The New Melanoma: a Novel Model for Disease Progression

DISS. ETH NO. 17606

The new melanoma: A novel model for disease progression

A dissertation submitted to SWISS FEDERAL INSTITUTE OF TECHNOLOGY ZÜRICH

for the degree of DOCTOR OF SCIENCES

presented by Natalie Schlegel Master of Science, Otago University (New Zealand)

born on January 20th 1976 citizen of Zürich (ZH)

accepted on the recommendation of

Professor Sabine Werner, examinor Professor Reinhard Dummer, co-examinor Professor Josef Jiricny, co-examinor

2008

22 Table of Contents Abstract ...... 6 Résumé...... 8 Abbreviations ...... 10 1. Introduction ...... 13 1.1 Definition ...... 14 1.2 Clinical features...... 14 1.3 Pathological features and staging...... 16 1.3.2 Clark’s level of invasion and Breslow’s thickness...... 16 1.3.3 TNM staging ...... 17 1.4 Epidemiology ...... 18 1.5 Genes and pathways...... 19 1.5.1 Genes involved in familial melanoma...... 19 1.5.2 PTEN...... 21 1.5.3 MAPK ...... 22 1.5.4 Wnt...... 23 1.5.5 Microphthalmia-associated transcription factor...... 24 1.6 The transforming growth factor-β super-family ...... 25 1.6.1 Receptors and Smad proteins ...... 27 1.6.2 Transforming growth factor-β...... 28 1.6.3 Activin...... 32 1.7 Id proteins...... 34 1.7.1 Ids, cell cycle regulation and cancer ...... 34 1.7.2 TGF-β and Id proteins...... 35 1.7.3 Id proteins and melanoma ...... 36 1.8 Aim...... 37 1.9 References ...... 38 2. Metastatic potential of melanomas defined by specific gene expression profiles with no BRAF signature...... 49 2.1 Abstract ...... 52 2.2 Introduction ...... 52 2.3 Results ...... 54 2.3.1 No correlation between BRAF/NRAS mutations and gene expression...... 54 2.3.2 Microarray analyses reveal three cohorts...... 56 2.3.3 Two groups of co-regulated genes define the cohorts...... 57 2.3.4 The cohorts reflect differences in metastatic potential...... 60 2.3.5 In vitro tests support the link between cohort distribution and metastatic potential...... 61 2.3.6 Wnt signalling controls Motif 1 ...... 62 2.4 Discussion ...... 63 2.5 Material and Methods...... 67 2.5.1 Cell Culture and Media ...... 67 2.5.2 Genotyping...... 68 2.5.3 Total RNA Extraction and Expression Profiling...... 68 2.5.4 Microarray Data Analysis ...... 68 2.5.5 Growth Inhibition Assays...... 70 2.5.6 Motility Assays ...... 70 2.5.7 Western Analyses and ELISA...... 70 2.5.8 Immunohistochemistry...... 71

3 2.6 References ...... 72 3. In vitro phenotype validation...... 79 3.1 Introduction ...... 80 3.1.1 Modulation of TGF-β signalling...... 80 3.1.2 Vasculogenic mimicry...... 83 3.2 Results ...... 84 3.2.1 Confirming TGF-β1 and activin A secretion ...... 84 3.2.2 Follistatin secretion does not correlate with activin secretion ...... 84 3.2.3 Smad2 and Smad3 are activated across all cohorts...... 85 3.2.4 Ski is not responsible for the differential TGF-β signalling ...... 86 3.2.5 The activation of the MAPK pathways does not correlate with the TGF-β signature ...... 87 3.2.6 Identifying vasculogenic mimicry as a discriminating phenotype...... 88 3.2.7 Phenotype switching ...... 89 3.3 Discussion ...... 90 3.4 Material and Methods...... 94 3.4.1 Cell culture ...... 94 3.4.2 Preparation of condition media ...... 95 3.4.3 ELISA...... 95 3.4.4 Preparation of total cell protein extracts ...... 95 3.4.5 Preparation of cytosolic and nuclear protein extracts ...... 95 3.4.6 Western blot analysis ...... 96 3.5 References ...... 98 4. In vivo switching of human melanoma cells between proliferative and invasive states 101 4.1 Abstract ...... 104 4.2 Introduction ...... 104 4.3 Results ...... 106 4.3.1 Phenotypic assignment of cell lines ...... 106 4.3.2 Mitf is a marker of proliferative phenotype ...... 107 4.3.3 Mitf expression reflects signature phenotype...... 108 4.3.4 Proliferative cells form fast growing tumours sooner than invasive cells ..... 109 4.3.5 Tumours derived from proliferative or invasive lines are indistinguishable...... 110 4.4 Discussion ...... 111 4.5 Material and Methods...... 117 4.5.1 Melanoma tissues and lines...... 117 4.5.2 In vitro motility and proliferation assays ...... 117 4.5.3 Recombinant adenovirus vector and siRNA ...... 117 4.5.4 Transfection and TGF-β challenge assay...... 118 4.5.5 Western blot analyses...... 118 4.5.6 Xenografts ...... 118 4.5.7 Immunohistochemistry...... 119 4.5.8 Statistical analysis ...... 119 4.6 References ...... 120 5. Id2 suppression of p15Ink4b abrogates TGF-β-mediated growth inhibition in melanoma ...... 123 5.1 Abstract ...... 125 5.2 Introduction ...... 125 5.3 Results ...... 127 5.3.1 Differential Id2 regulation and expression in human melanoma cultures ..... 127

4 5.3.2 Id2 overexpression protects proliferative cells from the growth inhibitory effects of TGF-β...... 129 5.3.3 Id2 regulates TGF-β-induced G1 cell-cycle arrest...... 131 5.3.4 Id2 restricts TGF-β-induced upregulation of p15Ink4b ...... 132 5.3.5 Id2 expression does not correlate with patient survival in melanoma ...... 132 5.4 Discussion ...... 133 5.5 Materials and Methods...... 135 5.5.1 Cell culture and Adenoviruses ...... 135 5.5.2 RNA extraction, cDNA synthesis and RT-PCR...... 136 5.5.3 Western blot analysis ...... 136 5.5.4 Growth inhibition assays...... 137 5.5.5 Cell cycle FACS analysis...... 137 5.5.6 Immunohistochemistry, cell culture and tissue array...... 137 5.6 References ...... 139 6. Discussion & Conclusions...... 143 Appendix A ...... 151 Acknowledgments...... 177 Curriculum vitae...... 178

5 Abstract Melanoma is an aggressively metastasising tumour originating from neural-crest derived melanocytes. It is recognised as being the most dangerous of skin cancers and over the last three decades its incidence has increased more rapidly than that of any other cancer. Although primary tumours can easily be removed with surgery, once the cancer has metastasised patient survival is dramatically low. It is therefore of primary importance to elucidate the mechanisms driving melanoma progression. The model of melanoma progression in which molecular lesions progressively accumulate is widely accepted and stands as the dominant paradigm for molecular studies of the disease. However, this model has some limitations. For example, although accumulation of irreversible genetic lesions during tumour progression has been reported for many tumours, the acquired capacity for invasion and metastasis has not been linked to recurrent mutations but rather to specific gene expression changes. We propose a new melanoma progression model which we formulated using gene expression array data acquired from three distinct libraries of melanoma cultures. Our model is based on reversible transcriptional changes and the concept of cohort specific expression. We contend that melanoma cells switch between two defined gene expression signatures, each underlying a distinct cell phenotype, which together drive disease progression. Presented in this thesis are the in vitro and in vivo experimental validations for this model, the investigation of the role of TGF-β-like signalling, predominantly its role in growth inhibition, and the identification of Id2 as a gene involved in TGF-β-induced growth inhibition response. After a literature review of genes identified to have phenotype-specific expression, we identified Wnt and TGF-β signalling as drivers of the identified transcriptional signatures. By in vitro characterisation of phenotypically opposed cells, we identified the two phenotypes as proliferative and invasive. As well as showing divergent proliferative and invasive behaviour, cell types could be discriminated based on their growth susceptibility to TGF-β and their capacity for vasculogenic mimicry. Reduced susceptibility to the growth inhibiting effects of TGF-β and the capacity for vasculogenic mimicry have both been associated with increased invasive and metastatic properties of melanoma cells. Our model suggests that both proliferative and invasive transcriptional signatures are important in disease progression and that each melanoma cell retains the capacity to express either signature given appropriate signalling. Our model also accounts for much observed gene expression heterogeneity in melanoma tumours. This heterogeneity and reversibility of transcription programs were also shown in vivo using a xenograft mouse model. We also investigated the motive forces behind

6 differential TGF-β signalling. Smad activation was present in all melanoma cultures irrespective of the presence of a TGF-β signature, which suggested Smad-independent TGF-β signalling. The TGF-β Smad-dependent pathway has long been considered as being central to TGF-β signalling but it is now recognised that TGF-β signals via crosstalk with alternative pathways. We investigated alternative pathways but could identify no link between the activation status of several MAPK pathways and the TGF-β signature. TGF-β is a multifunctional cytokine which controls aspects of cell proliferation, differentiation, migration, apoptosis, adhesion, angiogenesis, immune surveillance, and survival. TGF-β was initially defined as a transforming cytokine but it is now understood that TGF-β has dual roles both as tumour suppressor and tumour promoter. To better understand the regulation behind the expression of these opposite behaviours, we studied TGF-β’s cytostatic effect, which plays an important role in its tumour suppressing function and which is lost as melanoma cells become more invasive and metastatic. We identified the Id2 gene as differentially regulated by TGF-β and link the loss of its regulation to acquired resistance to TGF-β in invasive phenotype cells. We show that TGF-β induces cell cycle arrest through induction of p15Ink4b and repression of Id2. Furthermore, Id2 overexpression in proliferative phenotype cells counteracts p15Ink4b induction and consequently protects melanoma cells from TGF-β- mediated inhibition of proliferation. Treating tumours comprised of cells with variably expressing transcription signatures presents a difficult challenge. This is because specific therapies have targeted factors we identify here as being subject to repeated changes in regulation. It is therefore of primary importance we recognise that the existing paradigm for melanoma progression is insufficient for the design of effective therapies.

7 Résumé Le mélanome est un cancer agressif qui se développe à partir des mélanocytes, cellules dérivées de la crête neurale. Il est reconnu comme étant un des plus dangereux cancers de la peau. Depuis les trois dernières décennies, son incidence a augmenté plus que celui de tout autre cancer. Quoique les tumeurs primaires puissent être aisément retirées chirurgicalement, l’espérance de vie des patients présentant des métastases est abrégée de façon dramatique. Il est donc primordial d’élucider les mécanismes responsables de la progression du mélanome. Le modèle de progression du mélanome présentement accepté et sur lequel les études moléculaires sont fondées, est basé sur l’accumulation progressive de lésions moléculaires. Ce modèle a toutefois ses limites. Par exemple, quoique l’accumulation de lésions génétiques irréversibles au cours de la progression de la tumeur ait été reportée pour un nombre important de cancers, aucun lien entre l’acquisition des compétences permettant l’invasion et la métastase n’a été fait avec des mutations récurrentes mais plutôt avec des changements au niveau de l’expression de certains gènes. Nous proposons ici un nouveau modèle de progression du mélanome que nous avons formulé à partir d’études moléculaires d’expressions génétiques réalisées avec des cultures de mélanomes provenant de trois collections distinctes. Notre modèle est basé sur des modifications réversibles du transcriptome et sur le concept de la classification des tumeurs fondée sur les profils d’expressions uniques. Nous proposons un modèle dans lequel les cellules cancéreuses oscillent entre deux signatures transcriptomiques définissant deux phénotypes distincts, qui ensembles sont responsables de la progression du cancer. Dans cet ouvrage nous présentons la validation de ce modèle de façon in vitro et in vivo, l’étude du rôle de la signalisation du TGF-β, principalement son rôle comme inhibiteur de la prolifération cellulaire, et l’identification d’Id2 comme étant un gène impliqué dans l’inhibition de la prolifération cellulaire induite par le TGF-β. À la suite d’une revue littéraire des gènes spécifiques au phénotype, nous avons identifié la signalisation du Wnt et celle du TGF-β comme étant les moteurs des signatures transcriptomiques identifiées. Après avoir caractérisé des cellules phénotypiquement opposées, nous avons identifié les deux phénotypes comme étant prolifératif et invasif. En plus de posséder des caractères prolifératifs et invasifs divergents, les groupes de cellules se distinguent par leur résistance aux effets antiprolifératifs exercés par le TGF-β et par leur compétence pour le “vasculogenic mimicry”. Ces deux caractéristiques ont été associées avec une aggravation des qualités invasives et métastatiques des cellules de mélanome. Notre modèle suggère que les signatures transcriptomiques proliférative et

8 invasive sont toutes deux importantes pour la progression du mélanome et que chacune des cellules de mélanome est en mesure de les exprimer en recevant le signal approprié. Notre modèle explique également l’expression génétique hétérogène observée dans les mélanomes. Cette hétérogénéité et cette réversibilité des programmes de transcription ont aussi été démontrées à l’aide d’un modèle de xénogreffes sous-cutanées. Nous avons également tenté d’identifier les causes de la divergence observée dans la signalisation du TGF-β. Les R- Smads, Smad2 et Smad3, étaient activés dans toutes les cultures de mélanomes étudiées. Leur activation était donc indépendante de la présence d’une signature marquée par le TGF-β ce qui suggère la présence d’une signalisation ne dépendant pas des Smads comme transducteurs de signal. La famille des protéines Smads représente les messagers hautement spécifiques du TGF-β, mais il est maintenant évident que d’autres voies de signalisation peuvent être activées par le TGF-β. Nous avons étudié des voies de signalisation alternatives mais n’avons trouvé aucun lien entre la signature marquée par TGF-β et l’activation des voies des MAPKs. Le TGF-β est une cytokine multifonctionnelle qui contrôle un large spectre de réponses biologiques comme la prolifération, la différenciation, la motilité, l’apoptose, l’adhésion, l’angiogénèse, la surveillance immunitaire et la survie. Le TGF-β a été initialement décrit comme un peptide capable de provoquer la transformation réversible de cellules en culture mais il est maintenant reconnu qu’il joue des rôles antagonistes sur la tumorigénèse. Pour mieux comprendre la régulation de ces rôles antagonistes, nous avons étudié son effet antiprolifératif qui est important pour sa fonction suppressive du cancer et qui disparait dans les mélanomes présentant des caractères plus invasifs et métastatiques. Nous avons identifié le gène Id2 comme étant un gène contrôlé de façon différentielle par le TGF-β et nous associons la perte de ce contrôle à la perte de réponse au TGF-β observé chez les cellules de type invasif. Nous démontrons que le TGF-β provoque un arrêt du cycle cellulaire par l’induction transcriptionnelle de p15Ink4b et la répression de Id2. De plus, la surexpression d’Id2 chez les cellules de type prolifératif neutralise l’induction de p15Ink4b et protège les cellules de mélanome contre l’effet antiprolifératif de TGF-β. Le traitement de tumeurs formées de cellules comportant des signatures transcriptomiques variables représente un défi de taille. Les thérapies actuelles ciblent des facteurs que nous avons identifiés comme étant sujet à de nombreux changements au niveau de leur régulation transcriptionnelle. Il est donc primordial de reconnaître que le modèle actuel de progression du mélanome n’est pas adéquat pour la conception de thérapies efficaces.

9 Abbreviations ARF alternate reading frame ALM acral lentiginous melanoma bFGF basic fibroblast growth factor bHLH basic helix-loop-helix BMP bone morphogenic protein BRAF V-raf murine sarcoma viral oncogene homolog B1 CCND1 cyclin D1 CDKN2A cyclin-dependent kinase inhibitor 2A CDK cyclin dependent kinase CM conditioned media Co-Smad co-mediator Smad CTGF connective tissue growth factor E-cadherin epithelial cadherin EGF epidermal growth factor ERK extracellular signal-regulated kinase FAK focal adhesion kinase FCS foetal calf serum GSK3 glycogen synthase kinase 3 Id inhibitor of DNA binding/differentiation INK inhibitor of cyclin-dependent kinase I-Smad inhibitory Smad JNK c-jun N-terminal kinase LAP latent-associated peptide LEF1 lymphoid enhancer-binding factor 1 LMM lentigo maligna melanoma LOH loss of heterozygosity LTBP latent TGF-β binding proteins MAPK mitogen activated-protein kinase MC1R melanocortin 1 receptor MEK MAPK kinase MEKK MAPK kinase kinase MH Mad-homology

10 Mitf microphthalmia-associated transcription factor MMP matrix metalloproteinase MOI multiplicity of infection N-cadherin neuronal cadherin NM nodular melanoma PAI-1 plasminogen activator inhibitor-1 PAX3 paired box gene 3 PBS phosphate-buffered saline PDGF platelet-derived growth factor PI3K phosphatidylinositol-3-kinase PKC protein kinase C pRb retinoblastoma tumour suppressor protein PTEN phosphatase and tensin homolog RGP radial growth phase R-Smad receptor-regulated Smad RT-PCR reverse transcription polymerase chain reaction SBE Smad-binding element Smad homolog of mothers against decapentaplegic, drosophila Smurf Smad ubiquitination regulatory factor SOX10 SRY (sex-determining region Y)-box 10 SOM self-organised map SSM superficial spreading melanoma TAK1 TGF-β activated kinase-1 TβR TGF-β receptor TGF-β transforming growth factor beta TGIF TGF-β-induced factor TNM primary tumour/regional lymph node/metastases tPA tissue-type plasminogen activator uPA urokinase-type plasminogen activator UV ultraviolet VEGF vascular endothelial growth factor VGP vertical growth phase Wnt Wingless-type MMTV integration site family

11 12 Chapter 1

1 Introduction

13 Chapter 1 Introduction

1.1 Definition

Melanoma is an aggressively metastasising tumour originating from neural-crest derived melanocytes, specialised pigment cells found predominantly in the skin and eyes. While only 3-5% of melanomas are found in the eye, more than 90% occur in the skin (Houghton and Polsky, 2002), and these are the focus of this thesis. Cutaneous melanocytes, which account for approximately 1-3% of cells in the adult epidermis, reside on the epidermal basement membrane (Hoath and Leahy, 2003). Melanocytes produce melanin in endocytic vesicles known as melanosomes. Melanosomes are derived from smooth endoplasmic reticulum and contain a large number of enzymes involved in melanin synthesis (reviewed in Lin and Fisher, 2007). Skin pigmentation results from the transfer of mature melanosomes from melanocytes’ dendrites to neighbouring keratinocytes (Boissy, 2003). In keratinocytes, melanosomes are dispersed in the cytoplasm, and upon UV irradiation form cap-like structures around the nuclei to protect them from UV-induced DNA damage (reviewed in Lin and Fisher, 2007).

Figure 1.1. Melanocyte structure and function. Melanocytes reside on the basement membrane where they produce melanin within melanosomes. Skin pigmentation results from the transfer of mature melanosomes from melanocyte dendrites to neighbouring keratinocytes.

1.2 Clinical features

Clinically, primary melanomas present as any of four major subtypes. The first three, superficial spreading melanoma (SSM), nodular melanoma (NM) and lentigo maligna melanoma (LMM), were first described by Wallace Clark in 1967 (Clark, 1967). Acral

14 Chapter 1 Introduction lentiginous melanoma (ALM) was described ten years later by Arrington and co-workers (Arrington et al., 1977). Accounting for almost two thirds of all primary melanomas, SSM is the most common subtype (World Health Organization, 2006). It may arise in almost any part of the skin but is most frequent on sites with acute-intermittent sun exposure (World Health Organization, 2006). SSM presents as an asymmetrical flat skin lesion with irregular pigmentation and border as well as regression phenomena (Fig. 1.2 A) (Hengge and Dummer, 2006). NM is the second most common subtype and may also originate in any part of the skin, particularly on the trunk, head and neck (World Health Organization, 2006) It is often a papular lesion with pigmentation changes and irregularly spreading borders (Fig. 1.2B) (Hengge and Dummer, 2006). LMM is generally flat in appearance and occurs on chronic sun-exposed areas of elderly people, mostly on the face but also on extrafacial sites including the neck, upper back and forearm (Fig. 1.2C) (World Health Organization, 2006). ALM is a relatively rare subtype occurring in approximately 5% of all cases in caucasians but in more than 50% of cases in dark-skinned populations (Chu, 1999). This subtype of melanoma occurs on the palms, soles and subungual sites (Fig. 1.2D) (World Health Organization, 2006). In addition, there are several rare variants such as amelanotic, desmoplastic, verrucous, and polypoid melanoma, which together comprise approximately 5% of all cases (Hengge and Dummer, 2006). Although used clinically as descriptive tools, the different classifications have little prognostic value or therapeutic significance.

Figure 1.2. Clinical presentation of melanoma. (A) Superficial spreading melanoma. (B) Nodular melanoma. (C) Lentigo maligna melanoma. (D) Acral lentiginous melanoma

15 Chapter 1 Introduction

Clinicians have advocated the so called “ABCDE rule” in the early identification of melanoma. Clinical characteristics of melanoma include: Asymmetry of lesions, Border irregularity, Colour variations, Diameter greater than 6mm and Evolving (with respect to size, shape, shades of colour, surface features or symptoms) (Abbasi et al., 2004; Friedman et al., 1985). However, the diagnosis of pigment lesions is difficult for even the best clinicians and clinical diagnoses rarely exceed 60% accuracy, histopathological examination must be used for accurate diagnosis (Houghton and Polsky, 2002).

1.3 Pathological features and staging

1.3.1.1 Radial and vertical growth phase Location and depth of involvement are key features used in histopathological diagnosis. Primary melanomas typically progress through two well-defined phases. The initial stage of tumour progression, the radial growth phase (RGP), is characterised by a flat or plaque-like lesion expanding horizontally and confined to the epidermis (Clark et al., 1969; Gimotty et al., 2005). The second stage of progression, the vertical growth phase (VGP), is characterised by invasion through the basement membrane into the dermis and underlying subcutaneous tissues (Clark et al., 1975).

1.3.2 Clark’s level of invasion and Breslow’s thickness

Tumour thickness and depth of invasion are the most accurate prognostic features for melanomas. Clark’s level of invasion classification method correlates the anatomic level of invasion and the mitotic index (measure of cell proliferation, ratio between the number of cells in mitosis and the total number of cells) to prognosis (Clark, 1967). Level I is represented by melanoma in situ (tumour cells confined to the epidermis, RGP). Progression from RGP to VGP is seen in level II where melanoma cells have expanded to the papillary dermis. When the latter is filled and has expanded, it is considered level III. Level IV is marked by the invasion into the reticular dermis. Invasion of the subcutaneous tissue is characteristic of level V (Clark, 1967) (Fig. 1.3).

16 Chapter 1 Introduction

Breslow’s method is a direct measurement of the tumour thickness from the top of the granular cell layer to the deepest invasive tumour cell (Breslow, 1970) (Fig. 1.3). This second method has been accepted as a more reliable indicator of prognosis as it is more easily assessed and more objective than Clark’s method (reviewed in Chin et al., 1998).

Figure 1.3. Pathological classification of melanoma. Breslow thickness measures the melanoma thickness from the granular cell layer to the deepest level of invasion in millimetres. Clark’s classification system correlates prognosis to the anatomical level of involvement.

1.3.3 TNM staging

The widely used TNM staging system was developed by the American Joint Committee on Cancer and the International Union Against Cancer (AJCC/UICC). In its acronym, T stands for primary tumour, N for regional lymph nodes, and M for metastases. Stage 0 melanomas are non-invasive and have not broken the integrity of the epidermal basement membrane. Stage I (≤ 2 mm according to Breslow’s method) and stage II melanomas are localised primary tumours. Stage III is characterised by regional spread through lymphatic vessels and stage IV by distant metastasis (Balch et al., 2004; Thompson, 2002). The complete overview of this staging system can be found in the references mentioned above and a summary can be found in Table 1.1.

17 Chapter 1 Introduction

Table 1.1. TNM classification (Balch et al., 2004) T Primary Tumour

Tis In situ T1 ≤ 1,0 mm - a) without ulceration / b) with ulceration T2 1.01 – 2.0 mm a) without ulceration / b) with ulceration T3 2.01 – 4.0 mm a) without ulceration / b) with ulceration T4 > 4.0 mm a) without ulceration / b) with ulceration N Regional Lymph nodes N1 One lymph node a) micrometastasis1 b) macrometastasis2 N2 2-3 lymph nodes a) micrometastasis1 b) macrometastasis2 c) in-transit metastases / satellite metastases ------without metastatic lymph nodes N3 ≥ 4 metastatic lymph nodes, matted lymph nodes or combinations of in- transit metastases / satellite(s) or ulcerated melanoma and metastatic lymph nodes M Distant Metastases M1a Distant skin, subcutaneous, or lymph node metastases (normal LDH) M1b Lung metastases (normal LDH) M1c All other visceral (normal LDH) or any distant metastases (elevated LDH)

LDH = Lactate dehydrogenase 1 Micrometastases are diagnosed after elective or sentinel lymphadenectomy. 2 Macrometastases are defined as clinically detectable lymph node metastases confirmed by therapeutic lymphadenectomy or when any lymph node metastasis exhibits gross extracapsular extension.

1.4 Epidemiology

Despite growing awareness of the disease, the incidence of melanoma in developed countries continues to increase dramatically. Melanoma is recognised as being the most dangerous of skin cancers and five-year survival of patients with metastases is only 14% (Jemal et al., 2004).

As is the case for most cancers, the incidence of melanoma increases with age. However, cutaneous melanoma is the second most common cancer in young British adults aged 20-39 years of age (Giblin and Thomas, 2007). In Switzerland, the incidence of melanoma compared to other cancers in young adults is also high with an average rate of 13 per 100,000

18 Chapter 1 Introduction men and 19 per 100,000 women (20-49 years of age) for the 2001-2003 period (information available on www.asrt.ch).

Genetic and environmental factors are involved in the development of melanoma. Skin pigmentation and geographical location correlate with incidence. Caucasians have a 20-fold increased risk of developing cutaneous melanoma when compared to darker skin populations, and the risk increases as Caucasians live in countries with higher UV indices (Giblin and Thomas, 2007). The latter is well reflected in the incidence of cutaneous melanoma in New Zealand and Australia, which are the highest in the world (Giblin and Thomas, 2007).

The high incidence of melanoma in Caucasian populations living closer to the equator, together with the anatomical distribution of tumours, clearly highlights the importance of sun exposure in the development of the disease (Giblin and Thomas, 2007; Houghton and Polsky, 2002). Interestingly, short bursts of intense UV exposure are believed to be more harmful than chronic exposure. This would explain the higher incidence observed in professionals working indoors compared to outdoor workers (Houghton and Viola, 1981). Although sun exposure has been linked to melanoma incidence, the mechanisms involved remain unclear.

Together with environmental factors, genetic and phenotypic factors contribute greatly to melanoma incidence. Skin pigmentation, density and type of nevi, susceptibility to sunburn and a family history of melanoma are all important factors influencing the incidence of melanoma in a given individual (Houghton and Polsky, 2002).

1.5 Genes and pathways

1.5.1 Genes involved in familial melanoma

Individuals born into families with a history of melanoma have a 30-70% increase in the chance of developing melanoma (Pho et al., 2006). Although familial melanoma accounts for only 10% of all melanoma cases, it presents opportunities for understanding the genetic basis for melanoma susceptibility. Furthermore, diagnostic and prognostic methods for early detection and application of targeted therapies for all melanoma patients may be developed

19 Chapter 1 Introduction with the understanding of phenotypic and genotypic correlations in familial melanoma (Hayward, 2003; Pho et al., 2006).

Three melanoma-predisposing genes have been identified. The high-penetrance cell cycle related gene cyclin-dependent kinase inhibitor 2A (CDKN2A) is the most common mutated gene. It is located on chromosome 9p21 and encodes two cell cycle related proteins: inhibitor of cyclin-dependent kinase 4A (p16Ink4a) and ARF (alternate reading frame) through alternative splicing and translation of the products in different reading frames (Quelle et al., 1995). Mutations at this locus are present in 20-40% of melanoma families (reviewed in de Snoo et al., 2007). A second high-penetrance gene, although only present in a minority of melanoma families, is cyclin-dependent kinase 4 (CDK4) located on chromosome 12q13. CDK4 is the binding partner of p16Ink4a and mutations only occur at its binding domain (Soufir et al., 1998; Zuo et al., 1996). A third susceptibility-gene encoding the melanocortin 1 receptor (MC1R) shows low penetrance and therefore confers lower risk (Kennedy et al., 2001; Palmer et al., 2000; Valverde et al., 1996).

The cell cycle plays a critical role in cancer development and both high-penetrance genes, CDKN2A and CDK4, are involved in its control. Both mutations affecting p16Ink4a and CDK4 Ink4a disturb the G1 phase check point. Mutated p16 is no longer able to inhibit CDK4/6- mediated phosphorylation of the retinoblastoma protein (pRb) and, resulting in the same effect, mutated CDK4 is unable to bind functional p16Ink4a (Zuo et al., 1996). ARF induces cell cycle arrest by inhibiting MDM2-mediated ubiquitylation and degradation of p53 (Pomerantz et al., 1998; Stott et al., 1998).

The third melanoma-predisposing gene, MC1R, is not involved in cell cycle control but is involved in the regulation of the balanced production of brown/black eumelanin and yellow/red pheomelanin (Mountjoy et al., 1992). MC1R variants, particularly the ones associated with red hair, fair complexion, inability to tan and tendency to freckle, shift the balance to increase the amount of pheomelanin in the skin. Pheomelanin has diminished UV- light protective capacity, and therefore, melanocytes with such MC1R variants are more susceptible to the DNA damaging effects of UV radiation (Scott et al., 2002).

A single genetic defect is not solely responsible for familial melanoma but rather a group of genetic disorders individually contribute to the increase risk of melanoma observed in

20 Chapter 1 Introduction melanoma-prone families (Pho et al., 2006). Moreover, known germline mutations have not been identified in 50% of families with familial melanoma, indicating that unknown mutated loci are still to be identified (Pho et al., 2006).

1.5.2 PTEN

Loss of heterozygosity (LOH) and chromosomal rearrangements on chromosome 10 have been implicated in a large number of cancers, including melanoma (Wu et al., 2003). Loss of tumour suppressor genes on chromosome 10, including PTEN, is observed in 30-60% of sporadic melanomas (Bastian et al., 1998). Mutations or deletions of PTEN have been observed in 5-15% of uncultured melanomas and 30-40% of established cell lines (Guldberg et al., 1997; Teng et al., 1997). PTEN germline mutations are associated with three clinically related inheritable cancer syndromes: Cowden disease, Lhermitte-Duclos disease and Bannayan-Zonana syndrome (Chin, 2003).

PTEN signals down the phosphatidylinositol-3-kinase (PI3K) pathway through its lipid phosphatase function (Wu et al., 2003). PTEN negatively regulates PI3K signalling by dephosphorylating PIP3, inhibiting phosphorylation and activation of AKT. AKT activation promotes the downregulation of antiapoptotic proteins such as Bcl-2 and the upregulation of proapoptotic signals. Since activation of AKT enhances cellular survival, growth and proliferation, loss of PTEN regulation of AKT results in reduced proliferative and apoptotic control (Wu et al., 2003).

PTEN has another phosphatase function which targets proteins. Dephosphorylation of focal adhesion kinase (FAK) has been shown to inhibit focal adhesion and migration, and further decrease PI3K activity (Tamura et al., 1999; Tamura et al., 1998). Additionally, PTEN is thought to interact with growth factor-stimulated mitogen activated-protein kinase (MAPK) signalling by dephosphorylating adapter proteins such as Shc and IRS, resulting in reduced MEK1/2 (MAPK kinase) and ERK1/2 (extracellular signal-regulated kinase) phosphorylation (Wu et al., 2003).

21 Chapter 1 Introduction

1.5.3 MAPK

Because of its role in the regulation of cell growth, survival and invasion, the mitogen- activated protein kinase (MAPK) pathway is implicated in a large number of cancers. In melanoma, activation of this pathway occurs through paracrine and autocrine growth factor stimulation, adhesion receptor signalling, and activating mutations in signalling molecules or in growth-factor receptors such as c-Kit (Gray-Schopfer et al., 2007; Meier et al., 2005).

Activating RAS mutations have been detected in approximately 9-15% of melanomas, with NRAS mutants accounting for the large majority (Meier et al., 2005). However, the most common mutation in the MAPK pathway is found in BRAF. BRAF somatic missense mutations have been detected in 66% of malignant melanomas (Davies et al., 2002). All detected mutations were found in the kinase domain, with a single substitution (V600E) accounting for 80% of all such cases (Davies et al., 2002). Interestingly, 82% of benign nevi also present with BRAF mutations, suggesting that although the activating mutation may be involved in melanoma development, it is not sufficient for malignant transformation (Pollock et al., 2003).

In vitro studies support an oncogenic role for mutant BRAF. Small interfering RNA against BRAFV600E was shown to inhibit MAPK activation, induce growth arrest and apoptosis, and inhibit colony formation in soft agar (Hingorani et al., 2003). Furthermore, expression of BRAFV600E was shown to transform murine melanocytes and induce tumorigenicity in nude mice (Wellbrock et al., 2004).

BRAFV600E is not a typical ultraviolet fingerprint mutation; the transversion of T to A resulting in the substitution at nucleotide 1799 is distinct from UV-induced pyrimidine dimer formation (Davies et al., 2002). Rather than a direct relationship between UV radiation and the presence of BRAF mutations, it has been shown that BRAF mutations in melanomas arising on intermittently sun-exposed skin are statistically more common than melanomas on chronically sun-damaged skin and tissues relatively or completely unexposed to sunlight (acral and mucosal melanomas) (Curtin et al., 2005; Maldonado et al., 2003).

Regulation of cell growth by the MAPK pathway involves the downstream target cyclin D1 (CCND1) (Bhatt et al., 2005). Interestingly, Curtin and co-workers showed that focused

22 Chapter 1 Introduction amplification resulting in increased number of copies of CCND1 was inversely correlated with mutations in BRAF (Curtin et al., 2005). Also, increased CCND1 protein expression was exclusively correlated with either mutations in BRAF or NRAS or increased copy number of CCND1, suggesting an important role for elevated levels of CCND1 in melanoma progression (Curtin et al., 2005). Furthermore, no mutations in BRAF or NRAS or amplification of CCND1 were detected in tumour samples with CDK4 amplifications, indicating that the functions of the MAPK and CCND1/CDK4 pathways overlap and their oncogenic functions in melanoma are independent (Curtin et al., 2005).

Aside from its role in MAPK signalling, RAS is a known positive upstream regulator of the PI3K pathway (Rodriguez-Viciana et al., 1994; Rodriguez-Viciana et al., 1996). Activation of RAS activates both the PI3K and the MAPK pathway, abrogating the need for PTEN inactivation and BRAF activation. Supporting this model, Tsao and co-workers showed that NRAS was exclusively mutated in melanoma, while BRAF and PTEN mutations were found simultaneously (Tsao et al., 2004).

1.5.4 Wnt

Wnt signalling promotes pigment cell formation from neural crest cells and is therefore crucial for the development of melanocytes from their neural crest precursors (Dorsky et al., 1998). The 19 members of the Wnt family of proteins, with the numerous receptors and co- receptors, underlie the complexity and diversity of Wnt signalling. Three distinct pathways have so far been identified; the canonical β-catenin pathway, the Wnt/Ca+ pathway, and the planar cell polarity pathway (Weeraratna, 2005).

Although a significant number of stabilising β-catenin mutations have shown to contribute to tumorigenesis in diverse cancers and such mutations were reported to be prevalent in melanoma cell lines (Rubinfeld et al., 1997), they are much less important in primary malignant melanoma (Rimm et al., 1999). However, evidence for the activation of the canonical β-catenin pathway through the immunohistochemical detection of nuclear β-catenin in a subset of primary melanomas suggests a role for this pathway in melanoma development (Rimm et al., 1999).

23 Chapter 1 Introduction

The recent evidence implicating WNT5A in melanoma invasion supports a role for the non- canonical Wnt/Ca+ pathway in melanoma progression. Bittner and co-workers, using microarray analysis, identified WNT5A as the gene that best characterised melanoma metastases with a highly motile and invasive in vitro phenotype (Bittner et al., 2000). In a follow-up study, overexpression of WNT5A was shown to increase the activity of protein kinase C (PKC), induce actin reorganisation and increase cell adhesion, motility and invasion (Weeraratna et al., 2002). Furthermore, WNT5A expression in human melanoma biopsies directly correlated to increasing tumour grade (Weeraratna et al., 2002). In a recent publication, the Weeraratna group showed that WNT5A/PKC stimulate cell motility through the regulation of genes associated with an epithelial to mesenchymal transition (EMT), such as vimentin, Snail and E-cadherin (Dissanayake et al., 2007).

The planar cell polarity pathway, through the activation of the GTPases Rho, Rac and cdc42, is a major activator of the Rho and the JNK/MAPK pathways. In melanoma, RhoC was shown to enhance metastasis in vivo (Clark et al., 2000), and cdc42 and Rac mediate the formation of invadopodia (Nakahara et al., 2003).

1.5.5 Microphthalmia-associated transcription factor

Microphthalmia-associated transcription factor (Mitf) is a member of the Myc-related family of basic helix-loop-helix leucine zipper transcription factors. It modulates the expression of various differentiation and cell-cycle progression genes such as Bcl2 and CDK2, as well as the expression of melanogenic proteins such as tyrosinase, silver homologue (gp100) and melanoma-associated antigen recognised by T cells-1 (MART-1 or melan-A)(Levy et al., 2006). Mitf is recognised as the master regulator of melanocyte development, function and survival (Chin et al., 2006; Levy et al., 2006).

The promoter directing melanocytic-specific expression of Mitf contains regulatory elements associated with paired box gene 3 (PAX3), SRY (sex-determining region Y)-box 10 (SOX10), lymphoid enhancer-binding factor 1 (LEF1) and Mitf itself (Jacquemin et al., 2001; Saito et al., 2002; Steingrimsson et al., 2004; Yasumoto et al., 2002). Post-translationally, Mitf is also modified by the c-Kit/MAPK pathway. Erk kinase phosphorylation of Mitf has been shown to enhance its affinity for the p300/CBP transcriptional coactivator (Hemesath et

24 Chapter 1 Introduction al., 1998; Price et al., 1998) as well as trigger its ubiquitylation and degradation (Wellbrock and Marais, 2005; Wu et al., 2000).

Mitf was first considered as an oncogene after Garraway and co-workers detected amplification of the Mitf gene in 10% of primary cutaneous and 15-20% of metastatic melanomas using a high-density single nucleotide polymorphism array (Garraway et al., 2005). In addition, Mitf amplification was associated with decreased 5-year survival in patients presenting with metastatic melanoma (Garraway et al., 2005). In genetically modified human melanocytes, in which the p53 and p16Ink4a/CDK4/pRb pathways were inactivated, expression of Mitf together with activated BRAFV600E conferred robust factor-independent and anchorage-independent growth, demonstrating that deregulated Mitf expression could cooperate with BRAFV600E to transform human melanocytes (Garraway et al., 2005). Also, reduction of Mitf expression sensitised melanoma cells to conventional chemotherapeutic agents (Garraway et al., 2005).

Mitf amplification is only seen in a subset of melanomas. Its expression is variable across specimens, with some studies reporting decreased Mitf expression in advanced melanoma (Salti et al., 2000; Selzer et al., 2002; Vachtenheim et al., 2001). The existence of different subsets of melanomas has been suggested to explain the different Mitf expression patterns observed across specimens (Levy et al., 2006). It has also been proposed that Mitf expression must be kept within narrow limits as it regulates distinct functions in melanocytic cells at different levels of expression; cell cycle and differentiation at high levels and cell cycle arrest and apoptosis at critically low levels (Carreira et al., 2005; Gray-Schopfer et al., 2007).

