenital Nevi To confirm activation of the Shh pathway in strains carrying the CdonNOD variant we stained the mouse skin for Gli1 and Gli2, markers of Shh activation, using multiplexed immunofluorescence (IF) staining with Tyramide-based amplification (Figure 3). We stained dorsal skin from Cdk4::NRAS mice with Sox10 (a melanocyte maker) and Gli1 and Gli2 antibodies, to examine protein expression in the vicinity of the early nevi (mice at P10 and P21), and melanocytoma (in adult mice). In CdonNOD strains, at P10 when nevi are beginning to form, Gli proteins were mostly in the hair bulb (Figure 3a). In susceptible adult mice with follicles in telogen, Gli1 was expressed in a much higher proportion of cells within the skin (Figure 3b). In contrast, in the skin of resistant strains there was low levels Gli1 expression at P10 and in adults (Figures 3c and 3d). This is likely due to the low (and

97 only scattered) Shh activity in normal skin (except during anagen for which Shh activity is essential). Quantitating the proportion of Gli1 positive cells in the skin, we found that both in young mice at P10 and P21 (Figure 3e), and in adult animals (Figure 3f) the CdonNOD- carrying strains contained significantly more Gli-positive keratinocytes.

In human skin, CDON is expressed mainly in the suprabasal epidermis, and in parts of the hair follicle, probably the inner root sheath (Figure S2). In mouse skin also Cdon is expressed in the epidermis and the hair follicle, although because of thin hair shaft it is more difficult to ascribe staining to a specific layer. We stained four human intradermal nevi (Figure 3g and S3a) with congenital features including deep extension into the reticular dermis, and peri-appendageal accentuation. CDON was mostly expressed in differentiated keratinocytes in upper epidermis, and in the hair follicle. The expression of GLI1 in keratinocytes is consistent with a role for activated SHH in exacerbating human dermal nevus development. In the mouse this effect largely influences the early stages of nevus formation, which occur prenatally in humans. Understanding the temporal development (beginning at ~ 10 days of age), and the skin compartment in which they are first observed (in the superficial dermis adjacent to hair follicles) in our mouse model, provides us with a unique opportunity to better study the mechanisms by which natural genetic variation influences congenital nevi formation.

Linking Keratinocyte Shh Signalling to Development of Dermal Nevi LDE225, a potent antagonist of the Shh pathway protein Smo, is an FDA-approved drug for topical treatment of BCC. We administered this topically in DMSO to the dorsal skin every day from P5 to P10. Skin samples harvested at the end of P10 from drug-treated and control groups were stained for Sox10 (Figure 3a). Whereas at P10 small groups of melanocytes (early nevi) were beginning to form in the control mice, these were not evident in the drug-treated animals. The proportion of Sox10 positive nuclei in the dorsal skin of LDE225-treated mice was significantly lower in the superficial dermis than in controls (Figure 3i). The drug treatment also suppressed melanocyte numbers in the hair bulb suggesting that less melanocytes may be escaping from it (Figure 3i). Thus, drug targeting of Shh pathway activity can significantly suppress nevus formation.

Shh-Edn1 Axis Exacerbates Nevogenesis Basal cell carcinomas in humans and mice are sometimes pigmented (Chow et al., 2011; Grachtchouk et al., 2003; Moore et al., 2012; Peterson et al., 2015). In mice with keratinocyte-specific inactivation of Ptch1 (Peterson et al., 2015), or activation of other Shh pathway components e.g. Smo (Grachtchouk et al., 2003), Gli2 (Grachtchouk et al., 2000), 98 and Shh (Ellis et al., 2003), hyperpigmentation is usually evident as pigmented melanocytes within or near BCCs. These transgenics were all generated to study BCC hence did not properly analyse the hypigmentation. Where hyperpigmented skin was reported, and adequate published images available, the lesions appeared to be blue nevi (Ellis et al., 2003), which also develop in the Cdon+/- mice.

We set out to ask which genes that influence melanocyte behaviour are over-expressed in the skin of these transgenics? We compared gene expression from RNA sequencing of Ptch1 knockout epidermis to that of controls. For additional confirmation we used published data from another mouse BCC model (Youssef et al., 2012). In this case YFP- positive keratinocytes were isolated from tamoxifen-treated and control K14-CreER::Rosa- Smo2 (K14-Smo2) mice, and microarray gene expression analysis performed. In both cases Shh target genes are up-regulated in keratinocytes e.g. Gli1, Shh, Ptch1, Hhip (Figure 5a). Given that transgenic over-expression of known melanocyte mitogens like hepatocyte growth factor (Hgf), endothelin 3 (Edn3), in murine keratinocytes induces similar melanocytic dermal lesions to that we see in our model (Aoki et al., 2009), we looked for up-regulation of genes encoding secreted cytokines known to be melanocyte mitogens; Edn1, Edn2, Edn3, Hgf, Fgf, Pomc (Msh and Acth), and Ngf (Hirobe 2005), and Wnt5a and Wnt7b (Figure 4a). Edn1 (endothelin 1) (Zhang et al., 2013; Hyter et al., 2013; Urtatiz et al., 2016; Chiriboga et al., 2016) was the only such gene statistically upregulated in both the Ptch1 and Smo mouse model epidermis. By IF staining of mouse dorsal skin we found that Edn1 is expressed in the epidermis and certain cells in the outer root sheath and hair bulb (Figure 4b). We quantitated the expression of Edn1 protein in susceptible and resistant strains. The number of positive cells per mm2 was significantly higher in the susceptible strains (Figure 4c), suggesting that Edn1 secretion is a plausable mechanism for exacerbation of nevogenesis. To confirm this, we utilized an inhibitor of the Edn1 receptor, Bosentan. Daily treatments of Cdk4-NRAS mice (from P5 to P10 intraperitoneally and topically) with Bosentan significantly reduced the appearance of developing bands of nevi in the superficial dermis (Figures 4f and 5g).

Discussion Although NRASQ61K mutation is the major somatic mutation carried in giant congenital nevi, virtually nothing else is known about the mechanism of formation of these 99 lesions. Despite the evolutionary differences in the pigmentary system between species (mouse melanocytes are usually limited to hair follicles whereas in humans they are also in the epidermis), virtually all genes involved in regulating mouse melanocyte behaviour (i.e. pigmentation) are relevant in humans (Walker et al., 2011). Given that we have gone to great lengths to diagnose the murine nevi precisely according to the histopathological criteria used for human nevi, and shown that the pathway we have discovered is active in the epidermis adjacent to human congenital nevi, the cell extrinsic mechanism we have delineated could be involved in the formation of human nevi. It is not known whether germline CDON variants influence the development of human congenital nevi, but our findings enable us to postulate that variants in any gene that results in up-regulation of a pathway that increases expression of a secreted keratinocyte cytokine that is a melanocyte mitogen, could be a candidate.

A role for paracrine Edn1 driving nevogenesis seems uncontroversial. It is a mitogen for human melanocytes in culture (Imokawa et al., 1992) and in human skin explants (Takeo et al., 2016), and is involved in wound-induced melanocyte activation (Park et al., 2015). Edn1, along with Edn2 and Edn3, are mostly active only in anagen, or growth phase of the hair cycle, when they assist either to activate melanocyte stem cells, and/or to assist in migration of melanocytes to the hair bulb (Mort 2015). Hence under stress such as increased endothlin secretion or the presence of a NRAS or BRAF mutation it is unsurprising that melanocytes escape their follicular niches and congregate in the dermis as nevi. Edn1 binds to the melanocyte Ednrb receptor and stimulates signalling through Gnaq/Gna11 and subsequently the MAPK pathway (Figure 6). GNAQ is somatically mutated in ~80% of human blue nevi, suggesting that signalling through it is important in dermal nevus development, and mice expressing an activated Gnaq mutant in their melanocytes exhibit essentially the same phenotype in terms of dermal melanocytic nevi that we observe in our model with activated keratinocyte Shh (Huang et al., 2015). Perhaps SHH-EDN1 signalling acts as a rheostat, via melanocyte EDN1-GNAQ-MAPK, in exacerbating blue nevi to form very large or giant nevi (Figure 6).

SHH is normally associated with BCC. PTCH1 is frequently mutated in BCC-prone patients, and the SHH pathway is mutated somatically in BCC (Tang et al., 2010; Briscoe et al., 2013). Notably, the mouse strains carrying the CdonNOD variant do not develop BCC, presumably since the associated keratinocyte Shh up-regulation is relatively subtle. But one would expect BCCs to be pigmented, and many are, with melanocytes unusually located around the edges or within the lesion. EDN1 is highly expressed adjacent 100 pigmented, but not non-pigmented human BCCs (Lan et al., 2005). Since 80% of Asian, but only 3% of Caucasian BCCs show pigmentation (Moore et al., 2012; Chow et al., 2011), it is likely that modifier alleles that differ between these ethnic backgrounds modulate the SHH-EDN1 axis. However murine models with keratinocyte-specific deregulation of Shh components e.g. Shh (Ellis et al., 2003), Ptch1 (Peterson et al., 2015), Smo (Grachtchouk et al., 2003; Eberl et al., 2012), and Gli2 (Grachtchouk et al., 2000), all exhibit melanocyte hyperplasia.

An important message is the notion that different genes act at different stages of tumor progression. We show (Figure 1c) that the strains carrying the CdonNOD allele and upregulated keratinocyte Shh components are not necessarily more prone to the development of melanoma, although the nevi are much larger. Since the transformation of an individual nevus (of which there are >10,000 per mouse on an FVB strain background) is extremely low (a 0.04% probability that a nevus will convert to melanoma; Chi et al., 20012), it must be modified by epigenetic, or other genetic background effects yet been elucidated.

Our strategy using mouse models could be helpful in gaining insights into genetic background influences in other rare diseases. Giant congenital nevi are only seen in about 1/20,000 births, and no genetic studies have been performed on large case-control cohorts. But they can be disfiguring, covering significant parts of the body surface, and with a high propensity to convert to melanoma. They are not always stable. Satellite lesions can form, and neural melanosis sometimes becomes manifest in these patients. Studying human lesions post-natally does not tell us how they are formed. In years to come germ-line DNA from large enough cohorts may become available for genome wide association studies into susceptibility. But not being a common disease it will be difficult to attract traditional research funding. To confirm our hypothesis that keratinocyte cytokines may feed the formation of giant nevi would entail sensitive gene expression and or immunohistochemical studies of representative skin samples from perhaps hundreds of patients and ethnic matched controls, again a difficult study to realize.

Targeting keratinocyte cytokines might provide hope for a drug therapy by which giant nevi could be controlled, or even reversed, since they can spontaneously regress (Arpaia et al., 2005; Nath et al., 2011; Cusak et al., 2009). Surgical excision is the gold standard treatment for most cutaneous tumours. But topical treatments are now available for solar keratosis and in situ squamous and basal cell carcinomas (e.g. fluorouracil,

101 imiquimod, Picato gels), lesions previously only removed by excision. The same may eventually be true for large/giant nevi. Topical treatment of pigmented BCCs with the SMO antagonist LDE225, and Bosentan an inhibitor of Edn signalling, which we used in our study to suppress the development of dermal nevi, results in the disappearance of the associated melanocytosis.

Methods

Mice and Phenotyping Mouse breeding and screening for nevus density modifier gene: By breeding Cdk4R24C/R24C::Tyr-NRASQ61K/+ males with females from 70 CC strains. Experiments were undertaken with institute animal ethics approval (A98004M). Cdon knockout mice were provided by Robert S. Krauss (Icahn School of Medicine at Mount Sinai, NY). Collaborative Cross mouse strains were provided by Grant Morahan (Harry Perkins Institute of Medical Research, Perth). By breeding Cdk4R24C/R24C::Tyr-NRASQ61K/+ males (FVB background) with CC females we generated progeny heterozygous for Cdk4 and NRAS (Chitsazan 2016). At P10 small nests of melanocytic cells (nevi) appear within the superficial dermis around hair follicles (Chai et al., 2013). Nevi are completely formed by P40.

RNA Sequencing Four weeks after tamoxifen injection, epidermal cells of K14-CreER::Ptch1fl/fl and control mice were recovered by digestion and flow sorting of EPCAM+ epidermal cells. RNA was isolated using RNeasy Kit and libraries generated with Illumina mRNA kit. Sequencing was performed using Ilumina HiSeq chemistry, with 50bp single reads. The RNA- seq samples were mapped to mouse genome MM10 using TopHat2, then mapped reads for each gene counted for every sample using HT-Seq. We used EdgeR and DEseq to estimate the significance of differential expression between groups.

Haematoxylin and Eosin Staining Protocol Slides from formalin-fixed paraffin-embedded blocks were dewaxed and stained with haematoxylin and eosin using Leica autostainer XL.

Dermoscopy of Mouse Skin

102

After shaving to remove hair, clinical images were acquired with a digital camera (Canon PowerShot G10; Canon, Tokyo, Japan). Detailed macroscopic images were acquired with the same camera equipped with a dermoscopic attachment (Dermlite©, 3Gen, USA).

Multiplexed Fluorescent Immunohistochemistry with Tyramide Signal Amplification Sections from paraffin-embedded lesions were dewaxed and treated with Dako low pH antigen retrieval solution at 1000C for 15 minutes. Endogenous peroxidase activity was quenched in 3% H2O2, and sections blocked with 5% bovine serum albumin. Primary Sox10 (sc-17342), Gli1 (ab151796), Gli2 (AF3635), Cdon (sc-134837) and Edn1 (ab117757) antibodies were applied, followed by the appropriate secondary antibodies. After washing, the signal was amplified using Perkin Elmer Tyramide amplificiation kit (FITC, CY3, CY3.5, CY5), slides were washed, counterstained with DAPI, and coverslipped using DAKO mounting medium. Slides were scanned using the Vectra spectral imaging system (PerkinElmer). Images were scanned at at 4X magnification. Multispectral images (MSI) were captured using Phenochart software and scanned from 420 nm to 720 nm using Vectra 3.1 Automated quantitative pathology imaging system. Three MSIs per sample was analysed using InForm analysis software. Positive cells were segmented and counted per 20X magnification field.

LDE225 Treatment To generate high nevus count mice we crossed NOD X Cdk4R24C/R24C::Tyr-NRASQ61K transgenics. We topically treated the dorsal area with LDE225-DMSO (10 mg/ml, Selleck Chemicals) every day from P5 to P10 (with skin harvested at P10). Controls were treated with DMSO alone. Skin was harvested from drug treated and control groups (n=21) stained for Sox10 and slides digitised with an Aperio AT-Turbo Brightfield slide scanner. Sox10 positive nuclei count was determined using the Aperio IHC nuclear algorithm V8.001, and nuclear staining stratified into three intensities (1+ to 3+). We scored the number of strongly positive nuclei (3+) per 2 mm2 section.

Bosentan Hydrate Treatment To generate nevus-susceptible mice crossed NOD mice with Cdk4R24C/R24C::Tyr-NRASQ61K transgenics. Pups were both injected intraperinatally with Bosentan hydrate-DMSO (10mM/1mL in DMSO, Selleck Chemicals) every day from P5 to P10 (skin harvested at P10). Controls were treated with DMSO. Skin was harvested from drug-treated and control groups (n=14). H&E-stained slides were digitized with Aperio AT-Turbo Brightfield slide

103 scanner, and intensity of pigmented cells analysed with Aperio-V9 software. We quantitated to intensity of strongly positive cells.

Acknowledgments

We thank members of the Oncogenomics laboratory (QIMR Berghofer Medical Research Institute), for helpful comments. We are grateful to Artur Zembowicz for diagnosis of melanocytic lesions, Brandon Wainwright and Kerstin Zoidl for their critiques of the manuscript. Thanks to Clay Winterford for assistance with multiplex-IHC set up, Nigel Waterhouse and Tam Nguyen for assistance with microscopy, and Sang-Hee Park for histology sample preparation. We thank Roya Farhadi for clinical images of giant congenital nevi, and Robert Krauss and Mingi Hong for providing Cdon knockout mice. We thank Geniad Pty Ltd for generously providing CC mice. This work was supported by Melanoma Research Alliance, Washington, DC, USA, and the Cancer Council of Queensland, Australia.

Author Contributions

A.C. conceived the project, designed and performed the experiments, analysed data and prepared the manuscript. B.F. provided technical support, animal breeding, sample collection, prepared H&E slides and digitized the slides for Figures 1. And 2 . H.Y.H. performed DNA genotyping and edited manuscript. P.M. provided bioinformatic support and analysed data. R.V. and K.K. provided K14::Ptch1 KO mice and performed the RNA Seq on Ptch1 KO epidermis. B.G. contributed to experimental design and supplied pathological specimens. D.L., H.P.S., and W.J.M. contributed to Pathological diagnosis. G.M. provided the CC mouse strains, and genetic analysis software. G.J.W. designed the project, analysed data and prepared manuscript.

104

References

Aoki H, Yamada Y, Hara A, Kunisada T. 2009. Two distinct types of mouse melanocyte: differential signaling requirement for the maintenance of non-cutaneous and dermal versus epidermal melanocytes. Development 136: 2511-21

Arpaia N, Cassano N, Filotico R, Laricchia F, Vena GA. 2005. Unusual clinical presentation of regression in a congenital melanocytic nevus. Dermatol Surg 31: 471-473.

Briscoe J, Therond PP. 2013. The mechanisms of Hedgehog signalling and its roles in development and disease. Nat Rev Mol Cell Biol 14: 416-429.

Castets M, Broutier L, Molin Y, Brevet M, Chazot G, Gadot N, Paquet A, Mazelin L, Jarrosson-Wuilleme L, Scoazec JY et al. 2011. DCC constrains tumour progression via its dependence receptor activity. Nature 482: 534-537.

Chai E, Ferguson B, Prow T, Soyer P, Walker G. 2014. Three-dimensional modelling for estimation of nevus count and probability of nevus-melanoma progression in a murine model. Pigment Cell Melanoma Res 27: 317-319.

Chiriboga L, Meehan S, Osman I, Glick M, de la Cruz G, Howell BS, Friedman-Jimenez G, Schneider RJ, Jamal S. 2016. Endothelin-1 in the tumor microenvironment correlates with melanoma invasion. Melanoma Res 26: 236-244.

Chitsazan A, Ferguson B, Ram R, Mukhopadhyay P, Handoko HY, Gabrielli B, Soyer PH, Morahan G, Walker GJ. 2016. A mutation in the Cdon gene potentiates congenital nevus development mediated by NRAS(Q61K). Pigment Cell Melanoma Res 29: 459-464.

Chow VL, Chan JY, Chan RC, Chung JH, Wei WI. 2011. Basal cell carcinoma of the head and neck region in ethnic chinese. Int J Surg Oncol 2011: 890908.

Cole F, Krauss RS. 2003. Microform holoprosencephaly in mice that lack the Ig superfamily member Cdon. Curr Biol 13: 411-415.

Cramer S, Fesyuk A. 2012. On the Development of Neurocutaneous Units—Implications for the Histogenesis of Congenital, Acquired, and Dysplastic Nevi. Am J Dermatopathol 34: 60-81.

105

Cusack C, Warde D, Lynch W, McDermott M, Watson R. 2009. Dramatic spontaneous regression of a medium-sized congenital melanocytic naevus. Clin Exp Dermatol 34: e34- 36.

Dankort D, Curley DP, Cartlidge RA, Nelson B, Karnezis AN, Damsky WE, Jr., You MJ, DePinho RA, McMahon M, Bosenberg M. 2009. Braf(V600E) cooperates with Pten loss to induce metastatic melanoma. Nat Genet 41: 544-552.

Delloye-Bourgeois C, Gibert B, Rama N, Delcros JG, Gadot N, Scoazec JY, Krauss R, Bernet A, Mehlen P. 2013. Sonic Hedgehog promotes tumor cell survival by inhibiting CDON pro-apoptotic activity. PLoS Biol 11: e1001623.

Ellis T, Smyth I, Riley E, Bowles J, Adolphe C, Rothnagel JA, Wicking C, Wainwright BJ. 2003. Overexpression of Sonic Hedgehog suppresses embryonic hair follicle morphogenesis. Dev Biol 263: 203-215.

Etnyre D, Stone AL, Fong JT, Jacobs RJ, Uppada SB, Botting GM, Rajanna S, Moravec DN, Shambannagari MR, Crees Z et al. 2014. Targeting c-Met in melanoma: mechanism of resistance and efficacy of novel combinatorial inhibitor therapy. Cancer Biol Ther 15: 1129-1141.

Flores-Sarnat L. 2013. Neurocutaneous melanocytosis. Handb Clin Neurol 111: 369-388.

Gandini S, Sera F, Cattaruzza MS, Pasquini P, Zanetti R, Masini C, Boyle P, Melchi CF. 2005. Meta-analysis of risk factors for cutaneous melanoma: III. Family history, actinic damage and phenotypic factors. Eur J Cancer 41: 2040-2059.

Gavriilidis P, Michalopoulou I, Chatzikakidou K, Nikolaidou A. 2013. Pigmented epithelioid melanocytoma: a new concept encompassing animal-type melanoma and epithelioid blue nevus. BMJ Case Rep 2013.

Gleason BC, Crum CP, Murphy GF. 2008. Expression patterns of MITF during human cutaneous embryogenesis: evidence for bulge epithelial expression and persistence of dermal melanoblasts. J Cutan Pathol 35: 615–22.

Grachtchouk M, Mo R, Yu S, Zhang X, Sasaki H, Hui CC, Dlugosz AA. 2000. Basal cell carcinomas in mice overexpressing Gli2 in skin. Nat Genet 24: 216-217.

106

Grachtchouk V, Grachtchouk M, Lowe L, Johnson T, Wei L, Wang A, de Sauvage F, Dlugosz AA. 2003a. The magnitude of hedgehog signaling activity defines skin tumor phenotype. Embo j 22: 2741-2751.

Hirobe T. (2005). Role of keratinocyte-derived factors involved in regulating the proliferation and differentiation of mammalian epidermal melanocytes. Pigment Cell Res 18: 2-12.

Huang JL, Urtatiz O, Van Raamsdonk CD. 2015. Oncogenic G Protein GNAQ Induces Uveal Melanoma and Intravasation in Mice. Cancer Res 75: 3384-97.

Hyter S, Coleman DJ, Ganguli-Indra G, Merrill GF, Ma S, Yanagisawa M, Indra AK. 2013. Endothelin-1 is a transcriptional target of p53 in epidermal keratinocytes and regulates ultraviolet-induced melanocyte homeostasis. Pigment Cell Melanoma Res 26: 247-258.

Imokawa G, Yada Y, Miyagishi M. 1992. Endothelins secreted from human keratinocytes are intrinsic mitogens for human melanocytes. J Biol Chem 267: 24675-80.

Lan CC, Wu CS, Cheng CM, Yu CL, Chen GS, Yu HS. 2005. Pigmentation in basal cell carcinoma involves enhanced endothelin-1 expression. Exp Dermatol 14: 528-34.

Martins da Silva VP, Marghoob A, Pigem R, Carrera C, Aguilera P, Puig-Butille JA, Puig S, Malvehy J. 2017. Patterns of distribution of giant congenital melanocytic nevi (GCMN): The 6B rule. J Am Acad Dermatol 76: 689-694.

Mehlen P, Guenebeaud C. 2010. Netrin-1 and its dependence receptors as original targets for cancer therapy. Curr Opin Oncol 22: 46-54.

Moore MG, Bennett RG. 2012a. Basal cell carcinoma in asians: a retrospective analysis of ten patients. J Skin Cancer 2012: 741397.

Morahan G, Balmer L, Monley D. 2008. Establishment of "The Gene Mine": a resource for rapid identification of complex trait genes. Mamm Genome 19: 390-393.

Mort RL, Jackson IJ, Patton EE. 2015. The melanocyte lineage in development and disease. Development 142: 620–632.

Nath AK, Thappa DM, Rajesh NG. 2011. Spontaneous regression of a congenital melanocytic nevus. Indian J Dermatol Venereol Leprol 77: 507-510.

107

Ouspenskaia T, Matos I, Mertz AF, Fiore VF, Fuchs E. 2016. WNT-SHH Antagonism Specifies and Expands Stem Cells prior to Niche Formation. Cell 164: 156-169.

Park PJ, Lee TR, Cho EG. 2015. Substance P stimulates endothelin 1 secretion via endothelin-converting enzyme 1 and promotes melanogenesis in human melanocytes. J Invest Dermatol 135: 551-559.

Patel AB, Harting MS, Smith-Zagone MJ, Hsu S. 2008. Familial basaloid follicular hamartoma: a report of one family. Dermatol Online J 14: 14.

Pawlikowski JS, McBryan T, van Tuyn J, Drotar ME, Hewitt RN, Maier AB, King A, Blyth K, Wu H, Adams PD. 2013. Wnt signaling potentiates nevogenesis. Proc Natl Acad Sci U S A 110: 16009-16014.

Peterson SC, Eberl M, Vagnozzi AN, Belkadi A, Veniaminova NA, Verhaegen ME, Bichakjian CK, Ward NL, Dlugosz AA, Wong SY. 2015. Basal cell carcinoma preferentially arises from stem cells within hair follicle and mechanosensory niches. Cell Stem Cell 16: 400-412.

Powell DR, Williams JS, Hernandez-Lagunas L, Salcedo E, O'Brien JH, Artinger KB. 2015. Cdon promotes neural crest migration by regulating N-cadherin localization. Dev Biol 407: 289-299.

Price HN, Schaffer JV. 2010. Congenital melanocytic nevi-when to worry and how to treat: Facts and controversies. Clin Dermatol 28: 293-302.

Ruiz i Altaba A. 1997. Catching a Gli-mpse of Hedgehog. Cell 90: 193-196.

Sato N, Leopold PL, Crystal RG. 1999. Induction of the hair growth phase in postnatal mice by localized transient expression of Sonic hedgehog. J Clin Invest 104: 855-864.

Shakhova O, Zingg D, Schaefer SM, Hari L, Civenni G, Blunschi J, Claudinot S, Okoniewski M, Beermann F, Mihic-Probst D et al. 2012. Sox10 promotes the formation and maintenance of giant congenital naevi and melanoma. Nat Cell Biol 14: 882-890.

Skvara H, Kalthoff F, Meingassner JG, Wolff-Winiski B, Aschauer H, Kelleher JF, Wu X, Pan S, Mickel L, Schuster C et al. 2011. Topical treatment of Basal cell carcinomas in nevoid Basal cell carcinoma syndrome with a smoothened inhibitor. J Invest Dermatol 131: 1735-1744.

108

Sowa J, Kobayashi H, Ishii M, Kimura T. 2008. Histopathologic findings in Unna's nevus suggest it is a tardive congenital nevus. Am J Dermatopathol 30: 561-566.

Taipale J, Cooper MK, Maiti T, Beachy PA. 2002. Patched acts catalytically to suppress the activity of Smoothened. Nature 418: 892-897.

Tang T, Tang JY, Li D, Reich M, Callahan CA, Fu L, Yauch RL, Wang F, Kotkow K, Chang KS et al. 2011. Targeting superficial or nodular Basal cell carcinoma with topically formulated small molecule inhibitor of smoothened. Clin Cancer Res 17: 3378-3387.

Takeo M, Lee W, Rabbani P, Sun Q, Hu H, Lim CH, Manga P,1 Ito I. 2016. EdnrB governs regenerative response of melanocyte stem cells by crosstalk with Wnt signaling. Cell Rep 15: 1291–1302.

Urtatiz O, Van Raamsdonk CD. 2016. Gnaq and Gna11 in the Endothelin Signaling Pathway and Melanoma. Front Genet 7: 59.

Viros A, Sanchez-Laorden B, Pedersen M, Furney SJ, Rae J, Hogan K, Ejiama S, Girotti MR, Cook M, Dhomen N et al. 2014. Ultraviolet radiation accelerates BRAF-driven melanomagenesis by targeting TP53. Nature 511: 478-482.

Walker GJ, Soyer HP, Terzian T, Box NF. 2011. Modelling melanoma in mice. Pigment Cell Melanoma Res 24: 1158-76.

Welsh CE, Miller DR, Manly KF, Wang J, McMillan L, Morahan G, Mott R, Iraqi FA, Threadgill DW, de Villena FP. 2012. Status and access to the Collaborative Cross population. Mamm Genome 23: 706-712.

Yang SH, Andl T, Grachtchouk V, Wang A, Liu J, Syu LJ, Ferris J, Wang TS, Glick AB, Millar SE et al. 2008. Pathological responses to oncogenic Hedgehog signaling in skin are dependent on canonical Wnt/beta3-catenin signaling. Nat Genet 40: 1130-1135.

Youssef KK, Lapouge G, Bouvree K, Rorive S, Brohee S, Appelstein O, Larsimont JC, Sukumaran V, Van de Sande B, Pucci D et al. 2012. Adult interfollicular tumour-initiating cells are reprogrammed into an embryonic hair follicle progenitor-like fate during basal cell carcinoma initiation. Nat Cell Biol 14: 1282-1294.

Zaal LH, Mooi WJ, Sillevis Smitt JH, van der Horst CM. 2004. Classification of congenital melanocytic naevi and malignant transformation: a review of the literature. Br J Plast Surg 57: 707-719. 109

Zembowicz A, Scolyer RA. 2011. Nevus/Melanocytoma/Melanoma: an emerging paradigm for classification of melanocytic neoplasms? Arch Pathol Lab Med 135: 300-306.

Zhang P, Liu W, Yuan X, Li D, Gu W, Gao T. 2013. Endothelin-1 enhances the melanogenesis via MITF-GPNMB pathway. BMB Rep 46: 364-369.

110

Figure legends

Figure 1. Comparison of human and murine melanocytic lesions. (a) The inner ring shows representative images of H&E staining of murine melanocytic lesions at each nevocyte density score (1-10). The outer circle shows representative images of H&E staining of corresponding human melanocytic lesions. (b) Distribution of 70 CC strains based on their nevocyte density score and histopathological diagnosis. SBN: Superficial blue nevus, DBN: Deep blue nevus, DPN: Deep penetrating nevus, EBN: Epithelioid blue nevus and M: Melanocytoma. (c) Heat map of pigmentation in megapixel per 20X field of digitized mouse skin from 40 pigmented CC strains bearing Cdk4::NRAS and their relevant 1-10 nevocyte density score.

Figure 2. Giant congenital nevus is composed of interconnected melanocytomas. (a) Lesions in Cdon+/- mice. By dermoscopy blue coloration is due to dermal melanocytes (blue nevus). H&E shows dermial dendritic melanocytes (yellow arrow), while the white arrow shows dermal epithelioid melanocytes. (b) Normal skin from wild type and Cdon+/- mice, showing higher density of hair follicles in wild type. (c) Visual appearance of dorsal surface of susceptible Cdk4::NRAS::CdonNOD/+ mouse (right), at 8 months of age, versus a resistant Cdk4::NRAS::Cdon+/+ (+/+ = wild type) mouse (left), at 24 months, with dermoscopic images of tail and dorsal skin, and H&E staining of dorsal skin. Both mice are hyperpigmented, but the animal carrying the Cdon NOD allele exhibits a large thick giant congenital nevus-like lesion covering the majority of the dorsal surface. All Scale bars = 300 μm. (d) Large human congenital nevus. Red box shows large hyperpigmented plaque within nevus area. (e) Giant human congenital nevus covering most of the child’s leg.

Figure3. Upregulation of Gli in keratinocytes potentiates nevogenesis. Images of 4- plex fluorescent immunohistochemistry with Tyramide signal amplification using Sox10/Gli1/Gli2 antibodies. Samples are of dorsal skin. (a-d) The magnification of A and C, and B and D, are the same. Images show susceptible Cdk4::NRAS::CdonNOD/+ mice and resistant Cdk4::NRAS::Cdon+/+ mice at P10 and adulthood. Yellow arrows denote melanocytoma. (e and f) Proportion of Gli1-positive cells per mm2 in dorsal skin from susceptible and resistant mouse strains at P10 (n=10), P21 (n=10) and adulthood (n=21), 111 quantitated using InForm software. (p-values calculated using two-tailed T test, ****=p<0.0001). Scale bars = 60 μm. (g) Human congenital nevus expressing Shh pathway components. Scale bars = 800 μm. (h) Sox10 staining of dorsal skin from susceptible Cdk4::NRAS::CdonNOD/+ mice treated with LDE225 versus (control left panel). Red dotted line denotes superficial blue nevus. Blue arrows denote melanocytes located ectopically in, and at risk of escape from, the hair bulb. Graphs to the right show Sox10-positive nuclei counts in hair follicle bulb (i) at 40X, and the superficial dermis (j), at 20X magnification, of LDE225-treated mice (n=11) and controls (n=10). Cells quantitated using the nuclear staining algorithm of the AperioTM. Scale bars = 60 μm. ****=p<0.0001, **=p<0.007.