1.6 The transforming growth factor-β super-family

The members of the transforming growth factor-β (TGF-β) super-family are multifunctional cytokines controlling diverse cellular processes, including cell proliferation and differentiation, modification of the microenvironment, as well as cell fate determination and patterning during embryogenesis (reviewed in Shi and Massague, 2003; Siegel and Massague, 2003). The TGF-β super-family of cytokines comprises over 40 human members including TGF-βs, activins, nodal, bone morphogenic proteins (BMPs), the Müllerian inhibiting

25 Chapter 1 Introduction substance (MIS), and a number of other structurally related ligands (reviewed in Massague and Gomis, 2006).

TGF-β family cytokines signal through two transmembrane serine/threonine kinase receptors, a type II ligand binding receptor and a type I signal transducing receptor. Upon ligand binding to a type II receptor at the cell surface, a type I receptor is recruited and phosphorylated by the type II receptor. Phosphorylation leads to the activation of its receptor kinase domain and subsequent signalling through the Smad proteins (reviewed in Shi and Massague, 2003; Siegel and Massague, 2003) (Fig. 1.4A).

Figure 1.4. TGF-β-like signalling. (A) Upon ligand binding, the type II receptor recruits and phosphorylates the type I receptor. Phosphorylation leads to the activation of its receptor kinase domain. The activated type I receptor recruits and phosphorylates receptor-regulated Smads (R- Smads). Phosphorylated R-Smads bind to the co-mediator Smad (Co-Smad) and translocate to the nucleus where the heteromeric complex interacts with other transcriptional factors (TF) as well as co- activators and co-represors (co). SBE: Smad-binding element, TBE: transcription factor binding element. (B) Combinations of type II and type I recepors.

26 Chapter 1 Introduction

1.6.1 Receptors and Smad proteins

Multiple combinations of the seven human type I receptors (ALKs 1-7) and the five type II receptors (TβR-II, ActR-IIA, ActR-IIB, BMPR-II, and AMHR-II) are responsible for the transmission of signals initiated by the binding of any of over 40 functionally relevant ligands identified in the human genome (Roberts and Derynck, 2001). (Fig. 1.4B). Activated type I receptors recruit and phosphorylate receptor-regulated Smads (R-Smads) for signal propagation (reviewed in Shi and Massague, 2003; Siegel and Massague, 2003). Smad2 and Smad3 R-Smads are effectors of TGF-β, activin and nodal signalling, while Smad1, Smad5 and Smad8 are involved in BMP signalling (reviewed in Shi and Massague, 2003; Siegel and Massague, 2003). Phosphorylated R-Smads bind to the co-mediator Smad (Co-Smad) Smad4 and translocate to the nucleus where the heteromeric complex, in conjunction with other nuclear factors, regulate the transcription of target genes (reviewed in Shi and Massague, 2003; Siegel and Massague, 2003) (Fig. 1.4A).

Smad proteins are characterised by two conserved Mad-homology (MH) domains, MH1 and MH2, bound by a variable proline-rich linker region. The MH2 domain is responsible for interaction with anchors for cytoplasmic retention, receptors for activation, nucleoporins for nucleocytoplasmic translocation, and partner Smad proteins and other nuclear factors for the formation of transcriptional complexes (reviewed in Massague and Gomis, 2006). Via their MH1 domain, Smad3 and Smad4 interact directly with the CAGA sequence, which is known as the Smad-binding DNA element (SBE) (Jonk et al., 1998; Zawel et al., 1998). R-Smads share more than 90% homology at the amino acid level, but an extra 30 amino acid insertion in the MH1 domain of Smad2 inhibits its ability to bind DNA (Dennler et al., 1999). The affinity of Smad3 and Smad4 for the SBE is too low to confer promoter selectivity, therefore additional transcriptional co-activators and co-repressors are required for selective transcription regulation (reviewed in Massague and Gomis, 2006). An additional subclass of Smad proteins is composed of the inhibitory Smads (I-Smads), Smad6 and Smad7. They have been shown to antagonise TGF-β signalling by a number of means, including competing with R-Smads for receptor binding, recruiting E3-ubiquitin ligases (known as Smad ubiquitination regulatory factor 1 and 2 (Smurf1 and Smurf2) leading to type I receptor degradation or by recruiting factors leading to type I receptor dephosphorylation (reviewed in Dijke ten and Hill, 2004).

27 Chapter 1 Introduction

1.6.2 Transforming growth factor-β

Three mammalian TGF-β isoforms, TGF-β1-3, have been identified. Although encoded by three distinct genes, they share 76-80% amino acid sequence homology. The TGF-β cytokine is synthesised as a pre-proTGF-β containing the C-terminal mature TGF-β and an N-terminal pro-domain, the latent-associated peptide (LAP), which are cleaved in the Golgi apparatus by a furin-like endoproteinase (reviewed in Koli et al., 2001). After cleavage, the LAP dimer remains associated with the mature TGF-β dimer through non-covalent interaction. This complex, known as the small latent complex, is secreted, or in most cases associates covalently through the LAP to one of four latent TGF-β binding proteins (LTBP) to be secreted as a large latent complex. Release of the mature TGF-β molecule is induced by cleavage of the LAP by various proteases or by physical interactions of the LAP with other proteins (reviewed in Javelaud and Mauviel, 2004; Koli et al., 2001). TGF-β then mediates its effect through the heteromeric receptor complex consisting of TβR-II and ALK5.

1.6.2.1 Transforming growth factor-β and cancer

TGF-β is a multifunctional cytokine which controls aspects of cell proliferation, differentiation, migration, apoptosis, adhesion, angiogenesis, immune surveillance, and survival. TGF-β was initially defined as a transforming cytokine, hence it’s name, based on its ability to induce anchorage independent growth of normal fibroblasts (Todaro et al., 1980). On the other hand, TGF-β was also considered a tumour suppressor when it was shown to have growth-suppressive effects in a number of cell types (reviewed in Wakefield and Sporn, 1990). It is now understood that TGF-β has dual roles both as tumour suppressor and tumour promoter.

TGF-β’s tumour suppressor role is well exemplified by its cytostatic function. TGF-β regulates a number of important proteins involved in controlling cell cycle progression from G1 to S phase. Among the activated TGF-β target genes identified are the cyclin-dependent kinase (CDK) inhibitors p21Cip1 (Datto et al., 1995), p27Kip1 (Kamesaki et al., 1998) and p15Ink4b (Hannon and Beach, 1994). p15Ink4b interacts and inactivates CDK4 and CDK6, and thus prevents progression through the G1/S restriction point (Hannon and Beach, 1994). It also binds cyclin D complexes of CDK4 or CDK6, inactivating them and displacing p21Cip1

28 Chapter 1 Introduction and p27Kip1 which are then free to bind and inactivate CDK2 complexes with cyclin A or E (Polyak et al., 1994). p21Cip1 also directly inhibits DNA synthesis by interacting with proliferating cell nuclear antigen (PCNA), a polymerase delta accessory factor (Waga et al., 1994). Activation of the gene encoding p15Ink4b is further enhanced by the downregulation of c-Myc. TGF-β signalling represses the expression of c-Myc and therefore inhibits its recruitment to the p15Ink4b promoter, relieving c-Myc induced repression of p15Ink4b (Seoane et al., 2001; Staller et al., 2001). p57Kip2, another member of the Kip/Cip family of CDK inhibitors, has been shown to play similar roles as p21Cip1 and p27Kip1 in hematopoietic cells, in which it is transcriptionally induced by TGF-β (Scandura et al., 2004).

In addition to inhibiting proliferation, TGF-β may also induce apoptosis (reviewed in Schuster and Krieglstein, 2002). For example, TGF-β has been shown to regulate the expression of pro- and anti-apoptotic molecules, including p53, Bad, Bax, Bik, Bcl-2 and Bcl-XL (reviewed in Jakowlew, 2006).

Although only a small fraction of tumour cells resistant to TGF-β cytostatic effects have mutations in genes encoding TGF-β receptors or Smad proteins, the presence of these inactivating mutations in certain cancers suggests a tumour suppressor role for TGF-β. Inactivating mutations in the gene for TβR-II are found in colon and gastric cancers, as well as in glioma (Chung et al., 1996; Izumoto et al., 1997; Ku et al., 2007; Markowitz et al., 1995). Inactivating mutations in the gene encoding ALK5 are less frequent but are found in breast, colorectal, ovarian, head and neck, and pancreatic cancers, and T-cell lymphoma (Chen et al., 1998; Chen et al., 2001a; Chen et al., 2001b; Goggins et al., 1998; Ku et al., 2007; Schiemann et al., 1999).

Mutations in Smad2 are relatively rare and are limited to a few cases in colon, lung, gastric, and head and neck cancers (Han et al., 2004; Qiu et al., 2007; Riggins et al., 1996; Tsang et al., 2002; Uchida et al., 1996). Smad4 mutations occur in approximately half of cases of pancreatic cancer and are also detected at a lesser frequency in other carcinomas including colon carcinomas (Hahn et al., 1996; Schutte et al., 1996). Loss of Smad4 also correlates with cancer progression and metastasis (Losi et al., 2007; Luttges et al., 2001; Xu et al., 2000). Recently, the first Smad3 mutation inhibiting the nuclear translocation of the encoded

29 Chapter 1 Introduction protein and contributing to human carcinogenesis was detected in colorectal cancer (Ku et al., 2007).

Genetically modified mouse models have strengthened the evidence for a tumour suppressor role for TGF-β signalling. For example, re-expression of TβR-II or ALK5 in cancers with low levels of the corresponding receptors reduced tumorigenicity in a colon cancer mouse model (Wang et al., 1996; Wang et al., 1995).

Experimental and clinical observations also support a pro-oncogenic role for TGF-β signalling. Interestingly, TGF-β’s tumour suppressive and pro-oncogenic roles are not dependent on specific molecular etiology of the tumour or the specific cell type of origin. Tang and co-workers have shown that TGF-β signalling can switch roles during the course of carcinogenic progression in a given cell lineage with a defined initiating oncogenic event (Tang et al., 2003). The introduction of a dominant negative TβR-II could cooperate with an initiating oncogenic lesion to increase the frequency of malignant transformation of breast epithelial cells and increase the aggressiveness of a low-grade breast carcinoma (Tang et al., 2003). Conversely, the loss of TGF-β signalling did not affect primary tumorigenesis, but suppressed metastasis in a high-grade breast carcinoma (Tang et al., 2003). Strengthening the “switch hypothesis”, TGF-β’s dual role in carcinogenesis has also been demonstrated in a single animal model. In a murine Neu-induced mammary tumorigenesis model, Siegel and co- workers showed that activation of TGF-β signalling delayed the appearance of primary mammary tumours, but enhanced the frequency of lung metastasis (Siegel et al., 2003). On the other hand, expression of the dominant negative TβR-II decreased the latency of tumour formation while significantly reducing the incidence of extravascular lung metastases (Siegel et al., 2003). Interestingly, TGF-β seems to use the same effector proteins for its two opposite functions, as Tian and co-workers showed that interference with endogenous Smad2/3 signalling enhanced tumorigenesis but also strongly suppressed lung metastasis of breast cancer cell lines (Tian et al., 2003). Together, these data suggest that while TGF-β signalling suppresses proliferation processes it drives metastatic processes.

30 Chapter 1 Introduction

1.6.2.2 Transforming growth factor-β and melanoma

Melanomas produce and secrete all three TGF-β isoforms (Krasagakis et al., 1999; Rodeck et al., 1994; Van Belle et al., 1996). While melanomas express TGF-β constitutively, melanocytes require exogenous growth factor stimulation for TGF-β production (Rodeck et al., 1994). Comparable to what is described for other cancers, normal melanocytes are growth inhibited by TGF-β, while melanomas show various degrees of resistance to TGF-β-induced growth inhibition (Krasagakis et al., 1999; Rodeck et al., 1994). However, in addition to decreased susceptibility some melanomas show increased proliferation rates in response to TGF-β (Rodeck et al., 1999). In contrast to some epithelial malignancies which have acquired their resistance to the cytostatic effect of TGF-β through mutations of TGF-β signalling components, melanomas present intact Smad-dependent regulation of gene expression, regardless of their proliferative response to TGF-β (Rodeck et al., 1999). In the same study, a number of melanomas presented constitutive Smad-dependent transcription, which is partly thought to be the result of endogenously produced TGF-β (Rodeck et al., 1999).

As is the case for other cancers, TGF-β has both tumour suppressor and tumour promoter roles in melanoma. On one hand, TGF-β has been shown to strongly induce plasminogen activator inhibitor-1 (PAI-1) synthesis, leading to a significant decrease in tissue-type plasminogen activator (tPA) and urokinase-type plasminogen activator (uPA) secretion, which in turn led to tumour growth inhibition in a mouse melanoma model (Ramont et al., 2003). On the other hand, metastasis has also been shown to be facilitated by TGF-β’s induction of matrix metalloproteinase-9 (MMP9) and β1 and β3 integrins, downregulation of E-cadherin expression (Janji et al., 1999), as well as by enhancement of melanoma cell adhesion to the endothelium (Teti et al., 1997). Also, TGF-β stimulation of neighbouring stromal fibroblasts which translated into an increased production and deposition of extracellular matrix proteins was shown to increase survival and metastasis of melanoma in mice (Berking et al., 2001).

Overexpression of the inhibitory Smad protein Smad7 has also been shown to inhibit tumorigenicity in vitro and in vivo (Javelaud et al., 2005). Stable expression of Smad7 reduced MMP-2 and MMP-9 secretion which resulted in reduced in vitro invasion without affecting motility or adhesion. Furthermore, it reduced both anchorage-independent growth in vitro and the capacity to form a primary tumour in a xenograft transplantation model

31 Chapter 1 Introduction

(Javelaud et al., 2005). Using the same Smad7-overexpressing melanoma cell line, Javelaud and co-workers demonstrated the role of TGF-β in promoting bone metastasis (Javelaud et al., 2007).

TGF-β signalling has also been shown to play a role in melanogenesis. Using an immortalised mouse melanocyte cell line, Kim and co-workers demonstrated that TGF-β reduced Mitf promoter activity, resulting in decreased Mitf, tyrosinase, tyrosinase-related protein-1 (Trp-1) and Trp-2 protein production, which in turn led to reduced tyrosinase activity and melanin synthesis (Kim et al., 2004). The inhibition of these effects by a specific Erk-pathway inhibitor suggests this pathway mediates TGF-β’s regulation of melanogenesis (Kim et al., 2004).

1.6.3 Activin

Activin, another TGF-β-family cytokine, was initially isolated based on its ability to stimulate follicle-stimulating hormone release by the anterior pituitary, hence its name (Ling et al., 1986). Activin has now been shown to regulate a variety of events, including cell proliferation, differentiation, apoptosis, homeostasis, immune response, wound healing and endocrine function (reviewed in Chen et al., 2006). Activins are dimeric proteins composed of two β subunits linked by a single disulfide bond. In human, the two most common subunits are βA and βB which, when linked, form the two homodimers activin A and activin B, and the heterodimer activin AB. Additional β subunits (βC, βD and βE) have been identified, however the in vivo functions of the encoded polypeptides are still poorly defined. Activin signals through the type II activin receptors (ActRII and ActRIIB) and the type I receptors Alk2, ALK4, and Alk7, with Alk4 being the most important signal transducing receptor (reviewed in Chen et al., 2006).

1.6.3.1 Activin and cancer

As has been described for TGF-β, activin’s effects on growth and apoptosis have been linked to a putative role in cancer development. In human hepatoma cells, activin has been shown to induce cell cycle arrest through the induction of p15Ink4b expression (Ho et al., 2004) and to

32 Chapter 1 Introduction down-regulate the expression of the anti-apoptotic gene Bcl-xL (Kanamaru et al., 2002) in a Smad-dependent manner. Also, in breast cancer cells, activin has been shown to enhance the expression of p15Ink4b, reduce the expression of cyclin A and phosphorylate pRb (Burdette et al., 2005). Activin also inhibits proliferation of cells derived from other human tumours, including cells from gall bladder, prostate and pituitary gland (reviewed in Chen et al., 2006). However, activin has been shown to increase proliferation of ovarian cancer cell lines (Steller et al., 2005), highlighting a cell specific characteristic of activin’s effects on proliferation. Interestingly, activin A has also been suggested to inhibit TGF-β1-mediated inhibition of growth in differentiated human endometrial adenocarcinoma cells (Tanaka et al., 2004).

In opposition to the pro-angiogenic role described for TGF-β, activin is thought to function as an inhibitor of angiogenesis. Reintroduction of the activin gene in neuroblastoma cells has been shown to reduce the proliferation rate and induce the formation of smaller tumours with reduced vascularity in a xenograft mouse model (Panopoulou et al., 2005). Furthermore, both ALK4 overexpression and activin treatment of endothelial cells resulted in an increase in the expression of p15Ink4b, p21Cip1 and p27Kip1, and a decrease in the expression of vascular endothelial growth factor receptor 2 (VEGF receptor-2), a key angiogenic factor (Panopoulou et al., 2005).

A role for activin in cell invasion has also been suggested. Overexpression of activin A in esophageal carcinoma was shown to increase the expression of N-cadherin, a feature associated with depth of invasion and poor prognosis (Yoshinaga et al., 2004). Furthermore, activin A expression was significantly associated with N-cadherin in clinical samples (Yoshinaga et al., 2004).

1.6.3.2 Activin and melanoma

While the interest in the role of TGF-β in melanoma is significant, interest given to the role of the closely related activin molecule is limited. Hashimoto and co-workers reported on the differential expression of the βB subunit in high- and low-metastatic melanoma cell lines and consequently suggested that activin may play a role in metastasis (Hashimoto et al., 1996). Stove and co-workers identified follistatin as a secreted molecule inhibiting activin signalling in melanoma cells and characterised the activin/activin receptor system in a number of

33 Chapter 1 Introduction melanoma cell lines and melanocytes (Stove et al., 2004). They reported that all studied melanocytes and melanoma cell lines expressed mRNA for activin receptors and for the β subunits, primarily the βA subunit, suggesting a role for activin A. They further showed that melanocytes but not melanomas were growth inhibited by activin, and suggested that the secretion of follistatin by melanoma cells may represent an effective way to circumvent activin’s negative regulatory effects by binding to activin and preventing it from accessing its cell surface receptor (Stove et al., 2004).

1.7 Id proteins

Inhibitor of DNA binding/differentiation (Id) proteins were first recognised as regulators of differentiation but their role in a number of other biological processes, such as proliferation, cell-cycle regulation, angiogenesis, invasion and migration, has now been well documented (reviewed in Lasorella et al., 2001).

Id proteins act as dominant negative inhibitors of basic helix-loop-helix (bHLH) transcription factors. As opposed to the positively acting members of the bHLH family of transcription factors to which they bind, Id proteins lack a basic DNA binding domain and therefore prevent the dimers from binding to DNA and regulating transcription (reviewed in Lasorella et al., 2001). Id proteins can also bind to and alter the activities of other regulatory proteins, including the ternary complex factor subfamily of ETS-domain transcription factors (Yates et al., 1999) and the Pax-2/5/8 subfamily of transcription factors (Roberts et al., 2001). In addition, one member of the Id proteins, Id2, directly interacts with pRb and the related proteins p107 and p130 via its HLH domain (Iavarone et al., 1994; Lasorella et al., 1996).

1.7.1 Ids, cell cycle regulation and cancer

34 Chapter 1 Introduction

1 (Yates et al., 1999), and the CDK inhibitors p16Ink4a (Alani et al., 2001; Ohtani et al., 2001), p21Cip1 (Prabhu et al., 1997), p57Kip2 (Rothschild et al., 2006) and probably p27Kip1 (Lyden et al., 1999; Mori et al., 2000). Secondly, Id2 binds to the tumour suppressor proteins of the Rb family and when in large excess, abolishes their growth-suppressing activity by causing the release of E2F transcription factors required for cell cycle progression (Iavarone et al., 1994; Lasorella et al., 1996). In normal cells, Id2 is a downstream target of pRb and its family members which restrain its functions on natural targets. However, Id2 overexpressed by tumour cells can override the Rb pathway and deprive the cell of its most important cell cycle checkpoint (Lasorella et al., 1996). Furthermore, Rb pathway inhibition is believed to be achieved directly through the binding of Id2 but also indirectly through CDK inhibitor downregulation by all Id members (Ohtani et al., 2001).

Id2 has been shown to be a direct target of Myc transcription factors (Lasorella et al., 2000). The ability of Myc to promote cell cycle entry in the absence of growth factors and its ability to transform normal fibroblasts in cooperation with RAS have been shown to be strictly dependent on the presence of Id2 (Lasorella et al., 2000). Also, human neuroblastoma tumours presenting N-Myc amplification were shown to invariably overexpress Id2 (Lasorella et al., 2000). Furthermore, Lasorella and co-workers demonstrated the dependence of Myc on Id2 to overcome Rb-induced cell cycle block (Lasorella et al., 2000).

Id mRNA and protein levels have been reported to be elevated in a number of human tumours including carcinomas of the prostate, breast, ovary, colon, rectum, pancreas, liver, endometrium and thyroid, in squamous cell carcinomas of the nasopharynx, esophagus and oral cavity, in neural tumours, Ewing’s sarcoma, leukaemia, as well as melanoma (reviewed in Perk et al., 2005). In some cases, highs levels are associated with disease severity and poor prognosis.

1.7.2 TGF-β and Id proteins

Members of the TGF-β super-family of cytokines are known regulators of the Id genes. BMPs have been shown to upregulate the expression of Id genes in a wide range of cell lines and in embryonic stem cells (reviewed in Miyazono and Miyazawa, 2002). Conversely, TGF-β and

35 Chapter 1 Introduction activin A have been shown to repress the expression of Id genes in a number of epithelial cell lines and in keratinocytes (Kang et al., 2003; Kondo et al., 2004; Kowanetz et al., 2004; Ling et al., 2002; Rotzer et al., 2006). Furthermore, TGF-β-mediated repression of Id1 expression has been suggested to be involved in TGF-β-induced cell growth inhibition in keratinocytes and prostate epithelial cells (Di et al., 2006 ; Kang et al., 2003).

1.7.3 Id proteins and melanoma

In melanoma, Id levels have only been reported for Id1 and only limited publications report on Id2 and melanoma. Polsky and co-workers have demonstrated a correlation between Id1 expression and loss of p16Ink4a expression in early melanoma (Polsky et al., 2001). In a recent publication, the same group evaluated melanoma cell lines from different stages of progression for protein levels of Id1 with unconvincing results (Ryu et al., 2007). While two out of three RGP melanomas expressed high levels of Id1, only one out of four VGP melanomas but two out of three metastatic melanomas expressed high levels of Id1 (Ryu et al., 2007). Using tissue microarray, Straume and Akslen evaluated the expression of Id1 in 119 cases of nodular melanoma and showed that strong Id1 expression was significantly associated with increased tumour thickness and reduced survival (Straume and Akslen, 2005).

Id2 was first mentioned in association with melanoma when it was identified as a downregulated gene in melanomas with BRAFV600E or NRASQ61R mutations (Bloethner et al., 2005) and later in melanomas with homozygous deletion of the CDKN2A locus genes (Bloethner et al., 2006) in two global gene expression studies. In a microarray gene expression analysis of uveal melanoma, which generated two subgroups representing tumours with low and high risk of metastatic death, Id2 was one of the top class discriminating genes (Onken et al., 2006). Id2 was strongly downregulated in melanomas with high risk of metastatic death and suppression of Id2 expression in tumours with low risk of metastatic death rendered them more aggressive (Onken et al., 2006).

36 Chapter 1 Introduction

1.8 Aim

The aim of this thesis is to validate a new melanoma progression model we formulated from gene expression arrays performed on three distinct sets of melanoma cultures and to investigate the role of TGF-β-like signalling as a major player in this newly defined model.

37 Chapter 1 References

1.9 References

Abbasi, N.R., Shaw, H.M., Rigel, D.S., Friedman, R.J., McCarthy, W.H., Osman, I., Kopf, A.W. and Polsky, D. (2004) Early Diagnosis of Cutaneous Melanoma: Revisiting the ABCD Criteria. JAMA, 292, 2771-2776. Alani, R.M., Young, A.Z. and Shifflett, C.B. (2001) Id1 regulation of cellular senescence through transcriptional repression of p16/Ink4a 10.1073/pnas.141235398. Proceedings of the National Academy of Sciences, 98, 7812-7816. Arrington, J.H., 3rd, Reed, R.J., Ichinose, H. and Krementz, E.T. (1977) Plantar lentiginous melanoma: a distinctive variant of human cutaneous malignant melanoma. Am J Surg Pathol, 1, 131-143. Balch, C.M., Soong, S.J., Atkins, M.B., Buzaid, A.C., Cascinelli, N., Coit, D.G., Fleming, I.D., Gershenwald, J.E., Houghton, A., Jr., Kirkwood, J.M., McMasters, K.M., Mihm, M.F., Morton, D.L., Reintgen, D.S., Ross, M.I., Sober, A., Thompson, J.A. and Thompson, J.F. (2004) An evidence-based staging system for cutaneous melanoma. CA Cancer J Clin, 54, 131-149; quiz 182-134. Bastian, B.C., LeBoit, P.E., Hamm, H., Brocker, E.-B. and Pinkel, D. (1998) Chromosomal Gains and Losses in Primary Cutaneous Melanomas Detected by Comparative Genomic Hybridization. Cancer Res, 58, 2170-2175. Berking, C., Takemoto, R., Schaider, H., Showe, L., Satyamoorthy, K., Robbins, P. and Herlyn, M. (2001) Transforming Growth Factor-{beta}1 Increases Survival of Human Melanoma through Stroma Remodeling. Cancer Res, 61, 8306-8316. Bhatt, K.V., Spofford, L.S., Aram, G., McMullen, M., Pumiglia, K. and Aplin, A.E. (2005) Adhesion control of cyclin D1 and p27Kip1 levels is deregulated in melanoma cells through BRAF-MEK-ERK signaling. Oncogene, 24, 3459-3471. Bittner, M., Meltzer, P., Chen, Y., Jiang, Y., Seftor, E., Hendrix, M., Radmacher, M., Simon, R., Yakhini, Z., Ben-Dor, A., Sampas, N., Dougherty, E., Wang, E., Marincola, F., Gooden, C., Lueders, J., Glatfelter, A., Pollock, P., Carpten, J., Gillanders, E., Leja, D., Dietrich, K., Beaudry, C., Berens, M., Alberts, D. and Sondak, V. (2000) Molecular classification of cutaneous malignant melanoma by gene expression profiling. Nature, 406, 536-540. Bloethner, S., Chen, B., Hemminki, K., Muller-Berghaus, J., Ugurel, S., Schadendorf, D. and Kumar, R. (2005) Effect of common B-RAF and N-RAS mutations on global gene expression in melanoma cell lines 10.1093/carcin/bgi066. Carcinogenesis, 26, 1224-1232. Bloethner, S., Hemminki, K., Thirumaran, R.K., Chen, B., Mueller-Berghaus, J., Ugurel, S., Schadendorf, D. and Kumar, R. (2006) Differences in global gene expression in melanoma cell lines with and without homozygous deletion of the CDKN2A locus genes. Melanoma Res, 16, 297-307. Boissy, R.E. (2003) Melanosome transfer to and translocation in the keratinocyte. Exp Dermatol, 12 Suppl 2, 5-12. Breslow, A. (1970) Thickness, cross-sectional areas and depth of invasion in the prognosis of cutaneous melanoma. Ann Surg, 172, 902-908. Burdette, J.E., Jeruss, J.S., Kurley, S.J., Lee, E.J. and Woodruff, T.K. (2005) Activin A mediates growth inhibition and cell cycle arrest through Smads in human breast cancer cells. Cancer Res, 65, 7968-7975. Carreira, S., Goodall, J., Aksan, I., La Rocca, S.A., Galibert, M.-D., Denat, L., Larue, L. and Goding, C.R. (2005) Mitf cooperates with Rb1 and activates p21Cip1 expression to regulate cell cycle progression. Nature, 433, 764-769.

38 Chapter 1 References

Chen, T., Carter, D., Garrigue-Antar, L. and Reiss, M. (1998) Transforming Growth Factor {beta} Type I Receptor Kinase Mutant Associated with Metastatic Breast Cancer. Cancer Res, 58, 4805-4810. Chen, T., Triplett, J., Dehner, B., Hurst, B., Colligan, B., Pemberton, J., Graff, J.R. and Carter, J.H. (2001a) Transforming Growth Factor-{beta} Receptor Type I Gene Is Frequently Mutated in Ovarian Carcinomas. Cancer Res, 61, 4679-4682. Chen, T., Yan, W., Wells, R.G., Rimm, D.L., McNiff, J., Leffell, D. and Reiss, M. (2001b) Novel inactivating mutations of transforming growth factor-beta type I receptor gene in head-and-neck cancer metastases. Int J Cancer, 93, 653-661. Chen, Y.G., Wang, Q., Lin, S.L., Chang, C.D., Chuang, J. and Ying, S.Y. (2006) Activin signaling and its role in regulation of cell proliferation, apoptosis, and carcinogenesis. Exp Biol Med (Maywood), 231, 534-544. Chin, L. (2003) The genetics of malignant melanoma: lessons from mouse and man. Nat Rev Cancer, 3, 559-570. Chin, L., Garraway, L.A. and Fisher, D.E. (2006) Malignant melanoma: genetics and therapeutics in the genomic era. Genes Dev., 20, 2149-2182. Chin, L., Merlino, G. and DePinho, R.A. (1998) Malignant melanoma: modern black plague and genetic black box. Genes Dev, 12, 3467-3481. Chu, A.C. (1999) Malignant tumors of the skin. Arnold, London. Chung, Y.J., Song, J.M., Lee, J.Y., Jung, Y.T., Seo, E.J., Choi, S.W. and Rhyu, M.G. (1996) Microsatellite instability-associated mutations associate preferentially with the intestinal type of primary gastric carcinomas in a high-risk population. Cancer Res, 56, 4662-4665. Clark, E.A., Golub, T.R., Lander, E.S. and Hynes, R.O. (2000) Genomic analysis of metastasis reveals an essential role for RhoC. Nature, 406, 532-535. Clark, W. (1967) A classification of malignant melanoma in man correlated with histogenesis and biologic behaviour. In Montagna, W. and Hauhu, F. (eds.), Advances in Biology of the Skin, The Pigmentary System. Paramount Press, London, Vol. 8, pp. 621-647. Clark, W.H., Jr., Ainsworth, A.M., Bernardino, E.A., Yang, C.H., Mihm, C.M., Jr. and Reed, R.J. (1975) The developmental biology of primary human malignant melanomas. Semin Oncol, 2, 83-103. Clark, W.H., Jr., From, L., Bernardino, E.A. and Mihm, M.C. (1969) The histogenesis and biologic behavior of primary human malignant melanomas of the skin. Cancer Res, 29, 705-727. Curtin, J.A., Fridlyand, J., Kageshita, T., Patel, H.N., Busam, K.J., Kutzner, H., Cho, K.H., Aiba, S., Brocker, E.B., LeBoit, P.E., Pinkel, D. and Bastian, B.C. (2005) Distinct sets of genetic alterations in melanoma. N Engl J Med, 353, 2135-2147. Datto, M.B., Li, Y., Panus, J.F., Howe, D.J., Xiong, Y. and Wang, X.F. (1995) Transforming growth factor beta induces the cyclin-dependent kinase inhibitor p21 through a p53- independent mechanism. Proc Natl Acad Sci U S A, 92, 5545-5549. Davies, H., Bignell, G.R., Cox, C., Stephens, P., Edkins, S., Clegg, S., Teague, J., Woffendin, H., Garnett, M.J., Bottomley, W., Davis, N., Dicks, E., Ewing, R., Floyd, Y., Gray, K., Hall, S., Hawes, R., Hughes, J., Kosmidou, V., Menzies, A., Mould, C., Parker, A., Stevens, C., Watt, S., Hooper, S., Wilson, R., Jayatilake, H., Gusterson, B.A., Cooper, C., Shipley, J., Hargrave, D., Pritchard-Jones, K., Maitland, N., Chenevix-Trench, G., Riggins, G.J., Bigner, D.D., Palmieri, G., Cossu, A., Flanagan, A., Nicholson, A., Ho, J.W., Leung, S.Y., Yuen, S.T., Weber, B.L., Seigler, H.F., Darrow, T.L., Paterson, H., Marais, R., Marshall, C.J., Wooster, R., Stratton, M.R. and Futreal, P.A. (2002) Mutations of the BRAF gene in human cancer. Nature, 417, 949-954.

39 Chapter 1 References de Snoo, F.A., Kroon, M.W., Bergman, W., Ter Huurne, J.A., Houwing-Duistermaat, J.J., van Mourik, L., Snels, D.G., Breuning, M.H., Willemze, R., Frants, R.R. and Gruis, N.A. (2007) From sporadic atypical nevi to familial melanoma: Risk analysis for melanoma in sporadic atypical nevus patients. J Am Acad Dermatol. Dennler, S., Huet, S. and Gauthier, J.M. (1999) A short amino-acid sequence in MH1 domain is responsible for functional differences between Smad2 and Smad3. Oncogene, 18, 1643- 1648. Di, K., Ling, M.-T., Tsao, S.W., Wong, Y.C. and Wang, X. (2006 ) Id-1 modulates senescence and TGF-beta1 sensitivity in prostate epithelial cells. 10.1042/BC20060026. Biol. Cell 98, 523-533. Dijke ten, P. and Hill, C.S. (2004) New insights into TGF-beta-Smad signalling. Trends Biochem Sci, 29, 265-273. Dissanayake, S.K., Wade, M., Johnson, C.E., O'Connell, M.P., Leotlela, P.D., French, A.D., Shah, K.V., Hewitt, K.J., Rosenthal, D.T., Indig, F.E., Jiang, Y., Nickoloff, B.J., Taub, D.D., Trent, J.M., Moon, R.T., Bittner, M. and Weeraratna, A.T. (2007) The Wnt5A/Protein Kinase C Pathway Mediates Motility in Melanoma Cells via the Inhibition of Metastasis Suppressors and Initiation of an Epithelial to Mesenchymal Transition. J. Biol. Chem., 282, 17259-17271. Dorsky, R.I., Moon, R.T. and Raible, D.W. (1998) Control of neural crest cell fate by the Wnt signalling pathway. Nature, 396, 370-373. Friedman, R.J., Rigel, D.S. and Kopf, A.W. (1985) Early Detection of Malignant Melanoma: The Role of Physician Examination and Self-Examination of the Skin. CA Cancer J Clin, 35, 130-151. 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., Lee, C., Wagner, S.N., Li, C., Golub, T.R., Rimm, D.L., Meyerson, M.L., Fisher, D.E. and Sellers, W.R. (2005) Integrative genomic analyses identify MITF as a lineage survival oncogene amplified in malignant melanoma. Nature, 436, 117-122. Giblin, A.V. and Thomas, J.M. (2007) Incidence, mortality and survival in cutaneous melanoma. J Plast Reconstr Aesthet Surg, 60, 32-40. Gimotty, P.A., Van Belle, P., Elder, D.E., Murry, T., Montone, K.T., Xu, X., Hotz, S., Raines, S., Ming, M.E., Wahl, P. and Guerry, D. (2005) Biologic and prognostic significance of dermal Ki67 expression, mitoses, and tumorigenicity in thin invasive cutaneous melanoma. J Clin Oncol, 23, 8048-8056. Goggins, M., Shekher, M., Turnacioglu, K., Yeo, C.J., Hruban, R.H. and Kern, S.E. (1998) Genetic alterations of the transforming growth factor beta receptor genes in pancreatic and biliary adenocarcinomas. Cancer Res, 58, 5329-5332. Gray-Schopfer, V., Wellbrock, C. and Marais, R. (2007) Melanoma biology and new targeted therapy. Nature, 445, 851-857. Guldberg, P., thor Straten, P., Birck, A., Ahrenkiel, V., Kirkin, A.F. and Zeuthen, J. (1997) Disruption of the MMAC1/PTEN Gene by Deletion or Mutation Is a Frequent Event in Malignant Melanoma. Cancer Res, 57, 3660-3663. Hahn, S.A., Schutte, M., Hoque, A.T., Moskaluk, C.A., da Costa, L.T., Rozenblum, E., Weinstein, C.L., Fischer, A., Yeo, C.J., Hruban, R.H. and Kern, S.E. (1996) DPC4, a candidate tumor suppressor gene at human chromosome 18q21.1. Science, 271, 350-353. Han, S.U., Kim, H.T., Seong, D.H., Kim, Y.S., Park, Y.S., Bang, Y.J., Yang, H.K. and Kim, S.J. (2004) Loss of the Smad3 expression increases susceptibility to tumorigenicity in human gastric cancer. Oncogene, 23, 1333-1341. Hannon, G.J. and Beach, D. (1994) p15INK4B is a potential effector of TGF-beta-induced cell cycle arrest. Nature, 371, 257-261.

40 Chapter 1 References

Hashimoto, Y., Shindo-Okada, N., Tani, M., Takeuchi, K., Toma, H. and Yokota, J. (1996) Identification of genes differentially expressed in association with metastatic potential of K-1735 murine melanoma by messenger RNA differential display. Cancer Res, 56, 5266- 5271. Hayward, N.K. (2003) Genetics of melanoma predisposition. Oncogene, 22, 3053-3062. Hemesath, T.J., Price, E.R., Takemoto, C., Badalian, T. and Fisher, D.E. (1998) MAP kinase links the transcription factor Microphthalmia to c-Kit signalling in melanocytes. Nature, 391, 298-301. Hengge, U.R. and Dummer, R. (2006) Malignes Melanom Standards und Innovationen in Diagnostik und Therapie. Deutscher Ärzte-Verlag, Köln. Hingorani, S.R., Jacobetz, M.A., Robertson, G.P., Herlyn, M. and Tuveson, D.A. (2003) Suppression of BRAFV599E in Human Melanoma Abrogates Transformation. Cancer Res, 63, 5198-5202. Ho, J., de Guise, C., Kim, C., Lemay, S., Wang, X.-F. and Lebrun, J.-J. (2004) Activin induces hepatocyte cell growth arrest through induction of the cyclin-dependent kinase inhibitor p15INK4B and Sp1. Cellular Signalling, 16, 693-701. Hoath, S.B. and Leahy, D.G. (2003) The Organization of Human Epidermis: Functional Epidermal Units and Phi Proportionality. 121, 1440-1446. Houghton, A.N. and Polsky, D. (2002) Focus on melanoma. Cancer Cell, 2, 275-278. Houghton, A.N. and Viola, M.V. (1981) Solar radiation and malignant melanoma of the skin. J Am Acad Dermatol, 5, 477-483. Iavarone, A., Garg, P., Lasorella, A., Hsu, J. and Israel, M.A. (1994) The helix-loop-helix protein Id-2 enhances cell proliferation and binds to the retinoblastoma protein. Genes Dev., 8, 1270-1284. Izumoto, S., Arita, N., Ohnishi, T., Hiraga, S., Taki, T., Tomita, N., Ohue, M. and Hayakawa, T. (1997) Microsatellite instability and mutated type II transforming growth factor-beta receptor gene in gliomas. Cancer Lett, 112, 251-256. Jacquemin, P., Lannoy, V.J., O'Sullivan, J., Read, A., Lemaigre, F.P. and Rousseau, G.G. (2001) The transcription factor onecut-2 controls the microphthalmia-associated transcription factor gene. Biochem Biophys Res Commun, 285, 1200-1205. Jakowlew, S.B. (2006) Transforming growth factor-beta in cancer and metastasis. Cancer Metastasis Rev, 25, 435-457. Janji, B., Melchior, C., Gouon, V., Vallar, L. and Kieffer, N. (1999) Autocrine TGF-beta- regulated expression of adhesion receptors and integrin-linked kinase in HT-144 melanoma cells correlates with their metastatic phenotype. Int J Cancer, 83, 255-262. Javelaud, D., Delmas, V., Moller, M., Sextius, P., Andre, J., Menashi, S., Larue, L. and Mauviel, A. (2005) Stable overexpression of Smad7 in human melanoma cells inhibits their tumorigenicity in vitro and in vivo. Oncogene, 24, 7624-7629. Javelaud, D. and Mauviel, A. (2004) Mammalian transforming growth factor-betas: Smad signaling and physio-pathological roles. Int J Biochem Cell Biol, 36, 1161-1165. Javelaud, D., Mohammad, K.S., McKenna, C.R., Fournier, P., Luciani, F., Niewolna, M., Andre, J., Delmas, V., Larue, L., Guise, T.A. and Mauviel, A. (2007) Stable Overexpression of Smad7 in Human Melanoma Cells Impairs Bone Metastasis. Cancer Res, 67, 2317-2324. Jemal, A., Tiwari, R.C., Murray, T., Ghafoor, A., Samuels, A., Ward, E., Feuer, E.J. and Thun, M.J. (2004) Cancer statistics, 2004. CA Cancer J Clin, 54, 8-29. Jonk, L.J.C., Itoh, S., Heldin, C.-H., ten Dijke, P. and Kruijer, W. (1998) Identification and Functional Characterization of a Smad Binding Element (SBE) in the JunB Promoter That Acts as a Transforming Growth Factor-beta , Activin, and Bone Morphogenetic Protein- inducible Enhancer. J. Biol. Chem., 273, 21145-21152.

41 Chapter 1 References

Kamesaki, H., Nishizawa, K., Michaud, G.Y., Cossman, J. and Kiyono, T. (1998) TGF- {beta}1 Induces the Cyclin-Dependent Kinase Inhibitor p27Kip1 mRNA and Protein in Murine B Cells. J Immunol, 160, 770-777. Kanamaru, C., Yasuda, H. and Fujita, T. (2002) Involvement of Smad proteins in TGF-[beta] and activin A-induced apoptosis and growth inhibition of liver cells. Hepatology Research, 23, 211-219. Kang, Y., Chen, C.-R. and Massague, J. (2003) A Self-Enabling TGF[beta] Response Coupled to Stress Signaling: Smad Engages Stress Response Factor ATF3 for Id1 Repression in Epithelial Cells. Molecular Cell, 11, 915-926. Kennedy, C., ter Huurne, J., Berkhout, M., Gruis, N., Bastiaens, M., Bergman, W., Willemze, R. and Bavinck, J.N. (2001) Melanocortin 1 receptor (MC1R) gene variants are associated with an increased risk for cutaneous melanoma which is largely independent of skin type and hair color. J Invest Dermatol, 117, 294-300. Kim, D.S., Park, S.H. and Park, K.C. (2004) Transforming growth factor-beta1 decreases melanin synthesis via delayed extracellular signal-regulated kinase activation. Int J Biochem Cell Biol, 36, 1482-1491. Koli, K., Saharinen, J., Hyytiainen, M., Penttinen, C. and Keski-Oja, J. (2001) Latency, activation, and binding proteins of TGF-beta. Microsc Res Tech, 52, 354-362. Kondo, M., Cubillo, E., Tobiume, K., Shirakihara, T., Fukuda, N., Suzuki, H., Shimizu, K., Takehara, K., Cano, A., Saitoh, M. and Miyazono, K. (2004) A role for Id in the regulation of TGF-beta-induced epithelial-mesenchymal transdifferentiation. Cell Death Differ, 11, 1092-1101. Kowanetz, M., Valcourt, U., Bergstrom, R., Heldin, C.H. and Moustakas, A. (2004) Id2 and Id3 define the potency of cell proliferation and differentiation responses to transforming growth factor beta and bone morphogenetic protein. Mol Cell Biol, 24, 4241-4254. Krasagakis, K., Kruger-Krasagakes, S., Fimmel, S., Eberle, J., Tholke, D., von der Ohe, M., Mansmann, U. and Orfanos, C.E. (1999) Desensitization of melanoma cells to autocrine TGF-beta isoforms. J Cell Physiol, 178, 179-187. Ku, J.-L., Park, S.-H., Yoon, K.-A., Shin, Y.-K., Kim, K.-H., Choi, J.-S., Kang, H.-C., Kim, I.-J., Han, I.-O. and Park, J.-G. (2007) Genetic alterations of the TGF-[beta] signaling pathway in colorectal cancer cell lines: A novel mutation in Smad3 associated with the inactivation of TGF-[beta]-induced transcriptional activation. Cancer Letters, 247, 283- 292. Lasorella, A., Iavarone, A. and Israel, M.A. (1996) Id2 specifically alters regulation of the cell cycle by tumor suppressor proteins. Mol. Cell. Biol., 16, 2570-2578. Lasorella, A., Noseda, M., Beyna, M. and Iavarone, A. (2000) Id2 is a retinoblastoma protein target and mediates signalling by Myc oncoproteins. Nature, 407, 592-598. Lasorella, A., Uo, T. and Iavarone, A. (2001) Id proteins at the cross-road of development and cancer. Oncogene, 20, 8326-8333. Levy, C., Khaled, M. and Fisher, D.E. (2006) MITF: master regulator of melanocyte development and melanoma oncogene. Trends in Molecular Medicine, 12, 406-414. Lin, J.Y. and Fisher, D.E. (2007) Melanocyte biology and skin pigmentation. Nature, 445, 843-850. Ling, M.T., Wang, X., Tsao, S.W. and Wong, Y.C. (2002) Down-regulation of Id-1 expression is associated with TGF beta 1-induced growth arrest in prostate epithelial cells. Biochim Biophys Acta, 1570, 145-152. Ling, N., Ying, S.-Y., Ueno, N., Shimasaki, S., Esch, F., Hotta, M. and Guillemin, R. (1986) Pituitary FSH is released by a heterodimer of the [beta]-subunits from the two forms of inhibin. Nature, 321, 779-782.

42 Chapter 1 References

Losi, L., Bouzourene, H. and Benhattar, J. (2007) Loss of Smad4 expression predicts liver metastasis in human colorectal cancer. Oncol Rep, 17, 1095-1099. Luttges, J., Galehdari, H., Brocker, V., Schwarte-Waldhoff, I., Henne-Bruns, D., Kloppel, G., Schmiegel, W. and Hahn, S.A. (2001) Allelic loss is often the first hit in the biallelic inactivation of the p53 and DPC4 genes during pancreatic carcinogenesis. Am J Pathol, 158, 1677-1683. Lyden, D., Young, A.Z., Zagzag, D., Yan, W., Gerald, W., O'Reilly, R., Bader, B.L., Hynes, R.O., Zhuang, Y., Manova, K. and Benezra, R. (1999) Id1 and Id3 are required for neurogenesis, angiogenesis and vascularization of tumour xenografts. Nature, 401, 670- 677. 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. Markowitz, S., Wang, J., Myeroff, L., Parsons, R., Sun, L., Lutterbaugh, J., Fan, R.S., Zborowska, E., Kinzler, K.W., Vogelstein, B. and et al. (1995) Inactivation of the type II TGF-beta receptor in colon cancer cells with microsatellite instability. Science, 268, 1336- 1338. Massague, J. and Gomis, R.R. (2006) The logic of TGF[beta] signaling. FEBS Letters, 580, 2811-2820. Meier, F., Schittek, B., Busch, S., Garbe, C., Smalley, K., Satyamoorthy, K., Li, G. and Herlyn, M. (2005) The RAS/RAF/MEK/ERK and PI3K/AKT signaling pathways present molecular targets for the effective treatment of advanced melanoma. Front Biosci, 10, 2986-3001. Miyazono, K. and Miyazawa, K. (2002) Id: A Target of BMP Signaling 10.1126/stke.2002.151.pe40. Sci. STKE, 2002, pe40-. Mori, S., Nishikawa, S.I. and Yokota, Y. (2000) Lactation defect in mice lacking the helix- loop-helix inhibitor Id2. Embo J, 19, 5772-5781. Mountjoy, K.G., Robbins, L.S., Mortrud, M.T. and Cone, R.D. (1992) The cloning of a family of genes that encode the melanocortin receptors. Science, 257, 1248-1251. Nakahara, H., Otani, T., Sasaki, T., Miura, Y., Takai, Y. and Kogo, M. (2003) Involvement of Cdc42 and Rac small G proteins in invadopodia formation of RPMI7951 cells. Genes to Cells, 8, 1019-1027. Ohtani, N., Zebedee, Z., Huot, T.J.G., Stinson, J.A., Sugimoto, M., Ohashi, Y., Sharrocks, A.D., Peters, G. and Hara, E. (2001) Opposing effects of Ets and Id proteins on p16INK4a expression during cellular senescence. 409, 1067-1070. Onken, M.D., Ehlers, J.P., Worley, L.A., Makita, J., Yokota, Y. and Harbour, J.W. (2006) Functional Gene Expression Analysis Uncovers Phenotypic Switch in Aggressive Uveal Melanomas 10.1158/0008-5472.CAN-05-4196. Cancer Res, 66, 4602-4609. Palmer, J.S., Duffy, D.L., Box, N.F., Aitken, J.F., O'Gorman, L.E., Green, A.C., Hayward, N.K., Martin, N.G. and Sturm, R.A. (2000) Melanocortin-1 receptor polymorphisms and risk of melanoma: is the association explained solely by pigmentation phenotype? Am J Hum Genet, 66, 176-186. Panopoulou, E., Murphy, C., Rasmussen, H., Bagli, E., Rofstad, E.K. and Fotsis, T. (2005) Activin A Suppresses Neuroblastoma Xenograft Tumor Growth via Antimitotic and Antiangiogenic Mechanisms. Cancer Res, 65, 1877-1886. Perk, J., Iavarone, A. and Benezra, R. (2005) Id family of helix-loop-helix proteins in cancer. Nat Rev Cancer, 5, 603-614. Pho, L., Grossman, D. and Leachman, S.A. (2006) Melanoma genetics: a review of genetic factors and clinical phenotypes in familial melanoma. Curr Opin Oncol, 18, 173-179.

43 Chapter 1 References

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., Salem, G., Pohida, T., Heenan, P., Duray, P., Kallioniemi, O., Hayward, N.K., Trent, J.M. and Meltzer, P.S. (2003) High frequency of BRAF mutations in nevi. Nat Genet, 33, 19-20. Polsky, D., Young, A.Z., Busam, K.J. and Alani, R.M. (2001) The Transcriptional Repressor of p16/Ink4a, Id1, Is Up-Regulated in Early Melanomas. Cancer Res, 61, 6008-6011. Polyak, K., Kato, J.Y., Solomon, M.J., Sherr, C.J., Massague, J., Roberts, J.M. and Koff, A. (1994) p27Kip1, a cyclin-Cdk inhibitor, links transforming growth factor-beta and contact inhibition to cell cycle arrest. Genes Dev, 8, 9-22. Pomerantz, J., Schreiber-Agus, N., Liegeois, N.J., Silverman, A., Alland, L., Chin, L., Potes, J., Chen, K., Orlow, I., Lee, H.W., Cordon-Cardo, C. and DePinho, R.A. (1998) The Ink4a tumor suppressor gene product, p19Arf, interacts with MDM2 and neutralizes MDM2's inhibition of p53. Cell, 92, 713-723. Prabhu, S., Ignatova, A., Park, S. and Sun, X. (1997) Regulation of the expression of cyclin- dependent kinase inhibitor p21 by E2A and Id proteins. Mol. Cell. Biol., 17, 5888-5896. Price, E.R., Ding, H.-F., Badalian, T., Bhattacharya, S., Takemoto, C., Yao, T.-P., Hemesath, T.J. and Fisher, D.E. (1998) Lineage-specific Signaling in Melanocytes. c-Kit STIMULATION RECRUITS p300/CBP TO MICROPHTHALMIA. J. Biol. Chem., 273, 17983-17986. Qiu, W., Schonleben, F., Li, X. and Su, G.H. (2007) Disruption of transforming growth factor beta-Smad signaling pathway in head and neck squamous cell carcinoma as evidenced by mutations of SMAD2 and SMAD4. Cancer Lett, 245, 163-170. Quelle, D.E., Zindy, F., Ashmun, R.A. and Sherr, C.J. (1995) Alternative reading frames of the INK4a tumor suppressor gene encode two unrelated proteins capable of inducing cell cycle arrest. Cell, 83, 993-1000. Ramont, L., Pasco, S., Hornebeck, W., Maquart, F.X. and Monboisse, J.C. (2003) Transforming growth factor-beta1 inhibits tumor growth in a mouse melanoma model by down-regulating the plasminogen activation system. Exp Cell Res, 291, 1-10. Riggins, G.J., Thiagalingam, S., Rozenblum, E., Weinstein, C.L., Kern, S.E., Hamilton, S.R., Willson, J.K., Markowitz, S.D., Kinzler, K.W. and Vogelstein, B. (1996) Mad-related genes in the human. Nat Genet, 13, 347-349. Rimm, D.L., Caca, K., Hu, G., Harrison, F.B. and Fearon, E.R. (1999) Frequent nuclear/cytoplasmic localization of beta-catenin without exon 3 mutations in malignant melanoma. Am J Pathol, 154, 325-329. Roberts, A.B. and Derynck, R. (2001) Meeting report: signaling schemes for TGF-beta. Sci STKE, 2001, PE43. Roberts, E.C., Deed, R.W., Inoue, T., Norton, J.D. and Sharrocks, A.D. (2001) Id Helix- Loop-Helix Proteins Antagonize Pax Transcription Factor Activity by Inhibiting DNA Binding 10.1128/MCB.21.2.524-533.2001. Mol. Cell. Biol., 21, 524-533. Rodeck, U., Bossler, A., Graeven, U., Fox, F.E., Nowell, P.C., Knabbe, C. and Kari, C. (1994) Transforming growth factor beta production and responsiveness in normal human melanocytes and melanoma cells. Cancer Res, 54, 575-581. Rodeck, U., Nishiyama, T. and Mauviel, A. (1999) Independent regulation of growth and SMAD-mediated transcription by transforming growth factor beta in human melanoma cells. Cancer Res, 59, 547-550. Rodriguez-Viciana, P., Warne, P.H., Dhand, R., Vanhaesebroeck, B., Gout, I., Fry, M.J., Waterfield, M.D. and Downward, J. (1994) Phosphatidylinositol-3-OH kinase as a direct target of Ras. Nature, 370, 527-532.

44 Chapter 1 References

Rodriguez-Viciana, P., Warne, P.H., Vanhaesebroeck, B., Waterfield, M.D. and Downward, J. (1996) Activation of phosphoinositide 3-kinase by interaction with Ras and by point mutation. Embo J, 15, 2442-2451. Rothschild, G., Zhao, X., Iavarone, A. and Lasorella, A. (2006) E Proteins and Id2 Converge on p57Kip2 To Regulate Cell Cycle in Neural Cells. Mol. Cell. Biol., 26, 4351-4361. Rotzer, D., Krampert, M., Sulyok, S., Braun, S., Stark, H.J., Boukamp, P. and Werner, S. (2006) Id proteins: novel targets of activin action, which regulate epidermal homeostasis. Oncogene, 25, 2070-2081. Rubinfeld, B., Robbins, P., El-Gamil, M., Albert, I., Porfiri, E. and Polakis, P. (1997) Stabilization of beta-catenin by genetic defects in melanoma cell lines. Science, 275, 1790- 1792. Ryu, B., Kim, D.S., DeLuca, A.M., Healey, M.A., Dunlap, S., Fackler, M.J., Herman, J. and Alani, R.M. (2007) Id1 expression is transcriptionally regulated in radial growth phase melanomas. Int J Cancer, 121, 1705-1709. Saito, H., Yasumoto, K., Takeda, K., Takahashi, K., Fukuzaki, A., Orikasa, S. and Shibahara, S. (2002) Melanocyte-specific microphthalmia-associated transcription factor isoform activates its own gene promoter through physical interaction with lymphoid-enhancing factor 1. J Biol Chem, 277, 28787-28794. Salti, G.I., Manougian, T., Farolan, M., Shilkaitis, A., Majumdar, D. and Das Gupta, T.K. (2000) Micropthalmia Transcription Factor: A New Prognostic Marker in Intermediate- thickness Cutaneous Malignant Melanoma. Cancer Res, 60, 5012-5016. Scandura, J.M., Boccuni, P., Massague, J. and Nimer, S.D. (2004) Transforming growth factor beta-induced cell cycle arrest of human hematopoietic cells requires p57KIP2 upregulation. Proc Natl Acad Sci U S A, 101, 15231-15236. Schiemann, W.P., Pfeifer, W.M., Levi, E., Kadin, M.E. and Lodish, H.F. (1999) A deletion in the gene for transforming growth factor beta type I receptor abolishes growth regulation by transforming growth factor beta in a cutaneous T-cell lymphoma. Blood, 94, 2854-2861. Schuster, N. and Krieglstein, K. (2002) Mechanisms of TGF-beta-mediated apoptosis. Cell Tissue Res, 307, 1-14. Schutte, M., Hruban, R.H., Hedrick, L., Cho, K.R., Nadasdy, G.M., Weinstein, C.L., Bova, G.S., Isaacs, W.B., Cairns, P., Nawroz, H., Sidransky, D., Casero, R.A., Jr., Meltzer, P.S., Hahn, S.A. and Kern, S.E. (1996) DPC4 gene in various tumor types. Cancer Res, 56, 2527-2530. Scott, M.C., Wakamatsu, K., Ito, S., Kadekaro, A.L., Kobayashi, N., Groden, J., Kavanagh, R., Takakuwa, T., Virador, V., Hearing, V.J. and Abdel-Malek, Z.A. (2002) Human melanocortin 1 receptor variants, receptor function and melanocyte response to UV radiation. J Cell Sci, 115, 2349-2355. Selzer, E., Wacheck, V., Lucas, T., Heere-Ress, E., Wu, M., Weilbaecher, K.N., Schlegel, W., Valent, P., Wrba, F., Pehamberger, H., Fisher, D. and Jansen, B. (2002) The Melanocyte- specific Isoform of the Microphthalmia Transcription Factor Affects the Phenotype of Human Melanoma. Cancer Res, 62, 2098-2103. Seoane, J., Pouponnot, C., Staller, P., Schader, M., Eilers, M. and Massague, J. (2001) TGF[beta] influences Myc, Miz-1 and Smad to control the CDK inhibitor p15INK4b. Nat Cell Biol, 3, 400-408. Shi, Y. and Massague, J. (2003) Mechanisms of TGF-beta signaling from cell membrane to the nucleus. Cell, 113, 685-700. Siegel, P.M. and Massague, J. (2003) Cytostatic and apoptotic actions of TGF-beta in homeostasis and cancer. Nat Rev Cancer, 3, 807-821.

45 Chapter 1 References

Siegel, P.M., Shu, W., Cardiff, R.D., Muller, W.J. and Massague, J. (2003) Transforming growth factor beta signaling impairs Neu-induced mammary tumorigenesis while promoting pulmonary metastasis. Proc Natl Acad Sci U S A, 100, 8430-8435. Soufir, N., Avril, M.F., Chompret, A., Demenais, F., Bombled, J., Spatz, A., Stoppa- Lyonnet, D., Benard, J. and Bressac-de Paillerets, B. (1998) Prevalence of p16 and CDK4 germline mutations in 48 melanoma-prone families in France. The French Familial Melanoma Study Group [published erratum appears in Hum Mol Genet 1998 May;7(5):941]. Hum. Mol. Genet., 7, 209-216. Staller, P., Peukert, K., Kiermaier, A., Seoane, J., Lukas, J., Karsunky, H., Moroy, T., Bartek, J., Massague, J., Hanel, F. and Eilers, M. (2001) Repression of p15INK4b expression by Myc through association with Miz-1. Nat Cell Biol, 3, 392-399. Steingrimsson, E., Copeland, N.G. and Jenkins, N.A. (2004) Melanocytes and the microphthalmia transcription factor network. Annu Rev Genet, 38, 365-411. Steller, M.D., Shaw, T.J., Vanderhyden, B.C. and Ethier, J.F. (2005) Inhibin resistance is associated with aggressive tumorigenicity of ovarian cancer cells. Mol Cancer Res, 3, 50- 61. Stott, F.J., Bates, S., James, M.C., McConnell, B.B., Starborg, M., Brookes, S., Palmero, I., Ryan, K., Hara, E., Vousden, K.H. and Peters, G. (1998) The alternative product from the human CDKN2A locus, p14(ARF), participates in a regulatory feedback loop with p53 and MDM2. Embo J, 17, 5001-5014. Stove, C., Vanrobaeys, F., Devreese, B., Van Beeumen, J., Mareel, M. and Bracke, M. (2004) Melanoma cells secrete follistatin, an antagonist of activin-mediated growth inhibition. Oncogene. Straume, O. and Akslen, L.A. (2005) Strong expression of ID1 protein is associated with decreased survival, increased expression of ephrin-A1/EPHA2, and reduced thrombospondin-1 in malignant melanoma. Br J Cancer, 93, 933-938. Tamura, M., Gu, J., Danen, E.H., Takino, T., Miyamoto, S. and Yamada, K.M. (1999) PTEN interactions with focal adhesion kinase and suppression of the extracellular matrix- dependent phosphatidylinositol 3-kinase/Akt cell survival pathway. J Biol Chem, 274, 20693-20703. Tamura, M., Gu, J., Matsumoto, K., Aota, S., Parsons, R. and Yamada, K.M. (1998) Inhibition of cell migration, spreading, and focal adhesions by tumor suppressor PTEN. Science, 280, 1614-1617. Tanaka, T., Toujima, S. and Umesaki, N. (2004) Activin A inhibits growth-inhibitory signals by TGF-beta1 in differentiated human endometrial adenocarcinoma cells. Oncol Rep, 11, 875-879. Tang, B., Vu, M., Booker, T., Santner, S.J., Miller, F.R., Anver, M.R. and Wakefield, L.M. (2003) TGF-{beta} switches from tumor suppressor to prometastatic factor in a model of breast cancer progression. J. Clin. Invest., 112, 1116-1124. Teng, D.H.F., Hu, R., Lin, H., Davis, T., Iliev, D., Frye, C., Swedlund, B., Hansen, K.L., Vinson, V.L., Gumpper, K.L., Ellis, L., El-Naggar, A., Frazier, M., Jasser, S., Langford, L.A., Lee, J., Mills, G.B., Pershouse, M.A., Pollack, R.E., Tornos, C., Troncoso, P., Alfred Yung, W.K., Fujii, G., Berson, A., Bookstein, R., Bolen, J.B., Tavtigian, S.V. and Steck, P.A. (1997) MMAC1/PTEN Mutations in Primary Tumor Specimens and Tumor Cell Lines. Cancer Res, 57, 5221-5225. Teti, A., De Giorgi, A., Spinella, M.T., Migliaccio, S., Canipari, R., Onetti Muda, A. and Faraggiana, T. (1997) Transforming growth factor-beta enhances adhesion of melanoma cells to the endothelium in vitro. Int J Cancer, 72, 1013-1020. Thompson, J.A. (2002) The revised American Joint Committee on Cancer staging system for melanoma. Seminars in Oncology, 29, 361-369.

46 Chapter 1 References

Tian, F., DaCosta Byfield, S., Parks, W.T., Yoo, S., Felici, A., Tang, B., Piek, E., Wakefield, L.M. and Roberts, A.B. (2003) Reduction in Smad2/3 signaling enhances tumorigenesis but suppresses metastasis of breast cancer cell lines. Cancer Res, 63, 8284-8292. Todaro, G.J., Fryling, C. and De Larco, J.E. (1980) Transforming growth factors produced by certain human tumor cells: polypeptides that interact with epidermal growth factor receptors. Proc Natl Acad Sci U S A, 77, 5258-5262. Tsang, K.J., Tsang, D., Brown, T.N. and Crowe, D.L. (2002) A novel dominant negative Smad2 mutation in a TGFbeta resistant human carcinoma cell line. Anticancer Res, 22, 13- 19. Tsao, H., Goel, V., Wu, H., Yang, G. and Haluska, F.G. (2004) Genetic interaction between NRAS and BRAF mutations and PTEN/MMAC1 inactivation in melanoma. J Invest Dermatol, 122, 337-341. Uchida, K., Nagatake, M., Osada, H., Yatabe, Y., Kondo, M., Mitsudomi, T., Masuda, A., Takahashi, T. and Takahashi, T. (1996) Somatic in Vivo Alterations of the JV18-1 Gene at 18q21 in Human Lung Cancers. Cancer Res, 56, 5583-5585. Vachtenheim, J., Novotna, H. and Ghanem, G. (2001) Transcriptional Repression of the Microphthalmia Gene in Melanoma Cells Correlates with the Unresponsiveness of Target Genes to Ectopic Microphthalmia-Associated Transcription Factor. 117, 1505-1511. Valverde, P., Healy, E., Sikkink, S., Haldane, F., Thody, A.J., Carothers, A., Jackson, I.J. and Rees, J.L. (1996) The Asp84Glu variant of the melanocortin 1 receptor (MC1R) is associated with melanoma. Hum Mol Genet, 5, 1663-1666. Van Belle, P., Rodeck, U., Nuamah, I., Halpern, A.C. and Elder, D.E. (1996) Melanoma- associated expression of transforming growth factor-beta isoforms. Am J Pathol, 148, 1887-1894. Waga, S., Hannon, G.J., Beach, D. and Stillman, B. (1994) The p21 inhibitor of cyclin- dependent kinases controls DNA replication by interaction with PCNA. Nature, 369, 574- 578. Wakefield, L.M. and Sporn, M.B. (1990) Suppression of carcinogenesis: a role for TGF-beta and related molecules in prevention of cancer. Immunol Ser, 51, 217-243. Wang, J., Han, W., Zborowska, E., Liang, J., Wang, X., Willson, J.K.V., Sun, L. and Brattain, M.G. (1996) Reduced Expression of Transforming Growth Factor beta Type I Receptor Contributes to the Malignancy of Human Colon Carcinoma Cells. J. Biol. Chem., 271, 17366-17371. Wang, J., Sun, L., Myeroff, L., Wang, X., Gentry, L.E., Yang, J., Liang, J., Zborowska, E., Markowitz, S., Willson, J.K.V. and Brattain, M.G. (1995) Demonstration That Mutation of the Type II Transforming Growth Factor beta Receptor Inactivates Its Tumor Suppressor Activity in Replication Error-positive Colon Carcinoma Cells. J. Biol. Chem., 270, 22044- 22049. Weeraratna, A.T. (2005) A Wnt-er wonderland--the complexity of Wnt signaling in melanoma. Cancer Metastasis Rev, 24, 237-250. 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. Wellbrock, C. and Marais, R. (2005) Elevated expression of MITF counteracts B-RAF- stimulated melanocyte and melanoma cell proliferation. J Cell Biol, 170, 703-708. Wellbrock, C., Ogilvie, L., Hedley, D., Karasarides, M., Martin, J., Niculescu-Duvaz, D., Springer, C.J. and Marais, R. (2004) V599EB-RAF is an Oncogene in Melanocytes. Cancer Res, 64, 2338-2342. World Health Organization. (2006) Pathology and genetics , Skin Tumours. IARC, Lyon.

47 Chapter 1 References

Wu, H., Goel, V. and Haluska, F.G. (2003) PTEN signaling pathways in melanoma. Oncogene, 22, 3113-3122. Wu, M., Hemesath, T.J., Takemoto, C.M., Horstmann, M.A., Wells, A.G., Price, E.R., Fisher, D.Z. and Fisher, D.E. (2000) c-Kit triggers dual phosphorylations, which couple activation and degradation of the essential melanocyte factor Mi. Genes Dev, 14, 301-312. Xu, X., Brodie, S.G., Yang, X., Im, Y.H., Parks, W.T., Chen, L., Zhou, Y.X., Weinstein, M., Kim, S.J. and Deng, C.X. (2000) Haploid loss of the tumor suppressor Smad4/Dpc4 initiates gastric polyposis and cancer in mice. Oncogene, 19, 1868-1874. Yasumoto, K., Takeda, K., Saito, H., Watanabe, K., Takahashi, K. and Shibahara, S. (2002) Microphthalmia-associated transcription factor interacts with LEF-1, a mediator of Wnt signaling. Embo J, 21, 2703-2714. Yates, P.R., Atherton, G.T., Deed, R.W., Norton, J.D. and Sharrocks, A.D. (1999) Id helix- loop-helix proteins inhibit nucleoprotein complex formation by the TCF ETS-domain transcription factors. Embo J, 18, 968-976. Yoshinaga, K., Inoue, H., Utsunomiya, T., Sonoda, H., Masuda, T., Mimori, K., Tanaka, Y. and Mori, M. (2004) N-Cadherin Is Regulated by Activin A and Associated with Tumor Aggressiveness in Esophageal Carcinoma. Clin Cancer Res, 10, 5702-5707. Zawel, L., Dai, J.L., Buckhaults, P., Zhou, S., Kinzler, K.W., Vogelstein, B. and Kern, S.E. (1998) Human Smad3 and Smad4 are sequence-specific transcription activators. Mol Cell, 1, 611-617. 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.

Chapter 2-

2 Metastatic potential of melanomas defined by specific gene expression profiles with no BRAF signature

This paper is the fruit of the first years of research of our newly founded melanoma research group. Keith Hoek, as the key initiator of this project, generated the microarray data on which this new model of melanoma progression is founded. My task involved the establishment of a range of in vitro assays and the validation of this new model.

I optimised conditions for the generation of RT-PCR results shown in figure 2.3 and also performed a number of Western blot analyses, three of which are shown in figure 2.4. Figure 2.5 shows motility assays which I undertook and I carried out the TGF-β-mediated growth inhibition assays shown in table 2.2. Finally, I proofread the manuscript to ensure that the proper use of English was employed and that experimental results and concepts were displayed in a clear manner.

Metastatic potential of melanomas defined by specific gene expression profiles with no BRAF signature

Keith S. Hoek1, Natalie C. Schlegel1, Patricia Brafford2, Antje Sucker3, Selma Ugurel3, Rajiv Kumar4, Barbara L. Weber5, Katherine L. Nathanson6, David J. Phillips7, Meenhard Herlyn2, Dirk Schadendorf3, Reinhard Dummer1

1Department of Dermatology, University Hospital of Zürich, 8091 Zürich, Switzerland 2The Wistar Institute, Philadelphia PA 3Skin Cancer Unit of the German Cancer Research Center, University Hospital Mannheim, Mannheim, Germany 4Division of Molecular Genetic Epidemiology, German Cancer Research Center, Heidelberg, Germany 5Abramson Family Cancer Research Insitute, University of Pennsylvania Cancer Center, Philadelphia PA 6Department of Medicine, University of Pennsylvania School of Medicine, Philadelphia, PA 7Monash Institute of Medical Research, Monash University, Clayton, Victoria, Australia

Pigment Cell Research (2006)19, 290-302

Chapter 2 Abstract/Introduction

2.1 Abstract

The molecular biology of metastatic potential in melanoma has been studied many times previously and changes in the expression of many genes have been linked to metastatic behaviour. What is lacking is a systematic characterisation of the regulatory relationships between genes whose expression is related to metastatic potential. Such a characterisation would produce a molecular taxonomy for melanoma which could feasibly be used to identify epigenetic mechanisms behind changes in metastatic behaviour. To achieve this we carried out three separate DNA microarray analyses on a total of 86 cultures of melanoma. Significantly, multiple testing correction revealed that previous reports describing correlations of gene expression with activating mutations in BRAF or NRAS were incorrect and that no gene expression patterns correlate with the mutation status of these MAPK pathway components. Instead, we identified three different sample cohorts (A, B and C) and found that these cohorts represent melanoma groups of differing metastatic potential. Cohorts A and B were susceptible to TGF-β-mediated inhibition of proliferation and had low motility. Cohort C was resistant to TGF-β and demonstrated high motility. Meta-analysis of the data against previous studies linking gene expression and phenotype confirmed that cohorts A and C represent transcription signatures of weakly and strongly metastatic melanomas, respectively. Gene expression co-regulation suggested that signalling via TGF-β-type and Wnt/β-catenin pathways underwent considerable change between cohorts. These results suggest a model for the transition from weakly to strongly metastatic melanomas in which TGF-β-type signalling upregulates genes expressing vasculogenic/extracellular matrix remodeling factors and Wnt signal inhibitors, coinciding with a downregulation of genes downstream of Wnt signalling.

2.2 Introduction

Cutaneous melanoma is an often aggressively metastatic and fatal neoplasm that accounts for most skin cancer deaths (Balch et al., 2001). The early steps of melanoma progression are marked by molecular changes in networks of adhesion molecules such as those involving the cadherins, which mediate calcium-dependent intercellular adhesion through homotypic interaction (Takeichi, 1991). Epithelial cadherin (E-cadherin) is lost very early in transformation from melanocytes to melanoma. E-cadherin function is critical to melanocyte

Chapter 2 Introduction homeostasis and its loss precipitates changes in the expression of other adhesion factors (Haass et al., 2005; Hsu et al., 2000b; Li and Herlyn, 2000). Neuronal cadherin (N-cadherin) is upregulated and mediates gap junction formation with N-cadherin-expressing dermal fibroblasts (Hsu et al., 2000a), while placental cadherin is downregulated (Tsutsumida et al., 2004). Another early event in progression of melanoma is the upregulation of matrix metalloproteinase 2, which affects basal membrane integrity and dermal invasion (Vaisanen et al., 1996). Dermal invasion itself correlates with the production of immune modifying factors including interleukins, chemokines, and TGF-β. Later stages of metastasis see the induction of a host of additional growth factors including CTGF, VEGF, bFGF, and PDGF, which affect vascularisation and the growth of both melanoma and stromal cells (Bar-Eli, 2001; Hsu et al., 2002b) (Hsu et al., 2002a). Recent work has uncovered the role of transcription factors in orchestrating these changes. For example, the AP-2 transcription factor, associated with regulating adhesion factors and c-Kit, has been shown to be downregulated with malignant progression (Bar-Eli, 2001). In addition, microphthalmia-associated transcription factor (Mitf) regulates tissue-specific expression of cyclin-dependent kinase 2, a relationship that may be critical for melanoma growth (Du et al., 2004). Apart from changes in gene expression, melanoma progression also is marked by other changes. For example, a change in response to TGF-β marks a turning point in malignant progression. In some cancers, including melanoma, early stages are characterised as being susceptible to TGF-β-mediated growth inhibition, while later stages are increasingly less affected (Elliott and Blobe, 2005). Multiple studies show that melanomas resistant to TGF-β-mediated growth inhibition are more aggressively invasive and metastatic than variants that retain the growth-inhibited response (Heredia et al., 1996; Krasagakis et al., 1994; Roberts et al., 1985). In other cancer types, such as breast and colon, loss of growth control by TGF-β corresponds with increased motility and expression of factors promoting invasion and metastasis (Derynck et al., 2001; Gold, 1999; Padgett, 1999; Wright and Huang, 1996).

Recent analyses have focused on the serine/threonine kinase BRAF, which frequently undergoes an activating mutation in melanoma (Davies et al., 2002). This likely lies in BRAF’s role as a mediator of c-Kit activated RAS-ERK signalling (Linnekin, 1999) which is a regulator of melanocytic cell proliferation (Wellbrock and Marais, 2005). BRAF gene mutations, the most common being an activating V600E mutation, can occur early in melanoma

Chapter 2 Introduction/Results progression, as they are frequently recorded in lesions prior to neoplastic transformation (Pollock et al., 2003). Transformation of mouse melanocytes with mutant BRAF has yielded phenotypically altered cells with in vivo tumorigenicity when transplanted into nude mice (Wellbrock et al., 2004). The documented existence of patients with wild-type BRAF primary lesions who later develop BRAF-mutant metastases indicates that BRAF mutations are associated with metastasis (Shinozaki et al., 2004).

Melanomas exhibit a wide variety of characteristics and at least some are likely reflected in their gene expression patterns. The construction of a transcription signature taxonomy for melanoma and its association with elements of melanoma pathology would have significant value and contribute to understanding the gene regulation of melanoma. Our laboratories separately obtained the transcriptional profiles of three panels of melanoma cell cultures, uncovered the transcriptional variations responsible for intrapanel sample differences, and identified the coregulating gene sets among these. We then compared the results from clustering analyses of the two cell culture panels and found congruence for both cluster group (cohort) specific expression and co-regulation patterns in 223 genes. One of our laboratories (Zürich) went on to further characterise one panel of melanoma lines where we uncovered a relationship between sample organisation and in vitro characteristics for increased metastatic capacity.

2.3 Results

2.3.1 No correlation between BRAF/NRAS mutations and gene expression

The results from DNA microarray analyses carried out by individual groups working on similar problems often deviate appreciably due to differences in platform choice, methodology, and sample selection. We addressed this by combining post-clustering results from three laboratories working on different melanoma sets and filtering out data that did not qualitatively agree in all. One set contained the transcriptional profiles of three melanocyte cultures and 12 cell cultures derived from metastatic melanomas (Zürich data set). A second set comprised 29 melanoma cell cultures obtained from 12 in situ melanomas, 14 metastatic melanomas and three additional samples of unidentified origin (Philadelphia data set). A third

Chapter 2 Results set comprised 45 melanoma cell cultures obtained from metastatic melanomas (Mannheim data set). The transcriptional profiles were normalised within data sets and subsequent analyses were performed separately for each data set. These data sets have been deposited in NCBIs Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/) and are accessible using GEO Series accession GSE4845.

We were particularly interested in whether gene expression profiles could be used to determine pathways that were important in defining melanomas with metastatic potential and especially the potential involvement of activating mutations in BRAF and NRAS in determining metastatic potential. Having sufficient information concerning BRAF and NRAS mutation status in many of our samples we attempted to test the findings of others (Bloethner et al., 2005; Pavey et al., 2004) who describe melanoma gene expression patterns correlating with activating mutations in either or both of BRAF and NRAS genes. The Zürich data set provided four samples which were wild-type for both BRAF and NRAS, four samples which carried the BRAFV600E mutation, and four samples which carried either the NRASQ61K or NRASQ61R mutations. The Philadelphia dataset provided four samples which were wild-type for BRAF and ten samples which carried the BRAFV600E mutation, no NRAS mutation data is available for this set. The Mannheim dataset provided eleven samples which were wild-type for both BRAF and NRAS, fifteen samples which carried the BRAFV600E mutation only, and six samples which carried either the NRASQ61K or NRASQ61R mutations (Appendix A, Supplementary data 1). For each of the three data sets we applied ANOVA selection and multiple-testing correction (Benjamini et al., 2001) to look for gene expression changes correlating with activating mutations of BRAF or NRAS. We did not find any genes for which expression changes correlated with activating mutations in either of BRAF and NRAS. We obtained the data set used by Pavey et al. (2004) to identify genes which were co-regulated with BRAF. We used an identical protocol to the one they describe and we found a similar group of genes as they did. However, when we included multiple testing controls in their protocol we found only one gene whose expression changes correlated with BRAF mutation status (data not shown). We considered the possibility that correlations may instead be found within cohorts, analyses performed using samples within the same cohort would not be influenced by genes with cohort-specific expression. The Mannheim data set is large enough to look within each of the cohorts for correlations between BRAF mutation status and gene expression, however we did not find any statistically significant correlations. We find no

Chapter 2 Results genes whose changes in expression consistently correlate with activating mutations in either BRAF or NRAS across the data sets we tested.

2.3.2 Microarray analyses reveal three cohorts

Given the absence of a relationship between gene expression and activating mutations of BRAF or NRAS, we were interested in comparing the sample organisation between each of the three data sets. For the Zürich data set, normalisation and fold-change selection of probe sets, followed by unbiased hierarchical clustering of sample profiles revealed that the samples encompassed four stable cohorts (cohorts are here defined as groups of samples which have closely related transcriptional signatures): one cohort of melanocyte samples and three distinct cohorts of melanoma samples. An identical protocol applied to both the Philadelphia and Mannheim data sets yielded three stable melanoma cohorts each.

Figure 2.1. Identification of three distinct melanoma cohorts. DNA microarray data collected for melanocyte cultures and melanoma cultures were subjected to an unbiased hierarchical clustering protocol. Multiple clustering analyses using multiple pools of data showed that melanoma samples consistently cluster within one of three groups termed cohorts A (dark blue), B (yellow), and C (cyan). Displayed above are sample and gene clusterings performed using 296 probe sets with cohort-specific expression common to all data sets: Zürich (tree A; which includes the melanocyte cluster indicated by a red bar), Philadelphia (tree B) and Mannheim (tree C). Probe set coloring reflects normalised signal intensity, red being the highest and green the lowest.