Figure 4. Mechanism of exacerbation of nevogenesis. (a) Relative gene expression (Log fold change) in tamoxifen-induced versus control K14-Ptch1 and K14CREER/Rosa- SmoM2 epidermis. Differences significant except where marked (n.s=p> 0.05). (b) Expression of Edn1 in p10 anagen skin from Cdk4::NRAS mice carrying the CdonNOD variant. (c) Quantitative protein expression analysis of Edn1 in susceptible Cdk4::NRAS::CdonNOD/+ (n=14 mice) versus a resistant Cdk4::NRAS::Cdon+/+ (+/+ denotes wild type at Cdon) (n=7), as measured with InForm software (p=value calculated using the two-tailed T test (****= p<0.0001). (d) Edn1 expression in adult wt versus Cdon+/- in telogen. (e) Quantitative protein expression analysis of Edn1 in adult wt (n=3) versus Cdon+/- (n=4). (**= p<0.005). (f) Images of H&E staining of dorsal skin from control and Bosentan (an endothelin receptor inhibitor)-treated Cdk4::NRAS::CdonNOD/+ mice. Yellow arrowheads indicate blue nevi. White asterix denotes hair bulb filled with melanocytes in control but not Bosentan-treated mice. Graphs at right show pigmentation level in the dermis (g), and hair follicles (h) of control (n=7) and Bosentan-treated Cdk4::NRAS::CdonNOD/+ mice. There was significantly more pigmentation in the control skin (two-tailed T test, **= p<0.001), ****= p<0.0001). All Scale bar = 60 μm.

Figure 6. Model of nevus exacerbation. Proposed mechanism of action of the Cdon- Shh-Edn1 axis, exacerbating nevogenesis. Panel to the left depicts low innate Shh signaling activity in Cdk4::NRAS::Cdon+/+ resistant mouse strain (Cdon+/+), where NRASQ61K is the sole driver of nevogenesis. This leads to the formation of blue nevi. In Cdk4::NRAS::CdonNOD/+ mice with high innate Shh signaling, the Shh-Edn1 axis is activated by the CdonNOD variant. Upregulation and secretion of Edn1 from keratinocytes activates the Ednrb-Gnaq/Gna11 complex, which in turn augments the effect of NRASQ61K

112 on MAPK, resulting in the formation of melanocytoma (right panel). Targeting of the Cdon- Shh-Edn1 axis using Shh and endothelin pathway antagonists (LDE225 and Bosentan respectively) suppresses nevogenesis.

Suppl. Figure 1. Histopathological variation of melanocytic lesions with Cdk4::NRAS mutation. (a) Images of H&E staining of superficial and deep blue nevus in mouse and human. Nevocytes are comprised of pigmented dendritic cells having wedge shape architecture. (b) Images of H&E staining of mouse and human deep penetrating nevus. These lesions appear in deep dermis, superficial hypodermis, or adjacent to hair follicle bulb area. Pigmented balloon cell-like nevocytes sometimes appear in these lesions. Scale bars = 100 um. (c) Images of H&E staining of mouse non-bleached and bleached epithelioid blue nevus (left) and human epithelioid blue nevus (right). Scale bars = 400 μm. (d) Images of H&E staining of mouse non-bleached and bleached murine melanocytoma (left) and human melanocytoma (right). These lesions contain 3 cell types: dendritic cells, heavily pigmented balloon cells and epithelioid cells. Scale bars = 400 μm.

Supplementary Figure 2. CDON expression in non-lesional human and mouse skin IF staining of CDON in normal human (a) and mouse skin (b). CDON expression is mainly detected in the upper layer of epidermis and in the hair follicles. Scale bar = 100 μm.

Supplementary Figure 3. SHH activity in human dermal nevus and mouse melanocytoma (a) Three-Plex immunofluoresence with tyramide signal amplification (3Plex-TSA) of SOX10, CDON and GLI1 in human dermal nevus with congenital features. CDON is mainly expressed in keratinocytes of epidermis, hair follicle. CDON expression and GLI1 overlap follows similar pattern. Scale bars = 300 μm. (b) 3Plex-TSA of adult NOD mouse melanocytoma with Sox10-Cdon-Gli1. Murine lesion shows similar high Shh signalling. Scale bars = 100 μm.

Supplementary Figure 4. Alopecia is observed in high Shh signalling strains (a) H&E images of dorsal skin from wild type and Cdon heterozygous knockout littermates. Lower panels, H&E images of dorsal skin from Cdk4::NRAS+CC progeny mice on nevus- resistant and susceptible strain backgrounds respectively. Scale bars = 300 μm. (b)

113

Average count per strain of hair follicles per 1.5mm of dorsal skin (WT, n=4, susceptible strains, n=8, wt, resistant strains, n=8, Cdon+/-, n= 5). Statistical comparisons performed with a T-=test, ***=p<0.001, **=p<0.01, *=p<0.05).

114 a Chitsazan_Fig1

Human Human Superficial blue nevus Melanocytoma

1 10

2 9

3 Mouse 1-10 scores 8 for nevocyte density

4 7

5 6 Human Human Deep blue nevus Epithelioid blue nevus

Human Deep penetrating nevus

b c * 0 Mpixel

50 Mpixel

100 Mpixel

115

Chitsazan_Fig2 a b Dorsal skin - Cdon+/- Normal dorsal skin

H&E WT H&E Cdon+/- H&E Dermoscopy

c Tyr-NRAS::Cdk4

Dermoscopy Dermoscopy

Cdon+/+ CdonNOD/+

H&E H&E

Dermoscopy Dermoscopy

d e Human medium congenital nevus Human giant congenital nevus

116 a Sox10-CY5 Gli1-CY3.5 Gli2-CY3 Chitsazan_Fig3

Susceptible e 8 0 P10

2 ****

m 1

i 6 0

l ****

l l o ****

i 4 0

r s

P 2 0 o b Sox10-CY5 Gli1-CY3.5 Gli2-CY3 p

Resistant 0

s s s s s s n n n n P10 i in i in i i a a a a a a r r r r tr tr t t t t S S S S S S e t e t le t l n l n b n b a b a i ta ti t ti t t s p is p is p i e s e s e s c e c e c e s s s R u R u R u t 0 1 S l S S t u 0 1 1 2 l 1 P 2 P u d P P d A A

Sox10-CY5 Gli1-CY3.5 Gli2-CY3 c Susceptible f

8 0 Adult

2 ****

l m

6 0

G r

****

s n

l ****

i e

t 4 0

o t Sox10-CY5 r

Gli1-CY3.5 Gli2-CY3 s P

o 2 0

d p

Resistant Adult 0 s s s s s s n n n n n i in i i i i a a a a a a r r tr r t tr t tr t S S S S S S e t le t le t l n b n b n ib a i ta i ta t t t s t s p is p i p i e s e s e s c e c e c e s s s u R u R u R t S 0 S 1 S l t u 0 1 1 2 l 1 P 2 P u d P P d A A g SOX10-CY5 GLI1-CY3.5 Merge

GLI2-CY3 CDON-FITC

SOX10-CY5 GLI1-CY3.5 GLI2-CY3

h Control LDE225

d Cdk4::NRAS::

l 5 0 0

f Dorsal Dorsal skin

1 0 0

n 2

4 0 0

c o

SOX10 e 8 0

X e

c 3 0 0

r c

4 6 0 e

Control n

LDE225 u

r d

n **

–

Cdon

Sox10

x 2 0 0

c o

4 0 i

NOD

n + i **** u 1 0 0

3 2 0

/+ /+

n i

0 0 l 5 l o o 5 r 2 r 2 t 2 t 2 n E n E o D o D C L C L

117

Chitsazan_Fig4 a

4 0 Smo Transgenic mouse model Ptch1 KO mouse model Smo Transgenic mouse model 4 weeks post-induction 4 weeks post-induction 10 weeks post-induction

3 0 LogFC 2 0

1 0

n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. 0 a b 1 a b 1 a b 1 1 i1 i2 1 2 3 1 i1 i2 1 2 3 1 i1 i2 1 2 3 h l l n n n t5 t7 b h l l n n n t5 t7 b h l l n n n t5 t7 b tc G G d d d n n n tc G G d d d n n n tc G G d d d n n n E E E tn E E E tn E E E tn P W W P W W P W W C C C

Shh Edn Wnt Shh Edn Wnt Shh Edn Wnt Signaling Signaling Signaling Signaling Signaling Signaling Signaling Signaling Signaling

b 1 5 0 0 a WT Cdon+/- b c 6 0 Edn1-CY5 Edn1-CY5 Edn1-CY5 Edn1-CY5 ** DAPI DAPI 2 ****

1 DAPI DAPI

E e

1 0 0 0 i

n 4 0

v o

p 5 0 0

o e

o 2 0

o p

0 0 s s in in a a e /- r r p + t t ty n S S * o ld d t e i C n l W a b t ti is p s e e c R s u S f Control Bosentan

g h )

l 8

(

)

e e

6 l

e i

f 1 0

(

n **

t ****

t e

4 5

s 2

e t

* u

n s

* I 0 0 l l n o n o a r a tr t t t n n n n e o e o s C s C o o * * B B

118

Chitsazan_Fig5

Cdon+/+::Cdk4::Tyr-NRAS CdonNOD/+::Cdk4::Tyr-NRAS

LDE225

Bosentan

BLUE NEVUS GIANT NEVUS

119

Chitsazan_Supp Fig1

H&E – Human H&E - Human H&E - Superficial blue nevus - mouse a Superficial blue nevus Deep blue nevus

H&E - Deep blue nevus - mouse

H&E - Mouse H&E - Mouse H&E - Human b Deep Penetrating nevus Deep Penetrating nevus Deep Penetrating nevus

Bleached Bleached

H&E - Mouse H&E - Human c Epithelioid blue nevus Epithelioid blue nevus

Bleached Bleached

H&E - Mouse H&E - Human Melanocytoma Melanocytoma

Bleached Bleached Bleached

120

Chitsazan_SuppFig2 a

CDON DAPI

Cdon DAPI

121

Chitsazan_SuppFig3

a Human congenital nevus SOX10-CY5 CDON-FITC GLI1-CY3.5 DAPI DAPI DAPI

b Adult mouse melanocytoma Sox10-CY5 Cdon-FITC DAPI DAPI

Gli1-CY3.5 Merge DAPI

122

Chitsazan_SuppFig4

a WT Cdon+/-

NRAS-CDK4 NRAS-CDK4 Resistant Susceptible

b n.s. ** * *** ***

123

A QTL Study for the Development of Melanocytic Microsatellites

124

Chapter 4

4.1 Introduction

As described in chapter 2 and 3, by crossing many Collaborative cross (CC) strains with Tyr-NRAS::Cdk4 (NRAS-Cdk4) mouse model we observed tremendous diversity in the nevus density of superficial and deep dermal nevi. Another interesting diversity between CC strained crossed with (NRAS-Cdk4) was the number of melanocytic microsatellites (MSs) that appeared in the hypodermis layer of skin. Majority of these MSs appear near blood vessels (angiotropic features) (Wilmott 2012) (Figure 4.1a). MSs can sometimes be detected in the skin, associated with congenital nevi and melanoma. In congenital nevi and melanoma they represent groups of melanocytic cells away from the main lesion, often associated with regional lymph node metastasis (Day 1982, Harrist 1984) and are considered a poor prognosis factor. Presence of angiotropic MSs in stage III melanoma patients results in diminished 5-year survival rates (Barnhill 2002, Lugassy 2004, Van 2008, Lugassy 2011, Kokta 2013).

The majority of these patients have positive lymph node biopsies compared to other melanoma patients whose melanomas are negative for MS (Lugassy 1997, Lugassy 2014), supporting the existence of extravascular migratory metastasis (EVMM) or intravascular migratory metastasis (IVMM) culminating in lymph node positivity for melanoma cells (Figure 4.1b). It is unclear if MSs are only faithful to one of these two possible routes.

Wilmott et al. (2012) show that factors such as angiotropism, absence of regression, and Clark level V invasion, are strongly correlated with the presence of MSs in the skin. Hence, angiotropism in a sample predict the possibility of finding MSs and positive sentinel lymph node biopsies. Intriguingly, many other neural crest (NC) derived cancers e.g. glioblastoma multiforme and diffuse gliomas also exhibit angiotropic features (Lugassy 2005, Preusser 2007).

Virtually nothing is known about genes or mechanism conferring susceptibility to the development of MSs. To discover such genes, we conducted an unbiased genetic screen of MS counts in 45 CC strains. Our genetic analysis of MS count revealed a strong hit on murine chromosome 10, to an interval containing 22 candidate genes.

125

Allele segregation from 8 founders of CC strains revealed that the causal allele is inherited from NZO/HlLtJ (NZO) founder. Further, by using SNP cataloguing analysis we find mapped the region to an interval containing 11 genes. Our gene expression and protein expression studies suggest that Gja1 could be a strong candidate for further study of MS development.

4.2 Hypothesis

Germline genetic variation potentiates the development of melanocytic microsatellites in the hypodemirs of skin.

126

Figure 4.1 ǀ Microsatellite showing angiotropism. a, The image of H&E staining of mouse dorsal skin. A microsatellite with angiotropism (green arrows) appear 420 μm from a superficial blue nevus (left). Scale bar = 100 μm. b, Extravascular migratory metastasis (EVMM) model depicted in the diagram. EVMM is schematically depicted in the diagram by following steps: 1. tumor cells form an advancing front in the primary site; 2. tumor cells migrate near a vessel and cuff themselves to the vessel in order to from a colony (angiotropism); 3. tumor cells undergo epithelial mesenchymal transition (EMT) and obtain a pericytic phenotype; 4. pericytic cells migrate along vessels without intravasation; 5. formation of secondary colony in lymph nodes or distant site (Lugassy 2014).

127

4.3 Materials and Methods

Mice and phenotyping Experiments were undertaken with institute animal ethics approval (A98004M). Collaborative Cross mouse strains were provided by Grant Morahan (Harry Perkins Institute of Medical Research, Perth). By breeding Cdk4R24C/R24C::Tyr-NRASQ61K/+ males (FVB background) with females from 45 CC strains we generated progeny that are heterozygous for both the Cdk4 and NRAS mutations.

Genetic analyses

We tested 45 CC strains to discover quantitative trait loci (QTLs) that modify the density of microsatellite lesions in different mouse strains. The construction of the CC founder haplotypes is described in Ram et al. (2014). For mapping we used a logistic regression matrix model over the reconstructed haplotypes matrix to produce genome-wide distribution of P values (ANOVA chi squared). We used a false discovery rate of p=0.0001 to define significant genome wide linkage.

Gene expression study of genes differentially expressed during anagen and telogen

Dorsal skin from 20 CC strains collected during anagen (5 weeks after birth) and telogen (8 weeks after birth) in RNAlater™ stabilization solution. All mice were checked for the absence of melanocytic lesions with dermoscopy prior to collection of normal dorsal skin.

Assessment of RNA quantity and quality from frozen sections

The QIAGEN RNeasy® Micro Kit combined with our hands results in very good yield and quality of RNA. Quality control by direct analysis using Qsep100 analyzer is possible with quite large skin samples (often some 4 mm2 collected area from tissue sections of 8 μm thickness) and is limited to concentrations ≥5 ng/μl. Total RNA extracted from nevi and melanoma samples, with ribosomal RNA peaks ratio of ≥1.7 (28S:18S) are selected for downstream studies. cRNA Synthesis

128

We usually use 500 ng of total RNA for cRNA syntehsis using Illumina® TotalPrepTM RNA Amplification Kit. The yield and quality check is performed using Nanodrop Lite and Qsep100 analyzer respectively.

Analysis of cRNA size

The size distribution of cRNA is evaluated using Qsep 100 analyzer. The Qsep 100 can provide a fast and accurate size. The expected cRNA profile is a distribution of sizes from 250 to 5500 nucleotide (nt) with most of the cRNA at 1000 to 1500 nt.

Expression array analysis

RNA was subjected to one round of amplification/labeling with a MessageAmp kit (Ambion), then 1.5 μg labeled cRNA will be hybridized to MouseWG-6 v2.0 Expression BeadChips. The laboratory has used these systems extensively; labelling with 100 ng of total RNA gives high reproducibility (r2>0.99 across independent experiments). Data was extracted using GenomeStudio (Illumina) and imported into GeneSpring (Agilent) for normalisation, visualisation, and unsupervised hierarchical clustering. Analysis of variance (p-value ≤0.05 using Benjamini and Hochberg correction for multiple testing) was applied to identify genes whose expression differs between groups.

Haematoxylin and eosin staining protocol Slides from formalin-fixed paraffin-embedded blocks were dewaxed and stained with haematoxylin and eosin using Leica autostainer XL.

Multiplexed fluorescent immunohistochemistry with Tyramide signal amplification

Sections from paraffin-embedded lesions were dewaxed and treated with Dako low pH antigen retrieval solution at 1000C for 15 minutes. Endogenous peroxidase activity was 129 quenched in 3% H2O2, and sections blocked with 5% bovine serum albumin. Primary Sox10 (sc-17342), and Cx43 (ab11370) antibodies were applied, followed by the appropriate secondary antibodies. After washing, the signal was amplified using Perkin Elmer Tyramide signal amplification kit (FITC and CY5). Finally slides were washed, counterstained with DAPI, and coverslipped using DAKO mounting medium. The entire slides were scanned using the Vectra spectral imaging system (PerkinElmer) to accurately identify all regions. Images were scanned at 4X magnification. Multispectral images (MSI) were captured using Phenochart software and scanned from 420 nm to 720 nm using Vectra 3.1 automated quantitative pathology imaging system. Three MSIs per sample was analysed using InForm analysis software. Positive cells were segmented and counted per 20X magnification field.

130

Figure 4.2 ǀ MS usually appear in the vicinity of vessels. a, image of H&E staining of human congenital melanocytic nevus with multiple angiotropic MSs (asterisk). MSs invade deep into dermis and even hypodermis. b, image of H&E staining of dorsal skin from a susceptible NRAS-Cdk4 mouse at postnatal day 10 (P10). Only susceptible mouse strains develop MS by P10 in the hypodermis near the vessels. Blue arrows depict vessels. Scale bar = 100 μm. c, image of H&E staining of dorsal skin from a susceptible NRAS-Cdk4 mouse at postnatal day 60 (P60). Note that melanocytes cuff themselves around the vessel (V). Scale bar = 300 μm.

4.4 Results

Histopathology of murine MS

Cdk4R24C/R24C::Tyr-NRASQ61K/+ (NRAS-Cdk4) mice develop angiotropic MSs similar to human congenital nevi (Figure 4.2a) by postnatal day 10 (P10), from melanocytes escaping the confines of the hair follicles and by P60 they occupy majority of hypodermis (Figure 4.2b and 4.2c). The dorsal skin of NRAS-Cdk4 was immunostaied with a panel of melanocyte markers (Sox10 and Mitf), mast cell marker (Tryptase), and macrophage marker (F4/80). The majority of MSs are composed of Sox10 and Mitf positive cells. These melanocytes are interspersed with macrophages and mast cells which create a specific niche in the hypodermis of mouse with high MS counts (Figure 4.3).

NRAS-Cdk4 mice frequently have melanocytic cells in lymph nodes (Hacker et al, 2007), but almost never develop observable visceral metastasis. This may be partially because the animals must be euthanized when a melanoma reaches 100 mm in diameter, or because micrometastasis are not easily visible. It is possible that the melanocytic cells in lymph nodes could have migrated from MSs in the hypodermis.

Where possible we examined lymph nodes from mice with melanocytoma with high MS counts, and stage 3 malignant melanoma and found that metastasis is present in them as early as P60 (Figure 4.4a and 4.4b). Regional and distant lymph nodes are mainly populated by balloon like melanocytes in mice with melanocytoma (Figure 4.4 b lower left), while in mice with malignant melanoma these lymph nodes have high populations of melanophages (Figure 4.4b lower right). These colonies are histopathologically reminiscent of human pigmented balloon cell nevus (Figure 4.4c).

131

We also, investigated the development of nevi and MSs in a conditional transgenic mouse model of melanoma, Tyr::CreERT2;BrafCA;Ptenlox/lox (TBP). In this case the mutations are induced in adult mice. The mice develop nevi very quickly at day 10 post-mutation induction (D10), and stage 3 melanoma by D20. By D30 Melanomas that rapidly emanate from the pigmented nevi become non-pigmented (white melanoma), suggesting that they have switched phenotype to different differentation status, possibly to a Schwann cell-like phenotype (Handoko 2013) (Figure 4.4).

MS MS Sox10 Mitf

MS Tryptase MS F4/80

Ki67 Ki67 Figure 4.3MS ǀ MSs creates a special niche S4in the hypodermis. The immunoperoxidase staining of dorsal skin of NRAS-Cdk4 mice showsmelanoma that MSs are comprised of Sox10 and Mitf positive cells (melanocyte lineage markers top panels). The non-melanocytic cells that infiltrate the MSs are mainly composed of mast cells (Tryptase positive) and macrophages (F4/80 positive). Scale bars = 50 μm

Mice need to be euthanized by D35 due to size of tumors. However despite using a large cohort we were not able to capture the time points required to study the formation of dermal nevi and MSs. Understanding the spatio-temporal migratory patterns of melanocytes from epidermis to dermis was a crucial first step in the study in terms of trying 132 to validate mechanisms of nevus development using topical drug treatments. This is something that would be difficult or impossible to follow in humans since the formation of congenital nevi is only possible by studying samples obtained from spontaneous abortions or fetal deaths in utero, which would only facilitate a single time point (Cramer 2012) study of nevogenesis.

Figure 4.4 ǀ nevogenesis and melanomagenesis inTyr::CreERT2; BrafCA; Ptenlox/lox (TBP) mice. BN: Blue Nevus, MM: Malignant Melanoma, and WMM: White Malignant Melanoma. At Day 10 (D10) post induction mice develop BN and by D20 majority of the lesions progress to form MM. At D30 the enclosed area in blue is composed of BN, MM, and WMM. Scale bars: 0.5 cm.

Genetic background affects MS count

Male NRAS-Cdk4 were bred with females from many CC strains to generate progeny heterozygous for both the Cdk4 and NRAS mutant transgenes. Dorsal skin was collected from at least 3 mice of each CC cross, and the number of angiotropic MSs was scored for each sample from 1 (least dense/resistant) to 10 (most dense/susceptible). Linkage analysis was carried out based on the phenotypic score for hypodermal microsatellite 133 density in the dorsal skin of transgenic X CC strain progeny (Figure 4.6). We mapped a QTL to chromosome 10 which contains 20 genes (53-70 Mb) (Figure 4.6a). Analysis of the segregation of genomic segments from the 8 founder strains (129S1/SvImJ, A/J, C57BL/6J, CAST/EiJ, NOD/LtJ, NZO/HlLtJ, PWK/PhJ, and WSB/EiJ) through the linked interval showed that the NZO/HlLtJ (NZO) founder had contributed the causal allele. Allele segregation analysis and SNP cataloguing (Figure 4.7b and 4.7c) delimited the (-1- Log10(P) interval to a region (56-66 Mb) containing 11 genes (Dcbld1, Man1a, Gja1, Hsf2, Serinc1, Ascc1, Cdh23, Sgpl1, Hk1, Ccar1 and Tet1). Dcbld1 encodes a plasma membrane protein with unknown function which is strongly expressed in melanocytes, Man1a encodes Mannosidase alpha protein, Gja1 which encodes Cx43 the most ubiquitous connexin in mammals and previously suggested to be involved in melanoma progression (Villares 2009, Stoletov 2013, Sargen 2013, Zuker 2013). Hsf2 encodes a transcription factor for hsp70 which is considered to be a metastasis suppressor in prostate cancer (Björk 2016), Serinc1 encodes an enzyme that incorporates amino acid Serine into membrane (Inuzuka 2005), Ascc1 encodes activating signal co-integrator 1 complex subunit 1 which is known to be involved in development of esophageal adenocarcinoma (Orloff 2011), Cdh23 encodes cadherin-related 23 protein which is involved in breast cancer metastasis and development of familial and sporadic pituitary adenomas (Apostolopoulou 2012, Zhang 2017). Sgpl1 encodes Sphingosine-1-phosphate lyase 1 which acts as tumor suppressor in colon cancer (Oskouian 2006), Hk1 encodes Hexokinase-1which is an important enzyme in glucose metabolism and its upregulated in lymphatic metastasis and adverse prognosis in Chinese gastric cancer (Gao 2015), Ccar1 encodes a transcription factor which controls β-Catenin expression and controls the migration and invasion of gastric cells (Chang 2017). Finally, Tet1 encodes an epigenetic factor Ten-Eleven Translocation 1 which converts 5-methylcytosine (5-mC) to 5- hydroxymethylcytosine (5-hmC). It is usually absent in many cancers, and its expression induces the PTEN expression via demethylation of the PTEN promoter and impedes the invasion and metastasis of gastric cancer (Pei 2016).

134

60 days old 480 days old a Melanocytoma MM

b Lymph node Lymph node

c H&E – Human H&E - Mouse H&E - Human Melanoma metastasis lymph node balloon cell nevus

Bleached

Figure 4.5 ǀ Nodal metastasis appear early in mice with melanocytoma. a, Representative images of H&E staining of melanocyoma (left) and malignant melanoma (right) from susceptible NRAS-Cdk4 mice. b, H&E images of distal lymph nodes from 2 months old mouse with melanocytoma (left) and 21 months old mouse with malignant melanoma (right). High power image of H&E stained lymph nodes (lower right) show high number of melanophages (yellow arrows) and melanocytes (blue arrows) in mice with malignant melanoma (right). Scale bars = 300 μm. c, H&E images of human melanoma metastasis (right), bleached murine lymph node metastasis (middle) and human balloon cell nevus (left).

135

Gene expression analysis of candidate genes

NRAS-Cdk4 mice develop MSs during anagen phase of each hair cycle (Damsky 2015, Chitsazan 2016). Hence, significant up/down-regulation of the mapped genes on chromosome 10 could help us to narrow the number of candidate genes. 20 susceptible adult CC strain mice (n=20) euthanized during anagen phase of hair cycle and another cohort of 20 susceptible adult CC strain mice (n=20) were euthanized during telogen phase of hair cycle. Dorsal skin from all mice collected and gene expression study using RNA extracted from these samples revealed that only Dcbld1 and Gja1 expression are under control of hair cycle. The expression of Dcbld1 and Gja1 was upregulated 3 and 7 folds respectively during anagen vs. telogen phase. Hence, we picked Gja1 which is highly differentially expressed between anagen and telogen as the main candidate for further analysis. Correlation gene expression study revealed that Gja1 expression is positively linked to activation of Wnt signaling and pigmentation genes (Figure 4.7a).The strongest positively correlated genes are Fzd5 (r = 0.95), Tyr (r = 0.93) and Mlana (r = 0.90). Fzd5 encodes a receptor for Wnt ligand which is involved in metastasis in many cancers (Weeraratna 2002, Thiele 2011, Steinhart 2017). Tyr and Mlana are important melanocyte linage markers (Boyle 2002, Du 2003).

Gja1 expression is also positively correlated with other gap junctional subunit such as Gjb2 (r = 0.86) and Gjb6 (r = 0.78). Gjb2 and Gjb6 express Cx26 and Cx30, two important connexins in the skin that have been previously linked to melanomagenesis and metastasis (Saito-Katsuragi 2007, Riker 2008). Diversity in type of connexins could explain the intricate interactome of different cells in the skin during the anagen.Gja1 expression is also negatively correlated with expression of Bves (r = -0.71) and Hspb7 (r = -0.68). Bves encodes an important metastasis suppressor, blood vessel epicardial substance which is an integral protein which is involved in the formation of tight junctions. BVES loss is required for epithelial-mesenchymal transition (EMT) in human colorectal carcinoma (Williams 2011). HSPB is a also tumor suppressor which works in tandem with P53 (Lin 2014).

136

Figure 4.6 ǀ Genetic background dictates MS count of NRAS::Cdk4 mice. Representative images of H&E staining of MS at each score (1-10) across 45 CC strain backgrounds. Only MSs in the hypodermis have been counted. Scale bars = 500 μm.

137

In sum, Gja1 expression is positively correlated with expression of genes involved in metastasis and negatively correlated with metastasis inhibitor Bves and tumor suppressor Hspb7. Further analysis of the Cancer Genome Atlas (TCGA) data set also revealed that GJA1 is significantly differentially expressed between human metastatic melanoma (368 samples) and primary melanoma (103 samples). Other candidate genes in chromosome 10 region such as CCAR1, CDH23, MAN1A, SERINC1, and TET1 are also overexpressed in human metastatic melanoma vs. primary melanoma (Figure 4.8b).

In vivo study of Connexin 43 expression

To confirm that Gja1 is expressed in the mouse model, I performed multiplexed immunofluorescence staining with Tyramide-based amplification (IF-TSA). Dorsal skin from NRAS-Cdk4 mice was stained with Sox10 and Cx43 antibodies, to examine protein expression in baby and adult mice. Baby mice carrying the putative Gja1 causal allele showed high levels of Cx43 expression mainly in deep dermis and hypodermis where MSs are formed. Surprisingly, Cx43 is absent in epidermis and superficial dermis where dermal nevi are formed (Fig 4.9a). Dermal and hypodermal nevi do not express Cx43 and expression is limited to melanocytes residing in hair follicle bulb or bulge (Fig 4.9b and 4.9c).

Further, I stained 4 small and 4 medium human congenital nevi, and 4 human normal skin to study the expression pattern of CX43 in human melanocytic lesions. The CX43 expression in small human congenital nevi was similar to our mouse model. Only melanocytes residing in the hair bulb or bulge express CX43 (Figure 4.9c and 4.10a).

This is not the case in medium size human congenital nevi (Figure 4.10b). In these samples, CX43 expression is also detected in nevocytes. Superficial nevocytes do not express CX43 but deeper (more mature) nevocytes residing in the deep dermis express CX43. Also, deep nevocytes rarely show mislocalized expression of CX43 (Figure 4.7a) which is suggestive of functional gap junctions.

Melanocytes in normal skin do not express CX43 while keratinocytes express high levels of CX43 (Fig 4.10c). It is possible that change in CX43 expression in melanocytes be the result of nevus maturation (Pawlikowski 2013) since superficial and less mature nevocytes are CX43 negative.

138 a

139

c NZO Regulatory Linked to Hair cycle Genes changes metastasis dependent Dcbld1 1 NO YES

Man1a 1 NO NO Gja1 8 YES YES Hsf2 2 YES NO Serinc1 1 NO NO

Ascc1 2 NO NO Cdh23 1 YES NO sgpl1 1 NO NO Hk1 1 YES NO

Ccar1 1 YES NO

Tet1 1 YES NO

Figure 4.7 ǀ Mapping the modifier gene for development of MS. a, Genome-wide scan based on M density score in 45 CC strains. MS density score was treated as a multinomial analysis with values ranging from one to ten. The x-axis shows the chromosomal position, and the y-axis shows the 2log10(p) values; the p-values were derived from the linkage haplotype data. b, SNP cataloguing of the interval on murine chromosome 10 reveals only 11 genes that have NZO specific regulatory changes. c, List of candidate genes with NZO specific regulatory changes. Gja1, Hsf2, Cdh23, Hk1, Ccar1 and Tet1 have been previously linked to metastasis in literature. Gja1 has 8 regulatory SNPs in its promoter regions, Hsf2 and Ascc1 have 2 regulatory SNPs and the remaining genes each have only one regulatory SNP which could potentially change their expression. Analysis of gene expression of 20 CC strains revealed that only Dcbld1 and Gja1 expression is under influence of hair cycle.

140

4.5 Discussion

In this chapter we have conducted an unbiased genetic screen to map the modifier gene(s) for development of MSs. These melanocytic lesions that appear in the hypodermis layer of skin were previously described as dermal nevi (Viros 2014, Damsky 2015). We did not use this terminology to describe these lesions since they appear in the hypodermis. Also, we have categorized dermal lesions that appear in the deep dermis and superficial hypodermis as deep penetrating nevus (chapter 3). In this study we have only counted the number of MSs that are present in deep hypodermis and show angiotropic features. Based on our QTL study, allele segregation, SNP cataloguing, gene expression and gene correlation studies during anagen and telogen phases of hair cycle we picked Gja1 (encodes Cx43) as our candidate for further study. High expression of CX43 correlates to poor prognosis in many cancers (Sargen 2013, Stoletov 2013, Brockmeyer 2014, Poyet 2015, Tanaka 2016).

The high MS count CC strains are pigmented and majority of these cells have balloon cell structure both in the hypodermis and lymph nodes while albino CC strains develop lower number of MSs in the hypodermis and do not become balloon like, hence they can only be monitored using Sox10 or other melanocyte markers using IHC.

Each MS in high (pigmented) and low (albino) MS count mouse strains is comprised of a population of melanocytes, mast cells, and macrophages. Others have also shown that chronic UV exposure exacerbates the formation and number of MSs and angiotropic MSs (Viros 2014, Bald 2014). This positive effect in the formation of MSs has been attributed to the release of high mobility group box 1 (HMGB1) from epidermal keratinocytes exposed to damaging effects of UV (Bald 2014).

141

a c Genes Gja1

Fzd5 Dsg4 Elf5 Lap3 Ggt1 Stra6 S100a7a Golga7b Slc7a8

Gprc5d Slc7a5

Ppp1r3b Foxe1 Foxn1 Gjb2 Dct Tyrp1 Wif1 Shh Wnt5a Gjb6 b Kit

Neo1 Hspb7 Edn1 Bves Tgfa Ppp2r3a Kif7 Scin Sema6a Fgf11 Vdr Dusp13 Gli3 Ahr Sox13 Wdr41 Dusp6 Setd3 Ptch1

Figure 4.8 ǀ Gja1 expression is highly correlated with known metastatic genes. List of genes which their expression is positively (a) and negatively (b) correlated with Gja1 expression (r>0.65) in dorsal skin of 20 susceptible CC strains. Genes are sorted based on Gja1 expression using gene expression correlation matrix (top to bottom). Genes involved in metastasis are highlighted in green, melanocyte genes highlighted in yellow, Shh pathway genes highlighted in pink, and Wnt signaling genes highlighted in orange. View appendix 1 for complete list of genes highly correlated with Gja1 expression. c, Analysis of TCGA data set revealed that 6 out of 10 genes which mapped to murine 142 chromosome 10 are also differentially expressed in highly motile human metastatic melanomas (368 samples) versus primary melanomas (103 samples). GJA1 is significantly upregulated in primary melanoma VS metastatic melanoma while CCAR1, CDH23, MAN1A1, SERNC1 and TET1 are significantly upregulated in the metastatic melanoma VS primary melanoma. See appendix 2 for list of genes highly correlated with candidate genes from TCGA data set.

The expression of Cx43 in the melanocytes that are based in the hair follicle (Figure 4.8 and 4.9) is probably necessary for their cell-cell communications with other keratinocytes and hair follicle stem cells (HFSCs) that reside in the hair bulb. Intriguingly the expression of Gja1 in susceptible strains is positively correlated with many genes that have been previously linked to metastasis and negatively correlated with genes that block tumor progression and metastasis (Weeraratna 2002, Thiele 2011, Sargen 2013, Stoletov 2013, Brockmeyer 2014, Poyet 2015, Tanaka 2016, Steinhart 2017).

Cell-cell communications occur largely at gap junctions, where plasma membrane of neighbouring cells form 2-3 nm gaps that are bridged by channel-forming proteins called connexins. There are 21 human connexion genes. Connexins can also form hemi- channels which enable the cell to communicate with extracellular matrix (ECM) and microenvironment. They enable cell-cell, cell-ECM, and cell-microenvironment communications by allowing exchange of myriad of signals e.g. secondary messengers, electrical signals, microRNAs and small metabolites (Goodenough 2009, Naus 2010). Most cells express number of different connexins. Connexins can form hexameric assemblies (connexons) of homomeric or hetermeric connexins (Goodenough 1996, Falk 1997, Segretain 2004). Connexons can form gap junctions with identical (homotypic) or different (heterotypic) connexins (Dedek 2006).

Despite the limitation of assembly combinations, there are a large number of possible gap junction combinations since they can be composed of 21 family members (White 1994). CX43 is mainly expressed in hair follicles and nevocytes residing in deep dermis (mature nevocytes). Melanocytes residing in the hair follicle also express Cx43 both in NRAS-Cdk4 mice and in human hair follicles near the nevus. MSs in baby mice also show weak Cx43 expression.

143

Figure 4.9 ǀ Connexin 43 is expressed in the murine deep dermis and hypodermis. a, Immunofluoresence with tyramide signal amplification (IF-TSA) staining of dorsal skin from susceptible baby mice at P10 with Cx43 and Sox10. Cx43 expression is absent in the epidermis and superficial dermis (dotted lines depict this area). The majority of the deep dermis and hypodermis express high levels of Cx43. Higher power image of this staining with bright field presentation (b) shows the formation of an early MS (enclosed with dotted lines). Melanocytes with Cx43 expression are depicted with red arrow head and Cx43 negative melanocytes are depicted with black arrowhead. c, IF-TSA staining of human and

144 murine hair follicles in anagen (growth phase of hair cycle) shows similar pattern of expression of connexin 43 with most of expression present in the hair bulb. Note that dermal papilla (DP) in human hair follicle is weakly CX43 positive while mouse DP is Cx43 negative. Melanocytes residing in the hair bulb are connexin 43 positive both in mouse and human (dotted arrows) while escaped melanocytes in the dermis do not express connexin 43 (white arrow). Scale bars = 100 μm.

Although many studies support the notion that connexins are tumour suppressors, recent evidence suggests that, in some tumour types, they may facilitate specific stages of tumour progression. Sargen et al. studied the expression of CX43 in 34 nevi and melanoma samples. CX43 immunoreactivity was much higher in the malignant melanoma samples (Sargen 2013). Both melanoma and breast cancer cells can metastasize to brain. The homing mechanism appeared to be controlled by upregulation of Cx43 via Twist (Stoletov 2013). Knocking down Cx43 or blocking intercellular communication using carbenoxolone inhibits tumor cell extravasation, and brain colonization. Primary melanoma and breast cancer tumors in patients also verified the correlation between CX43 upregulation and increased recurrence and metastasis (Stoletov 2013).

The Cx43 expression pattern in NRAS-Cdk4 mice with high microsatellite counts is concordant with the location of microsatellites in the deep dermis and hypodermis with most of Cx43 being expressed in hair follicles. A similar pattern is evident in human dermal nevi and considering the positivity of proximal and distal lymph nodes in strains with high MS counts, we set out to study the gene expression of all candidate genes (on the murine chr10 mapped area) in highly migratory melanoma cells. CCAR1, CDH23, MAN1A, SERINC1 and TET1 are upregulated in human metastatic melanoma while GJA1 has higher expression in primary melanoma.

Melanoblasts and melanocytes are highly migratory during the early development and it is possible that two types of melanocytic lesions that appear in superficial dermis and deep dermis (MSs) have different molecular machinery which would confer different migratory potential to them. We have used laser capture microdissection (LCM) to capture MSs and superficial dermal nevi and isolated high quality RNA (RIN>7) from them which will be discussed in chapter 5.

145

Cells with high CX43 expression are known to secrete factors that are positive regulators of metastasis e.g. IL-6, IL-10, MCP-1 and TGFB (McLachlan 2006 Linnskog 2014 Dai 2009 Yoshimura 2013 Terai 2012). Understanding how these early MSs form is a first step towards understanding how their development could be clinically monitored in the future. This study will help to better understand the behaviour of melanocytes that become mobile and form MSs in the hypodermis.

146

Figure 4.10 ǀ CX43 expression in human melanocytic lesions. a and b IF-TSA staining of human small (a) and medium congenital nevi (b) with CX43 primary antibody. Small congenital nevocytes do not express CX43 while the adjacent keratinocytes are positive for CX43. Medium congenital nevi show more complex pattern of CX43 expression in both epidermis adjacent to the lesion and in less mature nevocytes of superficial dermis and mature nevocytes of deep dermis. CX43 expression in keratinocytes has an intermittent pattern compared to normal skin and small congenital nevi. Mature nevocytes (lower right) of medium congenital nevi show high expression of CX43 while less mature nevocytes (lower left) do not express CX43. Scale bars = 800 μm. c, IF-TSA staining of human normal skin (n=4) with CX43 and SOX10 primary antibodies. Keratinocytes in epidermis and hair follicle show strong positivity for CX43 while melanocytes (red arrows) do not express CX43.Scale bars = 300 μm.

147

Detection of Transformative Changes Driving Melanomagenesis

148

Chapter 5

5.1 Introduction

This chapter focuses on detecting gene expression changes during progression from nevus to dermal melanoma (melanocytoma) and malignant melanoma (MM) in the Cdk4R24C/R24C::Tyr-NRASQ61K (NRAS-Cdk4) mouse model (Ferguson 2010, Wurm 2012). They develop MM characterized by downward invasion into the dermis and hypodermis, cellular atypia, and other histological signs of malignancy. These mice carry an NRASQ61K mutation common in malignant melanoma, and a germline Cdk4R24C mutation, mimicking families who carry such mutations, or alterations in the gene encoding its binding partner, p16 (Zuo 1996). In genome-wide association study (GWAS) of melanoma susceptibility one of the strongest associated regions is at MTAP/CDKN2A (Law 2012). The best associated SNP maps near CDKN2A, in the 5’ UTR of the adjacent gene MTAP (Falchi 2009).

It is speculated that MTAP/CDKN2A-associated SNPs mark regulatory elements that influence the expression of CDKN2A. Further, in terms of gene expression signature our murine MMs are not dissimilar to many human lesions. Previous comparison of gene expression in NRAS-Cdk4 MMs with human MM revealed a significant overlapping gene signature with a group of well-differentiated human MMs that express MITF and melanocyte antigens (Sensi 2011). In earlier chapters we studied the non-cell autonomous effect of microenvironment (mainly keratinocytes) on development of melanocytoma (chapter 2 and 3), and melanocytic microsatellites that reside in the hypodermis of skin (chapter 4).

In this chapter we report our study on cell-autonomous changes that drive the transformation and progression of melanocytic lesions from stage 1 (S1/early melanocytoma) to stage 3 (S3) MM (Ferguson 2010; Wurm 2012) in the NRAS-Cdk4 mouse model. Considering the possibly skewing effect of RNA from keratinocyte populations in S1, we decided to first assess gene expression of normal skin vs. stage 3 to obtain the keratinocyte and MM-specific gene expression signatures. On using genome- wide gene expression arrays using whole tissue we found significant differences in surprisingly few genes between the respective murine MM stages. One of these genes was Dusp6, a negative regulator of MAPK pathway (Figure 5.1). Role of Dusp6 is controversial. While many studies support a tumor suppressor role for Dusp6 (Furukawa

149

2003, Xu 2005, Furukawa 2005, Okudela 2009, wong 2011), others suggest that it can act as an oncogene (Nunes-Xavier 2010, Messina 2011, Degl’Innocenti 2013, Shojaee 2015).

In our gene expression study the expression of Dusp6 increases with transformation and progression of MM. It was interesting for us to pursuit this further on protein level using mouse and human nevi and melanoma. To help put Dusp6 expression in some kind of biological context, I have correlated its expression with that of other proteins marking various aspects of melanoma cell behaviour and/or differentiation status. I have also studied the protein expression of ETS1 a transcription factor for DUSP6 (Ekerot 2008, Zhang 2010) and BRN2 a marker for melanoma invasiveness (Boyle 2011, Fane 2017, Simmons 2017) using immunohistochemistry (IHC) and IF with Tyramide signal amplification (IF-TSA).

Figure 5.1 ǀ Regulation of MAPK pathway via ERK dephosphorylation. Activated p- ERK can be inactivated by dephosphorylation via cytoplasmic DUSP6 or nuclear DUSP5 (from Caunt 2012).

5.2 Hypothesis

150

That during the transformation of benign moles to melanocytoma, and malignant melanoma there is a gradual accumulation of molecular changes that individually are insufficient, but when combined are responsible for transformation to melanoma and progression to more aggressive behaviour.

5.3 Methods

Mouse melanoma model

The Cdk4R24C/R24C::Tyr-NRASQ61K mouse model has been previously described (Ferguson 2010; Wurm 2012). Tyr-NRAS mice carry a melanocyte-specific activating mutation (Q61K) in human NRAS cDNA. They occasionally develop MM at an advanced age, but they do so earlier when crossed onto the Cdk4R24C/R24C background. The average age of onset is 150 days. In our hands, Tyr-NRASQ61K mice on an FVB background rarely develop nevi or MM until crossed on to the mutant Cdk4R24C background. Most studies on murine MM are conducted using pigmented backgrounds (e.g. BL6). Use of such strains with the various ‘onco-transgenic’ melanoma models (e.g. carrying oncogenic Nras, Braf, Hgf, Ret, Grm1) results in extreme hyperpigmentation, particularly within the lesions, that obscures the histopathology. Bleaching to remove excess melanin has rarely been reported, although it is simple to do. Unfortunately, bleaching destroys many antigens, hence normally prevents immunohistochemistry staining of the lesions. Thus the Cdk4R24C/R24C::Tyr-NRASQ61K were bred onto an albino FVB background.

Histopathological classification of melanocytic lesions

Benign nevus classification has provided in length in chapter 3. Stage 1 (S1) is reminiscent of thin melanocytoma in humans. Although the vast majority of these lesions never progress to melanoma, some do, and thus they can be thought of as a transitional stage between benign nevus early melanoma. They are made up of dense proliferations of nevoid cells residing in the superficial dermis and extend to mid dermis. These cells are somehow polymorphic and have low mitotic rate and few apoptotic cells.

Stage 2 (S2) or late melanocytoma is reminiscent of thick human melnocytoma. S2 melanocytoma is comprised of less densely-packed sheets of cells with atypia, still nevoid- 151 like in appearance, polymorphic, and with higher mitotic rate compared to S1. They tend to populate the whole dermis.

Stage 3 (S3) is considered MM. They occupy the majority of hypodermis and reach the muscle layer or panniculus carnosus (fascia in humans). S3 MM cells are no longer nevoid in appearance and show higher mitotic rates than S2. Some clusters of nevocytes can be only detected at the edges of dermal-hypodermal boundary.

NRAS-Cdk4 mice also develop hypodermal malignant melanoma metastasis which are reminiscent of human subcutaneous metastasis and are not in continuum with previous stages. These lesions are comprised of a well-circumscribed nodule which is only located in the hypodermis which is covered by normal nevocytes in deep and superficial dermis. These cells are highly polymorphic, show high mitotic rates and are engulfed by high number of blood vessels (Wurm 2012). All experiments were carried out with institute animal ethics approval. A98004M. Mice were euthanized before tumors exceeded 10 mm in diameter.

Non contact laser capture microdissection

With PALM MicroBeam a verity of biological material can be microdissected and lifted directly from glass slides using the “AutoLPC” function of PALM RoboSoftware.

Preparation of slides

Samples on nuclease free Membrane Slide: Membrane slides NF are glass slides covered with a membrane on one side which is certiﬁed to be free of DNase, RNase and human DNA. These slides are covered with polyethylene naphthalate (PEN)-membrane. This PEN-membrane is highly absorptive in the UV-A range, which facilitates laser cutting. This membrane is easily cut together with the sample and acts as a stabilizing backbone during lifting. Therefore even large areas can be lifted by a single laser pulse without affecting the morphological integrity (Zhang 2010).

152

Sample collection

An adhesive Cap is used as a collection device for all RNA experiments. Adhesive Caps are produced with cleanroom technology and can therefore be considered as RNase-free. The intention of these Adhesive Caps is to allow LCM without applying any capturing liquid into the caps prior to LCM. This minimizes possible RNase activities. Beside the quick relocation of the lifted samples in the cap due to instant immobilization means that there is no risk of evaporation and crystal formation during extended specimen harvesting.

Formalin Fixed Parafﬁn Embedded (FFPE) and PAXgene Fixed Paraffin Embedded (PFPE) Sections

Sections are mounted onto Membrane Slides the same way as routinely done using glass slides. After mounting slides are dried overnight in a drying oven at 56°C. Parafﬁn will reduce the efﬁciency of the laser, sometimes strongly inhibiting cutting and lifting. Deparaffination is carried out using standard protocols for normal glass slides. Most standard staining procedures can be used for FFPE and PFPE sections. We usually perform Hematoxylin or Cresyl Violet staining. Skipping the Eosin staining of the HE- procedure helps additionally against RNA degradation.

Flash frozen sections

Sections are mounted onto Membrane Slides similarly as when using glass slides. As first step the dehydration in ice-cold 70% ethanol for 2-3 minutes is performed in order to reduce RNase activity. It is important to remove tissue freezing medium before laser cutting because such media will interfere with laser efficiency. Removing this medium is done by dipping the slide 5-6 times into ice-cold RNase-free water. Using frozen sections endogenous RNases may still be active after the short fixation step. Therefore we keep all incubation steps as short as possible and use ice-cold RNase free water and solutions for all steps. We usually perform Hematoxylin or Cresyl Violet staining. Skipping the Eosin staining of the HE-procedure helps additionally against RNA degradation.

Assessment of RNA quantity and quality from FFPE and PFPE sections

153

We apply the QIAGEN RNeasy® FFPE Kit and PAXgene® Tissue RNA Kit with some LCM-specific modifications. These procedures are very effective and allow a high final concentration of RNA due to a small elution volume (12 μl). Genomic DNA contamination is minimized by a DNase I digest on the purification column. RNA from FFPE and PFPE material will frequently not show clear profiles and only rough estimations of the size distribution and quality of the RNA are possible.

Assessment of RNA quantity and quality from frozen sections

The QIAGEN RNeasy® Micro Kit combined with Adhesive Cap in our hands results in very good yield and quality of RNA. Quality control by direct analysis using Qsep100 analyzer is possible with quite large microdissected samples (often some 4 mm2 collected area from tissue sections of 8 μm thickness) and is limited to concentrations ≥5 ng/μl. Total RNA extracted from nevi and melanoma samples, with ribosomal RNA peaks ratio of ≥1.7 (28S:18S) are selected for downstream studies.

cRNA Synthesis

Analysis of cRNA size

Expression array analysis

RNA was subjected to one round of amplification/labeling with a MessageAmp kit (Ambion), then 1.5 μg labeled cRNA will be hybridized to MouseWG-6 v2.0 Expression BeadChips. Labelling with 100 ng of total RNA gives high reproducibility (r2>0.99 across independent experiments). Data was extracted using GenomeStudio (Illumina) and

154 imported into GeneSpring (Agilent) for normalisation, visualisation, and unsupervised hierarchical clustering. Analysis of variance (p-value ≤0.05 using Benjamini and Hochberg correction for multiple testing) was applied to identify genes whose expression differs between groups.

Immunohistochemistry and multiplexed fluorescent immunohistochemistry with Tyramide signal amplification

Sections from paraffin-embedded lesions will be dewaxed, and treated with Dako low pH antigen retrieval solution at 1000C for 15 minutes. Endogenous peroxidase activity will be quenched in 3% H2O2, and sections blocked with 5% bovine serum albumin. Primary SOX10 (sc-17342), BRN2 (ab151796), ETS1 (sc-55581) and DUSP6 (sc-134837) antibodies will be applied, followed by appropriate secondary antibodies. After washing, the signal was amplified using FITC, CY3, CY3.5 and CY5. Finally slides were washed, counterstained with DAPI and coverslipped using DAKO mounting medium. Entire slides scanned using Vectra spectral imaging system (PerkinElmer) to accurately identify all regions at 4X magnification. Multispectral image (MSI) captured using phenochart software and imaged from 420 nm to 720 nm using Vectra 3.1 Automated quantitative pathology imaging system. Three MSI per sample was analysed using InForm analysis software for a fluorescent 4-plex IHC panel of SOX10, BRN2, ETS1 and DUSP6. Positive cells were segmented and counted per 20 magnification field.

155

a H&E H&E Normal skin Malignant melanoma Stage 3

900 μm 100 μm

b c

RNA RNA

Gene expression Gene expression study study

Keratinocyte S3 Melanoma gene expression gene expression signature signature

Figure 5.2 ǀ Whole-genome gene expression study of stage 3 MM vs. normal skin. a, H&E staining of murine normal tail skin and stage 3 (S3) MM that were used for gene expression study of normal skin versus MM using Cdk4::NRAS mouse model. b and c, gene expression of normal skin versus S3 MM using Illumina® Mouse WG-6 v2.0 Expression BeadChips. Selective list of genes expressed differentially in S3 MM versus normal skin. Metrn, Vim, and Sox10 are highly expressed in S3 MM while Krt84, Expi, and Krt15 are not expressed in S3 MM.

156

5.4 Results

Whole-genome gene expression study of murine melanoma progression

As mentioned previously the use of the NRAS-Cdk4 mouse model allows access to melanocytic lesions over the spectrum of stages of progression from benign nevus to MM (Ferguson 2010). First, since the samples we used were not microdissected (only trimmed), meaning that there could be keratinocyte contamination in the early lesions, we set out to clarify exactly which genes are likely to be differentially expressed in the mouse skin keratinocytes and melanocytic lesions. Figure 5.2a shows the H&E staining of normal tail skin (n=3, left panel) versus the S3 MM (n=3, right panel) that was used in this study. About 3,000 genes were significantly differentially expressed between skin and melanoma (p<0.05). A selection of genes expressed higher in normal skin vs. MM are shown in figure 5.2b (Krt84, Expi, Krt15, Sox21, Fgfr2, Krt33a, Col17a1, Irf6, and Sox7) (appendix 3). All of the keratinocyte genes can be disregarded from future analyses in this chapter. Figure 5.1c shows a selection of genes expressed higher in the MM vs. normal skin (Metrn, Vim, Sox10, Mc1r, Dusp4, Dkk3, Dusp6, Cd44, Dct, S100a1, Tgfbr2, Ednrb, Ets1, Gna11, Mapk9, Tgfb1, and Tgfbr1).

Functional enrichments of gene expression of genes highly expressed in S3 MM vs. normal skin reveals that these genes are involved in extracellular exosome and vesicle formation, amino acid and sugar metabolism, and development of pancreatic cancer and regulation of pigmentation (while genes that have been downregulated in S3 melanoma vs. normal skin are involved in single organism process, fatty acid degradation, and peroxisome formation (https://string-db.org).

This allows us to infer which candidates genes would be likely to be expressed in the melanocyte/melanoma cells with highest population in malignant melanoma versus normal skin tissue mostly comprised of keratinocytes.

157

Stage 1 Stage 2 Stage 3

Trp1 Trp1 Trp1

Trp1

b 10 9 8 7

6 5 4 LogFC 3 2

1 0

c 12

8 6

LogFC 4

Figure 5.3 ǀ Whole-genome gene expression study of melanoma progression. a, Immunoperoxidase staining of stage 1 (S1), stage 2 (S2), and stage 3 (S3) MM with the melanocyte marker Trp1. Arrows point to the advancement of tumor edge in dermis, hypodermis and muscle. Scale bars = 300 μm. b and c, Selective list of genes that are highly expressed in S3 (b) and S1 melanoma (c).Gene expression study of S1 versus S2 and S3 determined using Illumina® MouseWG-6 v2.0 expression Bead Chips. Keratinocyte signature (KCS) was deducted from all stages prior to analysis. 158

Next we used genes expression comparison between S1 early melanocytoma, S2 late melanocytoma, and S3 MM and deducted the keratinocyte specific (KCS) gene expression signature in order to negate the masking effect of keratinocyte RNA. In Figure 5.3a, Trp1, a melanocyte marker has been used to highlight the melanocytic cells, to better visualise the progression of melanoma. Unsupervised hierarchical clustering revealed 300 genes significantly differentially expressed (p<0.05) in S3 MM (n=4) versus S1 early melanocytoma (n=3), and S2 late melanocytoma (n=3). Figure 5.2b shows a selective list of genes highly expressed in S3 MM (Car2, Qpct, Sema6d, Ednrb, Dusp6, H2-Ab1, Vcam1, Usp18, Dusp23, Hist1h2be, Hist1h2bf and Hist1h2bh). A selective list of genes highly expressed in S1 vs. S3 melanoma has been presented in Figure 5.2c (Krt16, Chi3l1, Stfa1, Lamb3, S100a9, Junb, Rab25, Anxa11, Lypd3, Mmp9, Gjb6 (which are all keratinocyte genes except Anxa11 and Mmp9) (appendix 4). Genes that are highly expressed in S3 melanoma after deduction of KCS are involved in nucleosome structure, developmental pigmentation, and immune responses (Figure 5.3c) while genes that are highly expressed in S1 after deduction of KCS are involved in skin and epidermal cell differentiation, and membrane bound vesicle formation (https://string-db.org). Despite our effort to nullify the masking effect of keratinocyte, the majority of genes differentially expressed between S1 and S2 were keratinocyte genes. One way to avoid much of the influence of stromal contamination in tumours is to use laser capture microdissection. Hence, we moved on to collect melanocytes, nevocytes, early and late melanocytoma, and MM cells from normal skin, nevi, early and late melanocytomas, and MM samples using laser capture microdissection.

Laser capture microdissection

Gene expression studies performed using nevus and melanocytoma were highly masked with KCS. In order to get a better understanding of cell autonomous genes whose expression changes during MM progression we devised a new gene expression study using laser capture microdissected samples (Figure 5.4). Laser capture microdissection was performed using formalin fixed paraffin embedded (FFPE), Paxgene-fixed paraffin- embedded (PFPE) and flash frozen samples. RNA yield and RNA purity was higher in flash frozen samples. RNA extracted from laser-captured materials from FFPE and PFPE did not show 18s and 28s peaks. Only RNAs extracted from flash frozen samples revealed 159 high RNA integrity number (RIN>7, Figure 5.5). So very pure high quality RNA was isolated from lesions at different stages of progression, but we were unable to continue this part of project since Illumina ceased the manufacturing of murine chips. All high quality RNAs have been stored in -80 freezer.

LCM. using samples 5.4 ǀ Figure

Whole

- genome gene ofexpression study

microdissected nevi and microdissectedand nevi melanoma

160

Figure 5.5 ǀ Qsep 100 analysis of total RNA extracted from flash frozen blocks (a), RNA isolated post-LCM (b), and RNA ladder (c) versus negative control (d).

In vivo validation of Dusp6 expression changes during melanocytic lesion progression

Dusp6 was the only gene consistently up-regulated during transformation and progression of melanoma (Figure 5.2 and 5.3), therefore we selected Dusp6 (an inhibitor of MAPK pathway) as the best candidate for further study. The expression level of Dusp6 increases with progression which is intriguing considering its known effect as a negative regulator of MAPK pathway. It was interesting for us to study this finding further and validate the result on protein level. In order to validate the gene expression result on protein level, we stained murine nevi and melanoma with Dusp6 antibody and quantitated its expression using Aperio software. Dusp6 is highly expressed in melanoma compared to nevus confirming its overexpression in malignant stage (Figure 5.6a and b). Further, we found that Dusp6 expression is up-regulated with progression and formation of cutaneous metastasis. The highest expression was observed in cutaneous metastasis (Figure 5.6c and d). This result confirms our gene expression array results (figure 5.2 and 5.3) suggesting that Dusp6 expression is correlated with melanoma progression and metastasis.

161

DUSP6, ETS1, and BRN2, expression in human nevi and thin melanoma

To confirm the role DUSP6 in melanoma progression in human, we used multiplexed immunofluorescence staining with Tyramide-based amplification (IF-TSA) and quantitated the protein expression using InForm software 2.2. To help put changes in DUSP6 staining in the context of the behaviour of the melanocytic cells, we used

a Dusp6 Analysis NC NC b

**** * S1 S1

c S2 S2 Slow melanoma NC Fast melanoma

Dusp6

S3 S3

Analysis c Slow melanoma NC Fast melanoma d

**** Dusp6

Analysis

Figure 5.6 ǀ Dusp6 is up-regulated with melanoma progression. a, Immunoperoxidase staining of Dusp6 d expression of S3 melanoma and its adjacent nevus of NRAS-Cdk4 transgenic mouse on FVB background.**** b, Aperio analysis of intensity of strong positive cells in murine congenital nevi (n=4) versus S3 melanoma (n=4). c and d, Immunoperoxidase staining and quantitative protein expression analysis of Dusp6 in S1

162

(n=4), S2 (n=3) and S3 (n=3) using Aperio software. This result confirms our gene expression array results (Figure 5.2 and 5.3) suggesting that Dusp6 expression is correlated with melanoma progression and metastasis. Analysis was performed on 20X magnification using Aperio IHC nuclear staining with Sox10 (500 cells counted per field) in S1, S2, S3, fast and slow samples. Dusp6 expression is upregulated with progression of melanoma (p=value calculated using the two-tailed T test (* = p<0.01, ** = p<0.001 and ****= p<0.0001). Scale bars = 300 μm.

ETS1 a transcription factor for DUSP6 and SOX10, and BRN2 a marker of invasiveness. Next, we analysed the expression of ETS1 in our panel of human melanocytic lesions and normal skin. Ets1 is an activator of Sox10 enhancer in mouse and chick and it loss or misssense mutations result in development of white belly patches in pigmented mice (Saldana-Caboverde 2015). It is known that Sox10 is crucial for maintance of congenital nevi and melanoma (Shakhova 2012) so it was intriguing to compare the protein expression of its upstream activator in human normal skin (n=4) to congenital nevi (n=5), benign nevi (n=5) and superficial spreading melanoma (n=8). We also stained t our panel with BRN2 which has been linked to an invasive phenotype in population of melanoma cells (Boyle 2013, Fane 2017, Simmons 2017). DUSP6 expression in human congenital nevi (n=5) is very heterogenous. There are populations of nevocytes that moderately express DUSP6 while other nevocytes are negative for DUSP6 expression. Benign nevi (n=5) express high levels of DUSP6 in almost all neovcytes. Both congenital nevi and benign nevi highly express SOX10 and ETS1 but only congenital nevi express BRN2 (Figure 5.7 and 5.9). Superficial spreading melanomas (SSMs) highly express DUSP6 which is homogenous through the lesion both in melanoma cells and in keratinocytes. Infiltrates express lower level of DUSP6 while expressing high levels of ETS1. Similar to benign nevi only few melanoma cells are BRN2 positive (Figure 5.8). Normal human skin epidermis is barely positive for DUSP6 and normal melanocytes with SOX10 positivity do not express ETS1 or BRN2 (Figure 5.7c). Much to our surprise, the human congenital nevi (n=5) have the highest expression of BRN2 vs. benign nevi (n=5) and SSMs (n=8) (Figure 5.9). Majority of BRN2 positive nevocytes also do not express DUSP6 which based on our murine and human study is a marker for transformation and progression. These BRN2 positive nevocytes are also ETS1 positive and entirely SOX10 positive (Figure 5.7a lower right panel). While all nevi (congenital and benign) express ETS1, SSMs do not express this transcription factor but in the infiltrates (Figure 5.8). 163

Figure 5.7 ǀ Immunofluoresence with tyramide signal amplification (IF-TSA) of human congenital nevi (a, n=5) and benign nevus (b, n=5) with a panel of DUSP6, SOX10, ETS1 and BRN2 antibodies. (a) Lower right panel shows the expression of a band of nevocytes near a hair follicle. Majority of nevocytes express ETS1 and BRN2 while only few express DUSP6. Note that hair follicles and adjacent epidermis do not express DUSP6, ETS1, and BRN2. (b) Lower right panel shows the expression of nevocytes near a hair follicle. All nevocytes express DUSP6 and ETS1 while only few are BRN2 positive. Scale bars = 500

164

μm. c, IF-TSA of human normal skin with a panel of DUSP6, SOX10, ETS1 and BRN2 antibodies. SOX10 positive melanocytes do not express DUSP6, ETS1 or BRN2. Scale bars = 100 μm.

Further the analysis of gene expression of DUSP6 and ETS1 in TCGA dataset (368 metastasis vs. 103 primary melanoma) revealed significant 8 and 7.8-fold increase in metastatic melanoma versus primary melanoma. However, POU3F2 (encodes BRN2) is not significantly differentially expressed between human metastatic melanoma and primary melanoma (appendix 5). We were not able to validate this result on protein level due to lack of thick and metastatic melanomas.

DUSP6 SOX10

Human Human SSM ETS1 BRN2

Merge

Figure 5.8 ǀ Immunofluoresence with tyramide signal amplification (IF-TSA) of thin melanoma with a panel of DUSP6, SOX10, ETS1 and BRN2 antibodies.Melanoma cells and keratinocytes express high levels of DUSP6. Only few melanoma cells express BRN2 which are also ETS1 positive. Scale bars = 200 μm.

165

90 80 70 60 50 40 30 20 10 0 SSM BN CMN DUSP6

70 30 60 25

50 20 40 15 30 10 20 5 10 0 0 SSM BN CMN SSM BN CMN BRN2 ETS1

Figure 5.9 ǀ Quantitation of DUSP6, BRN2, and ETS1 in human melanocytic lesions. Relative protein expression of DUSP6, BRN2 and ETS1 in congenital nevi (CMN, n=5), benign nevus (BN, n=5) and superficial spreading melanoma (SSM, n=8) per 20X field. DUSP6 expression is higher in SSMs than BN (p<0.01) and CMN (p<0.0001). BN also show higher DUSP6 expression than CMN (p<0.0001). On the other hand CMN have higher BRN2 expression than BN and SSMs (p<0.00001). BRN2 is not significantly differentially expressed beteen BN and SSMs (p=0.2). ETS1 is also not differentially expressed between CMN and BN (p=0.1) while SSM has less ETS1 expression compare to CMN (p<0.0001) and BN (p<0.0001).