Chapter 2 Results

Figure 2.1 shows the three data sets’ cohort distributions after reclustering using genes with cohort-specific expression (see next section). These data indicated that, from a transcriptional context, melanoma cell cultures are divisible into three recognisably discrete groups, which we call cohorts A, B, and C.

2.3.3 Two groups of co-regulated genes define the cohorts.

Having identified the cohort structure of the data sets we wanted to know if the genes which determined these similar distributions were the same. We used multiple testing controlled statistical methods to select probe sets with cohort-specific expression patterns and in this way identified 1973 probe sets in the Zürich data set, 1398 in the Philadelphia data set and 5753 in the Mannheim data set. Comparing these data sets identified 296 common probe sets, equivalent to 223 individual genes (Appendix A, Supplementary Data 2A). With a pool of shared genes having cohort specific expression patterns we were interested in clustering similarly regulated genes together so that we might determine the signalling pathways underlying the regulation of their expression. In order to identify which genes are co- regulated and how they are co-regulated with respect to the cohorts we used a clustering algorithm, self-organised map (SOM) analysis, to sort the cross-sample expression patterns in the Zürich data set. This revealed two major expression patterns of strong co-regulation involving many different genes (Fig. 2.2).

Figure 2.2. Self-organised map clustering. The 296 probe sets showing cohort-specific expression in both Zürich and Philadelphia data sets were subjected to SOM clustering. Two major expression patterns were revealed: Motif 1 (bottom right cell) and Motif 2 (top left cell), containing 70 and 77 probe sets, respectively. Displayed here is the SOM analysis performed using the Zürich samples. Data is presented as mean normalised signal intensity profile with 95% confidence boundaries. The Y- axis refers to normalised probe set signal intensity (0.1 to 100) and the X-axis denotes sample order arranged as depicted in Figure 2.1. The number of probe sets clustering within a cell is noted in its top-left corner. 57

Chapter 2 Results

For the 223 common genes we found that 105 were tightly linked to one of two distinct motifs (a motif is here defined as a pattern of expression shared by a number of correlating probe sets as observed across the samples), the members of which are likely co-regulated in melanoma (Appendix A, Supplementary Data 2B). These two patterns of co-regulated gene expression are sufficient to differentiate between the identified melanoma cohorts. Motif 1 shows downregulation of 51 co-regulated genes in cohort C compared to those in cohorts A and B. Motif 2 shows upregulation of 54 co-regulated genes in cohorts B and C compared to those in cohort A (Fig. 2.2.). These results were particularly interesting because many of the co-regulated genes within Motif 1 are involved in melanocytic and neural crest differentiation (Table 2.1).

Table 2.1. Motif genes regulated by Mitf/Sox10 (Motif 1) or TGF-β (Motif 2). Gene name Symbol Regulator 2.3.3.1.1.1 Reference tyrosinase related protein 2 DCT Mitf (Yasumoto et al., 2002) ocular albinism 1 GPR143 Mitf (Vetrini et al., 2004) absent in melanoma 1 MATP Mitf (Du and Fisher, 2002) melan-A MLANA Mitf (Du et al., 2003) gp100 / Pmel17 SILV Mitf (Du et al., 2003) melastatin TRPM1 Mitf (Miller et al., 2004) tyrosinase TYR Mitf (Hou et al., 2000) tyrosinase related protein 1 TYRP1 Mitf (Fang et al., 2002) endothelin receptor type B EDNRB Sox10 (Zhu et al., 2004) c-erbB3 ERBB3 Sox10 (Britsch et al., 2001) myelin basic protein MBP Sox10 (Stolt et al., 2002) microphthalmia-associated transcription factor Mitf Sox10 (Potterf et al., 2000) proteolipid protein 1 PLP1 Sox10 (Stolt et al., 2002) biglycan BGN TGF-β (Ungefroren et al., 2003) N-cadherin CDH2 TGF-β (Tuli et al., 2003) collagen type V, alpha 1 COL5A1 TGF-β (Lawrence et al., 1994) connective tissue growth factor CTGF TGF-β (Igarashi et al., 1993) fibrillin 1 FBN1 TGF-β (Lorena et al., 2004) basic fibroblast growth factor FGF2 TGF-β (Finlay et al., 2000) interleukin 6 / interferon β2 IL6 TGF-β (Park et al., 2003) inhibin βA INHBA TGF-β (Hubner and Werner, 1996) lysyl oxidase LOX TGF-β (Green et al., 1995) PAI-1 SERPINE1 TGF-β (Macfelda et al., 2002) transgelin TAGLN TGF-β (Chen et al., 2003) TGF-β-induced, 68kDa TGF-BI TGF-β (Skonier et al., 1992) thrombospondin 1 THBS1 TGF-β (Hugo, 2003) tumor necrosis factor receptor 11b TNFRSF11B TGF-β (Thirunavukkarasu et al., 2001) tropomyosin 1 (alpha) TPM1 TGF-β (Tada et al., 2000) tropomyosin 2 (beta) TPM2 TGF-β (Bakin et al., 2004)

Chapter 2 Results

On the other hand, members of the Motif 2 transcription profile are involved in modifying extracellular environments and a large fraction of these genes, as well as others in the motif, are subject to TGF-β regulation (Table 2.1). This is particularly interesting because many of these genes are thought to be involved in invasive and metastatic behaviours. To confirm gene expression trends shown by the motifs, we used reverse transcription polymerase chain reaction (RT-PCR) to validate relative expression levels for three Motif 1 genes and three Motif 2 genes (Fig. 2.3).

Figure 2.3. RT-PCR validation of select genes. Three genes from Motif 2 (BGN, CTGF, SERPINE1) and three from Motif 1 (MLANA, TYR, SOX10) were selected for validation by lightcycler RT-PCR. Samples come from the Zürich data set.

For the two regulatory proteins we hypothesised to have functional significance for the observed cohort-specific expression patterns, Motifs 1 and 2, we conducted western analyses. For Motif 1, in which expression changes are correlated with neural crest and pigmentation genes, we studied the Microphthalmia-associated transcription factor Mitf. We chose Mitf as it is a transcription factor crucial to processes of neural crest differentiation, proliferation and survival (Carreira et al., 2005; Garraway et al., 2005; Loercher et al., 2005). We find that Mitf protein levels correlate closely with Mitf gene expression (Fig. 2.4). For Motif 2, in which expression changes are correlated between genes regulated by TGF-β, we chose to examine the levels of secreted activin A into conditioned media. Activin A, a cytokine which functions similarly to TGF-β (Phillips et al., 2005), is a homodimer of a polypeptide encoded by the

Chapter 2 Results

INHBA gene present in Motif 2. We found that activin A secretion correlated well with INHBA gene expression (Fig. 2.4). The genes identified suggest that the cohorts may reflect differences in metastatic potential between melanomas as regulated by changes in Wnt and TGF-β-like signalling.

Figure 2.4 Comparison of expression and western blot patterns. Cell extracts, taken from the Zürich data set cell cultures, were subjected to western blotting against Mitf, while conditioned media (24 h) were similarly assessed for TGF-β1 and Activin A secretion. Activin A was also assessed by ELISA. Normalised signal intensity data for the Mitf and INHBA genes, extracted from Zürich data set DNA microarray experiments, are also shown.

2.3.4 The cohorts reflect differences in metastatic potential

Because the genes most responsible for the cohort distributions in our data sets seemed to include factors known to be involved in melanoma metastatic potential, we investigated the literature to find factors with known expression differences correlating with metastatic potential or patient prognosis and compared them with our dataset. We found references to 134 different genes (Appendix A, Supplementary Data 3b). We then examined these 134 genes for expression in the Zürich data set and generated Student’s two-sample t-test P-values for their differences in expression between cohorts A and C. We found that, of these, 54 genes were significantly (P<.05) and appropriately up- or downregulated between cohorts A and C for a gene expression model in which these cohorts are regarded as weakly and strongly metastatic, respectively (Appendix A, Supplementary Data 3a). Additionally, we find that 20

Chapter 2 Results genes are present in the 223 gene intersection from ANOVA analyses of the Zürich, Philadelphia and Mannheim data sets. The probability of at least 20 genes being selected by chance, as well as their having appropriate expression patterns, is calculated using hypergeometric distribution to be less than 2.7 x 10-27. This strongly suggests that the cohort distributions in our data sets are reflections of differences in metastatic potential.

2.3.5 In vitro tests support the link between cohort distribution and metastatic potential In order to better understand the characteristics of the different cohorts and provide further evidence for their association with metastatic potential, the Zürich cell cultures were assayed for their in vitro motility and their proliferation in the presence and absence of TGF-β. In the case of motility, tumour cell motility is a well understood aspect of metastatic potential (Raz, 1988) and cultures of aggressively metastatic melanomas have been reported to show greater in vitro motility relative to cultures of weaker metastatic character (Quinones and Garcia- Castro, 2004). A scratch wound assay was used to assess motility. In vitro motility assays showed that a cleared field repopulated within 24 hours with melanomas from cohorts A and C, while cohort B melanomas were appreciably slower (Fig. 2.5A-F). Time-lapse photography showed that melanomas from cohort A, rather than migrating across the substrate, repopulated the cleared field by detaching from the culture plate and diffusing over the field before reattaching (data not shown). No detachment/reattachment activity was observed with melanomas from cohorts B or C.

Figure 2.5 In vitro motility and in vivo Mitf/β-catenin. (A–F) Cultured melanomas from each of the three DNA microarray–identified cohorts were investigated for the potential to migrate into a cell-free scratch region. Shown here is migration at 0 and 24 h for cohort A (A,D), cohort B (B,E) and cohort C (C,F). (G–J) Paraffin-embedded tissue samples, the original sources for cohort B (G,H) and cohort C (I,J) were examined by histochemical staining for Mitf (G,I) and β-catenin (H,J). The cohort B melanoma shows nuclear staining for Mitf coinciding with strong staining for β-catenin, whereas the cohort C melanoma shows no nuclear staining for Mitf and little for β-catenin. Nuclear staining for Mitf coincides with strong peripheral staining for β-catenin.

Chapter 2 Results

Several researchers have shown the link between melanoma resistance to TGF-β and metastatic behaviour (Krasagakis et al., 1999; Medrano, 2003; Rodeck et al., 1994). While TGF-β stimulation of neural crest stem cells precipitates their transformation into smooth muscle cells (Shah et al., 1996), it induces apoptosis in neural crest derived melanocytes (Alanko and Saksela, 2000). Similarly, it has also been shown that TGF-β can inhibit the growth of some melanomas (Ladanyi et al., 2001). Heredia and coworkers showed that as melanoma cell lines become more differentiated (i.e. they resemble a neural crest derivative less and less), they are less sensitive to TGF-β-mediated growth inhibition (Heredia et al., 1996). The proliferation assays showed that in comparison to cohort A melanomas, most cohort C melanomas were only weakly or not at all inhibited by 5 ng/ml of exogenous TGF-β (Table 2.2). We also found that base growth rates of cohort A melanomas were almost two fold greater than cohort C melanomas (Table 2.2). These in vitro tests support the hypothesis that the identified cohorts are linked to metastatic potential.

Table 2.2. Cohort distribution, TGF-β-mediated growth inhibition and base proliferation rate. Relative Sample Cohort % inhibition proliferation M000921 46 1 M010817 A 53 0.53 M980513 36 0.49 M000216 28 0.34 M990514 B 19 0.48 M000907 11 0.44 M010119 3 0.21 M991121 10 0.18 M010322 51 0.91 C M010308 0 0.12 M990115 0 0.29 M010718 0 0.39

2.3.6 Wnt signalling controls Motif 1

The identity of the genes contributing to the Motif 1 expression pattern (Appendix A, Supplementary data 2b) suggest that in melanoma this gene set is regulated by Wnt signalling. To explore the possible relationship between Wnt signalling and neural crest/pigmentation gene expression we used immunohistochemistry. We compared the levels of β-catenin and Mitf in paraffin-embedded biopsy samples of one melanoma from each of cohorts B and C that show differential Mitf expression. We found that melanoma nuclear staining for Mitf correlated with strong immunoreactivity for β-catenin in cohort B, whereas in cohort C melanoma there was very little staining for Mitf and β-catenin (Fig. 2.5G-J). This is evidence

Chapter 2 Results / Discussion that in cohort C melanomas β-catenin is subject to increased turnover, suggesting that in these cells Wnt signalling has been deactivated.

2.4 Discussion

The patterns of gene regulation we identified and the signalling pathways they suggest, lead us to construct a model for the gene regulation of melanoma metastatic potential in melanoma (Fig. 2.6). Proliferative and weakly metastatic cells maintain a neural crest-like transcriptional signature through Wnt signalling. However, the induction of TGF-β-like signalling, possibly through microenvironmental changes brought about by hypoxia or inflammation, drives the expression of factors inhibitory to Wnt signalling and resulting in cells which are less proliferative but have high metastatic potential (Fig. 2.6).

Figure 2.6. A gene regulation model for melanoma metastatic potential. Left panel, the neural crest phenotype of cohort A cells is set by Wnt signalling which diverts β-catenin from ubiquitination to the nucleus where it participates in the regulation of neural crest genes, resulting in proliferative cells with weak metastatic potential. Right panel, the TGF-β-like signalling apparent in cohort C cells upregulates factors which drive a positive feedback signal, change the extracellular environment and inhibit Wnt signalling. Subsequent β-catenin degradation interrupts expression of neural crest genes, resulting in weakly-proliferative cells with high metastatic potential.

Chapter 2 Discussion

There have been two publications which have described correlations between activating mutations in BRAF and/or NRAS and changes in gene expression (Bloethner et al., 2005; Pavey et al., 2004). Interestingly, the gene lists these groups choose to highlight in their studies are entirely different. In contrast to their findings, we instead discovered that when multiple testing correction (Benjamini et al., 2001) is included in the analyses, activating mutations in BRAF and/or NRAS do not correlate with any gene expression changes in melanoma. We further discovered that when multiple testing correction is applied to the analysis originally performed by Pavey et al. (2004) on their data, no correlating genes are identified. Our findings indicate that the contribution to melanoma pathology by activating mutations of BRAF or NRAS does not involve their direct regulation of gene expression.

Subsequent to this we looked for a consistent molecular taxonomy of melanoma by assessing 86 different cultures of melanoma divided between three groups separately performing DNA microarray experiments. We found that melanomas can be categorised within a small number of cohorts based on their transcriptional signature. The cohorts distinguish melanomas which maintain a neural crest-type transcription signature (cohorts A and B) from melanomas which are significantly differentiated from this signature (cohort C). Identification of the genes with cohort-specific expression shows that twenty of them have previously recorded associations with metastatic potential (Appendix A, Supplementary Data 3a). This in combination with our in vitro experiments shows that the cohorts are representative of differences in metastatic potential. Additionally, our study has identified an additional 203 genes as having expression patterns linked to metastatic potential. An earlier study also used DNA microarrays to characterise the molecular signature of invasive melanomas (Bittner et al., 2000), identifying several genes which we also find similary regulated in melanomas we characterise as having high metastatic potential. Likewise, Seftor and co-workers compared the expression profiles of highly invasive and poorly invasive uveal melanoma cell cultures and observed many genes for whom expression was different in the two groups (Seftor et al., 2002). de Wit and co-workers also used expression profiling to compare two human melanoma cell cultures of differing capacity to metastasise to the lungs of nude mice (de Wit et al., 2002). We expand upon these findings, which help to confirm our assignment of cohort distribution to variations in metastatic potential, to show that many of these genes are co-regulated and that the signalling pathways behind the regulation of their expression are identifiable.

Chapter 2 Discussion

Our analysis of co-regulated gene expression, which specifically assessed pattern correlation, found that most of the cohort-differentiating genes were split between only two expression patterns (motifs) which dominate the cohort-specific data (Fig. 2.2). Close correlation of many different genes’ expression patterns implies that such genes are, within the system being studied, subject to a single transcriptional mechanism. This mechanism necessarily includes both the transcription factors and the signalling cascades which activate them. Our data implies that only two transcriptional programs underlie the molecular differences between strongly and weakly metastatic melanomas. Furthermore, these motifs are approximate mirror-images, which suggests that the transcriptional program of one may influence the other. To determine what these transcriptional programs might be we revisited the literature to evaluate what transcriptional regulators might link the genes in question together.

Motif 1 focuses on genes critical to neural crest differentiation and cell cycle control. Recent data suggest a connection between cell cycle control and melanoma genetics as the microphthalmia-associated transcription factor (Mitf), a master regulator for melanocytic differentiation (Steingrimsson et al., 2004), regulates the CDK2 gene in primary melanoma and subsequently plays an important role in melanoma cell cycle regulation (Du et al., 2004; Garraway et al., 2005). We found that CDK2 and all known melanocytic Mitf-regulated genes share a Motif 1 profile (Table 1), as does the Mitf gene itself. Having found that Mitf expression was also downregulated in cohort C melanomas, we examined the data to identify upstream transcription factors. Among the many other genes sharing the Motif 1 expression pattern, the neural crest cell regulator SOX10 was identified as a candidate for regulating Mitf expression and subsequent regulation of the CDK2/melanocytic differentiation set. Previous studies have shown that ectopically expressed SOX10 regulates Mitf expression (Huber et al., 2003; Lee et al., 2000; Potterf et al., 2000) and other SOX10-regulated genes are present within the Motif 1 expression pattern group (Table 1). Several other factors in the Motif 1 group likely are regulated by one or both of SOX10/Mitf. Among these is Rab38, which is expressed in melanocytes and abrogation of its function is responsible for oculocutaneous albinism in mice (Osanai et al., 2005). As the SOX10/Mitf axis is likely central to the established regulatory pattern of Motif 1, we then investigated the genes that might be regulating SOX10/Mitf. The downregulation of SOX10 and Mitf in cohort C melanomas is probably due to an interruption in Wnt/β-catenin signalling. In Xenopus, ectopic Xwnt-1 enhanced SOX10 expression and ectopic glycogen synthase kinase 3 (GSK3), an antagoniser of Wnt signalling, blocks SOX10 expression (Aoki et al., 2003; He et al., 1995). Also, 65

Chapter 2 Discussion canonical Wnt signalling through the Wnt/β-catenin pathway is a critical activator of Mitf expression (Dorsky et al., 2000; Shibahara et al., 2001; Yasumoto et al., 2002). Most convincingly, it was shown that Mitf expression in melanoma can be regulated by β-catenin (Widlund et al., 2002). Candidate genes for interrupting the Wnt/β-catenin pathway include several genes with the Motif 2 expression pattern. Perhaps the most interesting factor that can negatively regulate the Wnt signalling pathway is connective tissue growth factor, the gene for which is one of the most strongly expressed in cohort C melanomas. It was shown to bind the Wnt co-receptor LDL receptor–related protein 6 and inhibit Wnt signalling (Mercurio et al., 2004). Other candidate inhibitors whose genes are present in Motif 2 include Wnt-5a (Ishitani et al., 2003), Dickkopf 1 (Zorn, 2001), Dickkopf 3 (Mao et al., 2001), and cysteine- rich angiogenic inducer 61 (Latinkic et al., 2003). A further sign that the Wnt/β-catenin pathway has been disregulated in cohort C melanomas is the presence of the gene encoding transcription factor AP-2α (TFAP2A) in Motif 1. AP-2α has previously been shown to be downregulated in metastatic melanomas (Bar-Eli, 2001). Luo and co-workers linked Wnt signalling with increased TFAP2A expression in Xenopus development (Luo et al., 2003).

For Motif 2 genes (upregulated in melanoma of high metastatic potential) we found many regulated by TGF-β-type signalling (Table 1). This finding indicates that a difference exists in TGF-β-type signalling between cohorts. However, in the transcription record, we found no cohort-specific expression patterns among the TGF-β receptor or Smad genes. Likewise, no cohort-related change exists in expression of the three TGF-β variants (TGF-B1, TGF-B2, and TGF-B3). Furthermore, conditioned media samples showed little difference in TGF-β1 production across the samples (Fig. 2.4). While TGF-β production has been shown previously by several groups (Krasagakis et al., 1999; Moretti et al., 1997; Reed et al., 1994), and while increased resistance to TGF-β is associated with increased metastatic behaviour (Heredia et al., 1996), no group has clearly linked TGF-β production with increased metastatic behaviour. However, other autocrine effectors with appropriate expression patterns and TGF-β-like effects are present in our data. For example, activin A is a TGF-β-family signalling molecule that does have a cohort-specific expression pattern. The protein is a homodimer of inhibin beta A polypeptide whose gene (INHBA) is a member of the Motif 2 group (Data Supplement 1). The cohort specificity of this expression was supported by experiments that measured activin A secretion into conditioned media (Fig. 2.4). Additionally, the gene for thrombospondin 1 (THBS1), another member of Motif 2, encodes a

Chapter 2 Discussion / Material and Methods secreted matricellular factor that mediates changes in cell-cell and cell-matrix interactions (Bradshaw and Sage, 2001). Thrombospondin 1 also activates TGF-β, binding latent TGF-β and changing its conformation to an activated form (Schiemann et al., 2003). The secretion of these factors into the media is likely to contribute to an activated TGF-β-like signal.

We present here data that strongly suggests a transcriptional taxonomy for melanoma not related to neoplastic transformation (Hoek et al., 2004) or stage progression (Smith et al., 2005), but rather is separately concerned with metastatic potential. We note that immunohistochemical analyses of melanoma biopsies, using antibodies targeting the products of genes which we mention, often yield heterogeneous staining patterns among otherwise morphologically similar cells. This suggests to us that individual melanoma growths are comprised of cells from across the cohort spectrum, including populations which are proliferative but less metastatic than others less proliferative and yet more invasive, TGF-β- resistant, and secreting factors which change microenvironmental architecture and encourage neovasculogenesis. We suspect that microenvironmental cues, such as hypoxia and inflammation, may allow melanoma cells to switch epigenetically between cohort transcriptional signatures.

2.5 Material and Methods

2.5.1 Cell Culture and Media

Surplus material from cutaneous melanoma metastases removed by surgery were obtained after written informed consent. Clinical diagnosis was confirmed by histology and immunohistochemistry. Melanoma cells were released from tissue sections and grown as previously described (Geertsen et al., 1998). Melanocytes from infant foreskins were cultured in Medium 254 with Human Melanocyte Growth Supplement (Cascade Biologics, Portland,

OR) at 37°C in an atmosphere of 5% CO2.

Chapter 2 Material and Methods

2.5.2 Genotyping

DNA was extracted from each cell line using a QIAamp DNA Mini kit (Qiagen, Valencia, CA). Codons 597 and 600, and codons 12, 13, and 61 of BRAF and NRAS, respectively, were amplified using primers and amplification protocols as described by Pavey et al. (Pavey et al., 2004). Purified PCR products were submitted for sequencing to Microsynth (Balgach, Switzerland).

2.5.3 Total RNA Extraction and Expression Profiling

Total RNA was extracted from melanocyte and melanoma cell cultures using Trizol according to manufacturer instructions (Invitrogen, Carlsbad, CA). When melanin was present in the extract, we used a filter-based method, specifically the RNA-4PCR kit (Ambion, Austin, TX), to separate melanin from the total RNA. Extracted total RNA was used to synthesize poly(T)- primed double-stranded cDNA to provide a template for the transcription of biotin-labeled RNA according to recommendations provided by the manufacturer (Affymetrix, Santa Clara, CA). The biotin-labeled RNA was hybridized to Affymetrix HG-U133 set oligonucleotide microarrays following the manufacturer’s protocol. The Zürich group used both HG-U133A and HG-U133B chips, the Philadelphia group used the HG-U133A chip only, and the Mannheim group used the HG-U133 plus 2.0 chip. After hybridization, the microarrays were washed and stained using a GeneChip Fluidics Station 400 (Affymetrix), and scanned using a GeneArray Scanner (Agilent Technologies, Palo Alto, CA). The raw signal intensity data was scaled to an arbitrary mean value of 500 by MAS 5.0 software (Affymetrix). For select genes we performed real time quantitative PCR to validate gene expression using primers and conditions as outlined in Supplementary Data 4 (Appendix A).

2.5.4 Microarray Data Analysis

The Zürich, Philadelphia and Mannheim data sets were processed separately using identical protocols. All normalisations and analyses were performed using GeneSpring GX 7.3 (Agilent Technologies, Palo Alto, CA). Probe set data values below 0.01 were set to 0.01 and each measurement was divided by the 50th percentile of all measurements in that sample. Finally, each probe set measurement was divided by the median of its measurements in all

Chapter 2 Material and Methods samples. For unbiased sample clustering, hierarchical clustering was performed many times using different pools of probe sets and the results compared to find stable clusters of samples. Pools of probe sets were generated by using individual samples to act as denominators for fold-change comparison against the remaining samples. Probe sets with a two-fold change in normalised signal intensity between the denominator sample and at least half the remaining samples were selected. All samples were used in this way to generate separate pools of probe sets which were then individually subjected to sample clustering analyses. Hierarchical clustering of samples was performed using Standard, Pearson, and Spearman correlation algorithms for each pool in order to control for similarity measure artifacts. The results of hierarchical clustering experiments were compared to identify stable clusters of samples which were then termed “cohorts.” To determine the gene expression patterns differentiating between sample cohorts, a statistical analysis (ANOVA) was used to identify probe sets with cohort-specific expression patterns. The statistical analysis used a parametric test in which all variances were assumed to be equal, a p-value cut-off of 0.05 was used and the Benjamini Hochberg false discovery rate (Benjamini et al., 2001) was employed for multiple testing correction. Post Hoc testing was performed using the Tukey test to determine statistical significance of differences between the probe sets of specific cohort pairs. All statistical tests were two-sided. Cohort-specific expressing probe sets common to the Zürich, Philadelphia and Mannheim data sets were subjected to gene tree clustering using a Standard correlation algorithm. Self organized map (SOM) analysis was used to sort cohort-specific expression patterns across a 4×3 matrix over 15 000 iterations using a neighbourhood radius of 4. This process randomly distributes probe set expression patterns among a 4×3 matrix of cells. The signal values across the samples for each probe set are converted into a single n-dimensional vector (where n is the number of samples). The angular distance between probe set vectors within a cell is calculated and the probe sets are divided into tow equal groups. The first group represents the 50% of probe sets within the cell which form the population with the least angular difference from the centroid, the other group represents the remaining 50% of probe sets within the cell. The first group is retained within the cell and the remaining probe sets are randomly distributed between neighbour cells in the SOM. The reach of this distribution is determined by the user. This process is repeated over many thousand iterations, with closely similar vectors remaining closely linked in the SOM while dissimilar vectors tend to separate widely. The SOM was used to identify the major expression Motifs (across-sample expression patterns of groups of correlating probe sets) responsible for defining sample cohorts. Probe set

Chapter 2 Material and Methods identifications were confirmed using the NetAffx Analysis package accessible from Affymetrix (http://www.affymetrix.com).

2.5.5 Growth Inhibition Assays

Aliquots of 2×104 cells were seeded into 24-well microplates with 400 µL of standard media containing 0 or 5 ng/mL of TGF-β1 (Biosource, Camarillo, CA) and incubated for four days under conditions previously outlined in the cell culture section. Cell metabolic activity was determined with a standard colorimetric assay measuring 3-(4,5-dimethyldiazol-2-yl)-2,5 diphenyl tetrazolium bromide (MTT; Sigma–Aldrich, St Louis, MO) reactivity and used as an approximation of cell proliferation. Growth inhibition was expressed as a percentage of growth inhibition compared against growth in the absence of TGF-β1. Each experiment was performed four times.

2.5.6 Motility Assays

Cell migration in cultures was measured using a two-dimensional in vitro scratch motility assay. Cells were grown to confluence on tissue culture dishes under standard growth conditions. A wound of several millimetres in length and approximately 1 mm wide was made by scratching through the monolayer using a sterile Gilson 200-µL pipette tip. After washing with phosphate-buffered saline (PBS) and replacing the growth medium, cells were incubated under standard conditions for 24 h and then observed for repopulation of the cleared field.

2.5.7 Western Analyses and ELISA

Growth-conditioned media was prepared by growing cells to confluency in a 75-cm2 flask under previously described conditions(Geertsen et al., 1998), these were then washed with PBS and placed into fresh serum-free media for 24 h. Growth-conditioned media was collected, filtered through a 0.22-µm filter and stored at -80°C. 100 µL of growth-conditioned media was concentrated 10-fold by acetone precipitation. Proteins were separated by SDS- PAGE under reducing conditions and transferred onto nitrocellulose membrane (Invitrogen). 70

Chapter 2 Material and Methods

Western blotting used rabbit polyclonal to TGF-β1 used at 1:200 (Abcam) and rabbit anti- inhibin βA (gift of Prof. S. Werner) used at 1:2 000. Bound antibodies were detected using horseradish peroxidase-conjugated anti-rabbit secondary antibody (BioRad, Hercules, CA) and developed using ECL (GE Healthcare, Piscataway, NJ). Activin A was also measured in conditioned media samples using a specific enzyme linked immunosorbent assay (Knight et al., 1996) according to the manufacturer’s instructions (Oxford Bio-Innovations, Oxfordshire, UK), with some modifications for media samples as described previously (Buzzard et al., 2003). The average intra- and interplate coefficients of variation for this assay are routinely <8% and the lower limit of detection is 10 pg/ml. Cells were grown to confluency as previously described, washed twice with cold phosphate- buffered saline and lysed at 4°C in lysis buffer containing 20 mM Tris-HCl (pH 7.5), 1% Triton X-100, 150 mM NaCl, 10% glycerol, protease inhibitors (Roche) and phosphatase inhibitors (Sigma). After quantification, 10 µg of each sample was separated by SDS-PAGE under reducing conditions and transferred onto nitrocellulose membrane (Invitrogen).

Western blotting used mouse anti-Mitf AB-1(C5) used at 1:1 000 (Lab Visions). Goat polyclonal antibody against actin (Santa Cruz Biotechnology) was used at 1:1 000 as a control. Bound antibodies were detected using horseradish peroxidase-conjucated anti-mouse (Abcam) or anti-goat secondary antibodies (SantaCruz) and developed using ECL (GE Healthcare).

2.5.8 Immunohistochemistry

Paraffin-embedded tissue sections were stained using the alkaline phosphatase-anti-alkaline phosphatase technique and counterstained using hemotoxylin. Antibodies used were directed against β-catenin (Transduction Laboratories, Lexington, KY) and Mitf (DakoCytomation, Glostrup, Denmark).

Chapter 2 References

2.6 References

Alanko, T. and Saksela, O. (2000) Transforming growth factor beta1 induces apoptosis in normal melanocytes but not in nevus cells grown in type I collagen gel. J Invest Dermatol, 115, 286-291. Aoki, Y., Saint-Germain, N., Gyda, M., Magner-Fink, E., Lee, Y.H., Credidio, C. and Saint- Jeannet, J.P. (2003) Sox10 regulates the development of neural crest-derived melanocytes in Xenopus. Dev Biol, 259, 19-33. Bakin, A.V., Safina, A., Rinehart, C., Daroqui, C., Darbary, H. and Helfman, D.M. (2004) A critical role of tropomyosins in TGF-beta regulation of the actin cytoskeleton and cell motility in epithelial cells. Mol Biol Cell, 15, 4682-4694. Balch, C.M., Buzaid, A.C., Soong, S.J., Atkins, M.B., Cascinelli, N., Coit, D.G., Fleming, I.D., Gershenwald, J.E., Houghton, A., Jr., Kirkwood, J.M., McMasters, K.M., Mihm, M.F., Morton, D.L., Reintgen, D.S., Ross, M.I., Sober, A., Thompson, J.A. and Thompson, J.F. (2001) Final version of the American Joint Committee on Cancer staging system for cutaneous melanoma. J Clin Oncol, 19, 3635-3648. Bar-Eli, M. (2001) Gene regulation in melanoma progression by the AP-2 transcription factor. Pigment Cell Res, 14, 78-85. Benjamini, Y., Drai, D., Elmer, G., Kafkafi, N. and Golani, I. (2001) Controlling the false discovery rate in behavior genetics research. Behav Brain Res, 125, 279-284. Bittner, M., Meltzer, P., Chen, Y., Jiang, Y., Seftor, E., Hendrix, M., Radmacher, M., Simon, R., Yakhini, Z., Ben-Dor, A., Sampas, N., Dougherty, E., Wang, E., Marincola, F., Gooden, C., Lueders, J., Glatfelter, A., Pollock, P., Carpten, J., Gillanders, E., Leja, D., Dietrich, K., Beaudry, C., Berens, M., Alberts, D. and Sondak, V. (2000) Molecular classification of cutaneous malignant melanoma by gene expression profiling. Nature, 406, 536-540. Bloethner, S., Chen, B., Hemminki, K., Muller-Berghaus, J., Ugurel, S., Schadendorf, D. and Kumar, R. (2005) Effect of common B-RAF and N-RAS mutations on global gene expression in melanoma cell lines. Carcinogenesis, 26, 1224-1232. Bradshaw, A.D. and Sage, E.H. (2001) SPARC, a matricellular protein that functions in cellular differentiation and tissue response to injury. J Clin Invest, 107, 1049-1054. Britsch, S., Goerich, D.E., Riethmacher, D., Peirano, R.I., Rossner, M., Nave, K.A., Birchmeier, C. and Wegner, M. (2001) The transcription factor Sox10 is a key regulator of peripheral glial development. Genes Dev, 15, 66-78. Buzzard, J.J., Farnworth, P.G., De Kretser, D.M., O'Connor, A.E., Wreford, N.G. and Morrison, J.R. (2003) Proliferative phase sertoli cells display a developmentally regulated response to activin in vitro. Endocrinology, 144, 474-483. Carreira, S., Goodall, J., Aksan, I., La Rocca, S.A., Galibert, M.D., Denat, L., Larue, L. and Goding, C.R. (2005) Mitf cooperates with Rb1 and activates p21Cip1 expression to regulate cell cycle progression. Nature, 433, 764-769. Chen, S., Kulik, M. and Lechleider, R.J. (2003) Smad proteins regulate transcriptional induction of the SM22alpha gene by TGF-beta. Nucleic Acids Res, 31, 1302-1310. Davies, H., Bignell, G.R., Cox, C., Stephens, P., Edkins, S., Clegg, S., Teague, J., Woffendin, H., Garnett, M.J., Bottomley, W., Davis, N., Dicks, E., Ewing, R., Floyd, Y., Gray, K., Hall, S., Hawes, R., Hughes, J., Kosmidou, V., Menzies, A., Mould, C., Parker, A., Stevens, C., Watt, S., Hooper, S., Wilson, R., Jayatilake, H., Gusterson, B.A., Cooper, C., Shipley, J., Hargrave, D., Pritchard-Jones, K., Maitland, N., Chenevix-Trench, G., Riggins, G.J., Bigner, D.D., Palmieri, G., Cossu, A., Flanagan, A., Nicholson, A., Ho, J.W., Leung, S.Y., Yuen, S.T., Weber, B.L., Seigler, H.F., Darrow, T.L., Paterson, H., Marais, R.,

Chapter 2 References

Marshall, C.J., Wooster, R., Stratton, M.R. and Futreal, P.A. (2002) Mutations of the BRAF gene in human cancer. Nature, 417, 949-954. de Wit, N.J., Burtscher, H.J., Weidle, U.H., Ruiter, D.J. and van Muijen, G.N. (2002) Differentially expressed genes identified in human melanoma cell lines with different metastatic behaviour using high density oligonucleotide arrays. Melanoma Res, 12, 57-69. Derynck, R., Akhurst, R.J. and Balmain, A. (2001) TGF-beta signaling in tumor suppression and cancer progression. Nat Genet, 29, 117-129. Dorsky, R.I., Raible, D.W. and Moon, R.T. (2000) Direct regulation of nacre, a zebrafish MITF homolog required for pigment cell formation, by the Wnt pathway. Genes Dev, 14, 158-162. Du, J. and Fisher, D.E. (2002) Identification of Aim-1 as the underwhite mouse mutant and its transcriptional regulation by MITF. J Biol Chem, 277, 402-406. 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. Du, J., Widlund, H.R., Horstmann, M.A., Ramaswamy, S., Ross, K., Huber, W.E., Nishimura, E.K., Golub, T.R. and Fisher, D.E. (2004) Critical role of CDK2 for melanoma growth linked to its melanocyte-specific transcriptional regulation by MITF. Cancer Cell, 6, 565- 576. Elliott, R.L. and Blobe, G.C. (2005) Role of transforming growth factor Beta in human cancer. J Clin Oncol, 23, 2078-2093. Fang, D., Tsuji, Y. and Setaluri, V. (2002) Selective down-regulation of tyrosinase family gene TYRP1 by inhibition of the activity of melanocyte transcription factor, MITF. Nucleic Acids Res, 30, 3096-3106. Finlay, G.A., Thannickal, V.J., Fanburg, B.L. and Paulson, K.E. (2000) Transforming growth factor-beta 1-induced activation of the ERK pathway/activator protein-1 in human lung fibroblasts requires the autocrine induction of basic fibroblast growth factor. J Biol Chem, 275, 27650-27656. 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., Lee, C., Wagner, S.N., Li, C., Golub, T.R., Rimm, D.L., Meyerson, M.L., Fisher, D.E. and Sellers, W.R. (2005) Integrative genomic analyses identify MITF as a lineage survival oncogene amplified in malignant melanoma. Nature, 436, 117-122. Geertsen, R.C., Hofbauer, G.F., Yue, F.Y., Manolio, S., Burg, G. and Dummer, R. (1998) Higher frequency of selective losses of HLA-A and -B allospecificities in metastasis than in primary melanoma lesions. J Invest Dermatol, 111, 497-502. Gold, L.I. (1999) The role for transforming growth factor-beta (TGF-beta) in human cancer. Crit Rev Oncog, 10, 303-360. Green, R.S., Lieb, M.E., Weintraub, A.S., Gacheru, S.N., Rosenfield, C.L., Shah, S., Kagan, H.M. and Taubman, M.B. (1995) Identification of lysyl oxidase and other platelet-derived growth factor-inducible genes in vascular smooth muscle cells by differential screening. Lab Invest, 73, 476-482. Haass, N.K., Smalley, K.S., Li, L. and Herlyn, M. (2005) Adhesion, migration and communication in melanocytes and melanoma. Pigment Cell Res, 18, 150-159. He, X., Saint-Jeannet, J.P., Woodgett, J.R., Varmus, H.E. and Dawid, I.B. (1995) Glycogen synthase kinase-3 and dorsoventral patterning in Xenopus embryos. Nature, 374, 617-622. Heredia, A., Villena, J., Romaris, M., Molist, A. and Bassols, A. (1996) The effect of TGF- beta 1 on cell proliferation and proteoglycan production in human melanoma cells depends on the degree of cell differentiation. Cancer Lett, 109, 39-47.