166

5.6 Discussion

Different cellular activities e.g. proliferation, differentiation, survival and cellular death is controlled by a family of Ser-Thr protein kinases known as Mitogen-activated protein kinases (MAPKs) (Kim 2010). Extracellular signal-regulated kinase (ERK) which is activated via binding of extracellular ligand to the receptor tyrosine kinase (RTK) has several isoforms i.e. ERK1-8 (Dhillon 2007). The phosphorylated ERK then activates a variety of signalling cascades and transcription factors via phosphorylation. Controlling MAPK pathway involves the closely regulated interaction of stimulating and inhibiting kinases and phosphatases such as DUSP6, a dual specificity phosphatase. Melanocytes harbouring NRASQ61K are in constant hyperactivated state with respect to the MAPK pathway, leading to the propensity to form benign and eventually malignant melanocytic lesions. DUSPs (dual specificity phosphatases) comprise of a superfamily of protein phosphatases (Denua 1995, Denub 1998) of 61 members. One of the best-known subgroups of DUSPs that can target mitogen-activated protein kinases (MAPKs) is classified as MAPK phosphatases (MKPs). This subgroup has 11 members which can simultaneously dephosphorylate Ser-P/Thr-P and Tyr-P in the activation motifs of MAPK (The-X-Tyr) and subsequently deactivate the MAPK pathway (Kamata 2005). The activity of DUSPs is also under the control of cellular oxidative state since oxidation of active sites of these phosphatases lead to their inactivation (Tonks 2005, Lambeth 2004, Chiarugi 2003). DUSP1, DUSP2, DUSP4 and DUSP5 are specifically located in nucleous, DUSP6, DUSP7 and DUSP16 are located in cytosole, while DUSP8, DUSP9 and DUSP10 can reside both in cytosol and nucleus (Patterson 2009).

In this chapter we use NRAS-Cdk4 mouse model to study the gene and protein expression changes that occur during progression. After identification of melanocytic and keratinocytic specific gene expression signatures we were able to some extent search for cell autonomous changes in S1, S2 and S3 melanomas. Dusp6 is the only gene that its expression is constantly upregulated during transformation and progression.

Analysis of TCGA MM dataset revealed only two somatic DUSP6 mutations in 153 patients with malignant melanoma. Also, analysis of gene expression of human metastatic melanoma versus primary melanoma revealed a significant increase in DUSP6 expression in metastatic stage (Figure 5.3e). Neural crest derived tumors such as glioblastomas also show DUSP6 upregulation. High DUSP6 expression in human glioblastomas has been linked to the malignant phenotype and inhibition of DUSP6 expression both in vivo and in

167 vitro suppresses Cisplatin resistance (Messina 2011). In other cancers such as colorectal cancer DUSP6 expression opposes the effect of anti-apoptotic factor BCL-2. Overexpression of DUSP6 in human colorectal HCT116 sensitizes these cells to apoptotic factors such as 5-Fluorouracil (Piya 2012). These studies suggest a dual role for DUSP6 during transformation and progression which is cancer type dependent.

The exact role of Dusp6 in carcinogenesis and metastasis is controversial in many cancer types. It is possible that role of Dusp6 be tissue dependent and even controlled by tissue microenvironment. In vitro study on digestive tract cancer cells such as esophageal squamous cell carcinoma (11 cell lines) and nasopharyngeal carcinoma (7 cell lines) revealed complete and 71% loss respectively of DUSP6 expression in these cancer cell lines. DUSP6 expression was also downregulated in esophageal squamous cell carcinoma (40%) and nasopharyngeal carcinoma tumors (75%). Upregulation of DUSP6 reduces the cell migration ability of these cells in vitro (Wong 2011). In pancreatic cancer cell lines DUSP6 acts as tumor suppressor gene and its expression is downregulated compared to immortalised normal pancreatic ductal cells. However in vivo studies reveal a more complex dual role for DUSP6 since its expression is upregulated during proliferation phase of tumorogenesis and downregulated with invasion and metastasis (Furukawa 2003, Xu 2005, Furukawa 2005). In contradiction, upregulation of DUSP6 in papillary thyroid carcinoma cell lines increases the invasion and metastatic ability of these cells. Silencing DUSP6 expression resulted in less malignant phenotype (Degl’Innocenti 2013). In lung tumors only 17.7% of studied samples have shown low expression or loss of heterozygosity (Okudela 2009). Epigenetic changes such as methylation of CpG islands in intron 1 of DUSP6 has been linked to downregulation of DUSP6 expression (Xu 2005). Chemoresistance is a devastating factor in treatment of glioblastoma. Human primary and long term cell cultures of glioblastoma exhibit high DUSP6 expression both on RNA and protein levels. Similar to melanoma cell lines, DUSP6 upregulation in glioblastoma cells increased tumorogenesis in vivo (Messina 2011, Li 2012). A study on MCF-7 breast cancer cell line revealed that silencing of DUSP6 and/or DUSP5 result in hyperactivation of MAPK pathway which inhibit proliferation and migration of these cells while upregulation of DUSP6 and/or DUSP5 keeps MAPK pathway in transient activation state which is pro- growth and pro-migration (Nunes-Xavier 2010) which is in contrast to the pro-growth and anti-migration role of DUSP6 in pancreatic cancer cell lines (Wong 2011). A recent comprehensive study by Shojaee et al. depicts a more complex role for DUSP6 in tumorogenesis (Shojaee 2015). In human NRASG12D pre-B cells malignant transformation

168 is met with oncogene-induced apoptosis of majority of these cells while surviving cells that were not affected and transformed had high DUSP6 expression. Colony formation ability of DUSP6-/- B cell lineage leukemia was significantly reduced compared to the DUSP6+/+ cohorts (Shojaee 2015).

The role of ETS1 in transformation and progression is still illusive and controversial. Keehn et al. reported that expression of ETS1 in human invasive is significantly higher than benign nevi (Keehn 2003) while a larger study reported that ETS1 is widely expressed in benign and malignant melanocytic lesions such as dermal nevi, metastatic melanoma and primary melanomas (Torlakovic 2004). Expression of ETS1 is only nuclear in these two studies.

More recent functional studies reveal an oncogenic role for ETS1 but the expression patterns are only cytoplasmic which reveals mislocalization of ETS1. A functional study on the role of ETS1 revealed that it exacerbates melanomagenesis through the activation of ETS1-MET axis in combination with or without PAX3. Majority of ETS1 was expressed in cytoplasm while PAX3 is only expressed in nucleus (Kubic 2015). In concordat with this study, Potu et al. reported that Ets1 upregulation amplifies NRAS activity via its overexpression which leads to a malignant phenotype. ETS1 and NRAS were highly expressed in metastatic and primary melanoma vs. benign nevi. While benign nevi show clear nuclear staining of ETS1, primary and metastatic melanomas show only cytoplasmic staining (Potu 2017). In our study the expression of ETS1 is only limited to nucleus and no mislocalization was detected. ETS1 is a transcription factor for DUSP6 in fibroblasts, lung cancer cells and leukemia cells (Ekerot 2008, Zhang 2010, Shojaee 2015).

In our study the expression of BRN2 and ETS1 is positively correlated while there is no link between ETS1 and DUSP6 expression. It is possible that epigenetic changes that block the expression of DUSP6 be responsible for this in subset of nevi and melanomas.

Some studies point to the role of BRN2 as a developmental transcription factor which is mainly expressed in undifferentiated melanoblasts or dedifferentiated malignant melanoma cells (13, 38-43). Hence, differentiated melanocytes and nevocytes are negative for BRN2 and metastatic melanoma cells have the highest BRN2 expression. This is in direct contrast with findings of Richmond-Sinclair et al. that reported BRN2 is not expressed in 70% of 129 cutaneous melanomas that they studied. Also, this study revealed that BRN2 is expressed in melanomas that bear BRAFV600E mutation and originate from non sun exposed area of skin while majority of melanomas from sun exposed areas of skin do not 169 express BRN2. BRN2 positive group of tumors also show high pMAPK and P16 expression while the BRN2 negative group highly express P53 and pRB (Richmond- Sinclair 2008). While all SSMs that were used in our study are from sun exposed area of skin their mutation is unknown.

5.7 Future aims

1. High quality RNA from microdissected mouse nevi (n=6), stage 1(n=6), stage 2 (n=6) and metastasis (n=6) purified to be compared with isolated RNA from mouse nevi, stage 1(n=6), stage 2 (n=6) and metastasis (n=6). We were unable to complete this due to lack of funding. High cost of RNAseq or agilent chips vs original illumina chips. The isolated RNA materials will be kept in -80°C pending future funding.

2. Establishing melanoma cell lines from fast growing and slow growing melanoma strains, monitoring the growth of these melanoma cell lines in vitro, ablation of Dusp6 expression in fast growing cell lines and comparison of growth rate versus slow growing cell lines.

3. In vivo functional study of Dusp6 in melanoma progression using Tyr-Cre::Dusp6fl/fl mice. These mice will be crossed with NRAS-Cdk4 mice that develop fast growing melanomas to see if this will impede the formation of melanoma.

170

Discussion

171

Chapter 6

The overarching aim of this thesis was to examine how differences in genetic background influence dermal nevus development. This can be done in human populations using genetic association studies using the phenotype of nevus free and nevus prone, or the number of nevi per person. Genome wide association studies (GWAS) of nevus and melanoma have revealed many loci (Udayakumar 2009, eh 2009, Duffy 2010, Barrett 2011, Law 2015). However, these represent common variants that individually have a very small impact on risk, and in nearly all cases the true causal genes/variants are not yet identified. Hence no potential drug targets have resulted from such studies. Many of the single nucleotide polymorphisms (SNPs) found by GWAS are near pigmentation genes e.g. TYR, TRP1, MC1R, suggesting some degree of population stratification. In addition many of the implicated genes are of unknown function. Even when their function is known, how and where they act or interact during nevogenesis or melanoma progression has yet to be determined. It is difficult to confirm nevus or melanoma susceptibility genes and to understand their mode of action. These studies can be skewed by the effects of genetic ethnicity, diversity of lesion subtype, even within an indivdual patient with multiple nevi, let alone between patients, and it is very difficult to test gene-environment interactions. ,Further, it is very difficult to functionaly validate even a strong GWAS hit, not the least in the case where the gene has a cell extrinsic actions that may be able to be properly modelled in 2D/3D cell culture studies. Forward genetic strategies can be successful in understanding disease. An example of this is the Collaborative Cross (CC), a powerful and revolutionary approach for mapping genes for complex conditions (CC Consortium, 2012; Churchill, 2004). The CC can bring to light new genes and gene interactions involved in these complex diseases. For example it was recently found that one of the CC strains develops ulcerative colitis (Rogala, 2014). This disease is not present in any of the founders, i.e. the CC has uncovered genetic interactions that confer this phenotype. Genes which do not predispose to a disease, but which can influence its course or severity, are “modifier genes”. They are very difficult to identify, especially in humans (Nadeau, 2001). An excellent example of the power of the transgene modifier approach comes from the work of Hunter et al. (Lifsted, 1998). To find modifier genes of breast cancer metastasis, they crossed a relevant transgenic mouse with each of 27 strains. Recombinant inbred mouse lines are generally derived from just two parental strains. Despite the smaller size and lower heterogeneity of their strain panel compared to the CC, they discovered a set of interacting genes in a pathway that regulates metastasis; the

172

“Diaspora” pathway, alleles of which predict survival outcomes for patient genotype groups (Crawford, 2008). A similar strategy has been used to identify a modulator of breast cancer metastasis in mouse and humans by a non-cell-autonomous mechanism (Faraji, 2012). These studies used a small number of closely related common inbred strains. Instead, we can use the CC to harness most of the available genetic repertoire of the mouse. The mapping power of the CC has recently been demonstrated using coat colour as a known Mendelian trait (Ram, 2012).

The primary focus of this project was to use transgene modifier approach along with CC to discover modifier genes that exacerbate nevogenesis and melanocytic microsatellite formation in many CC strains. This project was stemmed from five overarching questions:

1. How melanocytes with NRAS-Cdk4 mutations reach the dermis and form dermal melanocytic lesions?

2. Does strain background modifies histopathology of melanocytic lesions with NRAS-Cdk4 mutations?

3. Does genetic background affects nevogenesis and is responsible for formation of large or giant dermal lesions?

4. Does genetic background affects the formation of angiotropic microsatellites in the hypodermis?

5. What cell intrinsic molecular changes drive the progression, and metastasis of melanocytic lesions?

These questions have been answered fully or partially in different chapters of this dissertation. In this chapter we provide the key findings, shortcomings, and limitations that hampered us to fully answer all five questions.

173

6.1 Summary of key findings and discussion

Melanocytes reach the dermis by breaking hair follicles

When we first asked the question “How do melanocytes with NRAS-Cdk4 mutations reach the dermis and form dermal melanocytic lesions?”. Normally melanocytes mainly reside in the hair follicles or epidermis of NRAS-Cdk4 mouse model would form dermal melanocytic lesions. We answered this question in chapter 2 by taking dorsal skin from NRAS-Cdk4 and wild type mice at different time points (at postnatal day 1, 3, 5, 10, 15, 21, 42, and 60) and subsequently staining these samples with Sox10 (a melanocyte marker) to track the transport of these lesions. We found that melanocytes escape hair bulb or hair bulge (Chitsazan 2016) and migrate to superficial or deep dermis where they congregate to form melanocytic lesions. Also, melanocytes breaking out of long hair follicles that are located in the hypdermis during anagen probably explain the derivation of melanocytic microsatellites in the hypodermis. These microsatellites usually form near blood vessels (angiotropism). The temporal development of dermal nevi begins at about post-natal day 5 (P5) and an early “band” of nevi appeared by P10. These tend to become fully-developed nevi by P42, hence the mouse model provides a unique teleological window into formation of these lesions. Understanding the spatio-temporal migratory patterns of melanocytes from epidermis to dermis was a crucial first step in the study in terms of trying to validate mechanisms of nevus development.

Somatic NRAS mutation is not the sole driver of nevogenesis, microenvironment can also play a role

Despite the evolutionary differences in the pigmentary system between species (mouse melanocytes are usually limited to hair follicles whereas in humans they are also located in the epidermis), virtually all genes involved in regulating mouse melanocyte behaviour (i.e. pigmentation) are relevant in humans (Walker et al., 2012). NRAS-Cdk4 mice represent a useful animal to model nevogenesis. One must ask “Does strain background modify the histopathology of melanocytic lesions in NRAS-Cdk4 mice?”. By crossing 70 different collaborative cross strains with NRAS-Cdk4 transgenic mice we observe tremendous diversity in histopathological architecture of melanocytic lesions. This was surprising since all progeny shared a similar NRAS-Cdk4 mutation. The histopathological classification of murine melanocytic lesions presented in Chapter 3.

174

Here we discovered that these lesions range from small superficial and deep blue nevi to giant lesions covering the whole dorsal skin of mice. These giant nevi were shown to be composed of interconnected units, each of which is histopathologically defined as pigmented epithelioid melanocytoma (melanocytoma), or giant congenital nevus.

In sum, this is the first methodical study of histopathological classification of NRAS-Cdk4 mouse model using 70 collaborative cross strains. Here we show that innate genetic background dictates the histopathology of NRAS-Cdk4 driven melanocytic lesions. This is also possible in human since a wide range of melanocytic lesions from small to large and giant nevi. Understanding the innate genetic diversity is also important in the case of melanomas that develop from these nevi or develop de novo in different patients with diverse genetic background which makes the treatment of melanoma so complicated.

The Cdon-Shh-Edn1 axis activation exacerbates nevogenesis

NRAS-Cdk4-CdonNOD/+ mice have activated Shh-Edn1 axis which result in exacerbation of nevogenesis. In these mice instead of many blue nevi developing on the dorsal area of skin, they exhibit large melanocytomas that cover the dorsal area of the mice which are in essence giant congenital nevi. Patients with giant congenital nevi have higher than normal risk for the development of malignant melanoma (85%) and neurocutaneous melanocytoma (60%) (Kinsler 2013, Martins de Silva 2017). Almost half of patients with neurocutaneous melanocytoma die from central/peripheral nervous system malignant melanoma (Kinsler 2013, Martins de Silva 2017).

The development of large and contiguous melanocytic lesions is linked to the activation of Shh-Edn1 axis in NRAS-Cdk4 susceptible and Cdon+/- mice while resistant NRAS-Cdk4 mice develop small and discontinuous melanocytic lesions. Also, the pattern of distribution of giant congenital nevi in susceptible mice strains is similar to bathing trunk pattern in human. These patients have the highest risk for development of MM and neurocutaneous melanocytoma (Martins de Silva 2017). Perhaps SHH-EDN1 signalling acts as a rheostat, via melanocyte EDN1-GNAQ-MAPK, in exacerbating blue nevi to form very large or giant nevi.

We propose that our strategy, using mouse models could be helpful also in gaining insights into genetic background influences in other rare diseases. In the case of giant congenital nevi, they are only seen in about 1/20,000 births, hence no large genetic 175 association studies have yet been performed on large cohorts of cases and controls. But giant congenital nevi can be very disfiguring, by definition covering significant parts of the body surface, and with a high propensity to convert to melanoma. Studying human lesions post-natally does not tell us how they are formed. In years to come DNA from large enough cohorts may become available for appropriate GWAS studies, which may shine some light onto genetic susceptibility, but not being a common disease it will be very difficult to attract traditional research funding. To confirm our hypothesis that keratinocyte cytokines may be involved in the formation of giant nevi would entail sensitive gene expression and or immunohistochemical studies of representative skin samples from perhaps hundreds of patients and ethnic matched controls, again a difficult study to realize. Thus we speculate that targeting keratinocyte cytokines might be a target for further investigation as a therapy for these nevi.

One would hope to have a drug therapy by which giant nevi could be controlled, or even reversed, since they can spontaneously regress (Arpaia et al., 2005; Nath et al., 2011; Cusak et al., 2009). Surgical excision is the gold standard for treatment of most cutaneous tumours. But topical treatments are now available for solar keratosis and in situ squamous and basal cell carcinomas (e.g. fluorouracil, imiquimod, Picato gels), lesions previously only removed by excision.

The same may eventually be true for large/giant nevi. We note that topical treatment of pigmented BCCs with the SMO antagonist LDE225, and Bosentan an inhibitor of Edn signalling, which we used in our study to suppress the development of dermal nevi, results in the disappearance of the melanocytosis associated with the lesions and could be a potential non-invasive treatment for removal of disfiguring large and giant nevi or patients with large number of nevi. It is not known whether germline CDON variants influence the development of human congenital nevi, but our findings enable us to postulate that variants in any gene that result in up-regulation of the SHH pathway, or any other pathway that increases expression of secreted keratinocyte cytokines with mitogenic effect on melanocytes, could also be candidates. The Shh-Edn1 axis is normally active only in anagen, when it most likely functions in the programmed activation of melanocyte stem cells, which supply daughter cells that migrate to the hair bulb (Rabbani et al., 2011). Hence in conditions of stress caused by environmental or genetic factors (such as Cdon or deregulation of other Shh pathway components) it is not surprising that melanocytes can escape their normal follicular niches and congregate in the dermis as nevi. Edn1 binds to the melanocyte Ednrb receptor and stimulates signalling through Gnaq/Gna11 and 176 subsequently the MAPK pathway. GNAQ is somatically mutated in ~80% of human blue nevi, suggesting that signalling through it is important in dermal nevus development, and mice expressing an activated Gnaq mutant in their melanocytes exhibit essentially the same phenotype in terms of dermal melanocytic nevi that we observe in our model with activated keratinocyte Shh (Huang et al., 2015).

Murine models with keratinocyte-specific deregulation of Shh pathway components e.g. Shh (Ellis 2003), Ptch1 (Peterson 2015), Smo (Grachtchouk 2003; Eberl 2012), and Gli2 (Grachtchouk 2000), also exhibit melanocyte hyperplasia. With regard to the role of Cdon gene variation in exacerbation of nevogenesis, another important aspect of our strategy is the notion that different genes act at different stages of tumor progression (Ferguson 2014). For instance we have shown before (Ferguson 2014) and confirmed now that the QTL at the Cdon gene is only detected for the phenotype of giant congenital nevus, not whether they proceed to melanoma.

Hence the strains carrying the CdonNOD allele are not necessarily more prone to the development of melanoma. In these mice it is fitting that keratinocyte-produced mitogens increase total nevocyte count per strain, whereas the transformation of an individual nevus (of which there are >10,000 per mouse on an FVB strain background) is extremely low (a 0.04% probability that a nevus will convert to melanoma; Chi 2014), but modified by other epigenetic or cell extrinsic effects that have not yet been elucidated.

A region on murine chromosome 10 is linked to the development of melanocytic microsatellite

The appearance of angiotropic melanocytic microsatellites (MSs) is an important prognostic marker for loco-regional nodal metastasis and distant metastasis in patients with a melanoma (Van 2008, Lugassy 2011). Patients with primary melanoma and matched loco-regional and distal metastasis did not show vascular or lymphatic invasion. Metastasis was only linked to the presence of angiotropism in these patients while angiotropism was absent in similar group of patients with primary melanoma and no loco- regional or distal metastasis (Lugassy 2011).

177

It should be noted that presence of MS or angiotropism is usually not reported in the pathological report hence there are limited information available in majority of pathological studies. However the presence of angiotropic MS is nearly ubiquitous in congenital melanocytc nevi both in newborn and adults (97%) (Barnhill 2010, Lugassy 2011), atypical spitzoid melanocytic nevi and atypical spitzoid dysplastic nevi (Barnhill 2011, Wilsher 2013). In this chapter we asked the following question “Does innate genetic background affect the formation of angiotropic microsatellites in the hypodermis?”.

Here we performed an unbiased genetic screen based on the microsatellite count and selected Gja1 (expressing Cx43) as the best candidate based on allele segregation, SNP cataloguing, and previous association with nevogenesis. Our gene and protein expression studies revealed that Gja1 and Cx43 expression, respectively, is much higher during anagen. Also, Cx43 is not expressed in the mouse epidermis, but in hair bulb and hair bulge areas where melanocytes interact with keratinocytes. However it is not yet clear how this interaction might affect the development of MSs and why early and fully develop MSs do not express Cx43.

An interesting result on the interaction of cells expressing connexion 43 comes from the work of Chen et al. on brain metastasis (Figure 6.1). Murine and human breast and lung cancer cells express high levels of protocadherin 7 (PCDH7) which facilitates their binding to normal astrocytes in the brain tissue. The formation of GAP junctions between Cx43 of cancer cells and astrocytes initiates an intercellular signiling between these cells. Cancer cells then transfer high levels of cGMP to astrocytes via the establish GAP junctions and promote the astrocytes to synthesise and release of mitogenic cytokines such as IFNα and TNF.

These cytokines activate STAT1 and NF-κB signiling in cancer cells which leads to their proliferation (Chen 2016 In our NRAS-Cdk4 mouse model only melanocytes that reside in the hair follicle bulge and bulb express Cx43, while Cx43 expression in telogen is limited to bulge melanocytes. Melanocyte stem cells could possibly use similar strategy to exploit metabolic machinery of keratinocytes and produce high population of differentiated cells which could escape hair follicle bulb and form MSs. Previous study by Chang et al. reveal a sophisticated interaction between melanocyte stem cells, differentiated melanocytes and hair bulb stem cells which activation of melanocyte stem cells could lead to apoptosis of hair bulb stem cells (Chang 2013).

178

(Chen 2016)

Figure 6.1 ǀ Cx43 mediates brain metastasis (from Chen 2016).

Very little is known about the relative expression of various connexins in skin, nevi and melanoma or how different connexins combine to form the connexon, the makeup of which have a large effect on cell behaviour. Dissecting the connexin interactome in mouse and human nevi and melanoma are important for understanding nevo/melanoma-genesis and also development of angiotropic microsatellites that promote metastasis. In terms of histopathology, murine MSs from NRAS-Cdk4 mice have balloon cell structures which are reminiscent of pigmented balloon cell nevi and metastatic balloon cell melanomas. While they lack the nuclear atypia of tumour and metastatic cells, they are enclosed by less collagen unlike the balloon cell nevi. It is not clear why these balloon cells with high melanin content are more mobile than the dendritic melanocytes. The appearance of balloon like melanocytes in the lymph node metastasis of has been recently been reported in BRAFCA/+PTEN-deficient mice (Young 2017). The appearance of these bulky cells in the lymph node metastasis of both NRAS-Cdk4 and BRAFCA/+PTEN suggest that these cells might go through epithelial-to-mesenchymal transition (EMT) before initiation of dissemination via extra/intra-vascular routes. 179

The in vivo imaging has shown that angiotropic melanoma cells follow the vessels and invade brain (Bentolila 2016). These mobile melanoma cells have gone through EMT and are mesenchymal in appearance during transition but look balloon like after arrival in the brain. These results in combination with our study reveal a more complex dissemination pattern in metastasis arising from benign nevus, melanocytoma and melanoma.

The dual role of Dusp6 in development of melanocytic lesions

Understanding early transformative changes that drive melanomagenesis is crucial for developing better diagnosis and treatment procedures for melanoma patients. Here we asked a very basic question: “What cell intrinsic molecular changes drive the transformation, progression, and metastasis of melanocytic lesions??”. By studying gene expression of melanocytes, nevocytes and different stages of melanoma (S1, S2 and S3) we found out that a gene that encodes Dusp6 is highly upregulated with transformation, progression, and metastasis.

Our mouse results is in line with previous study by Li et al. that show Dusp6 overexpression in a melanoma mouse model will increase tumorigenicity, enhance anchorage independent proliferation and metastasis (Li 2011). Interestingly overexpression of MEK reduced the ability of these melanoma cells to metastasize to distant organs. However, the increased DUSP6 expression was only studied on one human melanoma cell line (A375) which failed to validate the mouse results. A previous study on human thick melanoma revealed two types of melanomas with high and low expression of DUSP6. Both DUSP6 negative and positive melanomas have poor prognosis (Li 2011). However the expression of DUSP6 was not studied on normal skin and benign nevi which can’t help us to put the expression of DUSP6 in context. A challenge that is shared by many studies conducted on human samples including ours is the lack of comprehensive information on the samples e.g. mutation and lack of different time points during transformation of nevus to melanoma.

180

6.2 Challenges and shortcomings - Chapter 2

Scoring nevus density of pigmented CC strains

It is difficult to quantitate melanocytic cell number in pigmented strains because of the dense melanin deposits mask most nuclei. This is not helped even with Sox10 staining of melanocytes in pigmented lesions since fluorescence is also masked by heavy pigment deposits. For lightly pigmented low nevus count strains we assessed the number of lightly pigmented (brown) dendritic nevus cells. For heavily pigmented strains (i.e. dark pigmentation covering a large portion of the dermis), we also try to assess dendritic “nevus cells”, but essentially we scored based on the area covered by dense pigmentation as a marker for the density score. Using Aperio software to score the pigmentation levels on different background produced similar results to manual scoring.

6.3 Challenges and shortcomings - Chapter 3

Orthogonal approaches measuring gene and protein expression in skin

Orthogonal approaches such as qPCR and western blotting are very useful in quantitating gene and protein expression. However the result is highly dependent on the hair cycle phase and could simply be varied between strains not only by age but by the stage of hair cycle. Hence the gene expression and protein expression needs to be closely monitored and interpreted based on the aforementioned factor. Another factor that influences the result of gene and protein expression is the differences in hair follicle density between susceptible and resistant, and WT and Cdon KO mice. We were unable to use western blot or qPCR due to such limitations. We have focused on IF staining in which we could as best as possible select areas of skin with similar numbers of hair follicles for the quantitative expression comparisons. Western blotting and/or qPCR would not allow us to take into account the differences in hair follicle density or stage at all. The levels of expression of Glis is generally quite low adult mouse skin unless in anagen. Hence we thought best to quantitate our results using the very sensitive Perkin Elmer InForm software after tyramide signal amplification.

Lack of GWAS study on human large and giant congenital nevi

In terms of germline gene variation there are no GWAS or genetic associated studies conducted on human large or giant congenital nevi patients, nor familial studies that might 181 reveal rare mutations. In terms of somatic mutations, CDON missense mutations must be uncommon since NRAS is the only recurrently mutated gene in large and giant congenital nevi. Hence, we were unable to validate the occurrence of germ line or somatic mutations of CDON. But in essence, since the SHH-EDN1 effect is non cell autonomous, whether somatic mutations of CDON are present in the human nevi is irrelevant. Human congenital nevi do however carry mutations that affect the pathways in melanocytes downstream of the EDN1 receptor, EDNRB (e.g. GNAQ, GNA11, NRAS, BRAF etc). Also, it is not just EDN1 that could stimulate these melanocyte pathways, but also other secretory cytokines with mitogenic effect on melanocytes can activate MAPK pathway (e.g. KITL, EDN2, EDN3. HGF etc).

Hence any modifier gene that could increase the expression of one of these keratinocyte cytokines could potentially also exacerbate nevogenesis (Figure 6.2).

EDN1

Figure 6.2 ǀ Convergence of signaling pathways under the activity of MAPK could exacerbates the effect of somatic NRAS or BRAF mutations (from Etchevers 2014).

182

6.4 Challenges and shortcomings - Chapter 4

In chapter 4 we mapped an interval on murine chromosome 10, which, after fine mapping and SNP cataloguing, was narrowed to an area containing 11 genes. Surprisingly 6 of these genes have been previously linked to metastasis in different cancers including melanoma. The major shortcoming of this chapter is that while initial protein expression studies and comparisons were performed, Gja1 was not validated as the candidate in this genetic interval. This would require a large amount of extra work, such as testing more CC strains with recombinations within the interval to further fine map the region, and in vivo studies using knock out mouse model of our candidate gene Gja1. The latter work has commenced but it is not possible to complete within the timeframe of this PhD project, due to the large amount of back crossing, crossing with tissue specific Cre, and finally crossing with NRAS-Cdk4 mice.

6.5 Challenges and shortcomings - Chapter 5

Again, this chapter was based only upon initial follow-up of a large amount of work that my PhD panel thought was worth reporting. As discussed in Chapter 5 the gene expression comparisons of lesions at different stages of progression was so mplcated by keratinocyte contamination of the very early lesions. I put a large amount of effort in to working up and optimisation of microdissection methodologies to circumvent this contamination effect, but was not able to continue with global gene expression on the microdissected RNA. In addition, validation of changes in DUSP6 staining during human melanoma progression was hindered by the extreme difficulty in obtaining a clear progression series of human melanomas of the same subtype.

6.6 Future research direction

Future studies stem from chapter 3

In chapter 3, we used two FDA-cleared to inhibit the effect of Smo and EdnrB during nevogenesis. We will also use these drugs on mice with fully developed blue nevi, deep penetrating nevi, epithelioid blue nevi, melanocytoma and melanoma to observe the possible inhibitory effect. This study will provide more information about the possibility of set up a clinical trial. I have commenced initial studies on the nevus susceptible mouse

183 model to determine whether topical treatment with Shh or Edn1 pathway inhibitors can induce regression of nevi.

Future studies stem from chapter 4

In chapter 4, we mapped Gja1 as a modifier gene for development of angiotropic microsatellites. We have received the Cx43fl/fl mice and will cross them with K14-Cre and Tyr-Cre to design keratinocyte and melanocyte specific Cx43 deficient mouse models. This will help us to gain a better understanding of the role of Cx43 in development of microsatellites and metastasis on NRAS-Cdk4 background. Also, we will try to map the epidermis and hair follicular distribution of connexin monomers, and the makeup of connexons, by performing 6-plex immunofluorescence staining with tyramide signal amplification of mouse and human skin, nevus, melanoma, and metastasis sections with a panel of connexin (Cx26, Cx30, Cx43, Cx46), melanocyte (Sox10), keratinocyte (Krt14), and stem cell marker (Sox9). We will try to establish a better link between melanocytic microsatellite count per stain and the density of melanocytic or melanoma cells in the lymph nodes. Finally, we will use sixty paraffin-embedded blocks containing human melanocytic lesions including congenital naevi, benign nevi, dysplastic nevi, melanoma in situ, thin melanoma, and metastases to study the CX43 expression in human samples.

Future studies stem from chapter 5

In chapter 5, we have identified Dusp6 as a marker of fast cycling cells. We will continue to study this by establishing fast and slow growing nevi and melanoma. Next, we will inhibit Dusp6 expression in fast growing melanoma to impede their growth and increase Dusp6 expression in slow growing melanoma to potentially expedite their growth. We will apply the opposite strategy in the case of nevi cell lines and observe the potential changes.

We also will conduct a new gene expression study using RNA isolated from microdissected melanocytes, nevocytes and melanoma cells and compare their gene expression with RNA isolated from whole tissue samples.

184

6.7 Conclusions

In this study we used a multidisciplinary approach to study nevogenesis, development of angiotropic microsatellites, melanoma progression and metastasis. Our findings suggest that Cdon-Shh-Edn1 is a novel therapeutic target for patients with giant congenital nevi and melanocytoma. In addition, this work provides evidence that connexions modify the development of melanocytic microsatellites within the skin. The major lesson from this work is that the development of nevi is to a large degree controlled by the microenvironment surrounding the melanocytic lesion.