Chapter 2 References

Hoek, K., Rimm, D.L., Williams, K.R., Zhao, H., Ariyan, S., Lin, A., Kluger, H.M., Berger, A.J., Cheng, E., Trombetta, E.S., Wu, T., Niinobe, M., Yoshikawa, K., Hannigan, G.E. and Halaban, R. (2004) Expression profiling reveals novel pathways in the transformation of melanocytes to melanomas. Cancer Res, 64, 5270-5282. Hou, L., Panthier, J.J. and Arnheiter, H. (2000) Signaling and transcriptional regulation in the neural crest-derived melanocyte lineage: interactions between KIT and MITF. Development, 127, 5379-5389. Hsu, M., Andl, T., Li, G., Meinkoth, J.L. and Herlyn, M. (2000a) Cadherin repertoire determines partner-specific gap junctional communication during melanoma progression. J Cell Sci, 113 ( Pt 9), 1535-1542. Hsu, M.Y., Meier, F. and Herlyn, M. (2002a) Melanoma development and progression: a conspiracy between tumor and host. Differentiation, 70, 522-536. Hsu, M.Y., Meier, F.E., Nesbit, M., Hsu, J.Y., Van Belle, P., Elder, D.E. and Herlyn, M. (2000b) E-cadherin expression in melanoma cells restores keratinocyte-mediated growth control and down-regulates expression of invasion-related adhesion receptors. Am J Pathol, 156, 1515-1525. Hsu, S.C., Kirschenbaum, F., Miller, J., Cordell, B. and McCarthy, J.V. (2002b) Structural and functional characterization of the upstream regulatory region of the human gene encoding prostate apoptosis response factor-4. Gene, 295, 109-116. Huber, W.E., Price, E.R., Widlund, H.R., Du, J., Davis, I.J., Wegner, M. and Fisher, D.E. (2003) A tissue-restricted cAMP transcriptional response: SOX10 modulates alpha- melanocyte-stimulating hormone-triggered expression of microphthalmia-associated transcription factor in melanocytes. J Biol Chem, 278, 45224-45230. Hubner, G. and Werner, S. (1996) Serum growth factors and proinflammatory cytokines are potent inducers of activin expression in cultured fibroblasts and keratinocytes. Exp Cell Res, 228, 106-113. Hugo, C. (2003) The thrombospondin 1-TGF-beta axis in fibrotic renal disease. Nephrol Dial Transplant, 18, 1241-1245. Igarashi, A., Okochi, H., Bradham, D.M. and Grotendorst, G.R. (1993) Regulation of connective tissue growth factor gene expression in human skin fibroblasts and during wound repair. Mol Biol Cell, 4, 637-645. Ishitani, T., Kishida, S., Hyodo-Miura, J., Ueno, N., Yasuda, J., Waterman, M., Shibuya, H., Moon, R.T., Ninomiya-Tsuji, J. and Matsumoto, K. (2003) The TAK1-NLK mitogen- activated protein kinase cascade functions in the Wnt-5a/Ca(2+) pathway to antagonize Wnt/beta-catenin signaling. Mol Cell Biol, 23, 131-139. Knight, P.G., Muttukrishna, S. and Groome, N.P. (1996) Development and application of a two-site enzyme immunoassay for the determination of 'total' activin-A concentrations in serum and follicular fluid. J Endocrinol, 148, 267-279. Krasagakis, K., Garbe, C., Schrier, P.I. and Orfanos, C.E. (1994) Paracrine and autocrine regulation of human melanocyte and melanoma cell growth by transforming growth factor beta in vitro. Anticancer Res, 14, 2565-2571. Krasagakis, K., Kruger-Krasagakes, S., Fimmel, S., Eberle, J., Tholke, D., von der Ohe, M., Mansmann, U. and Orfanos, C.E. (1999) Desensitization of melanoma cells to autocrine TGF-beta isoforms. J Cell Physiol, 178, 179-187. Ladanyi, A., Gallai, M., Paku, S., Nagy, J.O., Dudas, J., Timar, J. and Kovalszky, I. (2001) Expression of a decorin-like moleculein human melanoma. Pathol Oncol Res, 7, 260-266. Latinkic, B.V., Mercurio, S., Bennett, B., Hirst, E.M., Xu, Q., Lau, L.F., Mohun, T.J. and Smith, J.C. (2003) Xenopus Cyr61 regulates gastrulation movements and modulates Wnt signalling. Development, 130, 2429-2441.

Chapter 2 References

Lawrence, R., Hartmann, D.J. and Sonenshein, G.E. (1994) Transforming growth factor beta 1 stimulates type V collagen expression in bovine vascular smooth muscle cells. J Biol Chem, 269, 9603-9609. Lee, M., Goodall, J., Verastegui, C., Ballotti, R. and Goding, C.R. (2000) Direct regulation of the Microphthalmia promoter by Sox10 links Waardenburg-Shah syndrome (WS4)- associated hypopigmentation and deafness to WS2. J Biol Chem, 275, 37978-37983. Li, G. and Herlyn, M. (2000) Dynamics of intercellular communication during melanoma development. Mol Med Today, 6, 163-169. Linnekin, D. (1999) Early signaling pathways activated by c-Kit in hematopoietic cells. Int J Biochem Cell Biol, 31, 1053-1074. Loercher, A.E., Tank, E.M., Delston, R.B. and Harbour, J.W. (2005) MITF links differentiation with cell cycle arrest in melanocytes by transcriptional activation of INK4A. J Cell Biol, 168, 35-40. Lorena, D., Darby, I.A., Reinhardt, D.P., Sapin, V., Rosenbaum, J. and Desmouliere, A. (2004) Fibrillin-1 expression in normal and fibrotic rat liver and in cultured hepatic fibroblastic cells: modulation by mechanical stress and role in cell adhesion. Lab Invest, 84, 203-212. Luo, T., Lee, Y.H., Saint-Jeannet, J.P. and Sargent, T.D. (2003) Induction of neural crest in Xenopus by transcription factor AP2alpha. Proc Natl Acad Sci U S A, 100, 532-537. Macfelda, K., Weiss, T.W., Kaun, C., Breuss, J.M., Zorn, G., Oberndorfer, U., Voegele- Kadletz, M., Huber-Beckmann, R., Ullrich, R., Binder, B.R., Losert, U.M., Maurer, G., Pacher, R., Huber, K. and Wojta, J. (2002) Plasminogen activator inhibitor 1 expression is regulated by the inflammatory mediators interleukin-1alpha, tumor necrosis factor-alpha, transforming growth factor-beta and oncostatin M in human cardiac myocytes. J Mol Cell Cardiol, 34, 1681-1691. Mao, B., Wu, W., Li, Y., Hoppe, D., Stannek, P., Glinka, A. and Niehrs, C. (2001) LDL- receptor-related protein 6 is a receptor for Dickkopf proteins. Nature, 411, 321-325. Medrano, E.E. (2003) Repression of TGF-beta signaling by the oncogenic protein SKI in human melanomas: consequences for proliferation, survival, and metastasis. Oncogene, 22, 3123-3129. Mercurio, S., Latinkic, B., Itasaki, N., Krumlauf, R. and Smith, J.C. (2004) Connective-tissue growth factor modulates WNT signalling and interacts with the WNT receptor complex. Development, 131, 2137-2147. Miller, A.J., Du, J., Rowan, S., Hershey, C.L., Widlund, H.R. and Fisher, D.E. (2004) Transcriptional regulation of the melanoma prognostic marker melastatin (TRPM1) by MITF in melanocytes and melanoma. Cancer Res, 64, 509-516. Moretti, S., Pinzi, C., Berti, E., Spallanzani, A., Chiarugi, A., Boddi, V., Reali, U.M. and Giannotti, B. (1997) In situ expression of transforming growth factor beta is associated with melanoma progression and correlates with Ki67, HLA-DR and beta 3 integrin expression. Melanoma Res, 7, 313-321. Osanai, K., Takahashi, K., Nakamura, K., Takahashi, M., Ishigaki, M., Sakuma, T., Toga, H., Suzuki, T. and Voelker, D.R. (2005) Expression and characterization of Rab38, a new member of the Rab small G protein family. Biol Chem, 386, 143-153. Padgett, R.W. (1999) TGFbeta signaling pathways and human diseases. Cancer Metastasis Rev, 18, 247-259. Park, J.I., Lee, M.G., Cho, K., Park, B.J., Chae, K.S., Byun, D.S., Ryu, B.K., Park, Y.K. and Chi, S.G. (2003) Transforming growth factor-beta1 activates interleukin-6 expression in prostate cancer cells through the synergistic collaboration of the Smad2, p38-NF-kappaB, JNK, and Ras signaling pathways. Oncogene, 22, 4314-4332.

Chapter 2 References

Pavey, S., Johansson, P., Packer, L., Taylor, J., Stark, M., Pollock, P.M., Walker, G.J., Boyle, G.M., Harper, U., Cozzi, S.J., Hansen, K., Yudt, L., Schmidt, C., Hersey, P., Ellem, K.A., O'Rourke, M.G., Parsons, P.G., Meltzer, P., Ringner, M. and Hayward, N.K. (2004) Microarray expression profiling in melanoma reveals a BRAF mutation signature. Oncogene, 23, 4060-4067. Phillips, D.J., Jones, K.L., Clarke, I.J., Scheerlinck, J.P. and de Kretser, D.M. (2005) Activin A: from sometime reproductive factor to genuine cytokine. Vet Immunol Immunopathol, 108, 23-27. 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., Salem, G., Pohida, T., Heenan, P., Duray, P., Kallioniemi, O., Hayward, N.K., Trent, J.M. and Meltzer, P.S. (2003) High frequency of BRAF mutations in nevi. Nat Genet, 33, 19-20. Potterf, S.B., Furumura, M., Dunn, K.J., Arnheiter, H. and Pavan, W.J. (2000) Transcription factor hierarchy in Waardenburg syndrome: regulation of MITF expression by SOX10 and PAX3. Hum Genet, 107, 1-6. Quinones, L.G. and Garcia-Castro, I. (2004) Characterization of human melanoma cell lines according to their migratory properties in vitro. In Vitro Cell Dev Biol Anim, 40, 35-42. Raz, A. (1988) Actin organization, cell motility, and metastasis. Adv Exp Med Biol, 233, 227- 233. Reed, J.A., McNutt, N.S., Prieto, V.G. and Albino, A.P. (1994) Expression of transforming growth factor-beta 2 in malignant melanoma correlates with the depth of tumor invasion. Implications for tumor progression. Am J Pathol, 145, 97-104. Roberts, A.B., Anzano, M.A., Wakefield, L.M., Roche, N.S., Stern, D.F. and Sporn, M.B. (1985) Type beta transforming growth factor: a bifunctional regulator of cellular growth. Proc Natl Acad Sci U S A, 82, 119-123. Rodeck, U., Bossler, A., Graeven, U., Fox, F.E., Nowell, P.C., Knabbe, C. and Kari, C. (1994) Transforming growth factor beta production and responsiveness in normal human melanocytes and melanoma cells. Cancer Res, 54, 575-581. Schiemann, B.J., Neil, J.R. and Schiemann, W.P. (2003) SPARC inhibits epithelial cell proliferation in part through stimulation of the transforming growth factor-beta-signaling system. Mol Biol Cell, 14, 3977-3988. Seftor, E.A., Meltzer, P.S., Kirschmann, D.A., Pe'er, J., Maniotis, A.J., Trent, J.M., Folberg, R. and Hendrix, M.J. (2002) Molecular determinants of human uveal melanoma invasion and metastasis. Clin Exp Metastasis, 19, 233-246. Shah, N.M., Groves, A.K. and Anderson, D.J. (1996) Alternative neural crest cell fates are instructively promoted by TGFbeta superfamily members. Cell, 85, 331-343. Shibahara, S., Takeda, K., Yasumoto, K., Udono, T., Watanabe, K., Saito, H. and Takahashi, K. (2001) Microphthalmia-associated transcription factor (MITF): multiplicity in structure, function, and regulation. J Investig Dermatol Symp Proc, 6, 99-104. Shinozaki, M., Fujimoto, A., Morton, D.L. and Hoon, D.S. (2004) Incidence of BRAF oncogene mutation and clinical relevance for primary cutaneous melanomas. Clin Cancer Res, 10, 1753-1757. Skonier, J., Neubauer, M., Madisen, L., Bennett, K., Plowman, G.D. and Purchio, A.F. (1992) cDNA cloning and sequence analysis of beta ig-h3, a novel gene induced in a human adenocarcinoma cell line after treatment with transforming growth factor-beta. DNA Cell Biol, 11, 511-522. Smith, A.P., Hoek, K. and Becker, D. (2005) Whole-Genome Expression Profiling of the Melanoma Progression Pathway Reveals Marked Molecular Differences between Nevi/Melanoma In Situ and Advanced-Stage Melanomas. Cancer Biol Ther, 4, 1018-1029.

Chapter 2 References

Steingrimsson, E., Copeland, N.G. and Jenkins, N.A. (2004) Melanocytes and the Microphthalmia Transcription Factor Network. Annu Rev Genet, 38, 365-411. Stolt, C.C., Rehberg, S., Ader, M., Lommes, P., Riethmacher, D., Schachner, M., Bartsch, U. and Wegner, M. (2002) Terminal differentiation of myelin-forming oligodendrocytes depends on the transcription factor Sox10. Genes Dev, 16, 165-170. Tada, A., Kato, H. and Hasegawa, S. (2000) Antagonistic effect of EGF against TGF beta1 on transformed phenotype and tropomyosin expression of human lung carcinoma A549 cells. Oncol Rep, 7, 1323-1326. Takeichi, M. (1991) Cadherin cell adhesion receptors as a morphogenetic regulator. Science, 251, 1451-1455. Thirunavukkarasu, K., Miles, R.R., Halladay, D.L., Yang, X., Galvin, R.J., Chandrasekhar, S., Martin, T.J. and Onyia, J.E. (2001) Stimulation of osteoprotegerin (OPG) gene expression by transforming growth factor-beta (TGF-beta). Mapping of the OPG promoter region that mediates TGF-beta effects. J Biol Chem, 276, 36241-36250. Tsutsumida, A., Hamada, J., Tada, M., Aoyama, T., Furuuchi, K., Kawai, Y., Yamamoto, Y., Sugihara, T. and Moriuchi, T. (2004) Epigenetic silencing of E- and P-cadherin gene expression in human melanoma cell lines. Int J Oncol, 25, 1415-1421. Tuli, R., Tuli, S., Nandi, S., Huang, X., Manner, P.A., Hozack, W.J., Danielson, K.G., Hall, D.J. and Tuan, R.S. (2003) Transforming growth factor-beta-mediated chondrogenesis of human mesenchymal progenitor cells involves N-cadherin and mitogen-activated protein kinase and Wnt signaling cross-talk. J Biol Chem, 278, 41227-41236. Ungefroren, H., Lenschow, W., Chen, W.B., Faendrich, F. and Kalthoff, H. (2003) Regulation of biglycan gene expression by transforming growth factor-beta requires MKK6-p38 mitogen-activated protein Kinase signaling downstream of Smad signaling. J Biol Chem, 278, 11041-11049. Vaisanen, A., Tuominen, H., Kallioinen, M. and Turpeenniemi-Hujanen, T. (1996) Matrix metalloproteinase-2 (72 kD type IV collagenase) expression occurs in the early stage of human melanocytic tumour progression and may have prognostic value. J Pathol, 180, 283-289. Vetrini, F., Auricchio, A., Du, J., Angeletti, B., Fisher, D.E., Ballabio, A. and Marigo, V. (2004) The microphthalmia transcription factor (Mitf) controls expression of the ocular albinism type 1 gene: link between melanin synthesis and melanosome biogenesis. Mol Cell Biol, 24, 6550-6559. Wellbrock, C. and Marais, R. (2005) Elevated expression of MITF counteracts B-RAF- stimulated melanocyte and melanoma cell proliferation. J Cell Biol, 170, 703-708. Wellbrock, C., Ogilvie, L., Hedley, D., Karasarides, M., Martin, J., Niculescu-Duvaz, D., Springer, C.J. and Marais, R. (2004) V599EB-RAF is an oncogene in melanocytes. Cancer Res, 64, 2338-2342. Widlund, H.R., Horstmann, M.A., Price, E.R., Cui, J., Lessnick, S.L., Wu, M., He, X. and Fisher, D.E. (2002) Beta-catenin-induced melanoma growth requires the downstream target Microphthalmia-associated transcription factor. J Cell Biol, 158, 1079-1087. Wright, J.A. and Huang, A. (1996) Growth factors in mechanisms of malignancy: roles for TGF-beta and FGF. Histol Histopathol, 11, 521-536. Yasumoto, K., Takeda, K., Saito, H., Watanabe, K., Takahashi, K. and Shibahara, S. (2002) Microphthalmia-associated transcription factor interacts with LEF-1, a mediator of Wnt signaling. Embo J, 21, 2703-2714. Zhu, L., Lee, H.O., Jordan, C.S., Cantrell, V.A., Southard-Smith, E.M. and Shin, M.K. (2004) Spatiotemporal regulation of endothelin receptor-B by SOX10 in neural crest-derived enteric neuron precursors. Nat Genet, 36, 732-737. Zorn, A.M. (2001) Wnt signalling: antagonistic Dickkopfs. Curr Biol, 11, R592-595.

Chapter 3

3 In vitro phenotype validation..

Chapter 3 Introduction

3.1 Introduction

We have characterised two different transcription signatures for melanoma cell lines which, based on known functions of the genes involved, defined their respective contributions to metastatic potential as either proliferative or invasive (Hoek et al., 2006). We have further identified two patterns of co-regulated gene expression, so called motifs, which are sufficient to differentiate between the identified melanoma cohorts. Motif 1 comprises a number of co- regulated genes involved in melanocytic and neural crest differentiation and is thought to be controlled by Wnt signalling (Hoek et al., 2006). On the other hand, members of the Motif 2 transcription profile are involved in modifying extracellular environments and a large fraction of these genes, as well as others in the motif, are known to be subject to TGF-β regulation (Hoek et al., 2006).

We have identified TGF-β response as a phenotype discriminating between melanoma cells with differing metastatic potential (Hoek et al., 2006). There is now a large body of published evidence showing that canonical Smad signalling does not account for all TGF-β effects and that a number of pathways, including MAPK pathways, may be either activated by TGF-β or modulate its signalling (reviewed in Javelaud and Mauviel, 2005; Massague, 2003). Furthermore, Smad signalling is modulated by nuclear repressors such as Ski (Xu et al., 2000).

We used motility and TGF-β susceptibility as in vitro phenotypes to support the link between transcription signature cohort distribution and metastatic potential (Hoek et al., 2006). Additionally, vasculogenic mimicry is another phenotype discriminating between highly invasive and poorly invasive melanoma (Maniotis et al., 1999).

3.1.1 Modulation of TGF-β signalling

3.1.1.1 Crosstalk between TGF-β signalling and the MAPK pathways

Interaction between mitogen activated protein kinase (MAPK) pathways and TGF-β signalling is two-fold. On one hand, Smad-dependent signalling is not the only way that TGF- β regulates cellular functions, as TGF-β activates other pathways including MAPK pathways.

Chapter 3 Introduction

On the other hand, Smad-dependent signalling is not solely regulated through TGF-β receptors, but can be activated by other pathways such as the MAPK pathways.

TGF-β activation of ERK, p38 and c-jun N-terminal kinase (JNK) MAPK pathways is complex. While it can lead to Smad activation in some instances, it may also induce responses unrelated to Smad-dependent transcription in others (Derynck and Zhang, 2003). For example, slow kinetics observed in the activation of these pathways suggest Smad-dependent transcription responses, while rapid activation (5-15 minutes) similar to those observed downstream of cytokine receptors suggests activities independent from transcription (reviewed in Derynck and Zhang, 2003). Furthermore, the existence of Smad-independent MAPK activation is demonstrated by the activation of MAPK pathways in Smad4 deficient cells and cells expressing dominant negative Smads (reviewed in Javelaud and Mauviel, 2005). Further evidence of Smad-independent MAPK activation comes from reports showing TGF-β activation of different MAPK pathways using a mutated ALK5 which, although retaining kinase activity, was unable to activate Smads (Itoh et al., 2003; Yu et al., 2002).

The mechanisms by which different MAPK pathways are activated by TGF-β are not well characterised. Activation of Ras has been suggested to be important in TGF-β induction of the ERK MAPK pathway (Yue and Mulder, 2000). On the other hand, JNK and p38 MAPK pathways are activated by various MAPK kinase kinases, including TGF-β activated kinase-1 (TAK1) which can stimulate both (Yamaguchi et al., 1995). The identification of TAK1 provided some of the earliest evidence for crosstalk between MAPK and TGF-β pathways. Furthermore, MAPK kinase kinase-1 (MEKK1) is also thought to mediate TGF-β activation of the JNK MAPK pathway (Brown et al., 1999). Repression, as well as activation, of MAPK pathways by TGF-β signalling has also been observed. For example, activin was shown to inhibit cell growth in human erythroleukemia cells by inhibiting ERK signalling and activating the p38 pathway (Huang et al., 2004).

MAPK pathways are not only activated by TGF-β but, by phosphorylating R-Smads, they also interfere in Smad-dependent transcriptional activation. ERK has been shown to phosphorylate Smad1, Smad2 and Smad3 at specific sites in their linker regions to inhibit nuclear translocation (Kretzschmar et al., 1997; Kretzschmar et al., 1999). Furthermore, inhibition of Ras-induced Smad3 phosphorylation by mutation of MAPK phosphorylation

Chapter 3 Introduction sites rescued the growth-inhibitory response to TGF-β in Ras-transformed cells (Kretzschmar et al., 1999). This mechanism was proposed to explain silencing of antimitogenic TGF-β functions in hyperactive Ras cells. In contrast, hepatocyte growth factor (HGF) and epidermal growth factor (EGF) signalling were shown to mediate Smad-dependent reporter gene activation and induce rapid phosphorylation of Smad1 and Smad2 by kinases downstream of MEK1 (de Caestecker et al., 1998).

JNK has also been shown to phosphorylate R-Smads. JNK phosphorylation of the Smad3 linker region facilitates both activation by TGF-β and nuclear accumulation (Engel et al., 1999). A recent report suggested that Ras transformation suppressed TGF-β-mediated Smad3 phosphorylation responsible for antimitogenic TGF-β responses and stimulated JNK signal- driven phosphorylation of the Smad3 linker region, leading to upregulation of PAI-1, MMP-1 MMP-2, and MMP-9 expression and increased tumour invasion (Sekimoto et al., 2007).

Modulation of TGF-β responses by MAPK pathways is also seen in the nucleus. EGF stimulation of the Ras-MEK-ERK pathway leads to phosphorylation and stabilisation of the Smad co-repressor TGF-β-induced factor (TGIF) which competes with the co-activator p300 for Smad2 association (Lo et al., 2001). Also, activator protein-1 (AP-1) family transcription factors, downstream components of MAPK signalling, interact with R-Smad/Smad4 complexes in the nucleus (Zhang et al., 1998). For example, Smad3/Smad4 has been shown to cooperate with c-Jun/c-Fos to mediate TGF-β-induced transcription (Zhang et al., 1998). Conversely, c-Jun and JunB have been shown to interact with Smad3 by forming off-DNA complexes, inhibiting Smad3-dependent transcription (Verrecchia et al., 2000; Verrecchia et al., 2003).

3.1.1.2 Ski

Ski is a negative regulator of TGF-β signalling. It has been shown to recruit various transcriptional co-repressors to TGF-β targeted promoters, as well as disrupt the binding of the Smad transcriptional complex to the coactivator p300/CBP, leading to repression of Smad-dependent TGF-β gene regulation (reviewed in Luo, 2004). Furthermore, as the Ski- binding surface on Smad4 significantly overlaps with the surface required for binding the R-

Chapter 3 Introduction

Smads, Ski interferes with the interaction between Smad4 and phosphorylated R-Smads (Wu et al., 2002).

In melanoma cells Ski counteracts TGF-β-induced growth inhibition by preventing the induction of p21Cip1 and by binding to pRb, repressing its activity (Reed et al., 2001). Moreover, when overexpressed, Ski has been shown to induce oncogenic transformation dependent on its ability to repress Smad regulation (He et al., 2003). Furthermore, Ski protein levels are reported to correlate with human melanoma progression. Its subcellular localisation changes from exclusively nuclear in preinvasive melanomas (melanomas in situ), to nuclear and cytoplasmic in primary invasive and metastatic melanomas (Reed et al., 2001). In this context, Ski was shown to bind Smad3 and inhibit its nuclear translocation (Reed et al., 2001).

3.1.2 Vasculogenic mimicry

Maniotis and colleagues reported the presence of patterned networks of interconnected loops of extracellular matrix in metastatic melanomas (Maniotis et al., 1999). Red blood cells were detected in the hollow channels formed by these networks, but endothelial cells were not identified within these matrix-embedded channels. They further reported the formation of patterned solid and hollow matrix channels in three-dimensional cultures of highly invasive primary and metastatic melanoma cells in Matrigel or dilute Type 1 collagen (Maniotis et al., 1999). It was hypothesised that aggressive melanoma cells may generate such channels to facilitate tumour perfusion independent of angiogenesis. This phenomenon was called vasculogenic mimicry (Maniotis et al., 1999). Interestingly, these in vitro channels are only formed by highly invasive cells and not by melanocytes or poorly invasive melanomas. Furthermore, the presence of microcirculatory loops and networks in uveal melanoma tumours correlated with a decreased survival rate for the patients (Maniotis et al., 1999).

Chapter 3 Results

3.2 Results

3.2.1 Confirming TGF-β1 and activin A secretion

The identification of TGF-β signalling as a driving force for Motif 2, described in chapter 2 and published in 2006 (Hoek et al., 2006), prompted us to investigate the mechanisms regulating this signalling pathway in cultured melanoma cells. Firstly, to complement the published data, secretion of activin A and TGF-β1 was measured in the conditioned media (CM) of an additional six cell lines derived from melanoma metastases and obtained from the Mannheim melanoma cell culture bank (Hoek et al., 2006) (Fig 3.1). Confirming the previously published data, little difference in TGF-β1 secretion across the samples was observed, while activin A was only secreted by cells belonging to what was described as cohort C or as being of “high metastatic potential” and which is now referred to as the invasive cohort (Hoek et al., 2008). Cohort A, which was referred to as weakly metastatic melanoma in chapter 2, is here described as the proliferative cohort (Hoek et al., 2008).

Figure 3.1. TGF-β1 and activin A secretion. Western blotting with antibodies against TGF-β1 and the inhibin βA subunit of 10x concentrated conditioned media (CM) of melanoma cultures representing the proliferative and invasive cohorts separated on a reducing gel. Six cell cultures from the Mannheim melanoma cell culture bank (MaMel) and four cell cultures from the Zürich melanoma cell culture bank are shown.

3.2.2 Follistatin secretion does not correlate with activin secretion

Follistatin has been reported to be secreted by melanoma cell lines and it was suggested that its secretion represented an effective way to neutralise activin’s effects (Stove et al., 2004). In order to verify if our cell cultures secreted follistatin, we performed an Enzyme-Linked ImmunoSorbent Assay (ELISA) to measure the levels of follistatin secretion in the CM. Only

Chapter 3 Results four of our 12 cell cultures secreted detectable levels of follistatin (lower detection limit of 0.97 ng/ml) and the secretion of follistatin did not correlate with activin secretion (Fig.3.2).

Figure 3.2. Follistatin and activin A secretion in CM. ELISA measuring follistatin and activin A secretion in CM of 12 melanoma cultures representing the proliferative, intermediate and invasive cohorts.

3.2.3 Smad2 and Smad3 are activated across all cohorts

After confirming activin A secretion by cells belonging to the invasive cohort, we further investigated the activation status of Smad2 and Smad3. Phospho-specific antibodies against Smad2 and Smad3 were used to detect carboxy-end phosphorylation in the two proteins. Surprisingly, phosphorylated Smad2 and Smad3 proteins were detected in all samples (Fig. 3.3), suggesting that the identified TGF-β signature was not dependent on Smad2 and Smad3 activation.

Chapter 3 Results

Figure 3.3. Smad2 and Smad3 activation status. Western blot analysis of nuclear protein extracts of melanoma cell cultures representing the proliferative, intermediate and invasive cohorts. Phospho- specific antibodies against Smad2 and Smad3 were used to evaluate their carboxy-terminus phosphorylation status. The detection of β-actin was used as a loading control.

3.2.4 Ski is not responsible for the differential TGF-β signalling

As Smad2 and Smad3 were found to be phosphorylated in all melanoma cultures regardless of the cohort they belonged to, the presence of Ski, a negative regulator of TGF-β, was investigated. Ski interacts with Smad proteins and could explain the apparent lack of TGF-β signalling in cells of the proliferative cohort. As Ski’s activity is concentrated in the nucleus, western blotting was performed with the nuclear extracts of all 12 melanoma cell cultures. All 12 cell cultures expressed Ski and we therefore excluded Ski as being involved in the differential TGF-β signature detected.

Chapter 3 Results

Figure 3.4. Ski expression. Western blot analysis with an anti-Ski antibody of the nuclear proteins of 12 melanoma cell cultures representing the proliferative, intermediate and invasive cohorts. The detection of β-tubulin was used as a loading control.

3.2.5 The activation of the MAPK pathways does not correlate with the TGF-β signature

To investigate the possible involvement of MAPK pathways in the TGF-β signature identified in our invasive cohort melanoma cell cultures, the phosphorylation status of three major MAPK factors was determined across our samples. The activation of the ERK pathway, which is induced by mitogens and growth factors via RAS and RAF, was determined using a phospho-specific antibody directed against ERK1 and ERK2 (Fig 3.5). Although the ERK phosphorylation status varied across our 12 melanoma cell cultures, no correlation with the cohort-specific TGF-β signalling activation could be seen. Activation of the two stress- induced MAPK pathways, p38 and JNK, was also investigated by western blotting with phospho-specific antibodies. Again, phosphorylation of p38 and JNK did not correlate with cohort-specific TGF-β signalling activation (Fig. 3.5).

Chapter 3 Results

Figure 3.5. MAPK activation. Western blot analysis of 12 melanoma cell cultures representing the proliferative, intermediate and invasive cohorts. Phospho-specific antibodies against ERK1/2. JNK and p38 were used to evaluate the activation of the MAPK pathways. The detection of β-tubulin was used as a loading control.

3.2.6 Identifying vasculogenic mimicry as a discriminating phenotype

TGF-β susceptibility and in vitro motility are two phenotypes discriminating between the proliferative and invasive cohorts discussed in the paper in chapter 2. To identify a third discriminating phenotype, which could be used in subsequent assays, the capacity of cells to form networks when seeded on Matrigel, a phenomenon referred to as vasculogenic mimicry, was investigated. When seeded on Matrigel, all cells belonging to the invasive cohort formed networks within 24 hours of seeding, while cells from the proliferative cohort formed dispersed clusters (Fig. 3.6.).

Chapter 3 Results

Figure 3.6. Vasculogenic mimicry. When seeded on Matrigel, invasive cells form networks, characteristic of vasculogenic mimicry (B), while proliferative cells form clusters (A). Cells were photographed 24 hours after being seeded in a Matrigel-coated well.

3.2.7 Phenotype switching

In contrast to other models implying “one-way” changes in gene expression leading to growth and metastasis of melanoma, we suggest a dynamic model in which gene expression is altered from a proliferative to an invasive signature and vice versa (Fig. 3.7.). This gene expression oscillation is thought to be caused by a signal induced by changes in the microenvironment and translates into phenotypic changes enabling the cells to proliferate or invade.

Chapter 3 Results / Discussion

We therefore explored the possibility of “switching” cells from one cohort to an other by inducing TGF-β signalling in the proliferative cohort or inhibiting TGF-β signalling in the invasive cohort. We considered activin A as a candidate effector as it is the only member of the TGF-β family of cytokines whose expression profile correlates with cohort distribution. Proliferative cells were treated with TGF-β1 and activin A and invasive cells with follistatin and SB431542 (an ALK4/5/7 inhibitor). Cells were grown on Matrigel for up to 72 hours but no treatment-induced change in phenotype was observed.

Figure 3.7. An integrated model for gene regulation of melanoma metastatic potential and progression. Early phase melanoma cells expressing the proliferative signature gene set proliferate to form the primary lesion. Subsequently, an unknown signal switch, likely brought about by altered microenvironmental conditions (e.g. hypoxia or inflammation), gives rise to cells with a significantly different invasive signature gene set. Invasive signature cells escape and, upon reaching a suitable distal site, revert to the proliferative state and nucleate a new metastasis where the cycle is repeated. Each switch in phenotype (state change) is accompanied by an exchange in expressed gene sets from proliferative to invasive and vice versa.

3.3 Discussion

TGF-β signalling has been extensively studied in cancer and its role in melanoma has been highlighted in a number of studies. On the other hand, the role of activin in melanoma has received very limited attention. In our recent publication in which we presented two different transcription signatures for melanoma cell cultures which, based on known functions of the genes involved, defined their respective contributions to metastatic potential as either

Chapter 3 Discussion proliferative or invasive, we proposed a role for activin A in maintaining the TGF-β signature (Hoek et al., 2006). In this chapter, we confirm some of the results presented in chapter 2 and our results suggest that a number of pathways that are regulated by TGF-β signalling are not involved in TGF-β signature modulation in melanoma.

In our recent paper, we demonstrated the cohort specificity of activin A secretion by detecting the presence of the inhibin βA subunit in the conditioned medium of our cell cultures (Hoek et al., 2006). Our results contrasted with those presented by Stove and coworkers who could not detect activin A concentrations exceeding 0.1 ng/ml in the CM of any of the 12 melanoma cell lines they studied (Stove et al., 2004). Therefore, we confirm those results using another set of cell cultures obtained from the Mannheim melanoma cell culture bank. Although our assay was ten times more sensitive than the one they used, we detected concentrations ranging from 0.326 ng/ml to 1.201 ng/ml, concentrations which should have been detected with the system they described (Hoek et al., 2006). As their RNA expression data supports the expression of the βA subunit and hence of activin A in their melanoma cell lines, the authors suggest the possible rapid degradation of the cytokine or its association with the cell surface to explain the unsuccessful detection of activin A in the CM. Our results do not support such claims but support their RNA expression data which suggests that a number of melanoma cell lines produce activin A.

Stove and coworkers further looked at the secretion of follistatin by their melanoma cell lines and suggested that its secretion may inhibit activin’s negative regulatory effects. We could detect follistatin in only four of our 12 studied melanoma cell cultures (lower detection limit of 0.97 ng/ml) and its secretion did not correlate with the secretion of activin A. We therefore exclude follistatin as an important player in our model.

Nodal, a potent embryonic morphogen from the TGF-β family, has been presented as a key regulator of melanoma plasticity and tumorigenicity (Topczewska et al., 2006). Nodal, like TGF-β and activin, signals through Smad2, Smad3 and Smad4, and, like activin, binds to ActR-IIB which dimerises with ALK4 and ALK7. Arguing that aggressive tumour cells share many properties with embryonic cells, Topczewska and co-workers used a zebrafish embryo model as a biosensor for metastatic melanoma expressing a plastic, stem cell-like phenotype to modulate an embryonic microenvironment. (Topczewska et al., 2006). They showed that a

Chapter 3 Discussion subpopulation of aggressive melanoma cells transplanted into the blastula-stage embryo secreted Nodal and consequently induced ectopic formation of the embryonic axis. They further demonstrated that Nodal was absent in normal human skin, absent or weakly present in primary lesions, but present in 60% of metastases examined, suggesting a correlation between Nodal expression and melanoma progression (Topczewska et al., 2006). Furthermore, inhibition of Nodal signalling promoted the reversion of melanoma cells towards a melanocytic phenotype and reduced invasiveness, colony formation in vitro and tumorigenicity in vivo (Topczewska et al., 2006). In our initial microarray analysis including the data from our collaborators in Mannheim, the expression of nodal does not differ between cohorts. It would be interesting to compare the gene targets of nodal and activin A and to hypothesise that activin A and nodal share similar effects on melanoma tumorigenicity in their respective models.

Because TGF-β and activin signal through Smad2 and Smad3, we investigated the possibility of their differential activation between cohorts. Also, as phosphorylated Smads need to translocate to the nucleus to regulate transcription and that nuclear translocation can be inhibited by a number of factors such as Ski (Reed et al., 2001) or MAPKs (Xu, 2006), we looked for phosphorylated Smad2 and Smad3 in the nucleus. Surprisingly, Smad2 and Smad3 were phosphorylated and nuclear in all investigated cell cultures, suggesting a TGF-β signature independent from the Smad-dependent pathway. It is possible that Smad activation is induced by growth factors present in the complete growth medium in which the cells were grown. For example, Rodeck and coworkers reported variable levels of Smad activation in melanoma grown in reduced-serum medium (Rodeck et al., 1999). However, the same growth conditions were used for the generation of the expression profiling data in which the different cohorts were identified, indicating that the observed TGF-β signature is not due to exogenous growth factors.

To explain the TGF-β signature highlighted in our microarray analysis, we also examined pathways known to interact with and modulate TGF-β signalling. Cross-signalling between the MAPK pathways and TGF-β signalling has been well described in diverse cell types and its implications in carcinogenesis are now clear (Javelaud and Mauviel, 2005). Activation of the MAPK pathways by TGF-β is not well characterised, but examples of TGF-β-dependent activation of the ERK, p38 and JNK pathways have been reported (Brown et al., 1999;

Chapter 3 Discussion

Yamaguchi et al., 1995; Yu et al., 2002). Furthermore, not only is the nuclear translocation of Smads regulated by non-TGF-β signalling including the MAPK pathways, but these pathways also regulate the ability of Smads to regulate gene transcription (Xu, 2006). We therefore looked at the activation status of several MAPK pathways, but could not correlate them with our TGF-β signature. However, we cannot rule out the presence of complex interactions between different MAPKs and the TGF-β pathway, which could not have been detected with the methods presented in this chapter. Furthermore, it is conceivable that multiple MAPKs could yield similar results, therefore a simple correlation between TGF-β signalling and one particular MAPK member might not be possible.

Ski’s capacity to disturb Smad signalling makes it another candidate for the source of the differential TGF-β signature. As for the MAPK pathways, Ski can both sequester R-Smads in the cytoplasm and also interfere with Smad transcription regulation in the nucleus. If Ski interference was responsible for differential TGF-β signalling, Ski would be expected to be differentially expressed and localised across cohorts. However, as Ski was present in the nucleus of melanomas of all cohorts, it had to be excluded as a candidate for the source of the differential TGF-β signature.

As we hypothesise that the transcription signatures of our two main cohorts, proliferative and invasive, represent distinct yet interchangeable states and that we are interested in identifying the source of the “switch” between them, in vitro phenotyping of cells was of primary importance for all subsequent experiments. Vasculogenic mimicry has been shown both in vivo and in vitro in melanoma (Maniotis et al., 1999). As Maniotis and coworkers have described vasculogenic mimicry as being a characteristic of invasive cells, we compared the ability of melanoma cells from invasive and proliferative cohorts to form networks when seeded on Matrigel. Interestingly, while all invasive cells formed distinct network patterns, proliferative cells generally formed clusters. This showed there was a strong link between gene expression patterns and vasculogenic mimicry, and confirmed that the cohorts were descriptive of proliferative or invasive phenotypes.

Maniotis and coworkers tried, without success, to induce or inhibit the formation of vasculogenic mimicry with CM of phenotypically opposed cells (Maniotis et al., 1999). They treated poorly invasive cells with basic fibroblast growth factor (bFGF), vascular endothelial

Chapter 3 Discussion / Material and Methods growth factor (VEGF), platelet-derived growth factor (PDGF), tumour necrosis factor alpha (TNF-α) and TGF-β, individually or in combination, but failed to induce the formation of networks.

With the TGF-β signature now characterised, we were interested in identifying how cells “switch” between cohort phenotypes. Using vasculogenic mimicry to identify the “switch”, we tested the possibility of inducing cell phenotype changes by activating TGF-β signalling in the proliferative cohort or by inhibiting it in the invasive cohort. Although Topczewska and coworkers reported the prevention of the formation of embryonic-like vasculogenic networks using SB431542 (Topczewska et al., 2006), we could not prevent or induce the phenotype by inhibiting or inducing TGF-β signalling, respectively. The different experimental conditions, for example the type I collagen versus the Matrigel matrix, and the different appearance of the phenotype, the three- versus the two-dimensional phenotype, could explain their success.

The data presented in this chapter illustrates the difficulty of identifying the source of transcriptional regulation identified by microarray analysis. This is further complicated as the transcriptional signature is associated with a multi-functional molecule such as TGF-β. Cross- interactions with this pathway are multiple and complex, and there may not be a unique modulator responsible for regulating TGF-β signature across melanoma cell phenotypes. However, the linking of a third discriminating phenotype, vasculogenic mimicry, supports the model as represented by two distinct melanoma cell phenotypes with distinct transcriptional signatures.

3.4 Material and Methods

3.4.1 Cell culture

Melanoma cell cultures were established from surplus material from cutaneous melanoma metastases removed by surgery after having obtained written informed consent of the patient. Clinical diagnosis was confirmed by histology and immunohistochemistry. Melanoma cells were released from tissue sections and grown as previously described (Geertsen et al., 1998). Cells were grown in RPMI 1640 (GIBCO, Invitrogen, Carlsbad, CA, USA) supplemented

Chapter 3 Material and Methods with 10% foetal calf serum (FCS), 4nM glutamine (Biochrom, Berlin, Germany) and 1nM sodium pyruvate (GIBCO), at 37°C in an atmosphere of 10% CO2.

3.4.2 Preparation of condition media

Growth-conditioned media was prepared by growing cells to confluency in a 75-cm2 flask in normal growth conditions, these were then washed with PBS and placed into fresh serum-free media for 24 hours. Growth-conditioned media was collected, filtered through a 0.22-µm filter and stored at -80°C. Growth-conditioned media was concentrated 10-fold by acetone precipitation.

3.4.3 ELISA

Activin A and follistatin were measured in conditioned media samples using a specific enzyme-linked immunosorbent assay (Knight et al., 1996) according to the manufacturer’s instructions (Oxford Bio-Innovations, Oxfordshire, UK), with some modifications for media samples as described previously (Buzzard et al., 2003). The average intra- and interplate coefficients of variation for this assay are routinely <8% and the lower limit of detection is 10 pg/ml for activin A and 1 ng/ml for follistatin.

3.4.4 Preparation of total cell protein extracts

Cells were washed twice with cold PBS and lysed at 4°C in lysis buffer containing 20 mM Tris-HCl (pH 7.5), 1% Triton X-100 (Sigma-Aldrich, Buchs, Switzerland), 137 mM NaCl, 10% glycerol, protease inhibitors (Complete Mini +EDTA, Roche, Basel, Switzerland) and phosphatase inhibitors (Sigma Phosphatase inhibitor cocktail 1 + 2). After rotating samples for 15 minutes and centrifugating for 20 minutes at 4°C, the supernatant was collected.

3.4.5 Preparation of cytosolic and nuclear protein extracts

Cells were washed twice with cold PBS and lysed at 4°C. Cells were lysed in lysis buffer A containing 10 mM HEPES (pH 7.9), 10 mM KCl, 1.5mM MgCl2, 0.625% NP-40, protease inhibitors (Complete Mini +EDTA, Roche) and phosphatase inhibitors (Sigma Phosphatase

Chapter 3 Material and Methods inhibitor cocktail 1 + 2). After a ten-minute incubation on ice, cells were centrifuged at 5,000 rpm for 5 minutes and the cytosolic fraction (supernatant) was transferred to a new tube. This fraction was centrifuged an additional three times to obtain maximal purity. For nuclear fraction extraction, the pellet was washed three times in buffer A before it was resuspended in lysis buffer B containing 20 mM HEPES (pH 7.9), 1.5mM MgCl2, 420mM NaCl, 25% glycerol, protease inhibitors (Complete Mini +EDTA, Roche) and phosphatase inhibitors (Sigma Phosphatase inhibitor cocktail 1 + 2). After rotating samples for 60 minutes and centrifugating for 20 minutes at 4°C, the nuclear fraction (supernatant) was collected. Purity of nuclear fractions were verified by western blot analysis using an antibody against the cytosolic protein copper and zinc-containing Superoxide Dismutase (Cu/Zn SOD) (Fig 3.8.).

Figure 3.8. The nuclear protein fraction was not contaminated. Western blot analysis of nuclear and cytosolic protein fractions probed with an antibody against cytosolic-specific Cu/Zn SOD.

3.4.6 Western blot analysis

Proteins were separated by SDS–PAGE under reducing conditions and transferred onto nitrocellulose membranes (Invitrogen, Basel, Switzerland). Membranes were probed with a specific primary antibody (Table 3.1.) followed by an appropriate horseradish peroxidase- conjugated goat anti-rabbit (Bio-Rad, Reinach, Switzerland), rabbit anti-goat (abcam, Cambridge, UK) or rabbit anti-mouse (SantaCruz, La Jolla, CA, USA) secondary antibodies. Bound antibodies were detected by chemiluminescence (ECL, GE Healthcare, Buckinghamshire, UK).

Chapter 3 Material and Methods

Table 3.1. Antibodies and Western blot conditions Immunogen Host Supplier # Conditions TGF-β rabbit abcam (Cambridge, UK) Ab9758 1:800, 2h, 5%milk Inhibin βA rabbit Gift from S. Werner (ETH, --- 1:2000, 2h, 5% milk Zürich) p-Smad2 rabbit Chemicon (Billerica, MA, AB3849 1:1000, O/Na, 5% milk USA) p-Smad3 rabbit Cell Signaling (Danvers, 9514 1:1000, O/Na, 5% BSA MA, USA) Ski rabbit Upstate (Lake Placid, NZ, 07-060 1:400, O/Na, 5% milk USA) p-ERK1/2 rabbit abcam (Cambridge, UK) ab4819 1:2000, 2h, 1% milk p-p38 rabbit abcam (Cambridge, UK) ab4822 1:1000, 2h, 1% milk p-SAPK/JNK rabbit Cell Signaling (Danvers, 9251 1:1000, O/Na, 5% BSA MA, USA) Actin goat Santa Cruz (La Jolla, CA, sc-1616 1:1000, 2h, 5% milk USA) β-Tubulin mouse Sigma (St-Louis, MO, USA) T4026 1:500, 2h, 5% milk p84 mouse abcam (Cambridge, UK) ab487 1:1000, 2h, 5% milk Cu/Zn SOD rabbit Stressgen (Victoria, Canada) SOD-100 1:400, 2h, 5% milk a)O/N = overnight

Chapter 3 References

3.5 References

Brown, J.D., DiChiara, M.R., Anderson, K.R., Gimbrone, M.A., Jr. and Topper, J.N. (1999) MEKK-1, a Component of the Stress (Stress-activated Protein Kinase/c-Jun N-terminal Kinase) Pathway, Can Selectively Activate Smad2-mediated Transcriptional Activation in Endothelial Cells. J. Biol. Chem., 274, 8797-8805. Buzzard, J.J., Farnworth, P.G., De Kretser, D.M., O'Connor, A.E., Wreford, N.G. and Morrison, J.R. (2003) Proliferative phase sertoli cells display a developmentally regulated response to activin in vitro. Endocrinology, 144, 474-483. de Caestecker, M.P., Parks, W.T., Frank, C.J., Castagnino, P., Bottaro, D.P., Roberts, A.B. and Lechleider, R.J. (1998) Smad2 transduces common signals from receptor serine- threonine and tyrosine kinases. Genes Dev., 12, 1587-1592. Derynck, R. and Zhang, Y.E. (2003) Smad-dependent and Smad-independent pathways in TGF-beta family signalling. Nature, 425, 577-584. Engel, M.E., McDonnell, M.A., Law, B.K. and Moses, H.L. (1999) Interdependent SMAD and JNK signaling in transforming growth factor-beta-mediated transcription. J Biol Chem, 274, 37413-37420. Geertsen, R.C., Hofbauer, G.F., Yue, F.Y., Manolio, S., Burg, G. and Dummer, R. (1998) Higher frequency of selective losses of HLA-A and -B allospecificities in metastasis than in primary melanoma lesions. J Invest Dermatol, 111, 497-502. He, J., Tegen, S.B., Krawitz, A.R., Martin, G.S. and Luo, K. (2003) The Transforming Activity of Ski and SnoN Is Dependent on Their Ability to Repress the Activity of Smad Proteins. J. Biol. Chem., 278, 30540-30547. Hoek, K.S., Eichhoff, O.M., Schlegel, N.C., Döbbeling, U., Kobert, N., Schaerer, L., Hemmi, S., Dummer, R. and (2008) In vivo switching of human melanoma cells between proliferative and invasive states. Cancer Res, in press. Hoek, K.S., Schlegel, N.C., Brafford, P., Sucker, A., Ugurel, S., Kumar, R., Weber, B.L., Nathanson, K.L., Phillips, D.J., Herlyn, M., Schadendorf, D. and Dummer, R. (2006) Metastatic potential of melanomas defined by specific gene expression profiles with no BRAF signature. Pigment Cell Res, 19, 290-302. Huang, H.-M., Chang, T.-W. and Liu, J.-C. (2004) Basic fibroblast growth factor antagonizes activin A-mediated growth inhibition and hemoglobin synthesis in K562 cells by activating ERK1/2 and deactivating p38 MAP kinase. Biochemical and Biophysical Research Communications, 320, 1247-1252. Itoh, S., Thorikay, M., Kowanetz, M., Moustakas, A., Itoh, F., Heldin, C.-H. and ten Dijke, P. (2003) Elucidation of Smad Requirement in Transforming Growth Factor-beta Type I Receptor-induced Responses. J. Biol. Chem., 278, 3751-3761. Javelaud, D. and Mauviel, A. (2005) Crosstalk mechanisms between the mitogen-activated protein kinase pathways and Smad signaling downstream of TGF-[beta]: implications for carcinogenesis. Oncogene, 24, 5742-5750. Knight, P.G., Muttukrishna, S. and Groome, N.P. (1996) Development and application of a two-site enzyme immunoassay for the determination of 'total' activin-A concentrations in serum and follicular fluid. J Endocrinol, 148, 267-279. Kretzschmar, M., Doody, J. and Massagu, J. (1997) Opposing BMP and EGF signalling pathways converge on the TGF-[beta] family mediator Smad1. Nature, 389, 618-622. Kretzschmar, M., Doody, J., Timokhina, I. and Massague, J. (1999) A mechanism of repression of TGFbeta/ Smad signaling by oncogenic Ras. Genes Dev, 13, 804-816. Lo, R.S., Wotton, D. and Massague, J. (2001) Epidermal growth factor signaling via Ras controls the Smad transcriptional co-repressor TGIF. Embo J, 20, 128-136.

Chapter 3 References

Luo, K. (2004) Ski and SnoN: negative regulators of TGF-[beta] signaling. Current Opinion in Genetics & Development, 14, 65-70. Maniotis, A.J., Folberg, R., Hess, A., Seftor, E.A., Gardner, L.M., Pe'er, J., Trent, J.M., Meltzer, P.S. and Hendrix, M.J. (1999) Vascular channel formation by human melanoma cells in vivo and in vitro: vasculogenic mimicry. Am J Pathol, 155, 739-752. Massague, J. (2003) Integration of Smad and MAPK pathways: a link and a linker revisited. Genes Dev, 17, 2993-2997. Reed, J.A., Bales, E., Xu, W., Okan, N.A., Bandyopadhyay, D. and Medrano, E.E. (2001) Cytoplasmic localization of the oncogenic protein Ski in human cutaneous melanomas in vivo: functional implications for transforming growth factor beta signaling. Cancer Res, 61, 8074-8078. Rodeck, U., Nishiyama, T. and Mauviel, A. (1999) Independent regulation of growth and SMAD-mediated transcription by transforming growth factor beta in human melanoma cells. Cancer Res, 59, 547-550. Sekimoto, G., Matsuzaki, K., Yoshida, K., Mori, S., Murata, M., Seki, T., Matsui, H., Fujisawa, J.-i. and Okazaki, K. (2007) Reversible Smad-Dependent Signaling between Tumor Suppression and Oncogenesis. Cancer Res, 67, 5090-5096. Stove, C., Vanrobaeys, F., Devreese, B., Van Beeumen, J., Mareel, M. and Bracke, M. (2004) Melanoma cells secrete follistatin, an antagonist of activin-mediated growth inhibition. Oncogene. Topczewska, J.M., Postovit, L.M., Margaryan, N.V., Sam, A., Hess, A.R., Wheaton, W.W., Nickoloff, B.J., Topczewski, J. and Hendrix, M.J. (2006) Embryonic and tumorigenic pathways converge via Nodal signaling: role in melanoma aggressiveness. Nat Med, 12, 925-932. Verrecchia, F., Pessah, M., Atfi, A. and Mauviel, A. (2000) Tumor Necrosis Factor-alpha Inhibits Transforming Growth Factor-beta /Smad Signaling in Human Dermal Fibroblasts via AP-1 Activation. J. Biol. Chem., 275, 30226-30231. Verrecchia, F., Tacheau, C., Wagner, E.F. and Mauviel, A. (2003) A Central Role for the JNK Pathway in Mediating the Antagonistic Activity of Pro-inflammatory Cytokines against Transforming Growth Factor-beta -driven SMAD3/4-specific Gene Expression. J. Biol. Chem., 278, 1585-1593. Wu, J.W., Krawitz, A.R., Chai, J., Li, W., Zhang, F., Luo, K. and Shi, Y. (2002) Structural mechanism of Smad4 recognition by the nuclear oncoprotein Ski: insights on Ski-mediated repression of TGF-beta signaling. Cell, 111, 357-367. Xu, L. (2006) Regulation of Smad activities. Biochimica et Biophysica Acta (BBA) - Gene Structure and Expression, 1759, 503-513. Xu, W., Angelis, K., Danielpour, D., Haddad, M.M., Bischof, O., Campisi, J., Stavnezer, E. and Medrano, E.E. (2000) Ski acts as a co-repressor with Smad2 and Smad3 to regulate the response to type beta transforming growth factor. Proc Natl Acad Sci U S A, 97, 5924- 5929. Yamaguchi, K., Shirakabe, K., Shibuya, H., Irie, K., Oishi, I., Ueno, N., Taniguchi, T., Nishida, E. and Matsumoto, K. (1995) Identification of a member of the MAPKKK family as a potential mediator of TGF-beta signal transduction. Science, 270, 2008-2011. Yu, L., Hebert, M.C. and Zhang, Y.E. (2002) TGF-beta receptor-activated p38 MAP kinase mediates Smad-independent TGF-beta responses. Embo J, 21, 3749-3759. Yue, J. and Mulder, K.M. (2000) Requirement of Ras/MAPK Pathway Activation by Transforming Growth Factor beta for Transforming Growth Factor beta 1 Production in a Smad-dependent Pathway. J. Biol. Chem., 275, 30765-30773. Zhang, Y., Feng, X.-H. and Derynck, R. (1998) Smad3 and Smad4 cooperate with c-Jun/c- Fos to mediate TGF-[beta]-induced transcription. Nature, 394, 909-913

100

Chapter 4

4 In vivo switching of human melanoma cells between proliferative and invasive states

101

For this second paper, I participated in the generation of in vitro growth curves and in the optimisation of motility assays shown in figure 4.1. I also took part in the optimisation of western blotting and TGF-β-mediated growth inhibition assays, as well as the setting up of the Mitf RT-PCR assay presented in figure 4.3. Finally, I constructed the pAd-H1∆lacZ-lnk1 by replacing the CMV promoter of pAd-CMV∆lacZ-lnk1 with the H1 promoter.

102

In vivo switching of human melanoma cells between proliferative and invasive states

Keith S. Hoek1*, Ossia Eichhoff1, Natalie C. Schlegel1, Udo Döbbeling1, Nikita Kobert1, Leo Schaerer2, Silvio Hemmi3, and Reinhard Dummer1.

1Department of Dermatology, University Hospital of Zurich, 8091 Zurich, Switzerland. 2Dermatohistopathologischen Gemeinschaftspraxis, 88048 Friedrichshafen, Germany. 3Institute of Molecular Biology, University of Zurich, 8057 Zurich, Switzerland.

Cancer Research (2008) in press

103

Chapter 4 Abstract/Introduction

4.1 Abstract

Metastatic melanoma represents a complex and heterogeneous disease for which there are no therapies to improve patient survival. Recent expression profiling of melanoma cell lines identified two transcription signatures respectively corresponding with proliferative and invasive cellular phenotypes. A model derived from these findings predicts that in vivo melanoma cells may switch between these states. Here, DNA microarray-characterised cell lines were subjected to in vitro characterisation before subcutaneous injection into immunocompromised mice. Tumour growth rates were measured and post-excision samples were assessed by immunohistochemistry to identify invasive and proliferative signature cells. In vitro tests showed that proliferative signature melanoma cells are faster growing but less motile than invasive signature cells. In vivo proliferative signature cells initiated tumour growth in 14 ± 3 days post injection. By comparison, invasive signature cells required a significantly longer (p < 0.001) period of 59 ± 11 days. Immunohistochemistry showed that regardless of the seed cell signature, tumours showed evidence for both proliferative and invasive cell types. Furthermore, proliferative signature cell types were detected most frequently in the peripheral margin of growing tumours. These data indicate that melanoma cells undergo transcriptional signature switching in vivo likely regulated by local microenvironmental conditions. Our findings challenge previous models of melanoma progression which evoke one-way changes in gene expression. We present a new model for melanoma progression which accounts for transcription signature plasticity and provides a more rational context for explaining observed melanoma biology.

4.2 Introduction

Metastatic stage melanoma is an aggressive disease that few patients survive for more than two years. Compounding this, scores of clinical trials testing different adjuvant therapies have brought no significant improvement in the survival outlook for these patients (Sasse et al., 2007). One possible explanation for this is that melanoma is a heterogeneous collection of different cells, and the differences between them are sufficient that some are missed by targeted therapies. The variety of phenotypic and behavioural features melanomas present range from distinct organ specificities during metastasis to changes in motility and invasiveness (Fidler and Kripke, 1977). Furthermore, melanoma tissues have various

104

Chapter 4 Introduction morphologies, from assorted macroscopic lesional structures to multiple microscopic cellular forms, which often complicate assessments of diagnosis and prognosis (Levene, 1980). Additionally, immunohistochemical staining regularly yields heterogeneous results. While most melanoma lesions will stain for a number of melanocytic markers, this is not necessarily true for all melanoma cells within a given lesion (Banerjee and Harris, 2000). Finally, DNA microarray examination of different lesions and melanoma cell line collections reveal among them consistent taxonomies of genomic aberrations and transcriptional signatures (Curtin et al., 2005; Haqq et al., 2005; Hoek et al., 2006). The source of heterogeneity is thought to rest in the combination of how melanoma cells respond to different microenvironments and the reciprocal influence of their own molecular states. This was an idea first conceptualised in Stephen Paget’s “seed and soil” model after his observation that particular cancer cells demonstrated tumorigenic preference for certain tissues over others (Paget, 1889; Ribatti et al., 2006). By comparison, current molecular models for melanoma progression are homogeneous. A generally accepted hypothesis assumes that progression is driven by a steady evolution of molecular changes, and this hypothesis provides the dominant paradigm for molecular studies (Miller and Mihm, 2006).

Of recent interest has been the activity of the microphthalmia-associated transcription factor (Mitf) in regulating melanoma cell proliferation. In normal melanocytes Mitf is critical for melanocytic differentiation, expression of melanogenic enzymes and upregulating cyclin- dependent kinase inhibitors to drive cell cycle exit (Carreira et al., 2005; Loercher et al., 2005; Steingrimsson et al., 2004). However, in melanoma Mitf is required for proliferation and has been identified as a “lineage survival” factor prone to amplification (Carreira et al., 2006; Du et al., 2004; Garraway et al., 2005). While the contrast in the activities of Mitf in normal and transformed cells remains unexplained, there is little doubt concerning its central role in melanoma biology.

We recently explored heterogeneity of gene expression in melanoma cells. Bittner and coworkers first suggested that there may be specific transcriptional signatures delineating melanoma cell subgroups (Bittner et al., 2000). We characterised two different transcription signatures for melanoma cell lines which, based on known functions of the genes involved, defined their respective contributions to metastatic potential as either proliferative or invasive (Hoek et al., 2006). We further hypothesised that the transcription signatures represent

105

Chapter 4 Introduction/Results distinct yet interchangeable states regulated by signalling from the microenvironment. Critically, Mitf expression is a central feature of the proliferative signature which is absent from the invasive form. Others’ in vitro work concerning Mitf gene regulation have corroborated the hypothesis that its expression is important for differentiating between proliferative and invasive states (Carreira et al., 2006). To test the validity of the proliferative signature we examined Mitf’s role in the proliferative signature phenotype and compared the in vivo tumorigenicity of these cells against those with an invasive signature. At the same time we used immunohistochemistry to monitor Mitf and the Ki67 antigen in the resulting tumours to provide evidence of in vivo switching between signatures.

4.3 Results

4.3.1 Phenotypic assignment of cell lines

To study the in vivo tumorigenic behaviour of melanoma cell lines with different transcriptional signatures, we selected pairs of proliferative and invasive signature melanoma cell lines based on previous genome-wide transcription profiling experiments (Hoek et al., 2006). We performed supervised hierarchical clustering of these samples using normalised signal intensity data from 105 genes shown to be tightly linked to signature (Fig. 4.1A). Earlier experiments had shown that proliferative signature lines were significantly less motile than invasive signature lines. Also, TGF-β challenge showed that proliferative signature cells were significantly more susceptible to TGF-β-mediated growth inhibition than invasive signature cells (Hoek et al., 2006). We performed additional motility and proliferation experiments to expand this range of in vitro characterisations. Cell growth experiments showed a significant (p < 0.001) difference in proliferation rates between proliferative and invasive signature cell lines (Fig. 4.1B). Conversely, invasive signature cell lines plated at subconfluent densities on microporous transwell filters migrated in significantly (p < 0.001) higher numbers towards the lower chamber than identically plated proliferative cell lines (Fig. 4.1C). With these experiments we concluded that signature assignments given to cell lines according to their gene expression signature correlate with in vitro data in the context of our model.

106

Chapter 4 Results

Figure 4.1. In vitro correlations with gene expression signatures. M980513 and M000907 proliferative signature melanoma lines, as well as M991121 and M010308 invasive signature melanoma lines, were chosen for this study. (A) A gene expression heatmap, generated by clustering samples based on the normalized expression of 105 metastatic potential genes, highlights subtype- specific signatures. In vitro growth (B) and motility (C) experiments correlate appropriately with proliferative and invasive signature assignments. Error bars indicate standard deviation.

4.3.2 Mitf is a marker of proliferative phenotype

To follow cell signatures in vivo, we selected immunohistochemical markers according to their signature specificity. Previous analysis indicated that Mitf mRNA and protein levels are high in proliferative signature lines and at low or undetectable levels in invasive signature samples (Hoek et al., 2006). We confirmed this by performing immunohistochemistry on paraffin-embedded cultures of proliferative and invasive signature melanoma lines. Immunohistochemical staining of the different signature cell line pellets with anti-Mitf antibodies showed that in proliferative signature cell lines 93% of cells were positive for nuclear staining for Mitf while invasive signature cell lines showed no positivity (data not shown). Because invasive melanoma cells have downregulated genes responsible for the melanocytic phenotype observable in proliferative signature melanoma cells there are no immunohistochemical markers which unequivocally identify them. Instead, the differential in growth rates for the signatures indicated that a general proliferation marker may be useful for

107

Chapter 4 Results immunohistochemical identification of signature type in vivo. While examination of previously published gene expression data shows that between transcriptional signature types there is no significant differential in the expression of mRNA encoding the proliferation marker Ki67 antigen, the significant difference in in vitro proliferation rates suggest that Ki67 antigen is likely to show a difference at the protein level. Accordingly, staining for Ki67 antigen showed that 94% of proliferative signature cells and 45% of invasive signature cells had positively stained nuclei (Fig. 4.2). These results indicate that Mitf is a good marker for specific identification of proliferative signature cells and that Ki67 antigen is a suitable marker for identifying regions undergoing differential rates of proliferation.

Figure 4.2. Immunohistochemical marker correlations with gene expression signatures. Immunohistochemical analysis of paraffin-embedded cell lines shows that proliferative and invasive signature lines have differential staining for Mitf (93% and 0%, respectively) and Ki67 antigen (94% and 45%, respectively).

4.3.3 Mitf expression reflects signature phenotype

To confirm that Mitf expression is functionally linked to signature phenotype we used siRNA to knockdown Mitf protein levels and assessed the effects in vitro. One in vitro characteristic which distinguishes between proliferative and invasive signature melanoma cells is a differential in susceptibility to TGF-β-mediated inhibition of proliferation, with proliferative signature cells being more sensitive to TGF-β than invasive signature cells (Hoek et al., 2006). Because proliferative signature cells express Mitf and invasive signature cells do not, we hypothesised that Mitf expression mediated the growth inhibitory effect of TGF-β on proliferative signature cells. We performed anti-Mitf siRNA knockdown experiments in a

108

Chapter 4 Results proliferative signature melanoma line and confirmed knockdown by western blot analyses (Fig. 4.3A). We found that Mitf-depletion from proliferative signature melanoma cells made them less susceptible to TGF-β-mediated growth inhibition (Fig. 4.3B), showing that Mitf mediates the growth inhibitory effect of TGF-β. Further, we showed that TGF-β treatment reduces Mitf mRNA and protein expression (Fig. 4.3C), suggesting that TGF-β-mediated growth inhibition may be effected by reduction of Mitf expression. This demonstrates that Mitf function is closely linked to the relationship between transcription signature and in vitro phenotype, confirming it as a useful in vivo marker for identifying different signature cells.

Figure 4.3. siRNA knockdown of Mitf protects against TGF-β-mediated growth inhibition. (A) siRNA-mediated knockdown of Mitf in a proliferative signature melanoma cell line (M000921) was confirmed by Western blot analysis. (B) The ratio of TGF-β-mediated inhibition of growth in cells treated with siRNA targeting Mitf over cells treated with a control siRNA is compared between proliferative (M000921) and invasive (M991121) signature melanoma cell lines. This shows that Mitf knockdown promotes resistance to TGF-β-mediated growth inhibition in a proliferative signature melanoma cell line while identical treatment does not change susceptibility in an invasive signature line. TGF-β-treatment of proliferative signature lines (M980513, M000907) results in reduction of Mitf mRNA (C).

4.3.4 Proliferative cells form fast growing tumours sooner than invasive cells

To test the relationship of cell line signature assignments with in vivo behaviour we performed subcutaneous injection of cell lines into the flanks of immunocompromised mice and recorded tumour growth characteristics. We found that proliferative melanoma lines consistently initiated tumours, measured as the time at which tumour volume exceeded 100 mm3, about 14 ± 3 days after being injected into the flanks of athymic nude mice. This was

109

Chapter 4 Results significantly (p < 0.001) shorter than for invasive lines which took 59 ± 11 days (Fig. 4.4). These data provide in vivo evidence for the significance of a proliferative signature in melanoma cells as predicted by in vitro experiments. The proliferative signature-seeded tumours all initiated growth at nearly the same time point. Contrasting this, initiation times for the invasive signature-seeded tumours was spread over a wider period. This suggests that, unlike proliferative signature-seeded initiation, invasive signature-seeded initiation may be dependent on microenvironmental variation.

Figure 4.4. Xenograft tumour growth. Human melanoma cell lines (M980513, M000907, M991121, M010308) were injected into both flanks of immunocompromised nude mice. Proliferation of melanoma cells led to tumour growth which was monitored daily. Proliferative melanoma cells (M980513, M000907) formed tumours rapidly, while invasive melanoma cells (M991121, M010308) took weeks longer to initiate tumour growth.

4.3.5 Tumours derived from proliferative or invasive lines are indistinguishable

Because both transcription signature melanoma cell types yielded tumours we were interested in examining these for signature-specific differences. Upon excision the tumours were stained for Mitf and Ki67 antigen expression. Tumours derived from invasive signature cell lines, which did not stain for Mitf, revealed melanoma cells with nuclei which were Mitf-positive and melanoma cells with nuclei which were Mitf-negative (Fig. 4.5A-E). Tumours derived from proliferative signature cell lines, which stained for Mitf, showed the same patterning of stained and unstained melanoma cell nuclei (Fig. 4.5F-J). Additionally, we found that Mitf- stained nuclei tended to concentrate within the peripheral margins of the tumours. Ki67 antigen staining patterns were similarly indistinguishable in tumours derived from proliferative or invasive signature lines. Also, it was apparent that tumour regions showing Mitf-positive nuclei were also enriched for Ki67-positive nuclei. These findings showed that

110

Chapter 4 Results/Discussion after removal tumours seeded with invasive or proliferative signature cell lines were not distinguishable and that homogeneous in vitro staining patterns yielded strikingly heterogeneous patterns in vivo, showing that signature patterns of melanoma cells change bi- directionally.

Figure 4.5. Immunohistochemistry of melanoma xenograft tumours. Human melanoma cell lines (M980513, M000907, M991121, M010308) were injected into the flanks of immunocompromised nude mice and allowed to grow tumours for a maximum of 75 days. After a tumour had formed it was removed and subjected to immunohistochemical analysis. (A) A day 75 tumour resulting from an invasive signature melanoma (M010308). (B) Mitf and Ki67 stains of fields 1 and 2. (C) A day 22 tumour resulting from a proliferative signature melanoma (M980513). (D) Mitf and Ki67 stains of fields 3 and 4. Black horizontal bars represent 200 µm.

4.4 Discussion

A feature of current models for gene expression involvement in melanoma progression is their explicitly one-way nature. It is typical to present gene expression changes proceeding concomitantly with stage progression, where a gene either increases or decreases expression as the disease evolves through clinically recognised stages to metastasis (Miller and Mihm,

111

Chapter 4 Discussion

2006). However, models of this design do not account for the broad molecular heterogeneity apparent in melanoma. Indeed it is clear that, in many melanoma cells within a given lesion, genes expected to be downregulated in late stages are found active and others expected to be upregulated are not. One possible answer is that many genes associated with metastatic potential do not undergo one-way modification of regulation and instead retain the potential to reverse changes in their expression.

Investigations into the gene expression signatures of melanoma cell lines taken from late stage tumours show that a given cell line will usually express one of two major transcription programs. It was also determined that the genes whose expression patterns respectively delineated the two signatures were likely involved in melanoma metastatic potential (Bittner et al., 2000; Hoek et al., 2006). One of these signatures (identified by us as proliferative) has Mitf and other melanocytic genes (e.g. TYR, DCT, MLANA) upregulated along with a number of additional neural crest-related factors (e.g. SOX10, TFAP1A, EDNRB). This signature is associated with high rates of proliferation, low motility and sensitivity to growth inhibition by TGF-β. A second signature (identified by us as invasive) downregulates these genes and instead upregulates others whose secreted products (e.g. INHBA, COL5A1, SERPINE1) are known to be involved in modifying the extracellular environment. This signature is associated with lower rates of proliferation, high motility and resistance to growth inhibition by TGF-β. Having identified these genes we found that many of the proliferative signature were frequent responders to Wnt signalling, and those of the invasive signature were commonly TGF-β signal-driven, and we proposed that the balance in activity of these signalling pathways are responsible for the different transcription signatures observed (Hoek et al., 2006). Among genes comprising the invasive signature are several (e.g. WNT5A, DKK1 and CTGF) known to negatively regulate Wnt signalling (Ishitani et al., 2003; Mercurio et al., 2004; Zorn, 2001), suggesting that activation of TGF-β signalling may precipitate deactivation of Wnt signalling. Similar cross-talk opposition between TGF-β and Wnt signalling has already been noted in gastrointestinal cancer (Mishra et al., 2005). This possible link between the signatures indicated they may be reversible given appropriate signals and further suggested that proliferation and invasion are program states which melanoma cells activate according to microenvironmental cues (Hoek et al., 2006). The results of our in vitro proliferation and motility analyses were consistent with signature assignments inferred from earlier DNA microarray experiments (Hoek et al., 2006). In order

112

Chapter 4 Discussion to immunohistochemically differentiate signatures in vivo we used nuclear Mitf as our marker for the proliferative signature and nuclear Ki67 antigen as a general indicator for proliferation activity. We found in vitro that while both invasive and proliferative signature melanoma cells expressed Ki67 antigen, less than half of the invasive signature cells expressed it, correlating with the relative growth differences observed between proliferative and invasive cells in vitro. Because Ki67 antigen is not detected in G0 (Braun et al., 1988; Bruno and Darzynkiewicz, 1992), and it may be an absolute requirement for cell proliferation (Schluter et al., 1993), we conclude that invasive signature cells spend more time in G0 (quiescence). Increasing Mitf expression in melanoma has been shown to be a proliferative factor and involved in Cdk2 production and activity (Carreira et al., 2006; Du and Fisher, 2002; Widlund et al., 2002). Our data also supports a proliferative role for Mitf in vitro as we find that Mitf-positive lines proliferate faster (Fig. 4.1) and express Ki67 antigen with greater frequency than Mitf- negative lines (Fig. 4.2).

We performed siRNA knockdown of Mitf in a proliferative signature cell line to show that this confers a TGF-β-resistance phenotype which others have shown is characteristic of invasive signature line melanomas (Heredia et al., 1996; Krasagakis et al., 1994; Roberts et al., 1985). Our DNA microarray data suggested that Mitf gene expression is central to the proliferative signature and may therefore have a role in mediating the growth-inhibitory response to TGF-β. After knockdown of Mitf expression in a proliferative signature line the cells gained resistance to TGF-β-mediated inhibition of proliferation (Fig. 4.3). This correlates with experiments by others who have shown that invasive characteristics are increased in melanoma cells treated with siRNA targeting Mitf expression (Carreira et al., 2006; Lekmine et al., 2007). Together these combined findings indicate that regulation of Mitf expression is critical to signature membership and supports our contention that in vivo changes in nuclear Mitf staining indicate proliferative/invasive signature switching.

Our xenograft experiments showed that tumour growth patterns correlated appropriately with the gene expression signatures. Proliferative signature lines initiated tumours about two weeks after injection while invasive signature lines lay dormant for an average of eight weeks before tumour growth began (Fig. 4.4). While these and the in vitro experiments further support the different signature assignments to different melanoma cell lines, immunohistochemical examination of the tumours showed evidence for signature switching during tumorigenesis.

113

Chapter 4 Discussion

Comparison of in vitro Mitf staining patterns with in vivo Mitf staining patterns shows that resultant tumours deriving from either proliferative signature or invasive signature lines reveal the presence of both Mitf-positive and Mitf-negative melanoma cells (Fig. 4.5). Concurrently, Ki67 antigen staining, while found throughout the tumours, was more frequent in regions positive for Mitf staining than in regions absent of Mitf. Furthermore, the distribution of Mitf staining and increased frequency of Ki67 antigen positivity shows a distinctly peripheral pattern. This confirms that, as expected, melanoma cells proximal to the interface between host tissues and the tumour are actively undergoing increased rates of proliferation. To address concerns that our invasive signature lines were contaminated with proliferative signature cells we monitored in vitro proliferation rates for invasive signature cells serially passaged over the same time frame as conducted for the xenograft experiments. We observed no change in proliferation rates (data not shown), and therefore we believe that the change in tumour growth in invasive signature line-seeded xenografts is not due to prior contamination with proliferative signature cells.

The long lag-time for tumour initiation observed with invasive signature cells and the presence of Mitf-positive nuclei in resulting tumours indicate that tumour growth was probably preceded by a switch in some cells to the proliferative signature type. Similar microenvironment-driven signature switching has been shown previously in other experiments. Recent studies investigating the effect of embryonic environments on melanoma cells has shown how these environments can affect the aggressive phenotype. Kulesa and co- workers, using the aggressive C8161 melanoma line, showed that transplantation of C8161 cells into chick embryonic tissues stimulated re-expression of melanocytic markers similar to poorly aggressive cells (Kulesa et al., 2006). Complimentary in vitro studies in which poorly aggressive cells, grown on 3D matrices preconditioned by aggressive lines, showed upregulation of extracellular matrix modifying genes and increased invasive ability (Seftor et al., 2006). Additional experiments in zebrafish revealed that in embryonic environments inhibition of the morphogen Nodal switched melanoma cells to a less aggressive phenotype, suggesting that Nodal signalling (which acts through TGF-β family receptors) was important to maintaining aggressive phenotypes in melanoma cells (Topczewska et al., 2006). Signature switching of cells in response to the microenvironment would explain why our xenograft tumours deriving from different signature lines were immunohistochemically indistinguishable.

114

Chapter 4 Discussion

That rapidly proliferating cells are found closer to the periphery of growing tumours also supports a role for the microenvironmental determination of activity. However, this location of proliferative phenotype melanoma cells at the tumour periphery directly contradicts the long-held assumption that these cells represent the tumour’s invasive front. This assumption is based primarily on observations linking primary tumour thickness and the frequency of subsequent metastatic disease (Breslow, 1978). We offer that the thickness of the primary tumour may, rather than serve to bring its peripheral cells closer to vascular egress, determine the extent to which cells deeper within it experience microenvironmental change which (as we contend) drives the switch to a more invasive phenotype.

Perhaps more interesting is the comparison of our model with the identification of Mitf as a lineage-specific oncogene. Garraway and coworkers have shown Mitf gene amplification in a fraction of melanomas and demonstrate its correlation with a poorer prognosis. They speculated that increased Mitf gene dosage may in part compensate affected melanoma cells in settings where Mitf activity is normally lost (Garraway et al., 2005). This suggests that with amplification of its gene Mitf expression in our xenograft model would be preserved in all cells. However while increased gene dosage may play a role when the gene is activated (and precipitate a worsened prognosis in patients), this does not mean the gene is necessarily immune from otherwise normal signalling responses.

These data are critical pieces of the melanoma progression puzzle because they suggest not only that invasion and proliferation are divisible aspects of metastatic potential, but that these different transcriptional states are interchangeable programs between which melanoma cells oscillate during progression in response to changing microenvironmental cues (Fig. 4.6). What these microenvironmental cues precisely are remains unknown, but there is growing evidence that hypoxia may be one (Holmquist et al., 2006) and inflammation another (de Visser et al., 2006). The model we use to describe melanoma progression is presented in binary terms of invasive versus proliferative. However this is not to deny that intermediate transcription signatures exist between the described archetypes and we have indeed characterised other melanoma lines with intermediate signatures in our earlier study (Hoek et al., 2006). Therefore we believe that the two signatures discussed here represent opposite ends of a signature continuum between which melanoma cells slide in response to microenvironment-linked changes in signalling.

115

Chapter 4 Discussion

Figure 4.6. An integrated model for gene regulation of melanoma metastatic potential and progression. Early phase melanoma cells expressing the “proliferative signature” gene set proliferate to form the primary lesion. Following this an unknown signal switch, likely brought about by altered microenvironmental conditions (e.g. hypoxia or inflammation), gives rise to cells with a significantly different “invasive signature” gene set. Invasive signature cells escape and, upon reaching a suitable distal site, revert to the proliferative state and nucleate a new metastasis where the cycle is repeated. Each switch in phenotype (state change) is accompanied by an exchange in expressed gene sets from proliferative to invasive and vice versa.

The relevance of this model to clinical aspects of melanoma is that it may explain why metastatic melanoma is so refractory to chemo- and immunotherapeutic strategies. During treatment of metastatic melanoma so-called mixed responses are often observed and this phenomenon may derive from a heterogeneous distribution of proliferative and invasive signature cells within metastases. The heterogeneity inherent in tumours containing melanoma cells with biological activities dependent on the microenvironment suggests that while proliferating cells are susceptible to chemotherapy there are populations of cells which, though not proliferating, have the capacity to switch back to a proliferative program and successfully drive tumour progression once therapy has ceased. Finally, while our state- switching model for melanoma progression offers attractive answers for why the disease behaves as it does, we acknowledge the caveat that much of our hypothesis rests on expression signatures obtained in vitro and thus may not fully recapitulate melanoma's in vivo biology. Therefore, until the in vivo situation is resolved, we cannot yet advocate the abandonment of models which favour step-wise accumulation of genetic lesions as a driver for melanoma progression.

116

Chapter 4 Material and Methods

4.5 Material and Methods

4.5.1 Melanoma tissues and lines

Melanoma cell cultures were established from surplus material from cutaneous melanoma metastases removed by surgery after having obtained written informed consent of the patient. Clinical diagnosis was confirmed by histology and immunohistochemistry. Melanoma cells were released from tissue sections and grown as previously described (Geertsen et al., 1998). Cell lines were chosen according to their transcription pattern signatures as previously described (Hoek et al., 2006). Two proliferative signature (M980513, M000907) and two invasive signature (M991121, M010308) melanoma lines were used.

4.5.2 In vitro motility and proliferation assays

For the motility assays 2 x 104 melanoma cells were seeded on 8 µM transwell microporous filters (Becton Dickinson, Franklin Lakes, NJ) in 200 µl RPMI. As a chemoattractant RPMI containing 10% FCS was added to the lower chamber. After 18 hours of incubation cells on the upper side of the filter were removed with cotton swab. The membrane was then stained using a standard hematoxylin and eosin protocol and the cells were counted under a light microscope. For the proliferation assay melanoma cells were seeded to a density of 5 x 104 in each well of a six-well plate. After 24, 72 and 96 hours cells were counted in a Neubauer chamber to estimate cell-doubling times.