185

186

(2012). The genome architecture of the Collaborative Cross mouse genetic reference population. Genetics 190, 389-401.

Aasen, T., Mesnil, M., Naus, C.C., Lampe, P.D., and Laird, D.W. (2016). Gap junctions and cancer: communicating for 50 years. Nat Rev Cancer 16, 775-788.

Aoki, H., Yamada, Y., Hara, A., and Kunisada, T. (2009). Two distinct types of mouse melanocyte: differential signaling requirement for the maintenance of non-cutaneous and dermal versus epidermal melanocytes. Development 136, 2511-2521.

Apostolopoulou, M., and Ligon, L. (2012). Cadherin-23 mediates heterotypic cell-cell adhesion between breast cancer epithelial cells and fibroblasts. PLoS One 7, e33289.

Arpaia, N., Cassano, N., Filotico, R., Laricchia, F., and Vena, G.A. (2005a). Unusual clinical presentation of regression in a congenital melanocytic nevus. Dermatol Surg 31, 471-473.

Arteaga, C.L., Sliwkowski, M.X., Osborne, C.K., Perez, E.A., Puglisi, F., and Gianni, L. (2011). Treatment of HER2-positive breast cancer: current status and future perspectives. Nat Rev Clin Oncol 9, 16-32.

Arumi-Uria, M., McNutt, N.S., and Finnerty, B. (2003). Grading of atypia in nevi: correlation with melanoma risk. Mod Pathol 16, 764-771.

Arumugam, T., Ramachandran, V., Fournier, K.F., Wang, H., Marquis, L., Abbruzzese, J.L., Gallick, G.E., Logsdon, C.D., McConkey, D.J., and Choi, W. (2009). Epithelial to mesenchymal transition contributes to drug resistance in pancreatic cancer. Cancer Res 69, 5820-5828.

Bagnato, A., Rosano, L., Spinella, F., Di Castro, V., Tecce, R., and Natali, P.G. (2004). Endothelin B receptor blockade inhibits dynamics of cell interactions and communications in melanoma cell progression. Cancer Res 64, 1436-1443. Bagnyukova, T.V., Restifo, D., Beeharry, N., Gabitova, L., Li, T., Serebriiskii, I.G., Golemis, E.A., and Astsaturov, I. (2013a). DUSP6 regulates drug sensitivity by modulating DNA damage response. Br J Cancer 109, 1063-1071. 187

Bagnyukova, T.V., Restifo, D., Beeharry, N., Gabitova, L., Li, T., Serebriiskii, I.G., Golemis, E.A., and Astsaturov, I. (2013b). DUSP6 regulates drug sensitivity by modulating DNA damage response. Br J Cancer 109, 1063-1071.

Balch, C.M., Gershenwald, J.E., Soong, S.J., Thompson, J.F., Atkins, M.B., Byrd, D.R., Buzaid, A.C., Cochran, A.J., Coit, D.G., Ding, S., et al. (2009). Final version of 2009 AJCC melanoma staging and classification. J Clin Oncol 27, 6199-6206.

Balch, C.M., Soong, S.J., Smith, T., Ross, M.I., Urist, M.M., Karakousis, C.P., Temple, W.J., Mihm, M.C., Barnhill, R.L., Jewell, W.R., et al. (2001). Long-term results of a prospective surgical trial comparing 2 cm vs. 4 cm excision margins for 740 patients with 1-4 mm melanomas. Ann Surg Oncol 8, 101-108.

Bald, T., Quast, T., Landsberg, J., Rogava, M., Glodde, N., Lopez-Ramos, D., Kohlmeyer, J., Riesenberg, S., van den Boorn-Konijnenberg, D., Homig-Holzel, C., et al. (2014). Ultraviolet-radiation-induced inflammation promotes angiotropism and metastasis in melanoma. Nature 507, 109-113.

Barnhill, R., Dy, K., and Lugassy, C. (2002). Angiotropism in cutaneous melanoma: a prognostic factor strongly predicting risk for metastasis. In J Invest Dermatol (United States), pp. 705-706.

Barnhill, R.L., Chastain, M.A., Jerdan, M.S., Lebbe, C., Janin, A., and Lugassy, C. (2010). Angiotropic neonatal congenital melanocytic nevus: how extravascular migration of melanocytes may explain the development of congenital nevi. Am J Dermatopathol 32, 495-499.

Barnhill, R.L., Kutzner, H., Schmidt, B., Ali, L., Bagot, M., Janin, A., and Lugassy, C. (2011). Atypical spitzoid melanocytic neoplasms with angiotropism: a potential mechanism of locoregional involvement. Am J Dermatopathol 33, 236-243. Barrett, J.H., Iles, M.M., Harland, M., Taylor, J.C., Aitken, J.F., Andresen, P.A., Akslen, L.A., Armstrong, B.K., Avril, M.F., Azizi, E., et al. (2011a). Genome-wide association study identifies three new melanoma susceptibility loci. Nat Genet 43, 1108-1113.

188

Barrett, J.H., Iles, M.M., Harland, M., Taylor, J.C., Aitken, J.F., Andresen, P.A., Akslen, L.A., Armstrong, B.K., Avril, M.F., Azizi, E., et al. (2011b). Genome-wide association study identifies three new melanoma susceptibility loci. Nat Genet 43, 1108-1113.

Bauer, J., Curtin, J.A., Pinkel, D., and Bastian, B.C. (2007). Congenital melanocytic nevi frequently harbor NRAS mutations but no BRAF mutations. J Invest Dermatol 127, 179- 182.

Bentolila, L.A., Prakash, R., Mihic-Probst, D., Wadehra, M., Kleinman, H.K., Carmichael, T.S., Peault, B., Barnhill, R.L., and Lugassy, C. (2016). Imaging of Angiotropism/Vascular Co-Option in a Murine Model of Brain Melanoma: Implications for Melanoma Progression along Extravascular Pathways. Sci Rep 6, 23834.

Beyer, E.C., Paul, D.L., and Goodenough, D.A. (1990). Connexin family of gap junction proteins. J Membr Biol 116, 187-194.

Bishop, D.T., Demenais, F., Iles, M.M., Harland, M., Taylor, J.C., Corda, E., Randerson- Moor, J., Aitken, J.F., Avril, M.F., Azizi, E., et al. (2009). Genome-wide association study identifies three loci associated with melanoma risk. Nat Genet 41, 920-925. Bjork, J.K., Akerfelt, M., Joutsen, J., Puustinen, M.C., Cheng, F., Sistonen, L., and Nees, M. (2016). Heat-shock factor 2 is a suppressor of prostate cancer invasion. Oncogene 35, 1770-1784.

Boyle, G.M., Woods, S.L., Bonazzi, V.F., Stark, M.S., Hacker, E., Aoude, L.G., Dutton- Regester, K., Cook, A.L., Sturm, R.A., and Hayward, N.K. (2011). Melanoma cell invasiveness is regulated by miR-211 suppression of the BRN2 transcription factor. Pigment Cell Melanoma Res 24, 525-537. Boyle, J.L., Haupt, H.M., Stern, J.B., and Multhaupt, H.A. (2002). Tyrosinase expression in malignant melanoma, desmoplastic melanoma, and peripheral nerve tumors. Arch Pathol Lab Med 126, 816-822.

Briscoe, J., and Therond, P.P. (2013). The mechanisms of Hedgehog signalling and its roles in development and disease. Nat Rev Mol Cell Biol 14, 416-429.

Brockmeyer, P., Jung, K., Perske, C., Schliephake, H., and Hemmerlein, B. (2014a). 189

Membrane connexin 43 acts as an independent prognostic marker in oral squamous cell carcinoma. Int J Oncol 45, 273-281.

Brockmeyer, P., Jung, K., Perske, C., Schliephake, H., and Hemmerlein, B. (2014b). Membrane connexin 43 acts as an independent prognostic marker in oral squamous cell carcinoma. Int J Oncol 45, 273-281.

Brown, C.J., Lain, S., Verma, C.S., Fersht, A.R., and Lane, D.P. (2009). Awakening guardian angels: drugging the p53 pathway. Nat Rev Cancer 9, 862-873.

Cano, A., Perez-Moreno, M.A., Rodrigo, I., Locascio, A., Blanco, M.J., del Barrio, M.G., Portillo, F., and Nieto, M.A. (2000). The transcription factor snail controls epithelial- mesenchymal transitions by repressing E-cadherin expression. Nat Cell Biol 2, 76-83. Castelli, M., Camps, M., Gillieron, C., Leroy, D., Arkinstall, S., Rommel, C., and Nichols, A. (2004). MAP kinase phosphatase 3 (MKP3) interacts with and is phosphorylated by protein kinase CK2alpha. J Biol Chem 279, 44731-44739.

Castets, M., Broutier, L., Molin, Y., Brevet, M., Chazot, G., Gadot, N., Paquet, A., Mazelin, L., Jarrosson-Wuilleme, L., Scoazec, J.Y., et al. (2011). DCC constrains tumour progression via its dependence receptor activity. Nature 482, 534-537.

Cengiz, F.P., Emiroglu, N., Ozkaya, D.B., Su, O., and Onsun, N. (2017). Dermoscopic Features of Small, Medium, and Large-Sized Congenital Melanocytic Nevi. Ann Dermatol 29, 26-32. Chai, E., Ferguson, B., Prow, T., Soyer, P., and Walker, G. (2014). Three-dimensional modelling for estimation of nevus count and probability of nevus-melanoma progression in a murine model. Pigment Cell Melanoma Res 27, 317-319.

Chan, D.W., Liu, V.W., Tsao, G.S., Yao, K.M., Furukawa, T., Chan, K.K., and Ngan, H.Y. (2008). Loss of MKP3 mediated by oxidative stress enhances tumorigenicity and chemoresistance of ovarian cancer cells. Carcinogenesis 29, 1742-1750.

Chang, C.Y., Pasolli, H.A., Giannopoulou, E.G., Guasch, G., Gronostajski, R.M., Elemento, O., and Fuchs, E. (2013). NFIB is a governor of epithelial-melanocyte stem cell behaviour in a shared niche. Nature 495, 98-102. 190

Chang, T.S., Wei, K.L., Lu, C.K., Chen, Y.H., Cheng, Y.T., Tung, S.Y., Wu, C.S., and Chiang, M.K. (2017). Inhibition of CCAR1, a Coactivator of beta-Catenin, Suppresses the Proliferation and Migration of Gastric Cancer Cells. Int J Mol Sci 18.

Chen, H.Y., Yu, S.L., Chen, C.H., Chang, G.C., Chen, C.Y., Yuan, A., Cheng, C.L., Wang, C.H., Terng, H.J., Kao, S.F., et al. (2007a). A five-gene signature and clinical outcome in non-small-cell lung cancer. N Engl J Med 356, 11-20. Chen, H.Y., Yu, S.L., Chen, C.H., Chang, G.C., Chen, C.Y., Yuan, A., Cheng, C.L., Wang, C.H., Terng, H.J., Kao, S.F., et al. (2007b). A five-gene signature and clinical outcome in non-small-cell lung cancer. N Engl J Med 356, 11-20.

Chen, H.Y., Yu, S.L., Chen, C.H., Chang, G.C., Chen, C.Y., Yuan, A., Cheng, C.L., Wang, C.H., Terng, H.J., Kao, S.F., et al. (2007c). A five-gene signature and clinical outcome in non-small-cell lung cancer. N Engl J Med 356, 11-20.

Chen, Q., Boire, A., Jin, X., Valiente, M., Er, E.E., Lopez-Soto, A., Jacob, L., Patwa, R., Shah, H., Xu, K., et al. (2016a). Carcinoma-astrocyte gap junctions promote brain metastasis by cGAMP transfer. Nature 533, 493-498.

Chen, Q., Boire, A., Jin, X., Valiente, M., Er, E.E., Lopez-Soto, A., Jacob, L.S., Patwa, R., Shah, H., Xu, K., et al. (2016b). Carcinoma-astrocyte gap junctions promote brain metastasis by cGAMP transfer. Nature 533, 493-498.

Chia, S., Norris, B., Speers, C., Cheang, M., Gilks, B., Gown, A.M., Huntsman, D., Olivotto, I.A., Nielsen, T.O., and Gelmon, K. (2008). Human epidermal growth factor receptor 2 overexpression as a prognostic factor in a large tissue microarray series of node-negative breast cancers. J Clin Oncol 26, 5697-5704.

Chiarugi, P., and Cirri, P. (2003). Redox regulation of protein tyrosine phosphatases during receptor tyrosine kinase signal transduction. Trends Biochem Sci 28, 509-514. Chin, L. (2003). The genetics of malignant melanoma: lessons from mouse and man. Nat Rev Cancer 3, 559-570.

Chiriboga, L., Meehan, S., Osman, I., Glick, M., de la Cruz, G., Howell, B.S., Friedman- 191

Jimenez, G., Schneider, R.J., and Jamal, S. (2016). Endothelin-1 in the tumor microenvironment correlates with melanoma invasion. Melanoma Res 26, 236-244.

Chitsazan, A., Ferguson, B., Ram, R., Mukhopadhyay, P., Handoko, H.Y., Gabrielli, B., Soyer, P.H., Morahan, G., and Walker, G.J. (2016). A mutation in the Cdon gene potentiates congenital nevus development mediated by NRAS(Q61K). Pigment Cell Melanoma Res 29, 459-464.

Chow, V.L., Chan, J.Y., Chan, R.C., Chung, J.H., and Wei, W.I. (2011). Basal cell carcinoma of the head and neck region in ethnic chinese. Int J Surg Oncol 2011, 890908.

Churchill, G.A., Airey, D.C., Allayee, H., Angel, J.M., Attie, A.D., Beatty, J., Beavis, W.D., Belknap, J.K., Bennett, B., Berrettini, W., et al. (2004). The Collaborative Cross, a community resource for the genetic analysis of complex traits. Nat Genet 36, 1133-1137.

Cole, F., and Krauss, R.S. (2003). Microform holoprosencephaly in mice that lack the Ig superfamily member Cdon. Curr Biol 13, 411-415.

Cox, R.P., Krauss, M.R., Balis, M.E., and Dancis, J. (1970). Evidence for transfer of enzyme product as the basis of metabolic cooperation between tissue culture fibroblasts of Lesch-Nyhan disease and normal cells. Proc Natl Acad Sci U S A 67, 1573-1579.

Cramer, S.F., and Fesyuk, A. (2012). On the development of neurocutaneous units-- implications for the histogenesis of congenital, acquired, and dysplastic nevi. Am J Dermatopathol 34, 60-81.

Cusack, C., Warde, D., Lynch, W., McDermott, M., and Watson, R. (2009a). Dramatic spontaneous regression of a medium-sized congenital melanocytic naevus. Clin Exp Dermatol 34, e34-36.

Cusack, C., Warde, D., Lynch, W., McDermott, M., and Watson, R. (2009b). Dramatic spontaneous regression of a medium-sized congenital melanocytic naevus. Clin Exp Dermatol 34, e34-36. Dai, P., Nakagami, T., Tanaka, H., Hitomi, T., and Takamatsu, T. (2007). Cx43 mediates TGF-beta signaling through competitive Smads binding to microtubules. Mol Biol Cell 18, 192

2264-2273.

Damsky, W., Micevic, G., Meeth, K., Muthusamy, V., Curley, D.P., Santhanakrishnan, M., Erdelyi, I., Platt, J.T., Huang, L., Theodosakis, N., et al. (2015). mTORC1 activation blocks BrafV600E-induced growth arrest but is insufficient for melanoma formation. Cancer Cell 27, 41-56.

Dankort, D., Curley, D.P., Cartlidge, R.A., Nelson, B., Karnezis, A.N., Damsky, W.E., Jr., You, M.J., DePinho, R.A., McMahon, M., and Bosenberg, M. (2009). Braf(V600E) cooperates with Pten loss to induce metastatic melanoma. Nat Genet 41, 544-552.

Davies, H., Bignell, G.R., Cox, C., Stephens, P., Edkins, S., Clegg, S., Teague, J., Woffendin, H., Garnett, M.J., Bottomley, W., et al. (2002). Mutations of the BRAF gene in human cancer. Nature 417, 949-954.

Day, C.L., Jr., Harrist, T.J., Gorstein, F., Sober, A.J., Lew, R.A., Friedman, R.J., Pasternack, B.S., Kopf, A.W., Fitzpatrick, T.B., and Mihm, M.C., Jr. (1981). Malignant melanoma. Prognostic significance of "microscopic satellites" in the reticular dermis and subcutaneous fat. Ann Surg 194, 108-112.

Day, C.L., Jr., Mihm, M.C., Jr., Lew, R.A., Harris, M.N., Kopf, A.W., Fitzpatrick, T.B., Harrist, T.J., Golomb, F.M., Postel, A., Hennessey, P., et al. (1982). Prognostic factors for patients with clinical stage I melanoma of intermediate thickness (1.51 - 3.39 mm). A conceptual model for tumor growth and metastasis. Ann Surg 195, 35-43.

De Raeve, L.E., Claes, A., Ruiter, D.J., van Muijen, G.N., Roseeuw, D., and van Kempen, L.C. (2006). Distinct phenotypic changes between the superficial and deep component of giant congenital melanocytic naevi: a rationale for curettage. Br J Dermatol 154, 485-492. Dedek, K., Schultz, K., Pieper, M., Dirks, P., Maxeiner, S., Willecke, K., Weiler, R., and Janssen-Bienhold, U. (2006). Localization of heterotypic gap junctions composed of connexin45 and connexin36 in the rod pathway of the mouse retina. Eur J Neurosci 24, 1675-1686.

Degl'Innocenti, D., Romeo, P., Tarantino, E., Sensi, M., Cassinelli, G., Catalano, V., Lanzi, C., Perrone, F., Pilotti, S., Seregni, E., et al. (2013). DUSP6/MKP3 is overexpressed in 193 papillary and poorly differentiated thyroid carcinoma and contributes to neoplastic properties of thyroid cancer cells. Endocr Relat Cancer 20, 23-37.

Delloye-Bourgeois, C., Gibert, B., Rama, N., Delcros, J.G., Gadot, N., Scoazec, J.Y., Krauss, R., Bernet, A., and Mehlen, P. (2013). Sonic Hedgehog promotes tumor cell survival by inhibiting CDON pro-apoptotic activity. PLoS Biol 11, e1001623.

Delmas, V., Beermann, F., Martinozzi, S., Carreira, S., Ackermann, J., Kumasaka, M., Denat, L., Goodall, J., Luciani, F., Viros, A., et al. (2007). Beta-catenin induces immortalization of melanocytes by suppressing p16INK4a expression and cooperates with N-Ras in melanoma development. Genes Dev 21, 2923-2935.

Denu, J.M., and Dixon, J.E. (1995). A catalytic mechanism for the dual-specific phosphatases. Proc Natl Acad Sci U S A 92, 5910-5914.

Denu, J.M., and Dixon, J.E. (1998). Protein tyrosine phosphatases: mechanisms of catalysis and regulation. Curr Opin Chem Biol 2, 633-641.

Denu, J.M., Stuckey, J.A., Saper, M.A., and Dixon, J.E. (1996). Form and function in protein dephosphorylation. Cell 87, 361-364.

Dessars, B., De Raeve, L.E., Morandini, R., Lefort, A., El Housni, H., Ghanem, G.E., Van den Eynde, B.J., Ma, W., Roseeuw, D., Vassart, G., et al. (2009). Genotypic and gene expression studies in congenital melanocytic nevi: insight into initial steps of melanotumorigenesis. J Invest Dermatol 129, 139-147.

Dhomen, N., Reis-Filho, J.S., da Rocha Dias, S., Hayward, R., Savage, K., Delmas, V., Larue, L., Pritchard, C., and Marais, R. (2009). Oncogenic Braf induces melanocyte senescence and melanoma in mice. Cancer Cell 15, 294-303.

Dobrowolski, R., Hertig, G., Lechner, H., Worsdorfer, P., Wulf, V., Dicke, N., Eckert, D., Bauer, R., Schorle, H., and Willecke, K. (2009). Loss of connexin43-mediated gap junctional coupling in the mesenchyme of limb buds leads to altered expression of morphogens in mice. Hum Mol Genet 18, 2899-2911.

194

Domercq, M., Alberdi, E., Sanchez-Gomez, M.V., Ariz, U., Perez-Samartin, A., and Matute, C. (2011). Dual-specific phosphatase-6 (Dusp6) and ERK mediate AMPA receptor-induced oligodendrocyte death. J Biol Chem 286, 11825-11836.

Du, J., Miller, A.J., Widlund, H.R., Horstmann, M.A., Ramaswamy, S., and Fisher, D.E. (2003). MLANA/MART1 and SILV/PMEL17/GP100 are transcriptionally regulated by MITF in melanocytes and melanoma. Am J Pathol 163, 333-343.

Duffy, D.L., Zhao, Z.Z., Sturm, R.A., Hayward, N.K., Martin, N.G., and Montgomery, G.W. (2010). Multiple pigmentation gene polymorphisms account for a substantial proportion of risk of cutaneous malignant melanoma. J Invest Dermatol 130, 520-528.

Duffy, K., and Grossman, D. (2012). The dysplastic nevus: from historical perspective to management in the modern era: part II. Molecular aspects and clinical management. J Am Acad Dermatol 67, 19.e11-12; quiz 31-12.

Dumaz, N., Hayward, R., Martin, J., Ogilvie, L., Hedley, D., Curtin, J.A., Bastian, B.C., Springer, C., and Marais, R. (2006). In melanoma, RAS mutations are accompanied by switching signaling from BRAF to CRAF and disrupted cyclic AMP signaling. Cancer Res 66, 9483-9491. Eberl, M., Klingler, S., Mangelberger, D., Loipetzberger, A., Damhofer, H., Zoidl, K., Schnidar, H., Hache, H., Bauer, H.C., Solca, F., et al. (2012a). Hedgehog-EGFR cooperation response genes determine the oncogenic phenotype of basal cell carcinoma and tumour-initiating pancreatic cancer cells. EMBO Mol Med 4, 218-233.

Eberl, M., Klingler, S., Mangelberger, D., Loipetzberger, A., Damhofer, H., Zoidl, K., Schnidar, H., Hache, H., Bauer, H.C., Solca, F., et al. (2012b). Hedgehog-EGFR cooperation response genes determine the oncogenic phenotype of basal cell carcinoma and tumour-initiating pancreatic cancer cells. EMBO Mol Med 4, 218-233.

Ekerot, M., Stavridis, M.P., Delavaine, L., Mitchell, M.P., Staples, C., Owens, D.M., Keenan, I.D., Dickinson, R.J., Storey, K.G., and Keyse, S.M. (2008). Negative-feedback regulation of FGF signalling by DUSP6/MKP-3 is driven by ERK1/2 and mediated by Ets factor binding to a conserved site within the DUSP6/MKP-3 gene promoter. Biochem J 412, 287-298. 195

Ellis, T., Smyth, I., Riley, E., Bowles, J., Adolphe, C., Rothnagel, J.A., Wicking, C., and Wainwright, B.J. (2003). Overexpression of Sonic Hedgehog suppresses embryonic hair follicle morphogenesis. Dev Biol 263, 203-215.

Etchevers, H.C. (2014). Hiding in plain sight: molecular genetics applied to giant congenital melanocytic nevi. J Invest Dermatol 134, 879-882.

Etnyre, D., Stone, A.L., Fong, J.T., Jacobs, R.J., Uppada, S.B., Botting, G.M., Rajanna, S., Moravec, D.N., Shambannagari, M.R., Crees, Z., et al. (2014). Targeting c-Met in melanoma: mechanism of resistance and efficacy of novel combinatorial inhibitor therapy. Cancer Biol Ther 15, 1129-1141.

Falchi, M., Bataille, V., Hayward, N.K., Duffy, D.L., Bishop, J.A., Pastinen, T., Cervino, A., Zhao, Z.Z., Deloukas, P., Soranzo, N., et al. (2009). Genome-wide association study identifies variants at 9p21 and 22q13 associated with development of cutaneous nevi. Nat Genet 41, 915-919.

Falk, M.M., Buehler, L.K., Kumar, N.M., and Gilula, N.B. (1997). Cell-free synthesis and assembly of connexins into functional gap junction membrane channels. Embo j 16, 2703- 2716.

Fane, M.E., Chhabra, Y., Hollingsworth, D.E., Simmons, J.L., Spoerri, L., Oh, T.G., Chauhan, J., Chin, T., Harris, L., Harvey, T.J., et al. (2017). NFIB Mediates BRN2 Driven Melanoma Cell Migration and Invasion Through Regulation of EZH2 and MITF. EBioMedicine 16, 63-75.

Faraji, F., Pang, Y., Walker, R.C., Nieves Borges, R., Yang, L., and Hunter, K.W. (2012). Cadm1 is a metastasis susceptibility gene that suppresses metastasis by modifying tumor interaction with the cell-mediated immunity. PLoS Genet 8, e1002926.

Ferguson, B., Konrad Muller, H., Handoko, H.Y., Khosrotehrani, K., Beermann, F., Hacker, E., Soyer, H.P., Bosenberg, M., and Walker, G.J. (2010). Differential roles of the pRb and Arf/p53 pathways in murine naevus and melanoma genesis. Pigment Cell Melanoma Res 23, 771-780. 196

Ferguson, B., Ram, R., Handoko, H.Y., Mukhopadhyay, P., Muller, H.K., Soyer, H.P., Morahan, G., and Walker, G.J. (2015). Melanoma susceptibility as a complex trait: genetic variation controls all stages of tumor progression. Oncogene 34, 2879-2886.

Ferguson, J.P., Cho, J.H., Yang, C., and Zhao, H. (2013). Empirical Bayes correction for the Winner's Curse in genetic association studies. Genet Epidemiol 37, 60-68.

Fischer, K.R., Durrans, A., Lee, S., Sheng, J., Li, F., Wong, S.T., Choi, H., El Rayes, T., Ryu, S., Troeger, J., et al. (2015). Epithelial-to-mesenchymal transition is not required for lung metastasis but contributes to chemoresistance. Nature 527, 472-476.

Flores-Sarnat, L. (2013). Neurocutaneous melanocytosis. Handb Clin Neurol 111, 369- 388.

Friedman, R.J., Farber, M.J., Warycha, M.A., Papathasis, N., Miller, M.K., and Heilman, E.R. (2009). The "dysplastic" nevus. Clin Dermatol 27, 103-115.

Furukawa, T., Sunamura, M., Motoi, F., Matsuno, S., and Horii, A. (2003). Potential tumor suppressive pathway involving DUSP6/MKP-3 in pancreatic cancer. Am J Pathol 162, 1807-1815.

Gandini, S., Sera, F., Cattaruzza, M.S., Pasquini, P., Abeni, D., Boyle, P., and Melchi, C.F. (2005). Meta-analysis of risk factors for cutaneous melanoma: I. Common and atypical naevi. Eur J Cancer 41, 28-44.

Gao, Y., Xu, D., Yu, G., and Liang, J. (2015). Overexpression of metabolic markers HK1 and PKM2 contributes to lymphatic metastasis and adverse prognosis in Chinese gastric cancer. Int J Clin Exp Pathol 8, 9264-9271.

Garcia-Borron, J.C., Sanchez-Laorden, B.L., and Jimenez-Cervantes, C. (2005). Melanocortin-1 receptor structure and functional regulation. Pigment Cell Res 18, 393-410.

Garraway, L.A., Widlund, H.R., Rubin, M.A., Getz, G., Berger, A.J., Ramaswamy, S., Beroukhim, R., Milner, D.A., Granter, S.R., Du, J., et al. (2005). Integrative genomic 197 analyses identify MITF as a lineage survival oncogene amplified in malignant melanoma. Nature 436, 117-122.

Gavriilidis, P., Michalopoulou, I., Chatzikakidou, K., and Nikolaidou, A. (2013). Pigmented epithelioid melanocytoma: a new concept encompassing animal-type melanoma and epithelioid blue nevus. BMJ Case Rep 2013.

Gembarska, A., Luciani, F., Fedele, C., Russell, E.A., Dewaele, M., Villar, S., Zwolinska, A., Haupt, S., de Lange, J., Yip, D., et al. (2012). MDM4 is a key therapeutic target in cutaneous melanoma. Nat Med 18, 1239-1247.

Gerstenblith, M.R., Shi, J., and Landi, M.T. (2010). Genome-wide association studies of pigmentation and skin cancer: a review and meta-analysis. Pigment Cell Melanoma Res 23, 587-606.

Gill, M., Renwick, N., Silvers, D.N., and Celebi, J.T. (2004). Lack of BRAF mutations in Spitz nevi. In J Invest Dermatol (United States), pp. 1325-1326.

Gleason, B.C., Crum, C.P., and Murphy, G.F. (2008). Expression patterns of MITF during human cutaneous embryogenesis: evidence for bulge epithelial expression and persistence of dermal melanoblasts. J Cutan Pathol 35, 615-622.

Goldschneider, D., and Mehlen, P. (2010). Dependence receptors: a new paradigm in cell signaling and cancer therapy. Oncogene 29, 1865-1882.

Goodenough, D.A., Goliger, J.A., and Paul, D.L. (1996a). Connexins, connexons, and intercellular communication. Annu Rev Biochem 65, 475-502.

Goodenough, D.A., and Paul, D.L. (2009). Gap junctions. Cold Spring Harb Perspect Biol 1, a002576.

Grachtchouk, M., Mo, R., Yu, S., Zhang, X., Sasaki, H., Hui, C.C., and Dlugosz, A.A. (2000). Basal cell carcinomas in mice overexpressing Gli2 in skin. Nat Genet 24, 216-217.

Grachtchouk, V., Grachtchouk, M., Lowe, L., Johnson, T., Wei, L., Wang, A., de Sauvage, 198

F., and Dlugosz, A.A. (2003). The magnitude of hedgehog signaling activity defines skin tumor phenotype. Embo j 22, 2741-2751.

Gray-Schopfer, V.C., Cheong, S.C., Chong, H., Chow, J., Moss, T., Abdel-Malek, Z.A., Marais, R., Wynford-Thomas, D., and Bennett, D.C. (2006). Cellular senescence in naevi and immortalisation in melanoma: a role for p16? Br J Cancer 95, 496-505.

Gremel, G., Rafferty, M., Lau, T.Y., and Gallagher, W.M. (2009). Identification and functional validation of therapeutic targets for malignant melanoma. Crit Rev Oncol Hematol 72, 194-214.

Guaita, S., Puig, I., Franci, C., Garrido, M., Dominguez, D., Batlle, E., Sancho, E., Dedhar, S., De Herreros, A.G., and Baulida, J. (2002). Snail induction of epithelial to mesenchymal transition in tumor cells is accompanied by MUC1 repression and ZEB1 expression. J Biol Chem 277, 39209-39216.

Haass, N.K., Ripperger, D., Wladykowski, E., Dawson, P., Gimotty, P.A., Blome, C., Fischer, F., Schmage, P., Moll, I., and Brandner, J.M. (2010). Melanoma progression exhibits a significant impact on connexin expression patterns in the epidermal tumor microenvironment. Histochem Cell Biol 133, 113-124.

Haass, N.K., Wladykowski, E., Kief, S., Moll, I., and Brandner, J.M. (2006). Differential induction of connexins 26 and 30 in skin tumors and their adjacent epidermis. J Histochem Cytochem 54, 171-182.

Hagan, C.R., Knutson, T.P., and Lange, C.A. (2013). A Common Docking Domain in Progesterone Receptor-B links DUSP6 and CK2 signaling to proliferative transcriptional programs in breast cancer cells. Nucleic Acids Res 41, 8926-8942.

Handoko, H.Y., Boyle, G.M., Ferguson, B., Muller, H.K., Soyer, H.P., and Walker, G.J. (2013). Plasticity of melanoma in vivo: murine lesions resulting from Trp53, but not Cdk4 or Arf deregulation, display neural transdifferentiation. Pigment Cell Melanoma Res 26, 731-734.

Harrison, S.L., MacLennan, R., and Buettner, P.G. (2008). Sun exposure and the 199 incidence of melanocytic nevi in young Australian children. Cancer Epidemiol Biomarkers Prev 17, 2318-2324.

Harrist, T.J., Rigel, D.S., Day, C.L., Jr., Sober, A.J., Lew, R.A., Rhodes, A.R., Harris, M.N., Kopf, A.W., Friedman, R.J., Golomb, F.M., et al. (1984). "Microscopic satellites" are more highly associated with regional lymph node metastases than is primary melanoma thickness. Cancer 53, 2183-2187.

Hirayama, T., and Suzuki, T. (1991). A new classification of Ota's nevus based on histopathological features. Dermatologica 183, 169-172.

Hirobe, T. (2005). Role of keratinocyte-derived factors involved in regulating the proliferation and differentiation of mammalian epidermal melanocytes. Pigment Cell Res 18, 2-12.

Houben, R., Hesbacher, S., Schmid, C.P., Kauczok, C.S., Flohr, U., Haferkamp, S., Muller, C.S., Schrama, D., Wischhusen, J., and Becker, J.C. (2011). High-level expression of wild- type p53 in melanoma cells is frequently associated with inactivity in p53 reporter gene assays. PLoS One 6, e22096.