4.5.3 Recombinant adenovirus vector and siRNA

Recombinant first-generation, E1/E3-deleted Ad5-based vectors Ad-H1-siMitf and Ad-H1- siControl were generated as described previously (Hemmi et al., 1998). Briefly, homologous recombination was performed in human embryonic retinoblast line 911 cells between a transfer plasmid pAd-H1-siMitf encoding the Mitf-specific siRNA sequence under the control of the H1 promoter and a genomic ClaI DNA fragment isolated from AdMLP-lacZ. To construct pAd-H1-siMitf, the CMV promoter of pAd-CMV∆lacZ-lnk1 was replaced with the H1 promoter (Hasuwa et al., 2002). The H1 promoter fragment was PCR-amplified from genomic DNA of human 293T cells and cloned into SfiI/BamHI-restricted pAd-CMV∆lacZ- Ink1. Subsequently, oligonucleotides for the silencing cassette (Saydam et al., 2005) containing a 19-nucleotide siRNA sequence targeting Mitf (Busca et al., 2005) were cloned 117

Chapter 4 Material and Methods into NheI/SalI-restricted pAd-H1∆lacZ-lnk1 (Ad-H1-siMitf). For a mock control (Ad-H1- siControl), the siRNA sequence of Mitf was scrambled and blasted to ensure no human sequence is targeted. Recombinant adenoviruses were plaque-purified, amplified, and CsCl purified. Viral titers were determined by plaque assay, using 911 cells, and were 1.8 x 1010 PFU/ml for Ad-H1-siMitf and 1.3 x 1010 PFU/ml for Ad-H1-siControl.

4.5.4 Transfection and TGF-β challenge assay

Melanoma cells were seeded to a density of 4 x 104 cells in a 24-well plate one day before infection. The next day medium was changed to RPMI containing 2% FCS and cells were either infected with virus particles carrying the pAd-H1-siMitf or pAd-H1-siControl. For assessment of susceptibility to growth inhibition by TGF-β, cells were challenged with 5 ng/ml recombinant TGF-β (Biosource, Camarillo, CA) 24 hours after virus transduction. After a further 56 hours cell growth was estimated using a standard MTT assay.

4.5.5 Western blot analyses

Cells were solubilised in lysis buffer containing 20 mM Tris-HCl pH 7.5, 1% TritonX-100, 150 mM NaCl, 10% glycerol and Complete mini protease inhibitor (Roche Diagnostics GmbH, Mannheim, Germany). Proteins were separated on a NuPAGE 10% Bis-Tris gel (Invitrogen, Carlsbad, CA) under denaturing and reducing conditions followed by transfer onto a Nitrocellulose membrane (Invitrogen). Mitf protein was detected with a mouse anti- Mitf MAb (clone C5; LabVision, CA, USA) diluted 1:100 in 3% BSA at 4oC overnight. Secondary rabbit-anti-mouse antibodies (Abcam, Cambridge, UK) conjugated with peroxidase was used at a dilution of 1:10000. Detection by chemiluminescence used an enhanced chemiluminescence system (GE Healthcare, Buckinghamshire, UK).

4.5.6 Xenografts

For each melanoma line a total of 3 x 106 cells were injected into both flanks of eight-week old female athymic nude mice. Mice were kept in individually ventilated cages for a maximum of 75 days post injection. Volume of tumours was measured using vernier callipers (V = W2 x L x 0.5) once every 3-7 days until linear growth was detected, after which measurements were taken every 1-2 days. If at least one xenograft tumour reached 1 cm3 the mouse was sacrificed and tumours removed. If the condition of the mouse deteriorated (e.g.

118

Chapter 4 Material and Methods listlessness, loss of weight) the mouse was sacrificed and tumours removed. All remaining mice were sacrificed on the 75th day and tumours removed. Effective tumour initiation time was calculated on the day tumour volume reached 100 mm3. Tumours not reaching 100 mm3 within 75 days were not considered.

4.5.7 Immunohistochemistry

Cell lines were prepared for immunohistochemistry as follows. Briefly, cells were cultured, washed in PBS (Biochrom, Berlin, Germany) and then put into suspension by incubating in 2 mL trypsin/EDTA solution (Biochrom) at 37° C. Trypsin was inactivated by adding 18 mL of FCS-containing growth medium. Cell suspensions were centrifuged for 5 min at 2000 rpm. After removing the supernatant, four drops of plasma were added to the pellet and the solution was mixed. One drop of thrombin was added and after five minutes the coagulated material was encapsulated for fixation in 4% formalin and embedded in paraffin. Excised xenograft samples were fixed in 4% formalin and embedded in paraffin. Slides were cut from paraffin blocks and immunohistochemically stained using the alkaline phosphatase-anti-alkaline phosphatase technique and counter-stained using hematoxylin. Antibodies used were directed against Mitf (clone D5; DakoCytomation, Glostrup, Denmark) or Ki-67 (clone MIB-1; DakoCytomation). Counting of stained and unstained nuclei was done on a PC using the free UTHSCSA ImageTool program (developed at the University of Texas Health Science Center at San Antonio, Texas and available from the Internet by anonymous FTP from ftp://maxrad6.uthscsa.edu).

4.5.8 Statistical analysis

For all quantitative sample comparisons Student’s two-sample heteroscedastic t-test was used to calculate a t-statistic for comparison against a significance cutoff of p = 0.05.

119

Chapter 4 References

4.6 References

Banerjee, S.S. and Harris, M. (2000) Morphological and immunophenotypic variations in malignant melanoma. Histopathology, 36, 387-402. Bittner, M., Meltzer, P., Chen, Y., Jiang, Y., Seftor, E., Hendrix, M., Radmacher, M., Simon, R., Yakhini, Z., Ben-Dor, A., Sampas, N., Dougherty, E., Wang, E., Marincola, F., Gooden, C., Lueders, J., Glatfelter, A., Pollock, P., Carpten, J., Gillanders, E., Leja, D., Dietrich, K., Beaudry, C., Berens, M., Alberts, D. and Sondak, V. (2000) Molecular classification of cutaneous malignant melanoma by gene expression profiling. Nature, 406, 536-540. Braun, N., Papadopoulos, T. and Muller-Hermelink, H.K. (1988) Cell cycle dependent distribution of the proliferation-associated Ki-67 antigen in human embryonic lung cells. Virchows Arch B Cell Pathol Incl Mol Pathol, 56, 25-33. Breslow, A. (1978) Tumour thickness in evaluating prognosis of cutaneous melanoma. Ann Surg, 187, 440. Bruno, S. and Darzynkiewicz, Z. (1992) Cell cycle dependent expression and stability of the nuclear protein detected by Ki-67 antibody in HL-60 cells. Cell Prolif, 25, 31-40. Busca, R., Berra, E., Gaggioli, C., Khaled, M., Bille, K., Marchetti, B., Thyss, R., Fitsialos, G., Larribere, L., Bertolotto, C., Virolle, T., Barbry, P., Pouyssegur, J., Ponzio, G. and Ballotti, R. (2005) Hypoxia-inducible factor 1{alpha} is a new target of microphthalmia- associated transcription factor (MITF) in melanoma cells. J Cell Biol, 170, 49-59. Carreira, S., Goodall, J., Aksan, I., La Rocca, S.A., Galibert, M.D., Denat, L., Larue, L. and Goding, C.R. (2005) Mitf cooperates with Rb1 and activates p21Cip1 expression to regulate cell cycle progression. Nature, 433, 764-769. Carreira, S., Goodall, J., Denat, L., Rodriguez, M., Nuciforo, P., Hoek, K.S., Testori, A., Larue, L. and Goding, C.R. (2006) Mitf regulation of Dia1 controls melanoma proliferation and invasiveness. Genes Dev, 20, 3426-3439. Curtin, J.A., Fridlyand, J., Kageshita, T., Patel, H.N., Busam, K.J., Kutzner, H., Cho, K.H., Aiba, S., Brocker, E.B., LeBoit, P.E., Pinkel, D. and Bastian, B.C. (2005) Distinct sets of genetic alterations in melanoma. N Engl J Med, 353, 2135-2147. de Visser, K.E., Eichten, A. and Coussens, L.M. (2006) Paradoxical roles of the immune system during cancer development. Nat Rev Cancer, 6, 24-37. Du, J. and Fisher, D.E. (2002) Identification of Aim-1 as the underwhite mouse mutant and its transcriptional regulation by MITF. J Biol Chem, 277, 402-406. Du, J., Widlund, H.R., Horstmann, M.A., Ramaswamy, S., Ross, K., Huber, W.E., Nishimura, E.K., Golub, T.R. and Fisher, D.E. (2004) Critical role of CDK2 for melanoma growth linked to its melanocyte-specific transcriptional regulation by MITF. Cancer Cell, 6, 565- 576. Fidler, I.J. and Kripke, M.L. (1977) Metastasis results from preexisting variant cells within a malignant tumour. Science, 197, 893-895. 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., Lee, C., Wagner, S.N., Li, C., Golub, T.R., Rimm, D.L., Meyerson, M.L., Fisher, D.E. and Sellers, W.R. (2005) Integrative genomic analyses identify MITF as a lineage survival oncogene amplified in malignant melanoma. Nature, 436, 117-122. Geertsen, R.C., Hofbauer, G.F., Yue, F.Y., Manolio, S., Burg, G. and Dummer, R. (1998) Higher frequency of selective losses of HLA-A and -B allospecificities in metastasis than in primary melanoma lesions. J Invest Dermatol, 111, 497-502.

120

Chapter 4 References

Haqq, C., Nosrati, M., Sudilovsky, D., Crothers, J., Khodabakhsh, D., Pulliam, B.L., Federman, S., Miller, J.R., 3rd, Allen, R.E., Singer, M.I., Leong, S.P., Ljung, B.M., Sagebiel, R.W. and Kashani-Sabet, M. (2005) The gene expression signatures of melanoma progression. Proc Natl Acad Sci U S A, 102, 6092-6097. Hasuwa, H., Kaseda, K., Einarsdottir, T. and Okabe, M. (2002) Small interfering RNA and gene silencing in transgenic mice and rats. FEBS Lett, 532, 227-230. Hemmi, S., Geertsen, R., Mezzacasa, A., Peter, I. and Dummer, R. (1998) The presence of human coxsackievirus and adenovirus receptor is associated with efficient adenovirus- mediated transgene expression in human melanoma cell cultures. Hum Gene Ther, 9, 2363- 2373. Heredia, A., Villena, J., Romaris, M., Molist, A. and Bassols, A. (1996) The effect of TGF- beta 1 on cell proliferation and proteoglycan production in human melanoma cells depends on the degree of cell differentiation. Cancer Lett, 109, 39-47. Hoek, K.S., Schlegel, N.C., Brafford, P., Sucker, A., Ugurel, S., Kumar, R., Weber, B.L., Nathanson, K.L., Phillips, D.J., Herlyn, M., Schadendorf, D. and Dummer, R. (2006) Metastatic potential of melanomas defined by specific gene expression profiles with no BRAF signature. Pigment Cell Res, 19, 290-302. Holmquist, L., Lofstedt, T. and Pahlman, S. (2006) Effect of hypoxia on the tumour phenotype: the neuroblastoma and breast cancer models. Adv Exp Med Biol, 587, 179-193. Ishitani, T., Kishida, S., Hyodo-Miura, J., Ueno, N., Yasuda, J., Waterman, M., Shibuya, H., Moon, R.T., Ninomiya-Tsuji, J. and Matsumoto, K. (2003) The TAK1-NLK mitogen- activated protein kinase cascade functions in the Wnt-5a/Ca(2+) pathway to antagonize Wnt/beta-catenin signaling. Mol Cell Biol, 23, 131-139. Krasagakis, K., Garbe, C., Schrier, P.I. and Orfanos, C.E. (1994) Paracrine and autocrine regulation of human melanocyte and melanoma cell growth by transforming growth factor beta in vitro. Anticancer Res, 14, 2565-2571. Kulesa, P.M., Kasemeier-Kulesa, J.C., Teddy, J.M., Margaryan, N.V., Seftor, E.A., Seftor, R.E. and Hendrix, M.J. (2006) Reprogramming metastatic melanoma cells to assume a neural crest cell-like phenotype in an embryonic microenvironment. Proc Natl Acad Sci U S A, 103, 3752-3757. Lekmine, F., Chang, C.K., Sethakorn, N., Das Gupta, T.K. and Salti, G.I. (2007) Role of microphthalmia transcription factor (Mitf) in melanoma differentiation. Biochem Biophys Res Commun, 354, 830-835. Levene, A. (1980) On the histological diagnosis and prognosis of malignant melanoma. J Clin Pathol, 33, 101-124. Loercher, A.E., Tank, E.M., Delston, R.B. and Harbour, J.W. (2005) MITF links differentiation with cell cycle arrest in melanocytes by transcriptional activation of INK4A. J Cell Biol, 168, 35-40. Mercurio, S., Latinkic, B., Itasaki, N., Krumlauf, R. and Smith, J.C. (2004) Connective-tissue growth factor modulates WNT signalling and interacts with the WNT receptor complex. Development, 131, 2137-2147. Miller, A.J. and Mihm, M.C., Jr. (2006) Melanoma. N Engl J Med, 355, 51-65. Mishra, L., Shetty, K., Tang, Y., Stuart, A. and Byers, S.W. (2005) The role of TGF-beta and Wnt signaling in gastrointestinal stem cells and cancer. Oncogene, 24, 5775-5789. Paget, S. (1889) The distribution of secondary growths in cancer of the breast. Lancet, 1, 571- 573. Ribatti, D., Mangialardi, G. and Vacca, A. (2006) Stephen Paget and the 'seed and soil' theory of metastatic dissemination. Clin Exp Med, 6, 145-149.

121

Chapter 4 References

Roberts, A.B., Anzano, M.A., Wakefield, L.M., Roche, N.S., Stern, D.F. and Sporn, M.B. (1985) Type beta transforming growth factor: a bifunctional regulator of cellular growth. Proc Natl Acad Sci U S A, 82, 119-123. Sasse, A.D., Sasse, E.C., Clark, L.G., Ulloa, L. and Clark, O.A. (2007) Chemoimmunotherapy versus chemotherapy for metastatic malignant melanoma. Cochrane Database Syst Rev, CD005413. Saydam, O., Glauser, D.L., Heid, I., Turkeri, G., Hilbe, M., Jacobs, A.H., Ackermann, M. and Fraefel, C. (2005) Herpes simplex virus 1 amplicon vector-mediated siRNA targeting epidermal growth factor receptor inhibits growth of human glioma cells in vivo. Mol Ther, 12, 803-812. Schluter, C., Duchrow, M., Wohlenberg, C., Becker, M.H., Key, G., Flad, H.D. and Gerdes, J. (1993) The cell proliferation-associated antigen of antibody Ki-67: a very large, ubiquitous nuclear protein with numerous repeated elements, representing a new kind of cell cycle- maintaining proteins. J Cell Biol, 123, 513-522. Seftor, E.A., Meltzer, P.S., Kirschmann, D.A., Margaryan, N.V., Seftor, R.E. and Hendrix, M.J. (2006) The epigenetic reprogramming of poorly aggressive melanoma cells by a metastatic microenvironment. J Cell Mol Med, 10, 174-196. Steingrimsson, E., Copeland, N.G. and Jenkins, N.A. (2004) Melanocytes and the Microphthalmia Transcription Factor Network. Annu Rev Genet, 38, 365-411. Topczewska, J.M., Postovit, L.M., Margaryan, N.V., Sam, A., Hess, A.R., Wheaton, W.W., Nickoloff, B.J., Topczewski, J. and Hendrix, M.J. (2006) Embryonic and tumorigenic pathways converge via Nodal signaling: role in melanoma aggressiveness. Nat Med, 12, 925-932. Widlund, H.R., Horstmann, M.A., Price, E.R., Cui, J., Lessnick, S.L., Wu, M., He, X. and Fisher, D.E. (2002) Beta-catenin-induced melanoma growth requires the downstream target Microphthalmia-associated transcription factor. J Cell Biol, 158, 1079-1087. Zorn, A.M. (2001) Wnt signalling: antagonistic Dickkopfs. Curr Biol, 11, R592-595.

122

Chapter 5-

5 Id2 suppression of p15Ink4b abrogates TGF-β-mediated growth inhibition in melanoma

123

Id2 suppression of p15Ink4b abrogates TGF-β-mediated growth inhibition of melanoma

NC Schlegel1, OM Eichhoff1, S Hemmi2, D Mihic3, S Werner4, R Dummer1 and KS Hoek1

1Department of Dermatology, University Hospital of Zürich, 8091 Zürich, Switzerland 2Institute of Molecular Biology, Faculty of Mathematics and Natural Sciences, University of Zürich, 8057 Zürich, Switzerland 3Institute of Clinical Pathology, University Hospital of Zürich, 8091 Zürich, Switzerland 4Institute of Cell Biology, Department of Biology, ETH Zurich, 8093 Zurich, Switzerland

Manuscript to be submitted to Oncogene

124

Chapter 5 Asbtract / Introduction

5.1 Abstract

Growth inhibition resistance to transforming growth factor β (TGF-β) is a turning point in the malignant progression of many cancer types. In melanoma this resistance is associated with a more aggressive metastatic potential. Here, using proliferative and invasive phenotype pairs, we explored molecular responses involved in modulating susceptibility to TGF-β-mediated inhibition of proliferation. We identified the Id2 gene as differentially regulated by TGF-β and link the loss of this regulation to acquired resistance to TGF-β in invasive phenotype cells. We show that TGF-β induces cell cycle arrest through induction of p15Ink4b and repression of Id2 gene expression. Furthermore, Id2 overexpression in proliferative phenotype cells counteracts p15Ink4b induction and consequently protects melanoma cells from TGF-β- mediated inhibition of proliferation. We find that although Id2 expression modulates susceptibility to TGF-β, tissue microarray analysis of its expression shows no link with patient survival. We conclude that transition to increased aggressiveness in melanoma cells requires Id2 upregulation to overcome its suppression by TGF-β and thus circumvent TGF-β- mediated inhibition of proliferation

5.2 Introduction

Transforming growth factor β (TGF-β) has been shown to have both tumour suppressing and promoting activities. Its tumour suppressing function is demonstrated by its important role as a negative regulator of proliferation for most cell types. Evasion from cytostatic responsiveness to TGF-β is important in a number of cancers, including melanoma. Reduced susceptibility to the growth inhibiting effects of TGF-β has been associated with increased invasive and metastatic properties of melanoma cells (Heredia et al., 1996; Krasagakis et al., 1999; Rodeck et al., 1994).

Most of our understanding of the mechanisms involved in growth control by TGF-β derives from studies on epithelial cells. A number of these findings are also relevant for other cell types. TGF-β regulates genes involved throughout the cell cycle, but its cytostatic effects have been primarily attributed to the regulation of factors targeting G1 events (Flores et al., 1996; Massague and Gomis, 2006). TGF-β induces the expression of cyclin-dependent kinase

125

Chapter 5 Introduction

(CDK) inhibitors (Massague and Gomis, 2006) and represses growth-promoting transcription factors such as c-Myc (Pietenpol et al., 1990) and inhibitor of DNA binding (Id) proteins (Kowanetz et al., 2004; Ling et al., 2002)

Id proteins are positive regulators of cell growth and play a critical role in promoting G1/S cell cycle progression. Their role in the regulation of cell proliferation is thought to be driven by two mechanisms. Firstly, they interfere with bHLH, ETS and Pax factors and consequently regulate the expression of their target genes such as the CDK inhibitors (Alani et al., 2001; Lyden et al., 1999; Mori et al., 2000; Ohtani et al., 2001; Prabhu et al., 1997; Rothschild et al., 2006). Secondly, Id2 binds tumour suppressor proteins of the Rb family and when in large excess, abolishes their growth-suppressing activity by causing the release of E2F transcription factors required for cell cycle progression (Iavarone et al., 1994; Lasorella et al., 1996). In normal cells, Id2 is itself a downstream target of pRb and its family members, which inhibit its function against natural targets. However, Id2 overexpressed by tumour cells can saturate the Rb pathway and deprive the cell of its most important cell cycle checkpoint (Lasorella et al., 1996). Rb pathway inhibition is believed to be achieved directly through the binding of Id2 and indirectly through Id2-mediated downregulation of CDK inhibitor expression (Ohtani et al., 2001).

In melanoma, Id levels have primarily been reported for Id1 and few publications report on Id2 and melanoma. For example, Polsky and co-workers have demonstrated a correlation between Id1 expression and loss of p16Ink4a expression in early melanoma (Polsky et al., 2001). Using tissue microarrays, Straume and Akslen evaluated the expression of Id1 in 119 cases of nodular melanoma and showed that strong Id1 expression was significantly associated with increased tumour thickness and reduced survival (Straume and Akslen, 2005).

Id2 was first associated with melanoma when it was identified as the product of a downregulated gene in melanomas with homozygous deletion of the CDKN2A locus genes in a global gene expression study (Bloethner et al., 2006). In a microarray gene expression analysis of uveal melanoma, which generated two subgroups representing tumours with low and high risk of metastatic death, Id2 was one of the top class discriminating genes (Onken et al., 2006). Id2 was shown to be strongly downregulated in melanomas with high risk of metastatic death (Onken et al., 2006).

126

Chapter 5 Introduction / Results

Bittner and co-workers first suggested that there may be specific transcriptional signatures delineating melanoma cell subgroups (Bittner et al., 2000). We recently characterised two different transcription signatures for melanoma cell cultures which, based on known functions of the genes involved, suggested that their respective contribution to metastatic potential were either proliferative or invasive (Hoek et al., 2006). We hypothesised that melanoma cells may be able to switch back and forth between these phenotypes and thereby drive melanoma progression. Subsequently, we demonstrated phenotype switching in vivo using an immunocompromised mouse model (Hoek et al., 2008). Many of the genes expressed by invasive phenotype melanoma cells had previously been characterised as being positively responsive to TGF-β-like signalling in other systems. Furthermore, TGF-β-mediated growth inhibition, which has previously been shown to be absent or less pronounced in aggressively invasive melanoma cells, was also shown to be a discriminating phenotype for the phenotypes we characterised (Hoek et al., 2006). Here, we identify Id2 as a TGF-β regulated gene involved in the TGF-β growth inhibitory response and discuss its relevance for the invasive phenotype.

5.3 Results

5.3.1 Differential Id2 regulation and expression in human melanoma cultures

In previous DNA microarray experiments we identified two in vitro transcription phenotypes, proliferative and invasive, which together drive melanoma progression (Hoek et al., 2008; Hoek et al., 2006). We also found that proliferative and invasive phenotypes were differentially susceptible to TGF-β-mediated growth inhibition (Hoek et al., 2006).

We selected pairs of proliferative and invasive phenotype melanoma cell cultures. Supervised hierarchical clustering of these samples was performed using normalised signal intensity data from 105 genes whose expression was shown to be tightly linked to phenotype (Fig. 5.1A). We also performed TGF-β-susceptibility assays to establish that our candidates were appropriately susceptible or resistant to TGF-β-mediated growth inhibition. These culture pairs represent opposing TGF-β responses observed in melanomas; at least 50% growth

127

Chapter 5 Results inhibition was observed in proliferative phenotype cells while no inhibition or even a slight growth stimulation was seen in invasive phenotype cells (Fig. 5.1B).

Figure 5.1. Melanoma cell phenotype characteristics and endogenous Id2 expression levels. (A) A gene expression heatmap, generated by clustering samples based on the normalised expression of 105 metastatic potential genes, highlights phenotype-specific signatures of our four cell cultures. This shows that M000921 and M010817 (proliferative phenotype) cluster separately from M990115 and

M010119 (invasive phenotype). (B) 96 hours after the addition of 5ng/ml TGF-β1 cell proliferation was assessed by a colorimetric assay and expressed as a percentage of proliferatiom relative to untreated cells. Proliferative phenotype cells are susceptible to the growth inhibitory effects of TGF-β while invasive phenotype cells are not. (C) Id2 RNA expression measured by RT-PCR and (D) western blot analysis of Id2 protein expression in proliferative (M000921; M010817) and invasive (M990115; M010119) cell cultures. For RT-PCR Id2 mRNA levels are expressed as ratios to GAPDH expression, and for western blotting beta-actin protein expression is shown as a loading control. Proliferation assays were performed three times in quadruplicate and RT-PCR experiments were performed three times in triplicate. Error bars represent ± standard deviation.

After confirming the opposing phenotypes, we looked for differential endogenous expression of Id2 mRNA and protein in our representative cell cultures. Using reverse transcription- polymerase chain reaction (RT-PCR) and western blotting we showed that Id2 expression was diminished in proliferative phenotype cells when compared to invasive phenotype cells (Figure 5.1C-D).

As Id proteins have been shown to be regulated by TGF-β in epithelial cells (Kowanetz et al., 2004; Ling et al., 2002), we used RT-PCR to further investigate TGF-β modulation of Id2 in proliferative and invasive phenotype cells. As shown by RT-PCR, TGF-β1 downregulated Id2 mRNA expression in proliferative (M010817; M000921) but not in invasive (M990115;

128

Chapter 5 Results

M0100119) phenotype cell cultures (Fig. 5.2A). To confirm that Id2 protein is also differentially regulated, we showed Id2 protein expression on two cohort-representative TGF- β-treated cell cultures (Fig. 5.2B). Consistent with the mRNA data, Id2 protein was downregulated after a 24h treatment with TGF-β1 in proliferative phenotype cells.

Figure 5.2. TGF-β suppresses Id2 expression only in proliferative phenotype melanoma cells. (A) Id2 expression was measured by RT-PCR as a ratio to GAPDH expression and expressed in terms of TGF-β-treated cells relative to untreated cells at 24h normalised against data acquired at 0h. This shows that Id2 levels are more susceptible to TGF-β suppression in the proliferative phenotype than in the invasive phenotype. Data represent averages of three independent experiments performed in triplicate with standard deviations (error bars). (B) Western blot analysis of Id2 protein expression in M010817 and M010119 after the addition of TGF-β; beta-actin protein expression is shown as a loading control.

5.3.2 Id2 overexpression protects proliferative cells from the growth inhibitory effects of TGF-β Given that Id2 is differentially expressed and regulated across melanoma cell phenotypes, we next hypothesised that Id2 overexpression could counteract the negative effects of TGF-β on the growth of proliferative phenotype cells. We used a myc-tagged wild-type Id2 adenoviral construct (Ad-Id2) as used by others (Gleichmann et al., 2002; Kowanetz et al., 2004; Toma et al., 2000). We infected proliferative phenotype cells with either Ad-Id2 or a control adenoviral construct (Ad-luciferase) at a multiplicity of infection (MOI) of 3 prior to treatment with 5 ng/ml TGF-β1 followed by assessment of cell proliferation. Overexpression of Id2 protein was confirmed by western blot analysis (Fig. 5.3B). Confirming our hypothesis, proliferative phenoptye cells overexpressing Id2 were less growth inhibited upon addition of TGF-β compared to the cells transfected with the control construct (Fig. 5.3A). We therefore conclude that overexpression of Id2 in proliferative phenotype cells protects the cells from the

129

Chapter 5 Results growth inhibitory effects of TGF-β by inhibiting the TGF-β-induced downregulation of its expression.

Furthermore, we suppressed Id2 expression in invasive phenotype cells using an Id2-specific siRNA. Although Id2 protein expression could not be repressed by more than 50% (Fig. 5.3D) we could still significantly reduce TGF-β-induced growth in invasive phenotype cells (Fig. 5.3C). Taken together, as it has been shown for epithelial cells (Kowanetz et al., 2004), our results indicate that repression of Id2 is necessary for TGF-β-induced growth control in melanoma cells.

Figure 5.3. Id2 modulates TGF-β growth inhibition in melanoma. (A,B) Proliferative phenoptye melanoma cells were infected with an Id2 overexpressing adenovirus or a control adenovirus Ad- luciferase at an MOI of 3. Id2 protein levels in transfected cells were determined by western blot analysis. β-actin staining served as a loading control. (B). Id2 overexpression resulted in reduced TGF-β-mediated inhibition of cell growth (A). (C,D) Invasive phenotype cells were transfected with siRNA duplexes targeting Id2 or non-targeting negative control. Id2 protein levels in transfected cells were determined by western blot analysis. β-actin staining served as a loading control (D). 24h post- infection or 48h post-transfection, cells were treated with 5 ng/ml recombinant TGF-β1 and 50 h post- treatment, cell proliferation was evaluated. Growth inhibition (A) or stimulation (C) was determined by colorimetric assay and expressed as growth inhibition/stimulation relative to untreated cells infected with the respective adenovirus or transfected with the respective siRNA. Data was then normalised to the growth inhibition/stimulation calculated for cells infected with Ad-luciferase or transfected with the control siRNA. For each cell culture, data represents the results of four independent experiments performed in quadruplicate with standard deviations (error bars). Student’s paired t-test was used to calculate significance.

130

Chapter 5 Results

5.3.3 Id2 regulates TGF-β-induced G1 cell-cycle arrest

TGF-β-induced growth inhibition has been attributed to G1 cell-cycle arrest in a number of cell types including melanoma (Flores et al., 1996; Massague and Gomis, 2006). In order to investigate if Id2 modulation directly impacts on TGF-β-induced G1 cell-cycle arrest, we analysed the cell cycle distribution of TGF-β-treated proliferative phenotype cells transduced with overexpressing Id2 or control adenoviruses. 24 hours after infecting M010817 cells with Ad-Id2 or Ad-luciferase (MOI 3), we followed this with 5 ng/ml TGF-β1 for 48 hours and propidium iodide staining to analyse their DNA content using flow cytometry. We detected a reduction of cells in S phase and an increase in cells in G1 following TGF-β addition to control cells (Fig. 5.4A). Confirming our hypothesis, the G1 arrest was reversed when cells were initially transduced with an overexpressing Id2 construct.

Figure 5.4. Id2 regulates TGF-β-induced G1 cell-cycle arrest and suppresses TGF-β-induced induction of p15Ink4b. (A) Proliferative phenotype cells (M010817) were infected at an MOI of 3 with Ad-Id2 or Ad-luciferase in serum-reduced medium (3%). 24h post-infection, cells were treated with 5 ng/ml TGF-β1 and 48h post-treatment, cells were harvested and their DNA content was analysed by flow cytometry. Data is expressed as TGF-β-induced change in cell cycle content expressed in percentage and is represents averages of three independent experiments with standard deviations (error bars). Student’s paired t-test was used to calculate significance. (B) Proliferative phenotype cells (M010817) were infected at an MOI of 3 with Ad-Id2 or Ad-luciferase in serum-reduced medium (3%). 24h post-infection, cells were treated with 5 ng/ml TGF-β1 and 1, 8 and 24 h post-treatment cells were harvested for RNA extraction and CDK inhibitors expression was measure by RT-PCR. CDK inhibitors mRNA levels were calculated as a ratio to GAPDH expression and data was then normalised to untreated samples transduced with Ad-luciferase at 1 hour. TGF-β induces the expression of the CDK inhibitor p15Ink4b but not p21Cip1, p27Kip1 or p57Kip2. Moreover, Id2 represses the expression of p15Ink4b and concurrently restricts the induction of p15Ink4b by TGF-β. Data represent averages of three independent experiments performed in triplicate.

131

Chapter 5 Results

5.3.4 Id2 restricts TGF-β-induced upregulation of p15Ink4b

Having shown that Id2 regulates TGF-β-induced G1 arrest, we next wanted to identify the mechanism of this interaction. The G1 cell cycle arrest induced by TGF-β has been partially attributed to its role in driving the expression of cyclin-dependent kinase (CDK) inhibitors, which are negative regulators of the cell cycle (Massague and Gomis, 2006). Contrasting this, Ids have been shown to positively regulate cell cycle progression by inhibiting the expression of CDK inhibitors (Alani et al., 2001; Lyden et al., 1999; Mori et al., 2000; Ohtani et al., 2001; Prabhu et al., 1997; Rothschild et al., 2006). We therefore hypothesised that Id2 suppressed the induction of CDK inhibitors by TGF-β.

To verify this, we transduced proliferative phenotype cells with Ad-Id2 or Ad-luciferase, followed by TGF-β treatment and analysed the mRNA levels of CDK inhibitors after 1, 8 and 24 hours of treatment. Contrasting with previously published melanoma reports (Florenes et al., 1996; Reed et al., 2001), p21Cip1 expression was not induced by TGF-β treatment in our melanoma cells (Fig. 5.4B). Similarly, we did not detect significant induction of p27Kip1 or p57Kip2. However, we did find a significant increase in p15Ink4b expression by TGF-β treatment. Interestingly, Id2 overexpressing cells showed significantly reduced levels of p15Ink4b, which were not rescuable by TGF-β treatment. From this data we conclude that Id2 dampens TGF-β-induced upregulation of p15Ink4b and the consequent G1 arrest.

5.3.5 Id2 expression does not correlate with patient survival in melanoma

We were next interested in correlating Id2 expression with patient survival. Using cell culture and tissue arrays derived from metastatic tumours, we evaluated Id2 expression by immunohistochemistry. Id2 expression was scored as being less than 10%, between 10% and 85% or more than 85% present in 51 metastatic melanoma tumours (Fig. 5.5A). After comparing Id2 expression with survival data from the tumour donors we could conclude that survival was not significantly different between the three groups (Fig. 5.5B). To look more closely if the presence of Id2 was important for patient survival, we scored Id2 expression as being absent (<10%) or present (>10%) but again, did not find a correlation between Id2 expression and patient survival.

132

Chapter 5 Results / Discussion

Figure 5.5. Id2 expression in melanoma organ metastases. (A) Representative tissue microarray spots stained for Id2. Id2 expression was scored as being less than 10%, between 10% and 85% or more than 85%. (B) Kaplan-Meier analysis of Id2 expression in melanoma organ metastases. It shows that Id2 expression does not correlate with patient survival in melanoma.

5.4 Discussion

In the present study we show that TGF-β represses Id2 expression in proliferative phenotype melanoma cells, which are susceptible to the growth inhibitory effects of TGF-β. In contrast, invasive phenotype melanoma cells, which continue to thrive in the presence of TGF-β, express higher levels of Id2, which are not repressed by TGF-β. Although a number of reports have demonstrated the modulation of Id proteins by TGF-β in epithelial cells (Di et al., 2006 ; Kang et al., 2003; Kondo et al., 2004; Kowanetz et al., 2004), we show differential modulation of Id2 expression between melanoma cell phenotypes. Moreover, elevated expression of Id mRNA and protein have been reported for many different human tumours, including carcinomas, neural tumours, leukaemia, as well as melanoma, and in some cases high levels were associated with increased disease severity and poor prognosis (Perk et al., 2005). Complementing those findings, we saw that Id2 expression correlated with the two different transcription signatures we previously identified (Hoek et al., 2006), where Id2 expression is increased in invasive phenotype cells compared to proliferative phenotype cells.

133

Chapter 5 Discussion

We next demonstrated that we could protect proliferative phenotype cells from TGF-β- mediated growth inhibition by overexpressing Id2 to circumvent TGF-β-driven Id2 repression. To complement this, we mimicked TGF-β-induced repression of Id2 with human Id2-specific siRNA in Id2-expressing invasive phenotype cells. We found that silencing Id2 inhibited the weak TGF-β-driven induction of proliferation characteristic of these cells but did not contribute to measurable TGF-β-like growth inhibition. This suggests that Id2 downregulation may not be sufficient to induce a TGF-β-like suppression of growth. Rather, this complex response is likely to involve the modulation of multiple genes, which may compensate for Id2 loss.

The high levels of Id2 expression and the inability of TGF-β to modulate it in our invasive phenotype cells, correlate well with the idea that aggressive cancer cells lose their susceptibility to the growth inhibitory effects of TGF-β while retaining their responses to tumour-promoting TGF-β-induced effects such as invasion, evasion of immune surveillance and metastasis. Adding to this concept, we can hypothesise that the loss of Id2 modulation favours an invasive behaviour in melanoma as Id proteins have been shown to play a role in a number of cancer promoting processes, such as proliferation, angiogenesis, invasion and migration ( reviewed in Lasorella et al., 2001).

While TGF-β is a negative regulator of the cell cycle, Id proteins are positive regulators of cell growth and play a critical role in promoting G1/S cell cycle progression. TGF-β-induced growth suppression has been associated with increased expression of CDK inhibitors and a concomitant G1 cell cycle arrest (Massague and Gomis, 2006). In various cell types, TGF-β has been shown to transcriptionally induce p21Cip1 (Datto et al., 1995), p27Kip1 (Kamesaki et al., 1998) p57Kip2 (Scandura et al., 2004) and p15Ink4b (Hannon and Beach, 1994) expression. In contrast, Id proteins have been shown to regulate the expression of CDK inhibitors, for example p16Ink4a (Alani et al., 2001; Ohtani et al., 2001), p21Cip1 (Prabhu et al., 1997), p57Kip2 (Rothschild et al., 2006) and p27Kip1 (Lyden et al., 1999; Mori et al., 2000). Using proliferative signature cell types, which were significantly growth inhibited by TGF-β, we could show that TGF-β induced the expression of the gene coding for p15Ink4b but not p21Cip1, p27Kip1 or p57Kip2. Furthermore, reinforcing the link between TGF-β-induced growth inhibition and TGF-β-induced downregulation of Id2, we could show that only p15Ink4b was downregulated by Id2 overexpression. We therefore conclude from these results that Id2 134

Chapter 5 Discussion / Material and Methods subdues TGF-β-induced upregulation of p15Ink4b, leading to attenuated TGF-β-induced growth-inhibition response in our cell lines.

Although p15Ink4b has not directly been shown to be regulated by Id proteins, analysis of the promoter regions of p15Ink4b, p16Ink4a and p21Cip1 reveals the presence of E-boxes, which renders these genes competent for Id-mediated repression (Pagliuca et al., 2000). We also highlight the importance of p15Ink4b as a TGF-β target gene involved in the cytostatic effects observed in melanoma and reveal that upregulation of p21Cip1 is not, as shown by others, necessary for this response (Florenes et al., 1996; Reed et al., 2001). It is important to note that the two groups reporting on the relevance of p21Cip1 used cell lines lacking p15Ink4b in the first case and expressing undetectable levels in the second (Florenes et al., 1996; Reed et al., 2001). This suggests that TGF-β may induce the expression of both CDK inhibitors and that their functions are redundant.

Evaluating Id2 expression in cell culture and tissue arrays, we found no association between Id2 expression and survival in melanoma. We observed Id2 expression heterogeneity in most samples of the tissue microarrays. This observation highlights questions about the validity of using tissue arrays to evaluate phenotype-specific protein expression in melanoma samples. The transcription profile phenotype switching (proliferative and invasive), which drives progression and is described in our melanoma model, allows melanoma cells to change their gene expression programs to favour disease progression via alternating states of proliferation and invasiveness (Hoek et al., 2008; Hoek et al., 2006). This model predicts that melanomas are heterogeneous with respect to the phenotypes, and suggests that tissue microarrays (in sourcing extremely limited samples of tumours) may therefore represent an inappropriate medium for exploring phenotype-specific factors.

5.5 Materials and Methods

5.5.1 Cell culture and Adenoviruses

135

Chapter 5 Material and Methods were released from tissue sections and grown as previously described (Geertsen et al., 1998). Cell lines were chosen according to their transcription pattern signatures as previously described (Hoek et al., 2006). Two proliferative signature (M000921, M010817) and two invasive signature (M990115, M010119) melanoma lines were used. Myc-tagged Ad-Id2 was a generous gift from F.D. Miller, Toronto, Canada. Ad-luciferase and Ad-Id2 were amplified and titrated in 293 cells as described previously (Hemmi et al., 1998).