Howard, B., Ragsdale, B., and Lundquist, K. (2005). Pigmented epithelioid melanocytoma: two case reports. Dermatol Online J 11, 1.

Huang, J.L., Urtatiz, O., and Van Raamsdonk, C.D. (2015). Oncogenic G Protein GNAQ Induces Uveal Melanoma and Intravasation in Mice. Cancer Res 75, 3384-3397.

Hui, P., Perkins, A., and Glusac, E. (2001). Assessment of clonality in melanocytic nevi. J Cutan Pathol 28, 140-144.

Hunter, K.W., Broman, K.W., Voyer, T.L., Lukes, L., Cozma, D., Debies, M.T., Rouse, J., and Welch, D.R. (2001). Predisposition to efficient mammary tumor metastatic progression is linked to the breast cancer metastasis suppressor gene Brms1. Cancer Res 61, 8866- 8872.

Hyter, S., Coleman, D.J., Ganguli-Indra, G., Merrill, G.F., Ma, S., Yanagisawa, M., and 200

Indra, A.K. (2013). Endothelin-1 is a transcriptional target of p53 in epidermal keratinocytes and regulates ultraviolet-induced melanocyte homeostasis. Pigment Cell Melanoma Res 26, 247-258.

Ichii-Nakato, N., Takata, M., Takayanagi, S., Takashima, S., Lin, J., Murata, H., Fujimoto, A., Hatta, N., and Saida, T. (2006). High frequency of BRAFV600E mutation in acquired nevi and small congenital nevi, but low frequency of mutation in medium-sized congenital nevi. J Invest Dermatol 126, 2111-2118.

Imokawa, G., Yada, Y., and Miyagishi, M. (1992). Endothelins secreted from human keratinocytes are intrinsic mitogens for human melanocytes. J Biol Chem 267, 24675- 24680.

Jamal, S., and Schneider, R. J. (2002). UV-induction of keratinocyte endothelin-1 downregulates E-cadherin in melanocytes and melanoma cells. J Clin Invest 110, 443- 452.

Kamata, H., Honda, S., Maeda, S., Chang, L., Hirata, H., and Karin, M. (2005). Reactive oxygen species promote TNFalpha-induced death and sustained JNK activation by inhibiting MAP kinase phosphatases. Cell 120, 649-661.

Kang, H.C., Wakabayashi, Y., Jen, K.Y., Mao, J.H., Zoumpourlis, V., Del Rosario, R., and Balmain, A. (2013). Ptch1 overexpression drives skin carcinogenesis and developmental defects in K14Ptch(FVB) mice. J Invest Dermatol 133, 1311-1320.

Keehn, C.A., Smoller, B.R., and Morgan, M.B. (2003). Expression of the ets-1 proto- oncogene in melanocytic lesions. Mod Pathol 16, 772-777.

Kim, H.K., and Chanock, S.J. (2012). Genome-wide association studies in melanoma: off to a good start. Pigment Cell Melanoma Res 25, 231-233.

Kim, J.Y., Cho, S.W., Lee, M.J., Hwang, H.J., Lee, J.M., Lee, S.I., Muramatsu, T., Shimono, M., and Jung, H.S. (2005). Inhibition of connexin 43 alters Shh and Bmp-2 expression patterns in embryonic mouse tongue. Cell Tissue Res 320, 409-415.

201

Kinsler, V.A., Thomas, A.C., Ishida, M., Bulstrode, N.W., Loughlin, S., Hing, S., Chalker, J., McKenzie, K., Abu-Amero, S., Slater, O., et al. (2013). Multiple congenital melanocytic nevi and neurocutaneous melanosis are caused by postzygotic mutations in codon 61 of NRAS. J Invest Dermatol 133, 2229-2236.

Kinsler, V. A., and Larue, L. (2018). The patterns of birthmarks suggest a novel population of melanocyte precursors arising around the time of gastrulation. Pigment Cell Melanoma Res 31, 95-109.

Kokta, V., Hung, T., Al Dhaybi, R., Lugassy, C., and Barnhill, R.L. (2013). High prevalence of angiotropism in congenital melanocytic nevi: an analysis of 53 cases. Am J Dermatopathol 35, 180-183.

Kubic, J.D., Little, E.C., Lui, J.W., Iizuka, T., and Lang, D. (2015). PAX3 and ETS1 synergistically activate MET expression in melanoma cells. Oncogene 34, 4964-4974. Lambeth, J.D. (2004). NOX enzymes and the biology of reactive oxygen. Nat Rev Immunol 4, 181-189.

Lan, C.C., Wu, C.S., Cheng, C.M., Yu, C.L., Chen, G.S., and Yu, H.S. (2005). Pigmentation in basal cell carcinoma involves enhanced endothelin-1 expression. Exp Dermatol 14, 528-534.

Landi, M.T., Bauer, J., Pfeiffer, R.M., Elder, D.E., Hulley, B., Minghetti, P., Calista, D., Kanetsky, P.A., Pinkel, D., and Bastian, B.C. (2006). MC1R germline variants confer risk for BRAF-mutant melanoma. Science 313, 521-522.

Lane, D.P., Cheok, C.F., and Lain, S. (2010). p53-based cancer therapy. Cold Spring Harb Perspect Biol 2, a001222.

Larsen, J.E., Nathan, V., Osborne, J.K., Farrow, R.K., Deb, D., Sullivan, J.P., Dospoy, P.D., Augustyn, A., Hight, S.K., Sato, M., et al. (2016). ZEB1 drives epithelial-to- mesenchymal transition in lung cancer. J Clin Invest 126, 3219-3235.

Law, M.H., Bishop, D.T., Lee, J.E., Brossard, M., Martin, N.G., Moses, E.K., Song, F., Barrett, J.H., Kumar, R., Easton, D.F., et al. (2015). Genome-wide meta-analysis identifies 202 five new susceptibility loci for cutaneous malignant melanoma. Nat Genet 47, 987-995.

Law, M.H., Macgregor, S., and Hayward, N.K. (2012). Melanoma genetics: recent findings take us beyond well-traveled pathways. J Invest Dermatol 132, 1763-1774.

Lee, M.S., Jun, H.J., Cho, S.H., Lee, J.D., and Kim, H.S. (2017). Intense Pulsed Light Alone and in Combination with Erbium Yttrium-Aluminum-Garnet Laser on Small-to- Medium Sized Congenital Melanocytic Nevi: Single Center Experience Based on Retrospective Chart Review. Ann Dermatol 29, 39-47.

Li, G., Yu, M., Lee, W.W., Tsang, M., Krishnan, E., Weyand, C.M., and Goronzy, J.J. (2012). Decline in miR-181a expression with age impairs T cell receptor sensitivity by increasing DUSP6 activity. Nat Med 18, 1518-1524.

Li, W., and Melton, D.W. (2012). Cisplatin regulates the MAPK kinase pathway to induce increased expression of DNA repair gene ERCC1 and increase melanoma chemoresistance. Oncogene 31, 2412-2422.

Li, W., Song, L., Ritchie, A.M., and Melton, D.W. (2012). Increased levels of DUSP6 phosphatase stimulate tumourigenesis in a molecularly distinct melanoma subtype. Pigment Cell Melanoma Res 25, 188-199.

Lin, J., Deng, Z., Tanikawa, C., Shuin, T., Miki, T., Matsuda, K., and Nakamura, Y. (2014). Downregulation of the tumor suppressor HSPB7, involved in the p53 pathway, in renal cell carcinoma by hypermethylation. Int J Oncol 44, 1490-1498.

Lin, J.Y., and Fisher, D.E. (2007). Melanocyte biology and skin pigmentation. Nature 445, 843-850.

Linnskog, R., Jonsson, G., Axelsson, L., Prasad, C.P., and Andersson, T. (2014). Interleukin-6 drives melanoma cell motility through p38alpha-MAPK-dependent upregulation of WNT5A expression. Mol Oncol 8, 1365-1378.

Lito, P., Rosen, N., and Solit, D.B. (2013). Tumor adaptation and resistance to RAF inhibitors. Nat Med 19, 1401-1409. 203

Lucci, M.A., Orlandi, R., Triulzi, T., Tagliabue, E., Balsari, A., and Villa-Moruzzi, E. (2010). Expression profile of tyrosine phosphatases in HER2 breast cancer cells and tumors. Cell Oncol 32, 361-372.

Lugassy, C., Kleinman, H.K., Engbring, J.A., Welch, D.R., Harms, J.F., Rufner, R., Ghanem, G., Patierno, S.R., and Barnhill, R.L. (2004). Pericyte-like location of GFP- tagged melanoma cells: ex vivo and in vivo studies of extravascular migratory metastasis. Am J Pathol 164, 1191-1198.

Lugassy, C., Lazar, V., Dessen, P., van den Oord, J.J., Winnepenninckx, V., Spatz, A., Bagot, M., Bensussan, A., Janin, A., Eggermont, A.M., et al. (2011a). Gene expression profiling of human angiotropic primary melanoma: selection of 15 differentially expressed genes potentially involved in extravascular migratory metastasis. Eur J Cancer 47, 1267- 1275.

Lugassy, C., Vernon, S.E., Warner, J.W., Le, C.Q., Manyak, M., Patierno, S.R., and Barnhill, R.L. (2005). Angiotropism of human prostate cancer cells: implications for extravascular migratory metastasis. BJU Int 95, 1099-1103.

Lugassy, C., Zadran, S., Bentolila, L.A., Wadehra, M., Prakash, R., Carmichael, S.T., Kleinman, H.K., Peault, B., Larue, L., and Barnhill, R.L. (2014). Angiotropism, pericytic mimicry and extravascular migratory metastasis in melanoma: an alternative to intravascular cancer dissemination. Cancer Microenviron 7, 139-152.

Lynch, H.T., Fusaro, R.M., Pester, J., and Lynch, J.F. (1980). Familial atypical multiple mole melanoma (FAMMM) syndrome: genetic heterogeneity and malignant melanoma. Br J Cancer 42, 58-70.

Ma, H., Liao, M., Qiu, S., Luo, R., Lu, R., and Lu, C. (2015). The case of a boy with nevus of Ota, extensive Mongolian spot, nevus flammeus, nevus anemicus and cutis marmorata telangiectatica congenita: a unique instance of phacomatosis pigmentovascularis. An Bras Dermatol 90, 10-12.

Ma, J., Yu, X., Guo, L., and Lu, S.H. (2013). DUSP6, a tumor suppressor, is involved in 204 differentiation and apoptosis in esophageal squamous cell carcinoma. Oncol Lett 6, 1624- 1630.

Maldonado, J.L., Fridlyand, J., Patel, H., Jain, A.N., Busam, K., Kageshita, T., Ono, T., Albertson, D.G., Pinkel, D., and Bastian, B.C. (2003). Determinants of BRAF mutations in primary melanomas. J Natl Cancer Inst 95, 1878-1890.

Marchell, R., Marghoob, A.A., Braun, R.P., and Argenziano, G. (2005). Dermoscopy of pigmented Spitz and Reed nevi: the starburst pattern. Arch Dermatol 141, 1060.

Marchetti, M.A., Kiuru, M.H., Busam, K.J., Marghoob, A.A., Scope, A., Dusza, S.W., Cordova, M.A., Fonseca, M., Wu, X., and Halpern, A.C. (2014). Melanocytic naevi with globular and reticular dermoscopic patterns display distinct BRAF V600E expression profiles and histopathological patterns. Br J Dermatol 171, 1060-1065.

Marine, J.C., and Jochemsen, A.G. (2005). Mdmx as an essential regulator of p53 activity. Biochem Biophys Res Commun 331, 750-760.

Martin, M.J., Hayward, R., Viros, A., and Marais, R. (2012). Metformin accelerates the growth of BRAF V600E-driven melanoma by upregulating VEGF-A. Cancer Discov 2, 344- 355.

Martins da Silva, V.P., Marghoob, A., Pigem, R., Carrera, C., Aguilera, P., Puig-Butille, J.A., Puig, S., and Malvehy, J. (2017). Patterns of distribution of giant congenital melanocytic nevi (GCMN): The 6B rule. J Am Acad Dermatol 76, 689-694.

Mas, J.S., Gerritsen, I., Hahmann, C., Jimenez-Cervantes, C., and Garcia-Borron, J.C. (2003). Rate limiting factors in melanocortin 1 receptor signalling through the cAMP pathway. Pigment Cell Res 16, 540-547.

Mattia, G., Errico, M.C., Felicetti, F., Petrini, M., Bottero, L., Tomasello, L., Romania, P., Boe, A., Segnalini, P., Di Virgilio, A., et al. (2011). Constitutive activation of the ETS-1- miR-222 circuitry in metastatic melanoma. Pigment Cell Melanoma Res 24, 953-965.

McLachlan, E., Shao, Q., Wang, H.L., Langlois, S., and Laird, D.W. (2006). Connexins act 205 as tumor suppressors in three-dimensional mammary cell organoids by regulating differentiation and angiogenesis. Cancer Res 66, 9886-9894.

McNeal, A.S., Liu, K., Nakhate, V., Natale, C.A., Duperret, E.K., Capell, B.C., Dentchev, T., Berger, S.L., Herlyn, M., Seykora, J.T., et al. (2015). CDKN2B Loss Promotes Progression from Benign Melanocytic Nevus to Melanoma. Cancer Discov 5, 1072-1085.

Mehlen, P., and Guenebeaud, C. (2010). Netrin-1 and its dependence receptors as original targets for cancer therapy. Curr Opin Oncol 22, 46-54. Mehlen, P., and Tauszig-Delamasure, S. (2014). Dependence receptors and colorectal cancer. Gut 63, 1821-1829.

Messina, S., Frati, L., Leonetti, C., Zuchegna, C., Di Zazzo, E., Calogero, A., and Porcellini, A. (2011). Dual-specificity phosphatase DUSP6 has tumor-promoting properties in human glioblastomas. Oncogene 30, 3813-3820.

Michaloglou, C., Vredeveld, L.C., Soengas, M.S., Denoyelle, C., Kuilman, T., van der Horst, C.M., Majoor, D.M., Shay, J.W., Mooi, W.J., and Peeper, D.S. (2005). BRAFE600- associated senescence-like cell cycle arrest of human naevi. Nature 436, 720-724.

Moore, M.G., and Bennett, R.G. (2012). Basal cell carcinoma in asians: a retrospective analysis of ten patients. J Skin Cancer 2012, 741397.

Morahan, G., Balmer, L., and Monley, D. (2008). Establishment of "The Gene Mine": a resource for rapid identification of complex trait genes. Mamm Genome 19, 390-393.

Mort, R.L., Jackson, I.J., and Patton, E.E. (2015). The melanocyte lineage in development and disease. Development 142, 620-632.

Muthusamy, V., Hobbs, C., Nogueira, C., Cordon-Cardo, C., McKee, P.H., Chin, L., and Bosenberg, M.W. (2006). Amplification of CDK4 and MDM2 in malignant melanoma. Genes Chromosomes Cancer 45, 447-454.

Na, Y.R., Lee, J.S., Lee, S.J., and Seok, S.H. (2013). Interleukin-6-induced Twist and N- cadherin enhance melanoma cell metastasis. Melanoma Res 23, 434-443. 206

Nadeau, J.H. (2001). Modifier genes in mice and humans. Nat Rev Genet 2, 165-174. Nan, H., Qureshi, A.A., Prescott, J., De Vivo, I., and Han, J. (2011). Genetic variants in telomere-maintaining genes and skin cancer risk. Hum Genet 129, 247-253.

Nath, A.K., Thappa, D.M., and Rajesh, N.G. (2011). Spontaneous regression of a congenital melanocytic nevus. Indian J Dermatol Venereol Leprol 77, 507-510. Naus, C.C., and Laird, D.W. (2010). Implications and challenges of connexin connections to cancer. In Nat Rev Cancer (England), pp. 435-441.

Nunes-Xavier, C.E., Tarrega, C., Cejudo-Marin, R., Frijhoff, J., Sandin, A., Ostman, A., and Pulido, R. (2010). Differential up-regulation of MAP kinase phosphatases MKP3/DUSP6 and DUSP5 by Ets2 and c-Jun converge in the control of the growth arrest versus proliferation response of MCF-7 breast cancer cells to phorbol ester. J Biol Chem 285, 26417-26430.

Okamura, J., Huang, Y., Moon, D., Brait, M., Chang, X., and Kim, M.S. (2012). Downregulation of insulin-like growth factor-binding protein 7 in cisplatin-resistant non- small cell lung cancer. Cancer Biol Ther 13, 148-155.

Okudela, K., Yazawa, T., Woo, T., Sakaeda, M., Ishii, J., Mitsui, H., Shimoyamada, H., Sato, H., Tajiri, M., Ogawa, N., et al. (2009). Down-regulation of DUSP6 expression in lung cancer: its mechanism and potential role in carcinogenesis. Am J Pathol 175, 867-881.

Olsen, C.M., Carroll, H.J., and Whiteman, D.C. (2010). Estimating the attributable fraction for cancer: A meta-analysis of nevi and melanoma. Cancer Prev Res (Phila) 3, 233-245.

Orloff, M., Peterson, C., He, X., Ganapathi, S., Heald, B., Yang, Y.R., Bebek, G., Romigh, T., Song, J.H., Wu, W., et al. (2011). Germline mutations in MSR1, ASCC1, and CTHRC1 in patients with Barrett esophagus and esophageal adenocarcinoma. Jama 306, 410-419.

Oskouian, B., Sooriyakumaran, P., Borowsky, A.D., Crans, A., Dillard-Telm, L., Tam, Y.Y., Bandhuvula, P., and Saba, J.D. (2006). Sphingosine-1-phosphate lyase potentiates apoptosis via p53- and p38-dependent pathways and is down-regulated in colon cancer. Proc Natl Acad Sci U S A 103, 17384-17389. 207

Ouspenskaia, T., Matos, I., Mertz, A.F., Fiore, V.F., and Fuchs, E. (2016). WNT-SHH Antagonism Specifies and Expands Stem Cells prior to Niche Formation. Cell 164, 156- 169.

Palleschi, G.M., Urso, C., Torre, E., and Torchia, D. (2008). Histopathological correlates of the parallel-furrow pattern seen in acral melanocytic nevi at dermatoscopy. In Dermatology (Switzerland), pp. 356-358; author reply 359.

Park, P.J., Lee, T.R., and Cho, E.G. (2015). Substance P stimulates endothelin 1 secretion via endothelin-converting enzyme 1 and promotes melanogenesis in human melanocytes. J Invest Dermatol 135, 551-559.

Pasparakis, M., Haase, I., and Nestle, F. O. (2014). Mechanisms regulating skin immunity and inflammation. Nat Rev Immunol 14, 289-301.

Patel, A.B., Harting, M.S., Smith-Zagone, M.J., and Hsu, S. (2008). Familial basaloid follicular hamartoma: a report of one family. Dermatol Online J 14, 14.

Patrizi, A., Neri, I., Sabattini, E., Rizzoli, L., and Misciali, C. (2005). Unusual inflammatory and hyperkeratotic halo naevus in children. Br J Dermatol 152, 357-360.

Patterson, K.I., Brummer, T., O'Brien, P.M., and Daly, R.J. (2009). Dual-specificity phosphatases: critical regulators with diverse cellular targets. Biochem J 418, 475-489. Pawlikowski, J.S., McBryan, T., van Tuyn, J., Drotar, M.E., Hewitt, R.N., Maier, A.B., King, A., Blyth, K., Wu, H., and Adams, P.D. (2013). Wnt signaling potentiates nevogenesis. Proc Natl Acad Sci U S A 110, 16009-16014.

Pei, Y.F., Tao, R., Li, J.F., Su, L.P., Yu, B.Q., Wu, X.Y., Yan, M., Gu, Q.L., Zhu, Z.G., and Liu, B.Y. (2016). TET1 inhibits gastric cancer growth and metastasis by PTEN demethylation and re-expression. Oncotarget 7, 31322-31335.

Peterson, S.C., Eberl, M., Vagnozzi, A.N., Belkadi, A., Veniaminova, N.A., Verhaegen, M.E., Bichakjian, C.K., Ward, N.L., Dlugosz, A.A., and Wong, S.Y. (2015a). Basal cell carcinoma preferentially arises from stem cells within hair follicle and mechanosensory niches. Cell Stem Cell 16, 400-412. 208

Peterson, S.C., Eberl, M., Vagnozzi, A.N., Belkadi, A., Veniaminova, N.A., Verhaegen, M.E., Bichakjian, C.K., Ward, N.L., Dlugosz, A.A., and Wong, S.Y. (2015b). Basal cell carcinoma preferentially arises from stem cells within hair follicle and mechanosensory niches. Cell Stem Cell 16, 400-412.

Piya, S., Kim, J.Y., Bae, J., Seol, D.W., Moon, A.R., and Kim, T.H. (2012). DUSP6 is a novel transcriptional target of p53 and regulates p53-mediated apoptosis by modulating expression levels of Bcl-2 family proteins. FEBS Lett 586, 4233-4240.

Pollock, P.M., Harper, U.L., Hansen, K.S., Yudt, L.M., Stark, M., Robbins, C.M., Moses, T.Y., Hostetter, G., Wagner, U., Kakareka, J., et al. (2003). High frequency of BRAF mutations in nevi. Nat Genet 33, 19-20.

Potu, H., Peterson, L.F., Kandarpa, M., Pal, A., Sun, H., Durham, A., Harms, P.W., Hollenhorst, P.C., Eskiocak, U., Talpaz, M., et al. (2017). Usp9x regulates Ets-1 ubiquitination and stability to control NRAS expression and tumorigenicity in melanoma. Nat Commun 8, 14449.

Powell, D.R., Williams, J.S., Hernandez-Lagunas, L., Salcedo, E., O'Brien, J.H., and Artinger, K.B. (2015). Cdon promotes neural crest migration by regulating N-cadherin localization. Dev Biol 407, 289-299.

Poyet, C., Buser, L., Roudnicky, F., Detmar, M., Hermanns, T., Mannhard, D., Hohn, A., Ruschoff, J., Zhong, Q., Sulser, T., et al. (2015). Connexin 43 expression predicts poor progression-free survival in patients with non-muscle invasive urothelial bladder cancer. J Clin Pathol 68, 819-824.

Poynter, J.N., Elder, J.T., Fullen, D.R., Nair, R.P., Soengas, M.S., Johnson, T.M., Redman, B., Thomas, N.E., and Gruber, S.B. (2006). BRAF and NRAS mutations in melanoma and melanocytic nevi. Melanoma Res 16, 267-273.

Preusser, M., Hoischen, A., Novak, K., Czech, T., Prayer, D., Hainfellner, J.A., Baumgartner, C., Woermann, F.G., Tuxhorn, I.E., Pannek, H.W., et al. (2007). Angiocentric glioma: report of clinico-pathologic and genetic findings in 8 cases. Am J Surg 209

Pathol 31, 1709-1718.

Price, H.N., and Schaffer, J.V. (2010). Congenital melanocytic nevi-when to worry and how to treat: Facts and controversies. Clin Dermatol 28, 293-302.

Rabbani, P., Takeo, M., Chou, W., Myung, P., Bosenberg, M., Chin, L., Taketo, M.M., and Ito, M. (2011). Coordinated activation of Wnt in epithelial and melanocyte stem cells initiates pigmented hair regeneration. Cell 145, 941-955.

Richmond-Sinclair, N.M., Lee, E., Cummings, M.C., Williamson, R., Muller, K., Green, A.C., Hayward, N.K., and Whiteman, D.C. (2008). Histologic and epidemiologic correlates of P-MAPK, Brn-2, pRb, p53, and p16 immunostaining in cutaneous melanomas. Melanoma Res 18, 336-345.

Riker, A.I., Enkemann, S.A., Fodstad, O., Liu, S., Ren, S., Morris, C., Xi, Y., Howell, P., Metge, B., Samant, R.S., et al. (2008). The gene expression profiles of primary and metastatic melanoma yields a transition point of tumor progression and metastasis. BMC Med Genomics 1, 13.

Robinson, W.A., Lemon, M., Elefanty, A., Harrison-Smith, M., Markham, N., and Norris, D. (1998). Human acquired naevi are clonal. Melanoma Res 8, 499-503.

Rocha, C.R., Grazziotin, T.C., Rey, M.C., Luzzatto, L., and Bonamigo, R.R. (2013). Congenital agminated melanocytic nevus--case report. An Bras Dermatol 88, 170-172.

Rogala, A.R., Morgan, A.P., Christensen, A.M., Gooch, T.J., Bell, T.A., Miller, D.R., Godfrey, V.L., and de Villena, F.P. (2014). The Collaborative Cross as a resource for modeling human disease: CC011/Unc, a new mouse model for spontaneous colitis. Mamm Genome 25, 95-108.

Ross, A.L., Sanchez, M.I., and Grichnik, J.M. (2011). Molecular nevogenesis. Dermatol Res Pract 2011, 463184.

Ross, J.S., Slodkowska, E.A., Symmans, W.F., Pusztai, L., Ravdin, P.M., and Hortobagyi, G.N. (2009). The HER-2 receptor and breast cancer: ten years of targeted anti-HER-2 210 therapy and personalized medicine. Oncologist 14, 320-368.

Ruiz i Altaba, A. (1997). Catching a Gli-mpse of Hedgehog. Cell 90, 193-196.

Saito-Katsuragi, M., Asada, H., Niizeki, H., Katoh, F., Masuzawa, M., Tsutsumi, M., Kuniyasu, H., Ito, A., Nojima, H., and Miyagawa, S. (2007). Role for connexin 26 in metastasis of human malignant melanoma: communication between melanoma and endothelial cells via connexin 26. Cancer 110, 1162-1172.

Saldanha, G., Purnell, D., Fletcher, A., Potter, L., Gillies, A., and Pringle, J.H. (2004). High BRAF mutation frequency does not characterize all melanocytic tumor types. Int J Cancer 111, 705-710.

Sand, M., Sand, D., Brors, D., Altmeyer, P., Mann, B., and Bechara, F.G. (2008). Cutaneous lesions of the external ear. Head Face Med 4, 2.

Sandru, A., Voinea, S., Panaitescu, E., and Blidaru, A. (2014). Survival rates of patients with metastatic malignant melanoma. J Med Life 7, 572-576.

Sargen, M.R., Gormley, R.H., Pasha, T.L., Yum, S., Acs, G., Xu, X., and Zhang, P.J. (2013). Melanocytic tumors express connexin 43 but not 26: immunohistochemical analysis with potential significance in melanocytic oncogenesis. Am J Dermatopathol 35, 813-817.

Sato, N., Leopold, P.L., and Crystal, R.G. (1999). Induction of the hair growth phase in postnatal mice by localized transient expression of Sonic hedgehog. J Clin Invest 104, 855-864.

Schaffer, J.V., Orlow, S.J., Lazova, R., and Bolognia, J.L. (2001). Speckled lentiginous nevus: within the spectrum of congenital melanocytic nevi. Arch Dermatol 137, 172-178.

Scolyer, R.A., Thompson, J.F., Warnke, K., and McCarthy, S.W. (2004). Pigmented epithelioid melanocytoma. In Am J Surg Pathol (United States), pp. 1114-1115; author reply 1115-1116.

211

Scotting, P.J., Walker, D.A., and Perilongo, G. (2005). Childhood solid tumours: a developmental disorder. In Nat Rev Cancer (England), pp. 481-488.

Segretain, D., and Falk, M.M. (2004). Regulation of connexin biosynthesis, assembly, gap junction formation, and removal. Biochim Biophys Acta 1662, 3-21.

Sensi, M., Catani, M., Castellano, G., Nicolini, G., Alciato, F., Tragni, G., De Santis, G., Bersani, I., Avanzi, G., Tomassetti, A., et al. (2011). Human cutaneous melanomas lacking MITF and melanocyte differentiation antigens express a functional Axl receptor kinase. J Invest Dermatol 131, 2448-2457.

Shain, A.H., and Bastian, B.C. (2016). From melanocytes to melanomas. Nat Rev Cancer 16, 345-358.

Shakhova, O., Zingg, D., Schaefer, S.M., Hari, L., Civenni, G., Blunschi, J., Claudinot, S., Okoniewski, M., Beermann, F., Mihic-Probst, D., et al. (2012). Sox10 promotes the formation and maintenance of giant congenital naevi and melanoma. Nat Cell Biol 14, 882- 890.

Sharma, N., Ho, S., Bing, T.K., McCormack, C., Scolyer, R., and Lee, J. (2015). Eruptive disseminated Spitz naevus (EDSN) in a young girl of Indian origin. Australas J Dermatol 56, 134-136.

Shen, C.H., Kim, S.H., Trousil, S., Frederick, D.T., Piris, A., Yuan, P., Cai, L., Gu, L., Li, M., Lee, J.H., et al. (2016). Loss of cohesin complex components STAG2 or STAG3 confers resistance to BRAF inhibition in melanoma. Nat Med 22, 1056-1061.

Shiga, T., Nakajima, K., Tarutani, M., Izumi, M., Tanaka, M., and Sano, S. (2012). Blue nevus with a starburst pattern on dermoscopy. Dermatol Pract Concept 2, 204a207. Shin, J., and Kim, Y.C. (2013). Multiple agminated acquired melanocytic nevi. Ann Dermatol 25, 251-252.

Shojaee, S., Caeser, R., Buchner, M., Park, E., Swaminathan, S., Hurtz, C., Geng, H., Chan, L.N., Klemm, L., Hofmann, W.K., et al. (2015). Erk Negative Feedback Control Enables Pre-B Cell Transformation and Represents a Therapeutic Target in Acute 212

Lymphoblastic Leukemia. Cancer Cell 28, 114-128.

Silva, J.H., Sa, B.C., Avila, A.L., Landman, G., and Duprat Neto, J.P. (2011). Atypical mole syndrome and dysplastic nevi: identification of populations at risk for developing melanoma - review article. Clinics (Sao Paulo) 66, 493-499.

Skender-Kalnenas, T.M., English, D.R., and Heenan, P.J. (1995). Benign melanocytic lesions: risk markers or precursors of cutaneous melanoma? J Am Acad Dermatol 33, 1000-1007.

Skvara, H., Kalthoff, F., Meingassner, J.G., Wolff-Winiski, B., Aschauer, H., Kelleher, J.F., Wu, X., Pan, S., Mickel, L., Schuster, C., et al. (2011). Topical treatment of Basal cell carcinomas in nevoid Basal cell carcinoma syndrome with a smoothened inhibitor. J Invest Dermatol 131, 1735-1744. Smalley, K.S., Xiao, M., Villanueva, J., Nguyen, T.K., Flaherty, K.T., Letrero, R., Van Belle, P., Elder, D.E., Wang, Y., Nathanson, K.L., et al. (2009). CRAF inhibition induces apoptosis in melanoma cells with non-V600E BRAF mutations. Oncogene 28, 85-94.

Soares, A.R., Martins-Marques, T., Ribeiro-Rodrigues, T., Ferreira, J.V., Catarino, S., Pinho, M.J., Zuzarte, M., Isabel Anjo, S., Manadas, B., J, P.G.S., et al. (2015). Gap junctional protein Cx43 is involved in the communication between extracellular vesicles and mammalian cells. Sci Rep 5, 13243.

Socrier, Y., Lauwers-Cances, V., Lamant, L., Garrido, I., Lauwers, F., Lopez, R., Rochaix, P., Chevreau, C., Payoux, P., Viraben, R., et al. (2010). Histological regression in primary melanoma: not a predictor of sentinel lymph node metastasis in a cohort of 397 patients. Br J Dermatol 162, 830-834.

Sowa, J., Kobayashi, H., Ishii, M., and Kimura, T. (2008). Histopathologic findings in Unna's nevus suggest it is a tardive congenital nevus. Am J Dermatopathol 30, 561-566.

Steinhart, Z., Pavlovic, Z., Chandrashekhar, M., Hart, T., Wang, X., Zhang, X., Robitaille, M., Brown, K.R., Jaksani, S., Overmeer, R., et al. (2017). Genome-wide CRISPR screens reveal a Wnt-FZD5 signaling circuit as a druggable vulnerability of RNF43-mutant pancreatic tumors. Nat Med 23, 60-68. 213

Stoletov, K., Strnadel, J., Zardouzian, E., Momiyama, M., Park, F.D., Kelber, J.A., Pizzo, D.P., Hoffman, R., VandenBerg, S.R., and Klemke, R.L. (2013). Role of connexins in metastatic breast cancer and melanoma brain colonization. J Cell Sci 126, 904-913.

Subak-Sharpe, H., Burk, R.R., and Pitts, J.D. (1969). Metabolic co-operation between biochemically marked mammalian cells in tissue culture. J Cell Sci 4, 353-367.

Suram, A., Kaplunov, J., Patel, P.L., Ruan, H., Cerutti, A., Boccardi, V., Fumagalli, M., Di Micco, R., Mirani, N., Gurung, R.L., et al. (2012). Oncogene-induced telomere dysfunction enforces cellular senescence in human cancer precursor lesions. Embo j 31, 2839-2851.

Taipale, J., Cooper, M.K., Maiti, T., and Beachy, P.A. (2002). Patched acts catalytically to suppress the activity of Smoothened. Nature 418, 892-897.

Takeo, M., Lee, W., Rabbani, P., Sun, Q., Hu, H., Lim, C.H., Manga, P., and Ito, M. (2016). EdnrB Governs Regenerative Response of Melanocyte Stem Cells by Crosstalk with Wnt Signaling. Cell Rep 15, 1291-1302.

Tanaka, T., Kimura, M., Ishiguro, H., Mizoguchi, K., and Takeyama, H. (2016). Connexin 43 expression is associated with poor survival in patients with esophageal squamous cell carcinoma. Mol Clin Oncol 4, 989-993.

Tang, J.Y., Aszterbaum, M., Athar, M., Barsanti, F., Cappola, C., Estevez, N., Hebert, J., Hwang, J., Khaimskiy, Y., Kim, A., et al. (2010). Basal cell carcinoma chemoprevention with nonsteroidal anti-inflammatory drugs in genetically predisposed PTCH1+/- humans and mice. Cancer Prev Res (Phila) 3, 25-34.

Tang, T., Tang, J.Y., Li, D., Reich, M., Callahan, C.A., Fu, L., Yauch, R.L., Wang, F., Kotkow, K., Chang, K.S., et al. (2011). Targeting superficial or nodular Basal cell carcinoma with topically formulated small molecule inhibitor of smoothened. Clin Cancer Res 17, 3378-3387.

Taube, J.H., Herschkowitz, J.I., Komurov, K., Zhou, A.Y., Gupta, S., Yang, J., Hartwell, K., Onder, T.T., Gupta, P.B., Evans, K.W., et al. (2010). Core epithelial-to-mesenchymal 214 transition interactome gene-expression signature is associated with claudin-low and metaplastic breast cancer subtypes. Proc Natl Acad Sci U S A 107, 15449-15454.

Terai, M., Eto, M., Young, G.D., Berd, D., Mastrangelo, M.J., Tamura, Y., Harigaya, K., and Sato, T. (2012). Interleukin 6 mediates production of interleukin 10 in metastatic melanoma. Cancer Immunol Immunother 61, 145-155. Thiele, S., Rauner, M., Goettsch, C., Rachner, T.D., Benad, P., Fuessel, S., Erdmann, K., Hamann, C., Baretton, G.B., Wirth, M.P., et al. (2011). Expression profile of WNT molecules in prostate cancer and its regulation by aminobisphosphonates. J Cell Biochem 112, 1593-1600.

Thomas, N.E., Edmiston, S.N., Alexander, A., Millikan, R.C., Groben, P.A., Hao, H., Tolbert, D., Berwick, M., Busam, K., Begg, C.B., et al. (2007). Number of nevi and early- life ambient UV exposure are associated with BRAF-mutant melanoma. Cancer Epidemiol Biomarkers Prev 16, 991-997.

Tittarelli, A., Guerrero, I., Tempio, F., Gleisner, M.A., Avalos, I., Sabanegh, S., Ortiz, C., Michea, L., Lopez, M.N., Mendoza-Naranjo, A., et al. (2015). Overexpression of connexin 43 reduces melanoma proliferative and metastatic capacity. Br J Cancer 113, 259-267.

Tonks, N.K. (2005). Redox redux: revisiting PTPs and the control of cell signaling. Cell 121, 667-670.

Torlakovic, E.E., Bilalovic, N., Nesland, J.M., Torlakovic, G., and Florenes, V.A. (2004). Ets-1 transcription factor is widely expressed in benign and malignant melanocytes and its expression has no significant association with prognosis. Mod Pathol 17, 1400-1406.

Tsao, H., Bevona, C., Goggins, W., and Quinn, T. (2003). The transformation rate of moles (melanocytic nevi) into cutaneous melanoma: a population-based estimate. Arch Dermatol 139, 282-288.

Tucker, M.A., Fraser, M.C., Goldstein, A.M., Struewing, J.P., King, M.A., Crawford, J.T., Chiazze, E.A., Zametkin, D.P., Fontaine, L.S., and Clark, W.H., Jr. (2002). A natural history of melanomas and dysplastic nevi: an atlas of lesions in melanoma-prone families. Cancer 94, 3192-3209. 215

Tucker, M.A., Halpern, A., Holly, E.A., Hartge, P., Elder, D.E., Sagebiel, R.W., Guerry, D.t., and Clark, W.H., Jr. (1997). Clinically recognized dysplastic nevi. A central risk factor for cutaneous melanoma. Jama 277, 1439-1444.

Udayakumar, D., and Tsao, H. (2009). Melanoma genetics: an update on risk-associated genes. Hematol Oncol Clin North Am 23, 415-429, vii.

Uribe, P., Andrade, L., and Gonzalez, S. (2006). Lack of association between BRAF mutation and MAPK ERK activation in melanocytic nevi. J Invest Dermatol 126, 161-166.

Urtatiz, O., and Van Raamsdonk, C.D. (2016). Gnaq and Gna11 in the Endothelin Signaling Pathway and Melanoma. Front Genet 7, 59.

Van der Ploeg, I.M., Valdes Olmos, R.A., Kroon, B.B., and Nieweg, O.E. (2008). Tumor- positive sentinel node biopsy of the groin in clinically node-negative melanoma patients: superficial or superficial and deep lymph node dissection? Ann Surg Oncol 15, 1485-1491.

Van Es, S.L., Colman, M., Thompson, J.F., McCarthy, S.W., and Scolyer, R.A. (2008). Angiotropism is an independent predictor of local recurrence and in-transit metastasis in primary cutaneous melanoma. Am J Surg Pathol 32, 1396-1403.

Venesio, T., Chiorino, G., Balsamo, A., Zaccagna, A., Petti, C., Scatolini, M., Pisacane, A., Sarotto, I., Picciotto, F., and Risio, M. (2008). In melanocytic lesions the fraction of BRAF V600E alleles is associated with sun exposure but unrelated to ERK phosphorylation. Mod Pathol 21, 716-726.

Viana, A.C., Gontijo, B., and Bittencourt, F.V. (2013). Giant congenital melanocytic nevus. An Bras Dermatol 88, 863-878.

Villares, G.J., Dobroff, A.S., Wang, H., Zigler, M., Melnikova, V.O., Huang, L., and Bar-Eli, M. (2009). Overexpression of protease-activated receptor-1 contributes to melanoma metastasis via regulation of connexin 43. Cancer Res 69, 6730-6737.

Viros, A., Sanchez-Laorden, B., Pedersen, M., Furney, S.J., Rae, J., Hogan, K., Ejiama, 216

S., Girotti, M.R., Cook, M., Dhomen, N., et al. (2014). Ultraviolet radiation accelerates BRAF-driven melanomagenesis by targeting TP53. Nature 511, 478-482.

Vogelstein, B., Lane, D., and Levine, A.J. (2000). Surfing the p53 network. Nature 408, 307-310.

Vousden, K.H., and Prives, C. (2009). Blinded by the Light: The Growing Complexity of p53. Cell 137, 413-431.

Vredeveld, L.C., Possik, P.A., Smit, M.A., Meissl, K., Michaloglou, C., Horlings, H.M., Ajouaou, A., Kortman, P.C., Dankort, D., McMahon, M., et al. (2012). Abrogation of BRAFV600E-induced senescence by PI3K pathway activation contributes to melanomagenesis. Genes Dev 26, 1055-1069.

Vultur, A., and Herlyn, M. (2013). SnapShot: melanoma. Cancer Cell 23, 706-706.e701. Wade, M., and Wahl, G.M. (2009). Targeting Mdm2 and Mdmx in cancer therapy: better living through medicinal chemistry? Mol Cancer Res 7, 1-11.

Walker, G.J., Soyer, H.P., Terzian, T., and Box, N.F. (2011). Modelling melanoma in mice. Pigment Cell Melanoma Res 24, 1158-1176.

Webber, S.A., Siller, G., and Soyer, H.P. (2011). Pigmented spindle cell naevus of Reed: a controversial diagnostic entity in Australia. Australas J Dermatol 52, 104-108. Weeraratna, A.T., Jiang, Y., Hostetter, G., Rosenblatt, K., Duray, P., Bittner, M., and Trent, J.M. (2002). Wnt5a signaling directly affects cell motility and invasion of metastatic melanoma. Cancer Cell 1, 279-288.

Wellner, U., Schubert, J., Burk, U.C., Schmalhofer, O., Zhu, F., Sonntag, A., Waldvogel, B., Vannier, C., Darling, D., zur Hausen, A., et al. (2009). The EMT-activator ZEB1 promotes tumorigenicity by repressing stemness-inhibiting microRNAs. Nat Cell Biol 11, 1487-1495. Welsh, C.E., Miller, D.R., Manly, K.F., Wang, J., McMillan, L., Morahan, G., Mott, R., Iraqi, F.A., Threadgill, D.W., and de Villena, F.P. (2012). Status and access to the Collaborative Cross population. Mamm Genome 23, 706-712.

217

White, T.W., Bruzzone, R., Goodenough, D.A., and Paul, D.L. (1994a). Voltage gating of connexins. Nature 371, 208-209.

White, T.W., Bruzzone, R., Wolfram, S., Paul, D.L., and Goodenough, D.A. (1994b). Selective interactions among the multiple connexin proteins expressed in the vertebrate lens: the second extracellular domain is a determinant of compatibility between connexins. J Cell Biol 125, 879-892.

Williams, C.S., Zhang, B., Smith, J.J., Jayagopal, A., Barrett, C.W., Pino, C., Russ, P., Presley, S.H., Peng, D., Rosenblatt, D.O., et al. (2011). BVES regulates EMT in human corneal and colon cancer cells and is silenced via promoter methylation in human colorectal carcinoma. J Clin Invest 121, 4056-4069.

Wilmott, J., Haydu, L., Bagot, M., Zhang, Y., Jakrot, V., McCarthy, S., Lugassy, C., Thompson, J., Scolyer, R., and Barnhill, R. (2012). Angiotropism is an independent predictor of microscopic satellites in primary cutaneous melanoma. Histopathology 61, 889-898.

Wilsher, M.J., McCarthy, S.W., and Scolyer, R.A. (2013). Prominent angiotropism in a small atypical spitzoid dysplastic melanocytic naevus: what is its significance? Pathology 45, 701-703.

Wintermeier, K., von Poblotzki, M., Genzel-Boroviczeny, O., Vogel, S., Schotten, K., Berking, C., and Giehl, K.A. (2014). Neonatal blue light phototherapy increases cafe-au-lait macules in preschool children. Eur J Pediatr 173, 1519-1525.

Witt, C., and Krengel, S. (2010). Clinical and epidemiological aspects of subtypes of melanocytic nevi (Flat nevi, Miescher nevi, Unna nevi). Dermatol Online J 16, 1.

Wojnowski, L., Stancato, L.F., Larner, A.C., Rapp, U.R., and Zimmer, A. (2000). Overlapping and specific functions of Braf and Craf-1 proto-oncogenes during mouse embryogenesis. Mech Dev 91, 97-104.

Wong, V.C., Chen, H., Ko, J.M., Chan, K.W., Chan, Y.P., Law, S., Chua, D., Kwong, D.L., Lung, H.L., Srivastava, G., et al. (2012). Tumor suppressor dual-specificity phosphatase 6 218

(DUSP6) impairs cell invasion and epithelial-mesenchymal transition (EMT)-associated phenotype. Int J Cancer 130, 83-95.

Wu, J., Rosenbaum, E., Begum, S., and Westra, W.H. (2007). Distribution of BRAF T1799A(V600E) mutations across various types of benign nevi: implications for melanocytic tumorigenesis. Am J Dermatopathol 29, 534-537.

Wurm, E.M., Lin, L.L., Ferguson, B., Lambie, D., Prow, T.W., Walker, G.J., and Soyer, H.P. (2012a). A blueprint for staging of murine melanocytic lesions based on the Cdk4 ( R24C/R24C ) ::Tyr- NRAS ( Q ) ( 61K ) model. Exp Dermatol 21, 676-681.

Wurm, E.M., Lin, L.L., Ferguson, B., Lambie, D., Prow, T.W., Walker, G.J., and Soyer, H.P. (2012b). A blueprint for staging of murine melanocytic lesions based on the Cdk4 ( R24C/R24C ) ::Tyr- NRAS ( Q ) ( 61K ) model. Exp Dermatol 21, 676-681. Xu, S., Furukawa, T., Kanai, N., Sunamura, M., and Horii, A. (2005). Abrogation of DUSP6 by hypermethylation in human pancreatic cancer. J Hum Genet 50, 159-167.

Yang, S.H., Andl, T., Grachtchouk, V., Wang, A., Liu, J., Syu, L.J., Ferris, J., Wang, T.S., Glick, A.B., Millar, S.E., et al. (2008). Pathological responses to oncogenic Hedgehog signaling in skin are dependent on canonical Wnt/beta3-catenin signaling. Nat Genet 40, 1130-1135.

Yeh, I., and Bastian, B.C. (2009). Genome-wide associations studies for melanoma and nevi. Pigment Cell Melanoma Res 22, 527-528.

Yoshimura, T., Howard, O.M., Ito, T., Kuwabara, M., Matsukawa, A., Chen, K., Liu, Y., Liu, M., Oppenheim, J.J., and Wang, J.M. (2013). Monocyte chemoattractant protein-1/CCL2 produced by stromal cells promotes lung metastasis of 4T1 murine breast cancer cells. PLoS One 8, e58791.

Young, A., Ngiow, S.F., Madore, J., Reinhardt, J., Landsberg, J., Chitsazan, A., Rautela, J., Bald, T., Barkauskas, D.S., Ahern, E., et al. (2017). Targeting Adenosine in BRAF- Mutant Melanoma Reduces Tumor Growth and Metastasis. Cancer Res 77, 4684-4696.

Youssef, K.K., Lapouge, G., Bouvree, K., Rorive, S., Brohee, S., Appelstein, O., 219

Larsimont, J.C., Sukumaran, V., Van de Sande, B., Pucci, D., et al. (2012). Adult interfollicular tumour-initiating cells are reprogrammed into an embryonic hair follicle progenitor-like fate during basal cell carcinoma initiation. Nat Cell Biol 14, 1282-1294.

Zaal, L.H., Mooi, W.J., Sillevis Smitt, J.H., and van der Horst, C.M. (2004). Classification of congenital melanocytic naevi and malignant transformation: a review of the literature. Br J Plast Surg 57, 707-719.

Zalaudek, I., Docimo, G., and Argenziano, G. (2009). Using dermoscopic criteria and patient-related factors for the management of pigmented melanocytic nevi. Arch Dermatol 145, 816-826.

Zembowicz, A., Carney, J.A., and Mihm, M.C. (2004). Pigmented epithelioid melanocytoma: a low-grade melanocytic tumor with metastatic potential indistinguishable from animal-type melanoma and epithelioid blue nevus. Am J Surg Pathol 28, 31-40.

Zembowicz, A., Knoepp, S.M., Bei, T., Stergiopoulos, S., Eng, C., Mihm, M.C., and Stratakis, C.A. (2007). Loss of expression of protein kinase a regulatory subunit 1alpha in pigmented epithelioid melanocytoma but not in melanoma or other melanocytic lesions. Am J Surg Pathol 31, 1764-1775.

Zembowicz, A., and Scolyer, R.A. (2011). Nevus/Melanocytoma/Melanoma: an emerging paradigm for classification of melanocytic neoplasms? Arch Pathol Lab Med 135, 300-306.

Zhang, P., Liu, W., Yuan, X., Li, D., Gu, W., and Gao, T. (2013). Endothelin-1 enhances the melanogenesis via MITF-GPNMB pathway. BMB Rep 46, 364-369.

Zhang, Q., Peng, C., Song, J., Zhang, Y., Chen, J., Song, Z., Shou, X., Ma, Z., Peng, H., Jian, X., et al. (2017). Germline Mutations in CDH23, Encoding Cadherin-Related 23, Are Associated with Both Familial and Sporadic Pituitary Adenomas. Am J Hum Genet 100, 817-823.

Zhang, Z., Kobayashi, S., Borczuk, A.C., Leidner, R.S., Laframboise, T., Levine, A.D., and Halmos, B. (2010). Dual specificity phosphatase 6 (DUSP6) is an ETS-regulated negative feedback mediator of oncogenic ERK signaling in lung cancer cells. Carcinogenesis 31, 220

577-586.

Zhang, Y., Ye, Y., Shen, D., Jiang, K., Zhang, H., Sun, W., Zhang, J., Xu, F., Cui, Z., and Wang, S. (2010). Identification of transgelin-2 as a biomarker of colorectal cancer by laser capture microdissection and quantitative proteome analysis. Cancer Sci 101, 523-529.

Zheng, X., Carstens, J.L., Kim, J., Scheible, M., Kaye, J., Sugimoto, H., Wu, C.C., LeBleu, V.S., and Kalluri, R. (2015). Epithelial-to-mesenchymal transition is dispensable for metastasis but induces chemoresistance in pancreatic cancer. Nature 527, 525-530.

Zhou, B., Ritt, D.A., Morrison, D.K., Der, C.J., and Cox, A.D. (2016). Protein Kinase CK2alpha Maintains Extracellular Signal-regulated Kinase (ERK) Activity in a CK2alpha Kinase-independent Manner to Promote Resistance to Inhibitors of RAF and MEK but Not ERK in BRAF Mutant Melanoma. J Biol Chem 291, 17804-17815.

Zucker, S.N., Bancroft, T.A., Place, D.E., Des Soye, B., Bagati, A., and Berezney, R. (2013). A dominant negative Cx43 mutant differentially affects tumorigenic and invasive properties in human metastatic melanoma cells. J Cell Physiol 228, 853-859.

Zuo, L., Weger, J., Yang, Q., Goldstein, A.M., Tucker, M.A., Walker, G.J., Hayward, N., and Dracopoli, N.C. (1996). Germline mutations in the p16INK4a binding domain of CDK4 in familial melanoma. Nat Genet 12, 97-99.

221

Appendix 1: List of genes positively (red) and negatively (yellow) correlated with Gja1 expression

Gja1 Gja1 Gja1 1 Gjb2 0.86 Fzd5 0.95 Dbf4 0.86 Dsg4 0.94 Tgm6 0.86 Proser2 0.93 Zdhhc13 0.85 Cryba4 0.93 Krt35 0.85 Prr9 0.93 Shcbp1 0.85 Elf5 0.93 Gadl1 0.84 Tyr 0.93 Krt32 0.84 Krt82 0.93 Msx2 0.84 Lap3 0.92 Knstrn 0.84 Krtap26- 1 0.91 Krt34 0.84 Ggt1 0.91 Nek2 0.84 Fam26d 0.91 Etv4 0.84 Stra6 0.91 Rnf149 0.83 S100a7a 0.91 Dok4 0.83 Cpm 0.9 Cenpe 0.83 Sptssb 0.9 Dct 0.83 Golga7b 0.9 Krtap3-3 0.83 Mlana 0.9 Kcnh1 0.83 Slc7a5 0.89 Clspn 0.83 Vsig8 0.89 Chn2 0.83 Gprc5d 0.89 Cdk1 0.82 Slc7a8 0.89 Marcksl1 0.82 Ppp1r3b 0.89 Krtap11-1 0.82 Lrrc75b 0.89 Specc1 0.82 Fbp1 0.88 Krt31 0.82 Krt27 0.88 Spag5 0.82 Bdh1 0.88 Slc39a14 0.82 Slc16a7 0.88 Bub1 0.82 Unc5b 0.88 Ret 0.82 Slc39a8 0.87 Dcbld1 0.81 Krtap13- 1 0.87 Gpx2 0.81 Foxe1 0.87 Krt73 0.81 Krt83 0.87 Lig1 0.81 Ero1l 0.87 Krtap22-2 0.81 Fam49a 0.87 Cep55 0.81 Foxn1 0.86 Tmem229b 0.81 Dusp2 0.86 Tyrp1 0.81 Gja1 Gja1 222

Dtnb 0.81 Slc7a1 0.77 Gabrp 0.8 Kit 0.77 Bora 0.8 Fblim1 0.77 Nfe2l3 0.8 Aurka 0.76 Rasl11b 0.8 Arg2 0.76 Tspan6 0.8 Edn1 0.76 Krt28 0.8 Zdhhc21 0.76 Ckap2l 0.8 Mcm2 0.76 Tpx2 0.8 Sprr1a 0.76 Gnmt 0.8 Lurap1l 0.76 Hells 0.8 Arid3a 0.76 Rab3ip 0.8 Krtap4-7 0.76 Cdc25c 0.8 Crot 0.76 Col12a1 0.8 Ercc6l 0.76 Dera 0.79 Tnc 0.76 Ctse 0.79 Arl15 0.76 Spc25 0.79 Pitrm1 0.76 Cdca2 0.79 Rnf222 0.75 Krtap9-3 0.79 Nup210 0.75 Gclc 0.79 Pbk 0.75 Krt33b 0.79 Ror2 0.75 Bub1b 0.79 Slc45a2 0.75 Wif1 0.79 Kif20b 0.75 Gjb6 0.78 E2f8 0.75 Otub2 0.78 Mdga1 0.75 Birc5 0.78 Slc35g1 0.74 Cdc20 0.78 Spint1 0.74 Msx1 0.78 Mthfd2 0.74 Syce2 0.78 Rcc2 0.74 Shh 0.78 Smox 0.74 Krtap7-1 0.78 Kremen2 0.74 Gm10036 0.78 Gtse1 0.74 Rad51 0.78 Tgfa 0.74 Smc2 0.78 Ccser1 0.74 Mxd3 0.78 Rfc4 0.74 Wnt5a 0.78 Tom1l1 0.74 Gpnmb 0.78 Diaph3 0.74 Irx4 0.78 Kif20a 0.74 Fam46b 0.77 Alx4 0.74 Mis18bp1 0.77 Krt81 0.74 Neo1 0.77 Krt86 0.74

Gja1 Gja1

223

Tgm3 0.74 Espl1 0.68 Kif26a 0.74 Plk2 0.68 Prss53 0.73 Sox13 0.68 Krt72 0.73 Evl 0.68 Nop56 0.73 Traf4 0.68 Mtbp 0.73 Cep164 0.68 Basp1 0.73 Erc1 0.67 Bnc2 0.73 Tle3 0.67 Rad18 0.73 Ephb3 0.67 Hspa14 0.72 Gsdma 0.67 Nusap1 0.72 Dusp6 0.67 Nkd2 0.72 Dhfr 0.67 Cdk4 0.72 Pitpnm3 0.67 E2f2 0.72 Dnmt1 0.67 Cdc45 0.72 Rbp1 0.67 Psrc1 0.72 Cntfr 0.67 Igsf3 0.72 Lbr 0.67 Kif7 0.72 Ptch1 0.67 Slc6a19 0.71 Zfp553 0.67 Nfil3 0.71 Chaf1a 0.67 Kif23 0.71 Atad5 0.67 Epha2 0.7 Isyna1 0.67 H2afx 0.7 Rhpn2 0.66 Mcm4 0.7 Zfat 0.66 Cyp2s1 0.7 Shb 0.66 Prr11 0.7 Asf1b 0.66 Sema6a 0.7 Nkrf 0.66 Vdr 0.7 Cdca7 0.66 Ctns 0.69 Ppp2r1b 0.66 Tdp1 0.69 Mark1 0.66 Gli3 0.69 Hmmr 0.66 Plekhg1 0.69 Esco2 0.66 Pxylp1 0.69 Polr1e 0.65 Pola2 0.69 Pprc1 0.65 Mtap 0.69 Lyar 0.65 Serinc5 0.69 Mms22l 0.65 Dut 0.68 Trpv3 0.65 Mcm3 0.68 Kifc1 0.65 Spdl1 0.68 Atad2 0.65 Plxna1 0.68 Myo10 0.65

224

Gja1 Gja1 Cep97 0.65 Celsr2 0.62 Enc1 0.65 Ccnb2 0.62 Ccnd1 0.65 St6gal1 0.62 Pappa 0.65 Anp32b 0.61 Nedd9 0.65 Rnaseh2b 0.61 Pop1 0.64 Cenpi 0.61 Tmeff1 0.64 L3mbtl2 0.61 Ranbp1 0.64 Nup43 0.61 Noc2l 0.64 Eaf1 0.61 Ift74 0.64 Thoc6 0.61 Map3k5 0.64 Nmral1 0.61 Mad2l1 0.64 Plxnb1 0.61 Rab15 0.64 Hirip3 0.61 Car12 0.64 Ccnd2 0.61 Whsc1 0.64 Rab34 0.61 Lrp12 0.64 S1pr3 0.61 Adamts9 0.64 Smtn 0.61 Lin9 0.63 Gdf10 0.61 Ifitm10 0.63 Bach2 0.61 Gmnn 0.63 Tril 0.61 Klrg2 0.63 Ppp2r3a -0.61 Ndc1 0.63 Scin -0.61 Cdca4 0.63 Fgf11 -0.61 Slc44a1 0.63 Dusp13 -0.62 Tcf19 0.63 Ahr -0.64 Pdcd4 0.63 Wdr41 -0.65 Tjp2 0.63 Setd3 -0.68 Mcm5 0.63 Hspb7 -0.68 Braf 0.63 Bves -0.71 Slc16a11 0.63 Lhx2 0.63 Smarca4 0.62 Smn1 0.62 Mettl8 0.62 Ddx39 0.62 Rab44 0.62 Tipin 0.62 Ammecr1 0.62 Rhno1 0.62 Dctpp1 0.62 Mocos 0.62

225

Appendix 2: Top 10 genes correlated with candidate genes from angiotropism linkage

226

227

228

229

230

231

Appendix 3: List of gene differentaly expressed between S3 and normal tail skin

Symbol p Regulation FC Symbol p Regulation FC Metrn 7.58E-06 up 53.40093 Gbp3 0.002109 up 14.57312 Syt4 3.39E-06 up 39.45795 Hexa 2.36E-04 up 14.5372 C4a 0.004386 up 39.44178 LOC380706 4.29E-04 up 14.45502 Plp1 4.23E-08 up 36.38424 St6gal1 4.92E-08 up 14.27336 ctsz 6.96E-05 up 36.24376 Rhoc 5.12E-04 up 13.96646 Vim 1.38E-06 up 34.12374 Lyz2 1.76E-04 up 13.92605 Tmem14c 6.50E-06 up 32.8507 Syngr1 1.03E-04 up 13.79572 Nppb 6.59E-04 up 31.94804 Igf2bp2 5.26E-04 up 13.17086 C4a 0.003114 up 29.59298 2310040A07Rik0.006514 up 13.15387 Sox10 1.23E-07 up 29.13505 Cd63 1.03E-04 up 13.08937 Slc45a2 7.57E-04 up 28.8962 Sri 4.31E-05 up 12.95427 Trf 1.42E-06 up 28.82491 Ets1 4.31E-04 up 12.71202 E330034G19Rik1.59E-04 up 26.55375 Anxa6 2.37E-04 up 12.60618 Slc45a2 8.20E-04 up 25.54596 Gpr137b 1.49E-07 up 12.55539 Phlda1 1.45E-05 up 25.35037 Lgals3bp 0.002816 up 12.45218 Lhfpl2 2.81E-05 up 24.14974 Tspan4 5.08E-05 up 12.41375 Gstp1 2.57E-06 up 23.66661 Tyr 2.98E-04 up 12.31767 Fez1 7.08E-05 up 23.54545 LOC1000468831.86E-04 up 12.19021 Insc 8.99E-06 up 23.41725 Tmod1 3.58E-06 up 12.1823 Megf10 8.56E-08 up 22.91981 Gpnmb 9.37E-06 up 12.04973 LOC1000485544.08E-04 up 21.97288 Nrcam 3.76E-04 up 11.73007 Ifit3 0.007715 up 20.95441 Dpysl2 8.53E-08 up 11.66019 C2 3.14E-05 up 20.91254 Lbh 8.91E-04 up 11.60907 Megf10 1.15E-07 up 20.64805 Ube1l 0.007644 up 11.49753 Fez1 6.37E-05 up 20.50714 Dkk3 6.07E-04 up 11.45916 Mc1r 0.005062 up 19.93181 Cyba 0.003939 up 11.35533 Timp1 0.001701 up 19.36412 LOC1000469533.59E-05 up 11.14357 C1qb 7.82E-06 up 19.33357 Eng 1.62E-04 up 11.14255 Emp3 1.13E-04 up 18.3658 Mcam 0.002023 up 11.04342 Timp1 0.001415 up 18.24675 Tmem176b 6.76E-05 up 11.0222 Aadacl1 5.62E-07 up 17.91425 Armcx2 6.92E-04 up 10.91515 Emp3 2.44E-07 up 17.7067 Olfm1 0.002229 up 10.86113 Dusp4 4.81E-04 up 16.93547 2210010N04Rik8.32E-05 up 10.81521 Aebp1 1.86E-06 up 16.70963 Cugbp2 4.58E-07 up 10.65763 Man2b1 2.79E-05 up 16.45113 Plekhb1 0.001178 up 10.64552 LOC383981 5.21E-09 up 16.39825 Cd97 7.15E-05 up 10.64257 Gbp2 0.003085 up 16.07947 Bst2 0.003714 up 10.62123 Syngr1 1.25E-05 up 15.84899 Ctsz 3.99E-04 up 10.57499 Gbp3 9.37E-04 up 15.77024 Dctn4 2.27E-05 up 10.50772 Lgals1 7.28E-07 up 15.68164 C1qc 6.93E-05 up 10.50147 Ela1 8.11E-04 up 15.53233 LOC385699 7.97E-05 up 10.43302 Gstp1 6.04E-06 up 15.24677 Kcnn4 0.001312 up 10.42506 2310040A07Rik0.007065 up 15.17902 Iigp2 0.005218 up 10.37634 Cep170 2.98E-05 up 14.91736 2310014H01Rik1.38E-07 up 10.34196 Mgat4b 3.41E-08 up 14.85566 Lass2 9.92E-07 up 10.29307 Cd97 4.99E-05 up 10.26537

232

Symbol p Regulation FC Symbol p Regulation FC Fermt2 9.77E-05 up 10.23331 Dusp4 4.20E-05 up 8.04391 Dennd5a 3.21E-05 up 10.19302 5031436O03Rik3.89E-05 up 8.006043 Grb10 3.12E-04 up 9.98502 Sdpr 0.003495 up 7.982078 Tuba1a 2.73E-04 up 9.981905 LOC1000453434.96E-04 up 7.916644 Bace2 0.001839 up 9.958223 Calu 0.004825 up 7.881936 Mgp 0.001067 up 9.906899 Syde1 6.80E-05 up 7.854989 Sepp1 0.001443 up 9.870643 Vegfb 1.13E-05 up 7.69777 Gpnmb 0.00281 up 9.762015 Dusp6 1.98E-04 up 7.622442 Mcoln3 1.03E-05 up 9.749008 H2-D1 8.40E-04 up 7.558251 Renbp 1.68E-05 up 9.67749 Snx10 4.52E-04 up 7.547583 Ccnd1 8.68E-04 up 9.6745 2310047A01Rik7.58E-04 up 7.449744 Dkk3 3.97E-04 up 9.642129 Calu 0.002463 up 7.438108 Specc1 0.005331 up 9.615511 Ifitm1 0.004263 up 7.401457 St3gal6 8.89E-05 up 9.54871 2610200O14Rik3.84E-04 up 7.381403 Ctsz 9.18E-04 up 9.539115 Olfm1 4.11E-06 up 7.337728 Fxyd5 6.02E-04 up 9.441154 Gpx7 4.32E-06 up 7.317973 Epdr1 1.50E-04 up 9.40813 Tmsb10 2.76E-04 up 7.265918 Lyz 1.08E-06 up 9.387224 Plekho2 5.76E-04 up 7.217587 Hbb-b1 0.001878 up 9.385243 H2-D1 0.001116 up 7.142329 LOC674135 0.002787 up 9.329122 Galk1 5.65E-05 up 7.07013 Fxc1 0.001973 up 9.247446 Dusp6 0.001215 up 7.054006 Lbh 2.56E-05 up 9.163786 Trem2 0.002631 up 7.052371 Sytl2 9.68E-04 up 9.119543 Mcam 0.007053 up 6.892709 6720469N11Rik0.00112 up 9.104646 Hbb-b2 0.00624 up 6.85074 Scly 9.85E-05 up 9.087372 Ppm1f 0.004445 up 6.811772 Ms4a7 0.001334 up 8.954047 Acot7 0.002605 up 6.806245 Slc45a2 0.001873 up 8.94855 Ddit3 0.00334 up 6.729267 LOC383099 3.55E-07 up 8.888801 Cotl1 0.001805 up 6.70259 Ifi27 0.005205 up 8.790092 Gpnmb 7.59E-06 up 6.665006 Gusb 3.02E-04 up 8.789815 H2-K1 0.002552 up 6.63826 Samd9l 7.30E-04 up 8.760407 2610027C15Rik0.001087 up 6.626056 Eno2 1.21E-06 up 8.618735 Fscn1 0.006608 up 6.612679 Sdcbp 4.54E-06 up 8.533064 Cdk2 6.90E-04 up 6.560614 Tspan10 0.005127 up 8.531222 Fxyd5 0.002948 up 6.546064 EG630499 0.002406 up 8.435089 Csrp2 0.008061 up 6.515853 Uap1l1 3.18E-04 up 8.419153 Tmem176a 8.09E-04 up 6.442013 Enpp5 1.38E-04 up 8.326103 Cryab 2.45E-05 up 6.414673 G0s2 3.95E-04 up 8.267694 Ttc7b 0.001646 up 6.40431 St3gal5 0.003042 up 8.264798 Bin1 1.83E-04 up 6.39506 Sema3b 3.77E-04 up 8.256878 Cd44 1.44E-04 up 6.372951 Cpe 1.18E-04 up 8.242963 Lass2 1.42E-06 up 6.370818 Ftl1 1.18E-04 up 8.186755 Prdx4 4.35E-06 up 6.370151 Adm 0.001609 up 8.171984 Gsta1 0.001553 up 6.324022 Tacc1 2.02E-04 up 8.155523 Smpdl3a 6.30E-06 up 6.312695 Psen2 1.82E-04 up 8.067988 D12Ertd647e9.43E-07 up 6.303988 Sdcbp 1.16E-05 up 8.061649 BC046404 5.50E-04 up 6.30118

233

Symbol p Regulation FC Symbol p Regulation FC

Krt84 1.63E-10 down -228.719 Nuak2 7.11E-05 down -14.834 2310034C09Rik1.24E-09 down -186.567 Grhl1 0.002845 down -14.8157 Serpinb12 6.94E-05 down -69.1538 Krt33a 5.95E-04 down -14.4985 Expi 0.005073 down -52.0704 Col17a1 0.003957 down -14.4962 Dapl1 9.58E-06 down -45.6938 Tgm1 4.49E-04 down -14.3092 Sbsn 0.005906 down -38.4977 LOC1000467411.91E-04 down -14.0985 2310040C09Rik0.002216 down -37.773 LOC547380 0.001369 down -13.9101 Pkp1 0.004556 down -35.8935 Hal 0.001486 down -13.6513 Aqp3 0.005976 down -35.3096 Pdzk1ip1 0.003724 down -13.5987 Krt15 0.005073 down -33.4912 Hoxd10 2.50E-07 down -13.5077 Krtap16-3 7.08E-04 down -33.0938 Hspb1 0.001889 down -13.5039 Rarres1 0.00508 down -32.6279 Hoxd9 7.34E-08 down -13.3973 A030004J04Rik0.006753 down -32.3703 Lrrc15 0.006802 down -13.3619 Serpinb6c 1.64E-05 down -30.3599 4632417N05Rik3.79E-05 down -13.2702 Mt4 8.90E-05 down -29.6621 Pnpla5 4.47E-05 down -13.0036 Pp11r 5.71E-05 down -29.4936 Prss8 2.03E-05 down -12.9927 Ckmt1 0.003563 down -28.8227 OTTMUSG000000009714.79E-05 down -12.9824 Tcfap2c 0.001268 down -28.819 Ugt1a10 9.26E-04 down -12.8874 Nupr1 0.00103 down -28.619 Rdh9 3.35E-05 down -12.8567 Rdh12 9.41E-04 down -27.8775 Fxyd4 2.40E-06 down -12.818 2310005G13Rik5.23E-07 down -27.7382 Selenbp2 2.59E-06 down -12.4815 Barx2 3.93E-04 down -25.8593 Irf6 0.001298 down -12.4227 Hopx 0.003689 down -25.7787 Arhgef4 1.50E-04 down -12.4075 Cds1 7.68E-04 down -25.5323 Ctnnbip1 1.45E-04 down -12.1766 Sox21 1.50E-04 down -21.6202 Ugt1a10 6.37E-04 down -12.1404 Scgb1a1 9.37E-07 down -20.5111 Ank 0.005316 down -12.0932 Krtap13-1 7.43E-05 down -20.4657 Rnf208 5.42E-05 down -12.0112 Dgat2 0.005094 down -20.1006 Tuba8 1.64E-04 down -11.9838 LOC1000442047.36E-07 down -19.2325 Elovl6 0.003225 down -11.8895 Tmem45a 0.003363 down -18.6771 Mapk13 4.79E-04 down -11.7023 Tns4 0.00431 down -18.5138 Rbp7 1.15E-05 down -11.5594 Mycl1 2.57E-04 down -18.4657 Sema4a 3.94E-04 down -11.4354 LOC1000442043.73E-06 down -17.142 Slc40a1 1.23E-05 down -11.4305 Acox2 1.49E-04 down -16.8326 3300005D01Rik3.16E-04 down -11.3648 Bglap-rs1 0.004246 down -16.7992 Limk2 3.51E-06 down -11.0818 Cldn4 0.007573 down -16.6309 BC054059 3.71E-04 down -10.9351 Fgfr2 0.006327 down -16.5578 Igsf9 4.57E-05 down -10.9066 Hoxd8 5.68E-07 down -16.5505 Gldc 1.72E-04 down -10.7724 Calml3 3.48E-04 down -16.5091 Rasgrp1 4.59E-05 down -10.7434 Tmem79 9.83E-04 down -16.4106 Pdlim4 0.001233 down -10.5443 Krt86 3.66E-05 down -16.308 Atp12a 6.25E-04 down -10.4979 Clip4 7.08E-04 down -16.1237 Spon2 4.04E-04 down -10.4038 Cyb561 2.66E-04 down -15.4291 Tlcd1 3.64E-05 down -10.392 S100a14 0.001867 down -15.3824 Sult2b1 2.85E-04 down -10.234 Trim29 0.001012 down -15.3792 Tuba8 7.23E-05 down -10.1383 Ripk4 5.70E-04 down -10.1304

234

Symbol p Regulation FC Symbol p Regulation FC

LOC1000461200.007642 down -10.1299 Cpa4 5.51E-04 down -8.18543 2310033E01Rik4.33E-04 down -10.1226 Hebp2 4.99E-04 down -8.1665 Lrrc28 3.63E-05 down -10.1074 Rab3d 3.55E-04 down -8.12701 Tcfap2c 3.75E-07 down -10.0668 Pla2g4f 7.26E-04 down -8.12449 Nrarp 9.45E-04 down -10.0064 Abpb 3.07E-07 down -8.09764 Cmtm8 6.51E-07 down -9.84392 Cib2 5.85E-04 down -8.07548 Atg9b 0.002294 down -9.8106 Btbd11 1.04E-05 down -8.04566 Hal 0.001888 down -9.79336 Krtap16-4 6.02E-04 down -8.03861 Alox12b 3.08E-04 down -9.71492 A030007L17Rik0.005842 down -8.00316 Crabp2 2.18E-04 down -9.68989 Actb 0.001121 down -7.85892 OTTMUSG000000049664.94E-05 down -9.65565 Agpat4 3.71E-06 down -7.83517 Pank1 3.06E-05 down -9.63629 Dusp14 5.08E-06 down -7.81435 C79267 7.00E-06 down -9.35731 5830404H04Rik4.80E-05 down -7.7967 OTTMUSG000000107472.67E-06 down -9.28194 Ltb4r1 3.83E-04 down -7.7945 AA409316 3.82E-05 down -9.26138 Nfix 0.002034 down -7.76038 Klf4 0.003724 down -9.24234 Klf4 0.003576 down -7.64601 4930438A08Rik2.38E-04 down -9.23095 Tm4sf4 2.28E-06 down -7.62528 Itgb4 0.001041 down -9.2186 1110036O03Rik3.36E-05 down -7.57553 Bcl11b 0.001074 down -9.15926 Mtap7 0.001475 down -7.57262 Stard5 2.62E-06 down -9.12026 Gcat 1.82E-04 down -7.55018 BC039632 5.73E-04 down -9.11339 Hoxb7 3.14E-04 down -7.5157 Palmd 7.73E-04 down -9.11119 Acacb 1.07E-04 down -7.51152 Emp2 0.002682 down -9.08896 Rdh11 9.06E-05 down -7.49488 Atp6v1c2 0.001717 down -9.0337 Fabp5 0.001051 down -7.45233 Psca 1.29E-05 down -8.98341 Tuft1 1.47E-04 down -7.44765 Gjb4 3.57E-04 down -8.89423 Mansc1 2.56E-05 down -7.30295 Ccdc120 2.47E-04 down -8.88478 Actb 5.83E-04 down -7.19582 Egfr 2.59E-04 down -8.83902 5430433G21Rik9.36E-04 down -7.11196 Dm15 8.22E-06 down -8.83318 Pcdh1 1.68E-04 down -7.07048 Ovol1 4.04E-05 down -8.76665 Osbpl3 1.82E-04 down -7.04257 Ugt1a10 9.52E-05 down -8.76326 Cmtm8 1.45E-06 down -7.01137 Naprt1 1.88E-05 down -8.74061 Ppif 1.64E-05 down -6.98744 Sox7 3.74E-05 down -8.73445 Sgpp1 3.93E-05 down -6.98584 Mtap2 9.28E-04 down -8.71187 D9Ertd280e 2.71E-04 down -6.98491 Gpr109a 0.002622 down -8.6821 BC021614 1.95E-06 down -6.95961 Egr2 0.00146 down -8.6335 Klk10 8.83E-04 down -6.9586 Hsd17b7 0.005807 down -8.60018 Slco3a1 1.28E-04 down -6.95773 Gltp 0.004638 down -8.55485 Ell3 2.11E-06 down -6.90729 Dhcr24 7.32E-04 down -8.53531 2310047D13Rik0.002681 down -6.8834 Dlx3 0.006677 down -8.45682 Il13ra1 8.79E-04 down -6.88299 Sdc1 3.41E-04 down -8.35417 Pdcd4 1.21E-05 down -6.84901 AI661453 9.55E-05 down -8.32911 Itpkc 3.75E-06 down -6.82474 Id1 5.78E-06 down -8.31411 Klf5 0.004171 down -6.76364 D16Ertd472e0.001226 down -8.28168 4833409F13Rik1.18E-06 down -6.72272 4632427C23Rik8.39E-04 down -8.27199 Luzp1 0.002243 down -6.64781 Tiam1 1.70E-04 down -8.24444 Itgb4 0.001025 down -6.6416

235

Appendix 4: Gene expression profiling S3 vs. S1 melanoma

Gene LogFC (S3 vs. S1) P.Value Car2 10.247 0.000353215 Qpct 8.3566 7.64E-05 Ednrb 7.53 0.007328914 Sema6d 7.8965 0.019791965 Dusp6 7.6575 3.81529E-05 H2-Ab1 5.7851 0.010086523 Vcam1 4.51 0.007223759 Usp18 4.4758 0.006065279 Dusp23 3.2498 0.00351508 Hist1h2be 2.7485 0.007450028 Cyp7a1 2.58584 0.03041699 Sp5 2.534623 0.001737648 Oas1g 2.49453 0.013883523 Slc45a2 2.366916 0.011066997 Ifit3 2.357629 0.002992523 Lmo2 2.315036 0.009409501 Oas1g 2.232349 0.002482399 Hpse 2.18447 0.014862797 Fcgr4 2.164948 0.005662231 Oasl2 2.163185 0.003378932 EG667977 2.003416 0.021180351 Hist1h2bf 2.0001 0.000964453 Hist1h2bh 2 0.001057836 Usp18 1.986873 0.027110676 Tpd52l1 1.986593 0.024551409 Aif1 1.964149 0.004114213 Psmb8 1.963154 0.004913133 Cdkn2a 1.953197 0.035106226 Tpd52l1 1.922451 0.032441971 Pi4k2b 1.898004 0.001972563 Irgm1 1.885813 0.00036888 Sfrs5 1.881634 0.023778617 Bmp15 1.832675 0.015822738 Plac8 1.798208 0.007164127 Sbk 1.787414 0.024128764 Ly6c1 1.763215 0.011566003 Gimap8 1.742587 0.018168938 Maob 1.729111 0.00539632 Abp1 1.6336 0.014022096 Atp6ap2 1.613991 0.006714994 Kit 1.602053 0.037206167 Hist1h2bn 1.594874 0.001780597 LOC638301 1.579727 0.000673733

236

Lama1 1.576606 0.028928626 Klk1 1.568355 0.008681186 AI607873 1.566455 0.01274632 Shf 1.559632 0.010742157 Hist1h4a 1.552415 0.016700717 Calm2 1.532909 0.015538266 Eif2ak2 1.532724 0.017266002 Nckap1 1.528891 0.057627606 Dcn 1.511798 0.003014611 Hist1h2bj 1.510719 2.65E-05 Ube2l6 1.498342 0.007501191 Hist1h2bk 1.49794 0.01247838 Asphd2 1.492979 0.002770187 EG433865 1.491414 0.00233863 Clk1 1.482343 8.07E-05 Ube2l6 1.478832 0.004022044 Mgl2 1.477008 0.003264857 Ifit2 1.476846 0.02921549 Igh-6 1.476671 0.015091298 Atp6ap2 1.473051 0.007835933 Dock1 1.470035 0.012932706 Eif4a2 1.455725 0.008073212 mtDNA_ND4L 1.450778 0.075706147 1700021K14Rik 1.446563 0.00461036 Sec11c 1.44053 0.001842936 Atp6ap2 1.440167 0.006361434 Ly86 1.435562 0.006602544 Clec4b1 1.423205 0.002422402 Dyrk3 1.422951 0.020113521 Reg1 1.420714 0.014530664 Pi4k2b 1.418528 0.000325606 Rnf213 1.417633 0.007533679 Nudt4 1.413777 0.0051538 Ifi47 1.396672 0.037207365 Mpeg1 1.379831 0.00298173 Slfn1 1.351826 0.007291272 Igh-6 1.34264 0.004106978 Id4 1.332305 0.002928703 Cd59b 1.327686 0.00758957 Vat1l 1.321144 0.128484482 Nt5c3l 1.320889 0.003513603 Glrb 1.314218 0.016112275 Slc3a2 1.313625 0.02234367 EG434858 1.311997 0.001152834 Ifit2 1.301085 0.001714144 LOC381649 1.294801 0.001361739 237

Jam2 1.28893 0.021628236 Rab38 1.273039 0.009124414 Art3 1.272077 0.004697433 LOC100045617 1.271006 0.193432706 Hist1h4f 1.269084 0.016650414 Gvin1 1.26771 0.007288613 Clec4b1 1.262325 0.001171732 Ms4a6d 1.258328 0.014275581 Hnrnpk 1.257673 0.000659738 Tnfrsf19 1.254525 0.023316256 Cd72 1.245332 0.019639912 Ccr5 1.24297 0.03469485 Ddah1 1.23503 0.088172932 C130078N17Rik 1.23297 0.01680274 Siat9 1.230391 0.000102015 D14Ertd668e 1.225743 0.00057162 A630077B13Rik 1.216404 0.005821002 Gadd45a 1.216299 0.003285285 Laptm5 1.207136 0.005776025 Lin7a 1.19468 0.006024679 Rbm11 1.19391 0.018068267 3830408P04Rik 1.189474 0.01000013 Abi3 1.184213 0.007683282 Gch1 1.176219 0.005487877 Nt5c3l 1.175824 0.015200051 Mgl1 1.171912 0.013261118 Sla 1.169845 0.011983694 Tmf1 1.167106 0.029918238 Hist1h2bm 1.166768 0.001488257 Phf11 1.165267 0.074029956 Adk 1.164365 0.001608453 Hnrpa1 1.15676 0.005974908 Oca2 1.155388 0.004024013 Fam120a 1.152119 0.001266192 Cd84 1.150816 0.024072425 Phf11 1.145742 0.004646133 Rbbp7 1.145433 0.016310764 Dnajb6 1.142882 0.000245755 Ms4a7 1.138679 0.007167994 Sepx1 1.137757 0.004808616 Dnajb6 1.137513 0.009355766 scl0004175.1_57 1.136108 0.00389321 Hddc2 1.133477 0.063196518 Pgp 1.132096 0.034049134 Etv1 1.129342 0.006430401 Enpp2 1.126856 0.003746878 238

Sh3kbp1 1.124759 0.023248566 P2ry14 1.124599 0.00655613 Ccr5 1.123964 0.00011854 Gdi1 1.123754 0.009532062 Etv4 1.121907 0.019832061 Sccpdh 1.120859 0.043347648 Gsdmdc1 1.120308 0.01100383 Fgd6 1.116672 0.005362519 D11Lgp2e 1.109123 0.007266692 Pik3ap1 1.105506 0.079017504 Hnrnph1 1.102299 0.018528795 Gcnt1 1.101578 0.01018405 Agfg1 1.099212 0.002796846 Fbln1 1.098963 0.00066565 Mitd1 1.098594 0.014057595 LOC385825 1.097732 0.001543794 Tm7sf3 1.096383 0.000158332 Mdk 1.095958 0.006231457 Cxcl9 1.08961 0.030743242 Spry4 1.089157 0.006318275 Hist1h2bc 1.084501 0.037736917 Gdpd3 1.081506 0.01503541 Scye1 1.080613 0.001754112 Tlr13 1.076773 0.009109145 Tef 1.076283 0.001504315 Fbxo3 1.073393 0.041162751 LOC669053 1.072278 0.002750908 Hrb 1.065025 0.037906722 Psmd10 1.064372 0.032299909 Ss18 1.058281 0.001731957 Gyg 1.058203 0.001585124 Notum 1.052936 0.056048386 Flnb 1.050833 0.0044665 Stat1 1.049984 0.019290918 Ypel5 1.049162 0.000185608 Nhsl1 1.044471 0.022972712 Cyfip2 1.043102 0.023179811 Sh3kbp1 1.041412 0.000156788 Rbbp7 1.041234 0.016110986 D14Ertd668e 1.036076 0.000167968 Sdcbp 1.035544 0.029380369 Tlr7 1.034986 0.023856618 BC018465 1.031833 0.010882638 Rps3 1.030818 0.006443435 Adnp 1.030696 0.024306827 LOC232606 1.023441 0.000544238 239

Hist2h2aa2 1.021537 0.01247992 A730054J21Rik 1.017896 0.015947287 Mylc2b 1.013441 0.029476983 Enc1 1.012139 0.002174496 Bxdc2 1.009719 0.000736489 Mafg 1.009512 0.004143515 H28 1.006783 0.007166107 Tcf12 1.003864 0.019055107 Ptgds2 1.003636 0.016763922 EG240327 1.003346 0.033379819 LOC100048480 1.002382 0.004226916 Fgd6 1.000174 0.060523906 Lyplal1 0.995399 0.007804339 Hist1h4j 0.995047 0.002508784 Rnf135 0.99477 0.026079332 P2ry13 0.993931 0.003158486 Hps3 0.993036 0.001668311

240

Gene LogFC (S1 vs. S3) P.Value Krt16 12.7532 0.005032 Chi3l1 9.7847 0.0201 Stfa1 9.2 0.00017 Lamb3 10.841 0.017139 S100a9 8.575 0.028232 Junb 5.1112 0.0182 Rab25 4.32 0.004987 Anxa11 3.5 0.021359 Lypd3 3.2143 0.049677 Mmp9 3.3 0.005166 Gjb6 2.8 0.003205 Krt6b -3.17013 0.01171 Sprr1b -2.93938 0.02813 Stfa1 -2.65245 0.003259 Prss18 -2.54905 0.035152 Klk6 -2.46898 0.022445 BC100530 -2.44743 0.028968 EG408196 -2.40697 0.01071 Krt5 -2.27918 0.001914 Krt79 -2.27214 0.02608 Ly6d -2.17697 0.004217 Klk8 -2.1613 0.028412 Cnfn -2.15895 0.020735 Sprr2d -2.0804 0.000477 Sult2b1 -2.06972 0.037683 Clic3 -2.056 1.24E-05 Ly6d -1.93576 0.045648 Fxyd3 -1.9309 0.026425 Evpl -1.92449 0.034718 Serinc2 -1.87926 0.039488 Cebpb -1.78454 0.015301 Lad1 -1.7795 0.019416 Prom2 -1.76744 0.006043 Ercc2 -1.75682 0.021261 Ngfr -1.74093 0.02916 Spnb3 -1.73206 0.02452 Sprr1a -1.72597 0.00385 Gjb3 -1.6321 0.012552 S100a14 -1.62178 0.030312 Sphk1 -1.6091 0.030884 Slc6a8 -1.58628 0.007716 Casp14 -1.57157 0.022335 Cnfn -1.5672 0.019018 Fbxl11 -1.5631 0.04904 Car12 -1.54942 0.026437 241

9530027K23Rik -1.54744 0.019197 Defb14 -1.51712 3.97E-05 Sfn -1.51507 0.044272 Cd109 -1.51008 0.004003 Tmem54 -1.50269 0.003643 Hr -1.49388 0.049944 4631405K08Rik -1.4843 0.034602 Capg -1.47081 0.032695 Bcat2 -1.47059 0.019992 LOC245892 -1.47051 0.032731 Anpep -1.46201 0.024952 Psme3 -1.46144 0.030402 2610009E16Rik -1.45385 0.01887 Homer2 -1.44104 0.008905 Epgn -1.43645 0.005449 Abhd9 -1.43558 0.034177 Serpina3n -1.42414 0.001619 Foxq1 -1.41915 0.005145 H2-M2 -1.4044 0.005896 Aqp9 -1.40432 0.007537 Tmprss13 -1.40115 0.017771 Egln3 -1.39407 0.01542 Tmprss13 -1.39338 0.024388 Ubl7 -1.38641 0.028073 BC004728 -1.36887 0.031511 Klf13 -1.35979 0.02038 Cav1 -1.35944 0.039989 Tagln2 -1.35897 0.017697 Smox -1.35827 0.034553 BC004728 -1.35742 0.000783 Serpinb3b -1.35708 0.008947 Smox -1.35649 0.01596 LOC381860 -1.34376 0.006303 Cyp4f39 -1.34046 0.002042 Arg1 -1.32995 0.007977 Mllt4 -1.32861 0.025318 Ap2a1 -1.32516 0.042419 BC065085 -1.31516 0.010506 Scaf1 -1.30844 0.056213 Il17rc -1.30423 0.01671 1700025G04Rik -1.28732 0.00154 LOC100046483 -1.28085 0.006004 Flrt3 -1.28062 0.014602 Gcnt2 -1.27808 0.01672 Ltbp4 -1.27663 0.003267 Chit1 -1.27653 0.00083 242

Palm -1.27109 0.008527 Bok -1.26791 0.001168 Josd2 -1.26739 0.028931 Map3k6 -1.25575 0.023837 Ppard -1.25204 0.023833 Npm3-ps1 -1.24999 0.039548 Hdac5 -1.24789 0.000159 Furin -1.22823 0.036547 Ndrg1 -1.22771 0.002733 Cdc42ep5 -1.22738 0.009265 Npm3 -1.22599 0.000228 Itpr3 -1.22085 0.008321 Cdh13 -1.2201 0.0026 Prkcdbp -1.21356 2.00E-06 Blmh -1.21088 0.000835 2310007G05Rik -1.209 0.009909 Rnase4 -1.20857 0.007532 Bnipl -1.20735 0.001624 Plek2 -1.20719 0.018984 Synpo -1.20517 0.000266 Erdr1 -1.19684 0.008896 Endod1 -1.19455 0.000621 Lmna -1.19363 0.023602 A430096B05Rik -1.1926 0.038457 Angptl4 -1.1916 0.012158 Hdc -1.18914 0.001404 Il33 -1.17996 9.79E-05 Areg -1.17704 0.038027 Fkbp8 -1.17607 0.000899 Plec1 -1.17446 0.035376 Ndrl -1.17325 0.008997 Actb -1.17312 0.025676 Lmna -1.17276 0.010051 Fam129b -1.16896 0.037642 Eif6 -1.1688 0.000584 Mark2 -1.16786 0.000438 Thy1 -1.16779 0.000938 Cyb5r3 -1.16272 0.000543 Fam132a -1.15715 0.000593 Slc2a1 -1.15699 0.030428 Hgs -1.15474 0.000126 Tcfap2c -1.15087 0.003282 Ltbp4 -1.14974 0.008801 C130090K23Rik -1.14942 1.76E-06 Klk11 -1.14026 0.000876 Pde1b -1.13623 0.010734 243

Rapgefl1 -1.13572 0.000925 Bcl7c -1.13477 0.037969 C77080 -1.13439 0.012754 Dgka -1.13068 0.0359 Pde1b -1.12958 0.012987 Slco2a1 -1.12631 0.00123 Eps8l2 -1.12507 0.046682 Uck2 -1.12464 0.00095 Cyp2b10 -1.12133 0.000832 Sdc4 -1.12113 2.01E-05 LOC100048721 -1.12089 0.001251 Wnt10a -1.11213 4.60E-05 Fam115c -1.09434 0.040172 Aldh18a1 -1.09151 0.00209 4933407C03Rik -1.09011 0.02374 S100a16 -1.08709 0.018985 Vat1 -1.08153 0.045565 Ino80b -1.0783 0.030992 Fermt1 -1.06861 0.005408 AI481316 -1.06421 1.45E-06 Itpkb -1.05919 0.000385 Lce3c -1.05897 0.008794 Usp2 -1.03985 0.00945 Lcn2 -1.03849 0.009448 Ankrd22 -1.03786 0.025558 Lmna -1.03528 0.042832 Sf3b4 -1.03375 0.001096 Ccnt1 -1.03368 0.037171 Serpinb6a -1.02803 0.010004 Hdac5 -1.02284 0.002844 Uck2 -1.01231 0.024131 Slc25a1 -1.00375 0.000279 Akr1b8 -0.99937 0.005313 Aprt -0.99367 0.015485 1190002H23Rik -0.99173 0.006219 Ndufa3 -0.99169 0.00363 Trp53 -0.98761 0.031997 Ext1 -0.98596 0.003293 LOC100046483 -0.98595 0.033829 Orai1 -0.97984 0.004321 Ass1 -0.97975 0.000222 Aldoa -0.979 0.002665 Mall -0.97743 0.004942 Il1r2 -0.97522 4.90E-05 Tmem45b -0.97375 0.021281 Sh3bgrl3 -0.9732 0.002011 244

Pdgfa -0.97213 0.056719 Dnmt3l -0.97212 0.000814 Wnt3a -0.96547 0.009386 Ahnak2 -0.96348 0.029777 Atf3 -0.96109 0.033458 Tpm1 -0.95782 5.52E-05 Nbl1 -0.9533 0.031262 2610035D17Rik -0.95282 0.020442 2300004C15Rik -0.94888 0.016835 Myo1e -0.94656 0.034667 Cdc42ep4 -0.94588 0.040632 Gna13 -0.94522 0.018837 Ahnak -0.94399 0.013364 Purb -0.94331 0.02118 Npm3-ps1 -0.93862 0.004218 Cdca7 -0.93614 0.000833 LOC100047419 -0.93535 0.002864 Obfc2b -0.93222 0.010785 Pcdh7 -0.92937 0.019505 Wnt4 -0.92758 0.004564 Pkm2 -0.92314 0.002144 Stat3 -0.92224 0.011649 Arhgef1 -0.90764 0.039955 2210023G05Rik -0.90749 0.007757 Tomm6 -0.90412 0.031433 Uba1 -0.90399 0.001927 Entpd2 -0.90109 0.010215 Ndufa3 -0.89481 0.000996 Lss -0.89285 0.006301 Elovl1 -0.88925 0.011894 Polr2a -0.88819 0.001228 Lmna -0.88578 0.020845 9030619K07Rik -0.88564 0.005778 Styx -0.8837 0.004978 Rnu6 -0.87789 0.010314 Car13 -0.87629 0.026674 Fam162a -0.87559 0.019944 Klf6 -0.87045 0.012348

245

Appendix 5: List of genes differentially expressed between human metastasis and primary melanoma

Gene logCPM PValue DUSP6 8.111519103 0.000745292 EDN1 1.717414898 0.061659637 EDN3 6.178411092 2.91066E-09 EDNRA 3.254382162 1.96185E-06 POU3F2 3.390107634 0.264229636 GLI1 1.770661459 1.37062E-05 SHH 1.508576135 0.000202695 NTM 3.386307292 0.004575069 OPCML 1.485709919 4.08523E-05 ZEB1 4.654526018 4.11465E-13 ETS1 7.726328109 0.003022034 TGFB2 3.557590955 0.173681835 TGFBR1 6.673329074 0.346489974 TGFB3 4.287321324 9.68551E-09 TGFB1 6.135397725 0.000647391 TGFBI 8.192451729 0.002056265 TGFBR2 7.023262033 0.003217058 TGFBR3 5.116016274 0.014784709 DCBLD1 3.429120248 0.542622947 MAN1A1 4.513929597 5.29716E-14

GJA1 5.691504606 7.1852E-20 HSF2 3.874996846 0.264840689 SERINC1 7.535637571 0.008244481 ASCC1 4.420274513 0.303873625 CDH23 1.839840342 0.000108607 SGPL1 5.77268312 0.984316286 HK1 7.388268527 0.775551788 CCAR1 5.548297122 0.000420877 TET1 2.527577612 1.40877E-06

246

Appendix 6: Approval for research involving human subjects

247

Appendix 7: Approval for research involving animal subjects

P240 - Delineation of the key molecular events that underlie melanoma development Status: In Progress

Project Details

Information

PD001 Title A Short descriptive title for the project. Delineation of the key molecular events that underlie melanoma development PD002 Alternative Titles List any other titles that this research is referred to, including existing or prospective grant titles. not answered

PD003 Brief Description A brief description of the project, including animal and human work. Including lay terms where possible. Please limit to 250 words. This proposal is based on the following over-arching hypothesis that pRb acts as the pivotal point of convergence for each of the three major axes involved in melanoma development, namely the Ink4a/CDK4/pRb, RTK/Ras/Raf/MAPK, and Arf/mdm2/p53 pathways,Within the proposal we will test the following specific hypotheses:H1: Studying mice with dysregulation of one or more of these pathways will allow us to determine which components are most important for melanoma development, both constitutively and with neonatal exposure to ultraviolet radiation (UVR).H2: Melanocytes and melanomas from these genetically modified mice harbour a critical overlapping set of mutations and gene expression abnormalities that are necessary for melanoma development.H3: In vitro UVR exposure to melanocytes from genetically engineered mouse lines induces mutational, transcriptional or post- translational changes in components of the Ink4a/CDK4/pRb, RTK/Ras/Raf/MAPK, or Arf/mdm2/p53 pathways that represent a necessary additional “hit” required for melanoma development.These hypotheses will be tested with the following specific aims:A1: To characterise the phenotype of mouse mutants derived from crossing conditional tissue-specific Rb1 knockout mice, with tyrosinase-Ha-Ras (G12V), tyrosinase-Cdk4 transgenic mice, and Tp53-/- mice. A2: To challenge all single and compound knockout/transgenic animals with a 9 kJ/m2 neonatal dose of solar-simulated UVR to assess its role as a promoter of melanoma development in genetically susceptible mice.A3: To carry out microarray gene expression profiling on melanomas from these animals to identify a common set of genes that are dysregulated in melanoma development. A4: To assess molecular perturbations in cell cycle regulation in cultured melanocytes from these animals with and without UVR exposure.A5: To study cultured melanocytes from the above mice to look for differences in Cdk-inhibitor usage by Cdk4 and Cdk2, phosphorylation status of pRb, and expression of Ink4a, p21, p18, p27, E2Fs, cyclins, and Arf. Does this project involve an OGTR application? Please select relevant OGTR applications involved in this project.

No Yes

Title Notes Status Delineation of key molecular events that underlie melanoma development Approved

248

PD004 Expected Start Date Wednesday, 27 June 2001 Expected Finish Date Sunday, 27 March 2022

23 October 2017 Established GM Species Details

GMO Type knockout [KO] Full Strain Name Common Name knockin [KI] Identity and function of gene(s)responsible for modified trait conditional mutation [CM] transgenic [TG] Cdk4-R24C Cdk4 KO Cdk4-cell cycle Tyr-CreER(T2) Cre transgenic transgenic Cre for deletion of floxe alleles Tyr-NrasQ61K Nras transgenic Nras - Oncogenic signalling Gene mine strains Recombinant Inbred Mice - Natural genome variation between strains K14-Cre K14-Cre CRE transgenic Cre excision to generate KO mice Cdon-null (Cdon-tm1Rsk) Cdon-null KO Cdon-sonic hedgehog pathway member Patched floxed Ptch1 KO Cre excision to generate KO mice Cx43 floxed Cx43 KO Cre excision to generate KO mice Tgfbr2 floxed Tgfb receptor 2 floxed KO Cre excision to generate KO mice

249