5.5.2 RNA extraction, cDNA synthesis and RT-PCR

Total RNA was extracted from melanoma cell cultures using TRIzol according to manufacturer instructions (Invitrogen, Carlsbad, CA, USA). One microgram total RNA was used for cDNA synthesis using Promega’s Reverse Transcription System according to manufacturer instructions (Promega, Madison, WI, USA). PCR was performed on 1 µg template cDNA using Roche’s LightCycler DNA Master SYBR Green kit (Roche, Basel, Switzerland). Primers were 5’-CTGCCCAAGCTCTACCTTCC-3’ and 5’- CAGGTCCACATGGTCTTCCT-3’ (p21Cip1); 5’-CGTGCGAGTGTCTAACGGGAGC-3’ and 5’-TGCGTGTCCTCAGAGTTAGCC-3’ (p27Kip1); 5’-GCGGCGATCAAGAAGCTGTC- 3’ and 5’-CCGGTTGCTGCTACATGAAC-3’ (p57Kip2). Primers for p15Ink4b were purchased at Qiagen, QT00203147 (Qiagen, Hombrechtikon, Switzerland).

5.5.3 Western blot analysis

Cells were washed twice with cold phosphate-buffered saline (PBS) and lysed at 4°C in lysis buffer containing 20 mM Tris-HCl (pH 7.5), 1% Triton X-100 (Sigma-Aldrich), 137 mM NaCl, 10% glycerol and protease inhibitors (Roche). Proteins were separated by SDS–PAGE under reducing conditions and transferred onto nitrocellulose membranes (Invitrogen, Basel, Switzerland). Membranes were probed with a rabbit anti-Id2 monoclonal antibody (Zymed Laboratories, Invitrogen, San Francisco,CA, USA) or a goat anti-actin polyclonal antibody (SantaCruz, Biotechnology, La Jolla, CA, USA) followed by horseradish peroxidase- conjugated goat anti-rabbit or rabbit anti-goat secondary antibodies (Bio-Rad, Reinach, Switzerland and SantaCruz, respectively). Bound antibodies were detected by chemiluminescence (ECL, GE Healthcare, Buckinghamshire, UK).

136

Chapter 5 Material and Methods

5.5.4 Growth inhibition assays

Cells were seeded in 24-well microplates and left to settle overnight. Cells were then infected at an MOI of 3 with Ad-Id2 or Ad-luciferase in serum-reduced medium (3%). For siRNA treatment, cells were transfected with siRNA duplexes targeting Id2 (sense: GGUGGAGCGUGAAUACCAGtt and antisense: CUGGUAUUCACGCUCCACCtt) or non- targeting negative control (scrambled sequence) using INTERFERin (PolyPlus, Illkirch, France) according to manufacturer instructions. 24-hours post-infection or 48-hours post- transfection cells were treated with 5 ng/ml recombinant TGF-β1 (BioSource, Camarillo, CA, USA) and 50 hours post-treatment cell metabolic activity was determined with a standard colorimetric assay measuring 3-(4,5-dimethyldiazol-2-yl)-2,5 diphenyl tetrazolium bromide (MTT; Sigma-Aldrich, Buchs, Switzerland) reactivity and used as an approximation of cell proliferation. Growth inhibition was expressed as a percentage of growth inhibition compared against growth in the absence of TGF-β1.

5.5.5 Cell cycle FACS analysis

Cells were seeded in 24-well microplates and left to settle overnight. Cells were then infected at an MOI of 3 with Ad-Id2 or Ad-luciferase in serum-reduced medium (3%). 24-hours post- infection, cells were treated with 5 ng/ml recombinant TGF-β1 (BioSource) and 48 hours post-treatment, cells were harvested by trypsinisation and fixed in 70% ethanol at 4°C overnight. Cells were then washed with PBS, followed by incubation in 300 µl of 0.1% Triton X-100 (Sigma-Aldrich), 50 µg/ml propidium iodide (Fluka BioChemica, Switzerland) and 200 µg/ml DNase-free RNase A (Sigma-Aldrich) for 30 minutes before analysis on FACSCalibur (Becton-Dickinson, Switzerland).

5.5.6 Immunohistochemistry, cell culture and tissue array

Paraffin embedded tissue of 51 metastases was used for the preparation of a melanoma metastases tissue microarray. A morphologically representative region of the paraffin “donor” blocks was chosen. The representative region was taken with a core tissue biopsy (diameter 0.6mm, height 3-4mm) and precisely arrayed into a new “recipient” paraffin block using a customer built instrument (Kononen et al., 1998). After the block construction was completed, 4.0µm sections of the resulting tumour tissue microarray block were cut and

137

Chapter 5 Material and Methods immunohistochemically stained for Id2 using an anti-Id2 monoclonal antibody (Zymed Laboratories, Invitrogen).

138

Chapter 5 References

5.6 References

Alani, R.M., Young, A.Z. and Shifflett, C.B. (2001) Id1 regulation of cellular senescence through transcriptional repression of p16/Ink4a. Proceedings of the National Academy of Sciences, 98, 7812-7816. Bittner, M., Meltzer, P., Chen, Y., Jiang, Y., Seftor, E., Hendrix, M., Radmacher, M., Simon, R., Yakhini, Z., Ben-Dor, A., Sampas, N., Dougherty, E., Wang, E., Marincola, F., Gooden, C., Lueders, J., Glatfelter, A., Pollock, P., Carpten, J., Gillanders, E., Leja, D., Dietrich, K., Beaudry, C., Berens, M., Alberts, D. and Sondak, V. (2000) Molecular classification of cutaneous malignant melanoma by gene expression profiling. Nature, 406, 536-540. Bloethner, S., Hemminki, K., Thirumaran, R.K., Chen, B., Mueller-Berghaus, J., Ugurel, S., Schadendorf, D. and Kumar, R. (2006) Differences in global gene expression in melanoma cell lines with and without homozygous deletion of the CDKN2A locus genes. Melanoma Res, 16, 297-307. Datto, M.B., Li, Y., Panus, J.F., Howe, D.J., Xiong, Y. and Wang, X.F. (1995) Transforming growth factor beta induces the cyclin-dependent kinase inhibitor p21 through a p53- independent mechanism. Proc Natl Acad Sci U S A, 92, 5545-5549. Di, K., Ling, M.-T., Tsao, S.W., Wong, Y.C. and Wang, X. (2006 ) Id-1 modulates senescence and TGF-beta1 sensitivity in prostate epithelial cells. 10.1042/BC20060026. Biol. Cell 98, 523-533. Florenes, V.A., Bhattacharya, N., Bani, M.R., Ben-David, Y., Kerbel, R.S. and Slingerland, J.M. (1996) TGF-beta mediated G1 arrest in a human melanoma cell line lacking p15INK4B: evidence for cooperation between p21Cip1/WAF1 and p27Kip1. Oncogene, 13, 2447-2457. Flores, J.F., Walker, G.J., Glendening, J.M., Haluska, F.G., Castresana, J.S., Rubio, M.-P., Pastorfide, G.C., Boyer, L.A., Kao, W.H., Bulyk, M.L., Barnhill, R.L., Hayward, N.K., Housman, D.E. and Fountain, J.W. (1996) Loss of the p16INK4a and p15INK4b Genes, as well as Neighboring 9p21 Markers, in Sporadic Melanoma. Cancer Res, 56, 5023-5032. Geertsen, R.C., Hofbauer, G.F., Yue, F.Y., Manolio, S., Burg, G. and Dummer, R. (1998) Higher frequency of selective losses of HLA-A and -B allospecificities in metastasis than in primary melanoma lesions. J Invest Dermatol, 111, 497-502. Gleichmann, M., Buchheim, G., El-Bizri, H., Yokota, Y., Klockgether, T., Kugler, S., Bahr, M., Weller, M. and Schulz, J.B. (2002) Identification of inhibitor-of-differentiation 2 (Id2) as a modulator of neuronal apoptosis. J Neurochem, 80, 755-762. Hannon, G.J. and Beach, D. (1994) p15INK4B is a potential effector of TGF-beta-induced cell cycle arrest. Nature, 371, 257-261. Hemmi, S., Geertsen, R., Mezzacasa, A., Peter, I. and Dummer, R. (1998) The presence of human coxsackievirus and adenovirus receptor is associated with efficient adenovirus- mediated transgene expression in human melanoma cell cultures. Hum Gene Ther, 9, 2363- 2373. Heredia, A., Villena, J., Romaris, M., Molist, A. and Bassols, A. (1996) The effect of TGF- beta 1 on cell proliferation and proteoglycan production in human melanoma cells depends on the degree of cell differentiation. Cancer Lett, 109, 39-47. Hoek, K.S., Eichhoff, O.M., Schlegel, N.C., Döbbeling, U., Kobert, N., Schaerer, L., Hemmi, S., Dummer, R. and (2008) In vivo switching of human melanoma cells between proliferative and invasive states. Cancer Res, in press.

139

Chapter 5 References

Hoek, K.S., Schlegel, N.C., Brafford, P., Sucker, A., Ugurel, S., Kumar, R., Weber, B.L., Nathanson, K.L., Phillips, D.J., Herlyn, M., Schadendorf, D. and Dummer, R. (2006) Metastatic potential of melanomas defined by specific gene expression profiles with no BRAF signature. Pigment Cell Res, 19, 290-302. Iavarone, A., Garg, P., Lasorella, A., Hsu, J. and Israel, M.A. (1994) The helix-loop-helix protein Id-2 enhances cell proliferation and binds to the retinoblastoma protein. Genes Dev., 8, 1270-1284. Kamesaki, H., Nishizawa, K., Michaud, G.Y., Cossman, J. and Kiyono, T. (1998) TGF- {beta}1 Induces the Cyclin-Dependent Kinase Inhibitor p27Kip1 mRNA and Protein in Murine B Cells. J Immunol, 160, 770-777. Kang, Y., Chen, C.-R. and Massague, J. (2003) A Self-Enabling TGF[beta] Response Coupled to Stress Signaling: Smad Engages Stress Response Factor ATF3 for Id1 Repression in Epithelial Cells. Molecular Cell, 11, 915-926. Kondo, M., Cubillo, E., Tobiume, K., Shirakihara, T., Fukuda, N., Suzuki, H., Shimizu, K., Takehara, K., Cano, A., Saitoh, M. and Miyazono, K. (2004) A role for Id in the regulation of TGF-beta-induced epithelial-mesenchymal transdifferentiation. Cell Death Differ, 11, 1092-1101. Kononen, J., Bubendorf, L., Kallioniemi, A., Barlund, M., Schraml, P., Leighton, S., Torhorst, J., Mihatsch, M.J., Sauter, G. and Kallioniemi, O.P. (1998) Tissue microarrays for high-throughput molecular profiling of tumor specimens. Nat Med, 4, 844-847. Kowanetz, M., Valcourt, U., Bergstrom, R., Heldin, C.H. and Moustakas, A. (2004) Id2 and Id3 define the potency of cell proliferation and differentiation responses to transforming growth factor beta and bone morphogenetic protein. Mol Cell Biol, 24, 4241-4254. Krasagakis, K., Kruger-Krasagakes, S., Fimmel, S., Eberle, J., Tholke, D., von der Ohe, M., Mansmann, U. and Orfanos, C.E. (1999) Desensitization of melanoma cells to autocrine TGF-beta isoforms. J Cell Physiol, 178, 179-187. Lasorella, A., Iavarone, A. and Israel, M.A. (1996) Id2 specifically alters regulation of the cell cycle by tumor suppressor proteins. Mol. Cell. Biol., 16, 2570-2578. Lasorella, A., Uo, T. and Iavarone, A. (2001) Id proteins at the cross-road of development and cancer. Oncogene, 20, 8326-8333. Ling, M.T., Wang, X., Tsao, S.W. and Wong, Y.C. (2002) Down-regulation of Id-1 expression is associated with TGF beta 1-induced growth arrest in prostate epithelial cells. Biochim Biophys Acta, 1570, 145-152. Lyden, D., Young, A.Z., Zagzag, D., Yan, W., Gerald, W., O'Reilly, R., Bader, B.L., Hynes, R.O., Zhuang, Y., Manova, K. and Benezra, R. (1999) Id1 and Id3 are required for neurogenesis, angiogenesis and vascularization of tumour xenografts. Nature, 401, 670- 677. Massague, J. and Gomis, R.R. (2006) The logic of TGF[beta] signaling. FEBS Letters, 580, 2811-2820. Mori, S., Nishikawa, S.I. and Yokota, Y. (2000) Lactation defect in mice lacking the helix- loop-helix inhibitor Id2. Embo J, 19, 5772-5781. Ohtani, N., Zebedee, Z., Huot, T.J.G., Stinson, J.A., Sugimoto, M., Ohashi, Y., Sharrocks, A.D., Peters, G. and Hara, E. (2001) Opposing effects of Ets and Id proteins on p16INK4a expression during cellular senescence. 409, 1067-1070. Onken, M.D., Ehlers, J.P., Worley, L.A., Makita, J., Yokota, Y. and Harbour, J.W. (2006) Functional Gene Expression Analysis Uncovers Phenotypic Switch in Aggressive Uveal Melanomas 10.1158/0008-5472.CAN-05-4196. Cancer Res, 66, 4602-4609.

140

Chapter 5 References

Pagliuca, A., Gallo, P., De Luca, P. and Lania, L. (2000) Class A Helix-Loop-Helix Proteins Are Positive Regulators of Several Cyclin-dependent Kinase Inhibitors' Promoter Activity and Negatively Affect Cell Growth. Cancer Res, 60, 1376-1382. Perk, J., Iavarone, A. and Benezra, R. (2005) Id family of helix-loop-helix proteins in cancer. Nat Rev Cancer, 5, 603-614. Pietenpol, J.A., Stein, R.W., Moran, E., Yaciuk, P., Schlegel, R., Lyons, R.M., Pittelkow, M.R., Munger, K., Howley, P.M. and Moses, H.L. (1990) TGF-[beta]1 inhibition of c-myc transcription and growth in keratinocytes is abrogated by viral transforming proteins with pRB binding domains. Cell, 61, 777-785. Polsky, D., Young, A.Z., Busam, K.J. and Alani, R.M. (2001) The Transcriptional Repressor of p16/Ink4a, Id1, Is Up-Regulated in Early Melanomas. Cancer Res, 61, 6008-6011. Prabhu, S., Ignatova, A., Park, S. and Sun, X. (1997) Regulation of the expression of cyclin- dependent kinase inhibitor p21 by E2A and Id proteins. Mol. Cell. Biol., 17, 5888-5896. Reed, J.A., Bales, E., Xu, W., Okan, N.A., Bandyopadhyay, D. and Medrano, E.E. (2001) Cytoplasmic localization of the oncogenic protein Ski in human cutaneous melanomas in vivo: functional implications for transforming growth factor beta signaling. Cancer Res, 61, 8074-8078. Rodeck, U., Bossler, A., Graeven, U., Fox, F.E., Nowell, P.C., Knabbe, C. and Kari, C. (1994) Transforming growth factor beta production and responsiveness in normal human melanocytes and melanoma cells. Cancer Res, 54, 575-581. Rothschild, G., Zhao, X., Iavarone, A. and Lasorella, A. (2006) E Proteins and Id2 Converge on p57Kip2 To Regulate Cell Cycle in Neural Cells. Mol. Cell. Biol., 26, 4351-4361. Scandura, J.M., Boccuni, P., Massague, J. and Nimer, S.D. (2004) Transforming growth factor beta-induced cell cycle arrest of human hematopoietic cells requires p57KIP2 upregulation. Proc Natl Acad Sci U S A, 101, 15231-15236. Straume, O. and Akslen, L.A. (2005) Strong expression of ID1 protein is associated with decreased survival, increased expression of ephrin-A1/EPHA2, and reduced thrombospondin-1 in malignant melanoma. Br J Cancer, 93, 933-938. Toma, J.G., El-Bizri, H., Barnabe-Heider, F., Aloyz, R. and Miller, F.D. (2000) Evidence That Helix-Loop-Helix Proteins Collaborate with Retinoblastoma Tumor Suppressor Protein to Regulate Cortical Neurogenesis. J. Neurosci., 20, 7648-7656.

141

142

Chapter 6

6 Discussion & Conclusions..

143

Chapter 6 Discussion & Conclusions

Melanoma is recognised as being the most dangerous of skin cancers, and over the last three decades its incidence has increased more rapidly than that of any other cancer (Giblin and Thomas, 2007). Although primary tumours can be safely removed, once the cancer has metastasised patient outlook is dramatically reduced; five-year survival of metastatic melanoma patients is only 14% (Jemal et al., 2004). It is therefore of critical to elucidate the mechanisms driving melanoma progression.

The aim of this thesis was to validate a new melanoma progression model we formulated from gene expression arrays performed on three distinct sets of melanoma cultures, and to investigate the role of TGF-β-like signalling as a major factor in this newly defined model.

Clinically, melanoma progression is described in terms of increasing primary tumour thickness and extent of metastatic spread (Balch et al., 2004). A model of melanoma progression in which molecular lesions progressively accumulate enjoys universal acceptance and stands as the dominant paradigm for molecular studies of the disease (Miller and Mihm, 2006). However, this model has some limitations. For example, although accumulation of irreversible genetic lesions during tumour progression has been reported for many tumours, acquired invasion and metastasis have not been linked to recurrent mutations but rather to specific gene expression changes (reviewed in Feinberg et al., 2006).

We have proposed a new model in which melanoma cells switch between two defined gene expression signatures translating into phenotypes, which favour disease progression (Hoek et al., 2008; Hoek et al., 2006). Contrasting with the classical model, our model is not based on the accumulation of irreversible genetic lesions, but rather allows and requires reversible transcriptional changes, which are not dictated by genetic alterations. While genetic alterations, such as mutations, deletions, and amplification, likely play a role in melanoma, it should be remembered that genes affected by amplification or activating mutations are not necessarily exempt from transcriptional control.

Our model suggests that both proliferative and invasive transcriptional signatures are important in disease progression and that a single cell is capable of expressing either signature given appropriate signalling. Our model also accounts for gene expression heterogeneity in tumours and is supported by immunophenotypic variations observed in melanoma (Banerjee 144

Chapter 6 Discussion & Conclusions and Harris, 2000). This heterogeneity and the reversibility of transcription programs presented in our model may also offer an explanation for therapeutic failure against metastatic melanoma. In our model, any given targeted treatment may be successful against one phenotype of cells but leaves the other untouched and free to switch back to the targeted phenotype when treatment has ceased. We also hypothesise that primary tumours initially consist of proliferative signature cells and that by inhibiting transition to the invasive signature, we may inhibit escape of cells from the primary tumour and formation of metastases. However, since primary tumours are easily removed by surgery, and because metastases are often present at time of diagnosis, it is of primary importance to identify targets, which inhibit both groups of cells without enhancing the aggressive properties, proliferative or invasive, of one or the other.

In vitro experiments investigating the motility and proliferative properties of melanoma cells, as well as their capacity for vasculogenic mimicry, confirmed the phenotypic characterisations of proliferative and invasive that we had attributed to the two opposite transcriptional signatures (Hoek et al., 2006). Observed evasion from cytostatic responsiveness to TGF-β, an important characteristic in a number of cancers including melanoma, also supported our model; invasive signature cells were, in general, less sensitive to the growth inhibitory effect of TGF-β compared to proliferative signature cells. Reduced susceptibility to the growth inhibiting effects of TGF-β has been associated with increased invasive and metastatic properties of melanoma cells (Heredia et al., 1996; Krasagakis et al., 1999; Rodeck et al., 1994).

After a thorough literature review of genes with cohort-specific gene expression and the subsequent identification of Wnt and TGF-β signalling as drivers of the identified transcriptional signatures, we were interested in understanding the motive forces behind the differential TGF-β signalling. To our surprise, Smad2 and Smad3 proteins were phosphorylated in all melanoma cultures, irrespective of the presence of a TGF-β signature. Consistent Smad activation in melanoma cultures has been observed by others (Mauviel A. (Paris), personal communication) and suggests that the observed TGF-β signature is not dependent on Smad activation. Although the Smad-dependent pathway has long been considered as being central to TGF-β signalling, it is now recognised that TGF-β signals through alternative pathways such as the MAPK, PI3K or Rho pathways (Derynck and

145

Chapter 6 Discussion & Conclusions

Zhang, 2003). We could identify no link between the activation status of several MAPK pathways and the TGF-β signature. Analysis of phosphorylation and cellular signalling events by flow cytometry is an alternative method to western blotting and should be considered for further analyses of multiple interacting pathways as it offers simultaneous correlation of multiple active kinases involved in signalling cascades in rapid and scalable assays (Krutzik et al., 2004).

TGF-β’s cytostatic effect plays an important role in its tumour suppressing function. As mentioned above, when melanoma cells become more invasive and metastatic their responsiveness to TGF-β-induced inhibitory growth response is suppressed. Using both cell phenotypes, we investigated molecular mechanisms involved in the differential response observed between them. We identified Id2 as a TGF-β target gene differentially regulated in phenotypically opposed melanoma cells. Id proteins are positive regulators of cell proliferation and have been shown to regulate the expression of proteins involved in cell cycle regulation, such as the CDK inhibitors (Alani et al., 2001; Lyden et al., 1999; Mori et al., 2000; Ohtani et al., 2001; Prabhu et al., 1997; Rothschild et al., 2006). We showed that TGF- β represses the expression of Id2 only in TGF-β susceptible cells and that Id2 can repress TGF-β-induced upregulation of p15Ink4b in proliferative cells. As Id2 is a negative regulator of transcription by inhibiting transcriptional activation induced by bHLH, Ets and Pax factors, it would be of interest to identify its binding partners, which could help identify further regulated genes.

However, despite the role Id2 plays in modulating TGF-β susceptibility, loss of Id2 regulation is likely more a consequence and not a driver of the phenotypic switch. We have hypothesised that an environmental stimulus is responsible for triggering transcriptional change. The identification of the drivers of both signatures, marked by TGF-β and Wnt signalling, will hopefully bring us closer to the identification of this stimulus. We suggested activin A, whose expression shows cohort specificity, as the driving force of the TGF-β signature (Hoek et al., 2006). Although a tremendous overlap exists between activin and TGF-β transcriptomes (Ryu and Kern, 2003), different phenotypes are produced by engineered knockout of their respective receptor genes in mice (Gu et al., 1998; Oshima et al., 1996) Furthermore, melanoma cells are variably affected by activin A and the effects, in contrast to the TGF-β- induced growth inhibition, are not cohort specific. Also, exogenous activin A, as opposed to

146

Chapter 6 Discussion & Conclusions

TGF-β, did not suppress Mitf, whose expression is a central feature of the proliferative signature, in proliferative signature cells. The expression of activin A in invasive signature cells may also need to be considered as a consequence and not a driver of the TGF-β signature, in particular since TGF-β was shown to induce the expression of activin A in different cell types (Hubner and Werner, 1996)

TGF-β’s dual role as a tumour suppressor and tumour promoter has attracted significant attention. The multiple cytokines, receptors and interacting pathways contributing to the complexity of TGF-β signalling and the mechanisms involved in the expression of its dual role in cancer, enhance the challenge in developing therapeutics interfering with any element of this labyrinthine signalling complex. As discussed above when describing the challenge in treating cells with variably expressing transcription signatures without enhancing the tumorigenic behaviour of one group of cells over another, impairing TGF-β signalling raises similar concerns. The mechanisms involved in the change of response to TGF-β, which could also be responsible for the switch described in our model, need to be carefully elucidated to develop a safe and effective treatment.

147

Chapter 6 References

References Alani, R.M., Young, A.Z. and Shifflett, C.B. (2001) Id1 regulation of cellular senescence through transcriptional repression of p16/Ink4a. Proceedings of the National Academy of Sciences, 98, 7812-7816. Balch, C.M., Soong, S.J., Atkins, M.B., Buzaid, A.C., Cascinelli, N., Coit, D.G., Fleming, I.D., Gershenwald, J.E., Houghton, A., Jr., Kirkwood, J.M., McMasters, K.M., Mihm, M.F., Morton, D.L., Reintgen, D.S., Ross, M.I., Sober, A., Thompson, J.A. and Thompson, J.F. (2004) An evidence-based staging system for cutaneous melanoma. CA Cancer J Clin, 54, 131-149; quiz 182-134. Banerjee, S.S. and Harris, M. (2000) Morphological and immunophenotypic variations in malignant melanoma. Histopathology, 36, 387-402. Derynck, R. and Zhang, Y.E. (2003) Smad-dependent and Smad-independent pathways in TGF-beta family signalling. Nature, 425, 577-584. Feinberg, A.P., Ohlsson, R. and Henikoff, S. (2006) The epigenetic progenitor origin of human cancer. Nat Rev Genet, 7, 21-33. Giblin, A.V. and Thomas, J.M. (2007) Incidence, mortality and survival in cutaneous melanoma. J Plast Reconstr Aesthet Surg, 60, 32-40. Gu, Z., Nomura, M., Simpson, B.B., Lei, H., Feijen, A., van den Eijnden-van Raaij, J., Donahoe, P.K. and Li, E. (1998) The type I activin receptor ActRIB is required for egg cylinder organization and gastrulation in the mouse. Genes Dev, 12, 844-857. Heredia, A., Villena, J., Romaris, M., Molist, A. and Bassols, A. (1996) The effect of TGF- beta 1 on cell proliferation and proteoglycan production in human melanoma cells depends on the degree of cell differentiation. Cancer Lett, 109, 39-47. Hoek, K.S., Eichhoff, O.M., Schlegel, N.C., Döbbeling, U., Kobert, N., Schaerer, L., Hemmi, S., Dummer, R. and (2008) In vivo switching of human melanoma cells between proliferative and invasive states. Cancer Res, in press. Hoek, K.S., Schlegel, N.C., Brafford, P., Sucker, A., Ugurel, S., Kumar, R., Weber, B.L., Nathanson, K.L., Phillips, D.J., Herlyn, M., Schadendorf, D. and Dummer, R. (2006) Metastatic potential of melanomas defined by specific gene expression profiles with no BRAF signature. Pigment Cell Res, 19, 290-302. Hubner, G. and Werner, S. (1996) Serum Growth Factors and Proinflammatory Cytokines Are Potent Inducers of Activin Expression in Cultured Fibroblasts and Keratinocytes. Experimental Cell Research, 228, 106-113. Jemal, A., Tiwari, R.C., Murray, T., Ghafoor, A., Samuels, A., Ward, E., Feuer, E.J. and Thun, M.J. (2004) Cancer statistics, 2004. CA Cancer J Clin, 54, 8-29. Krasagakis, K., Kruger-Krasagakes, S., Fimmel, S., Eberle, J., Tholke, D., von der Ohe, M., Mansmann, U. and Orfanos, C.E. (1999) Desensitization of melanoma cells to autocrine TGF-beta isoforms. J Cell Physiol, 178, 179-187. Krutzik, P.O., Irish, J.M., Nolan, G.P. and Perez, O.D. (2004) Analysis of protein phosphorylation and cellular signaling events by flow cytometry: techniques and clinical applications. Clinical Immunology, 110, 206-221. Lyden, D., Young, A.Z., Zagzag, D., Yan, W., Gerald, W., O'Reilly, R., Bader, B.L., Hynes, R.O., Zhuang, Y., Manova, K. and Benezra, R. (1999) Id1 and Id3 are required for neurogenesis, angiogenesis and vascularization of tumour xenografts. Nature, 401, 670- 677. Miller, A.J. and Mihm, M.C., Jr. (2006) Melanoma. N Engl J Med, 355, 51-65. Mori, S., Nishikawa, S.I. and Yokota, Y. (2000) Lactation defect in mice lacking the helix- loop-helix inhibitor Id2. Embo J, 19, 5772-5781.

148

Chapter 6 References

Ohtani, N., Zebedee, Z., Huot, T.J.G., Stinson, J.A., Sugimoto, M., Ohashi, Y., Sharrocks, A.D., Peters, G. and Hara, E. (2001) Opposing effects of Ets and Id proteins on p16INK4a expression during cellular senescence. 409, 1067-1070. Oshima, M., Oshima, H. and Taketo, M.M. (1996) TGF-beta receptor type II deficiency results in defects of yolk sac hematopoiesis and vasculogenesis. Dev Biol, 179, 297-302. Prabhu, S., Ignatova, A., Park, S. and Sun, X. (1997) Regulation of the expression of cyclin- dependent kinase inhibitor p21 by E2A and Id proteins. Mol. Cell. Biol., 17, 5888-5896. Rodeck, U., Bossler, A., Graeven, U., Fox, F.E., Nowell, P.C., Knabbe, C. and Kari, C. (1994) Transforming growth factor beta production and responsiveness in normal human melanocytes and melanoma cells. Cancer Res, 54, 575-581. Rothschild, G., Zhao, X., Iavarone, A. and Lasorella, A. (2006) E Proteins and Id2 Converge on p57Kip2 To Regulate Cell Cycle in Neural Cells. Mol. Cell. Biol., 26, 4351-4361. Ryu, B. and Kern, S.E. (2003) The essential similarity of TGFbeta and activin receptor transcriptional responses in cancer cells. Cancer Biol Ther, 2, 164-170.

149

150

Appendix A

Supplementary Data to Chapter 2 Metastatic potential of melanomas defined by specific gene expression profiles with no BRAF signature

151

Zürich data set Melanoma BRAF NRAS cohort Mannheim data set cultures status status Melanoma BRAF NRAS cohort M980513 V600E wt A cultures status status M010817 wt Q61R A Ma-Mel 12 unknown unknown A M000921 V600E wt A Ma-Mel 15 wt wt A M990514 wt Q61K B Ma-Mel 27 wt G12D A M000907 wt wt B Ma-Mel 30 D594N unknown A M000216 V600E wt B Ma-Mel 35 wt wt A M991121 wt wt C Ma-Mel 36 V600E wt A M990115 wt wt C Ma-Mel 40 wt wt A M010718 wt Q61K C Ma-Mel 46 V600E wt A M010322 V600E wt C Ma-Mel 48a G469R wt A M010308 wt Q61K C Ma-Mel 52 V600K wt A M010119 wt wt C Ma-Mel 57 V600E wt A Ma-Mel 59a V600E wt A Philadelphia data set Ma-Mel 63a V600E Q61R A Melanoma BRAF Ma-Mel 67 V600K wt A cohort cultures status Ma-Mel 69 V600K wt A WM1321 WT A Ma-Mel 76 wt wt A Q61K WM1382 unknown A Ma-Mel 79b wt A WM1617 V599E A hom. WM164 unknown A Ma-Mel 80a V600E wt A WM1727A unknown A Ma-Mel 82 wt G12D A WM1799 unknown A Ma-Mel 39a wt wt B WM1819 unknown A Ma-Mel 60 wt Q61K B WM239A unknown A Ma-Mel 61a V600E wt B WM3268V unknown A Ma-Mel 71 wt wt B WM39 V599E A Ma-Mel 73a wt wt B WM51 unknown A Ma-Mel 84 unknown unknown B WM88 unknown A Ma-Mel 86b V600E wt B WM983A V599E A KNUD V600E wt C WM983B V599E A Ma-Mel 06 V600E wt C WM983C unknown A Ma-Mel 07 V600E wt C WM115 unknown B Ma-Mel 25 wt wt C WM3248 unknown B Ma-Mel 26a wt Q61R C WM3451 unknown B Ma-Mel 42a wt wt C WM35 unknown B Ma-Mel 45a V600E wt C 1205 Lu V599E C Ma-Mel 50b unknown unknown C Q61K, WM1346 unknown C Ma-Mel 53a wt C R68T WM1361A WT C Ma-Mel 54a V600E wt C WM1361B V599E C Ma-Mel 58 wt wt C WM1366 WT C Ma-Mel 65 wt Q61K C WM278 V599E C Ma-Mel 66b unknown unknown C WM3211 WT C Q61R Ma-Mel 74 wt C WM793B V599E C het. WM852 unknown C Ma-Mel 85 V600E wt C WM858 V599E C Ma-Mel 86a V600E wt C

152

Supplementary Data 2A

223 genes which have cohort-specific expression patterns in cultures of melanoma.

Probe Set ID Gene Title Gene Symbol 212387_at ------208161_s_at ATP-binding cassette, sub-family C (CFTR/MRP), member 3 ABCC3 43427_at acetyl-Coenzyme A carboxylase beta ACACB 204638_at acid phosphatase 5, tartrate resistant ACP5 202952_s_at ADAM metallopeptidase domain 12 (meltrin alpha) ADAM12 213790_at 222108_at adhesion molecule with Ig-like domain 2 AMIGO2 203002_at angiomotin like 2 AMOTL2 221009_s_at angiopoietin-like 4 ANGPTL4 205082_s_at aldehyde oxidase 1 AOX1 203299_s_at adaptor-related protein complex 1, sigma 2 subunit AP1S2 203300_x_at 205265_s_at aortic preferentially expressed gene 1 APEG1 204416_x_at apolipoprotein C-I APOC1 213553_x_at 203381_s_at 203382_s_at apolipoprotein E APOE 212884_x_at 210980_s_at 213702_x_at N-acylsphingosine amidohydrolase (acid ceramidase) 1 ASAH1 213902_at 205673_s_at ankyrin repeat and SOCS box-containing 9 ASB9 220948_s_at ATPase, Na+/K+ transporting, alpha 1 polypeptide ATP1A1 202685_s_at AXL receptor tyrosine kinase AXL 202686_s_at 217867_x_at beta-site APP-cleaving enzyme 2 BACE2 220488_s_at breast carcinoma amplified sequence 3 BCAS3 205681_at BCL2-related protein A1 BCL2A1 213905_x_at biglycan /// serologically defined colon cancer antigen 33 BGN /// SDCCAG33 210538_s_at baculoviral IAP repeat-containing 3 BIRC3 220451_s_at baculoviral IAP repeat-containing 7 (livin) BIRC7 221534_at basophilic leukemia expressed protein BLES03 Bles03 213246_at chromosome 14 open reading frame 109 C14orf109 217118_s_at chromosome 22 open reading frame 9 C22orf9 210944_s_at 211890_x_at calpain 3, (p94) CAPN3 214475_x_at 206837_at cartilage paired-class homeoprotein 1 CART1 204306_s_at CD151 antigen CD151 204726_at cadherin 13, H-cadherin (heart) CDH13 203440_at cadherin 2, type 1, N-cadherin (neuronal) CDH2 203441_s_at 204252_at cyclin-dependent kinase 2 CDK2 204995_at cyclin-dependent kinase 5, regulatory subunit 1 (p35) CDK5R1 204029_at cadherin, EGF LAG seven-pass G-type receptor 2 CELSR2 204266_s_at choline kinase alpha CHKA 204591_at cell adhesion molecule with homology to L1CAM CHL1 207144_s_at Cbp/p300-interacting transactivator, with Glu/Asp-rich carboxy-terminal domain, 1 CITED1 200998_s_at cytoskeleton-associated protein 4 CKAP4 201734_at Chloride channel 3 CLCN3 209235_at chloride channel 7 CLCN7 211343_s_at collagen, type XIII, alpha 1 COL13A1 203325_s_at 212488_at collagen, type V, alpha 1 COL5A1 212489_at

153

202110_at cytochrome c oxidase subunit VIIb COX7B 206256_at carboxypeptidase N, polypeptide 1, 50kD CPN1 202551_s_at cysteine rich transmembrane BMP regulator 1 (chordin-like) CRIM1 221541_at cysteine-rich secretory protein LCCL domain containing 2 CRISPLD2 209716_at colony stimulating factor 1 (macrophage) CSF1 209101_at connective tissue growth factor CTGF 204925_at cystinosis, nephropathic CTNS 209774_x_at chemokine (C-X-C motif) ligand 2 CXCL2 210764_s_at cysteine-rich, angiogenic inducer, 61 CYR61 209569_x_at DNA segment on chromosome 4 (unique) 234 expressed sequence D4S234E 209570_s_at 203139_at death-associated protein kinase 1 DAPK1 205337_at 205338_s_at dopachrome tautomerase DCT 216512_s_at 216513_at 219113_x_at dehydrogenase/reductase (SDR family) member 10 DHRS10 221031_s_at hypothetical protein DKFZp434F0318 DKFZP434F0318 204602_at dickkopf homolog 1 (Xenopus laevis) DKK1 202196_s_at dickkopf homolog 3 (Xenopus laevis) DKK3 214247_s_at 212838_at dynamin binding protein DNMBP 219648_at dilute suppressor DSU 204271_s_at endothelin receptor type B EDNRB 206701_x_at 206580_s_at EGF-containing fibulin-like extracellular matrix protein 2 EFEMP2 201983_s_at epidermal growth factor receptor EGFR 201984_s_at 221870_at EH-domain containing 2 EHD2 45297_at 222294_s_at Eukaryotic translation initiation factor 2C, 2 EIF2C2 214446_at elongation factor, RNA polymerase II, 2 ELL2 202454_s_at v-erb-b2 erythroblastic leukemia viral oncogene homolog 3 (avian) ERBB3 204363_at coagulation factor III (thromboplastin, tissue factor) F3 203206_at family with sequence similarity 53, member B FAM53B 202765_s_at fibrillin 1 (Marfan syndrome) FBN1 202766_s_at 201178_at F-box protein 7 FBXO7 204421_s_at fibroblast growth factor 2 (basic) FGF2 204422_s_at 219620_x_at hypothetical protein FLJ20245 FLJ20245 208614_s_at filamin B, beta (actin binding protein 278) FLNB 207876_s_at filamin C, gamma (actin binding protein 280) FLNC 218881_s_at FOS-like antigen 2 FOSL2 206307_s_at forkhead box D1 FOXD1 213056_at FERM domain containing 4B FRMD4B 204948_s_at follistatin FST 210220_at frizzled homolog 2 (Drosophila) FZD2 212256_at polypeptide N-acetylgalactosaminyltransferase 10 (GalNAc-T10) GALNT10 203397_s_at polypeptide N-acetylgalactosaminyltransferase 3 (GalNAc-T3) GALNT3 207116_s_at glyceraldehyde-3-phosphate dehydrogenase, spermatogenic GAPDHS 1598_g_at growth arrest-specific 6 GAS6 202177_at 217167_x_at glycerol kinase GK 207034_s_at GLI-Kruppel family member GLI2 GLI2 221447_s_at glycosyltransferase 8 domain containing 2 GLT8D2 35820_at GM2 ganglioside activator GM2A 204187_at guanosine monophosphate reductase /// guanosine monophosphate reductase GMPR 209167_at 209168_at glycoprotein M6B GPM6B 209169_at

154

209170_s_at 201141_at glycoprotein (transmembrane) nmb GPNMB 206673_at putative G protein coupled receptor GPR 206696_at G protein-coupled receptor 143 GPR143 206582_s_at G protein-coupled receptor 56 GPR56 212070_at 203632_s_at G protein-coupled receptor, family C, group 5, member B GPRC5B 205862_at GREB1 protein GREB1 210963_s_at glycogenin 2 GYG2 210964_s_at 212822_at HEG homolog 1 (zebrafish) HEG1 213069_at 204670_x_at 209312_x_at major histocompatibility complex, class II, DR beta 1 HLA-DRB1 215193_x_at 222020_s_at neurotrimin HNT 203914_x_at hydroxyprostaglandin dehydrogenase 15-(NAD) HPGD 54037_at Hermansky-Pudlak syndrome 4 HPS4 219985_at heparan sulfate (glucosamine) 3-O-sulfotransferase 3A1 HS3ST3A1 218971_s_at HSPC049 protein HSPC049 202070_s_at isocitrate dehydrogenase 3 (NAD+) alpha IDH3A 203851_at insulin-like growth factor binding protein 6 IGFBP6 205207_at interleukin 6 (interferon, beta 2) IL6 202859_x_at interleukin 8 IL8 211506_s_at 210511_s_at inhibin, beta A (activin A, activin AB alpha polypeptide) INHBA 205376_at inositol polyphosphate-4-phosphatase, type II, 105kDa INPP4B 204562_at interferon regulatory factor 4 IRF4 201389_at integrin, alpha 5 (fibronectin receptor, alpha polypeptide) ITGA5 203723_at inositol 1,4,5-trisphosphate 3-kinase B ITPKB 201362_at 201363_s_at influenza virus NS1A binding protein IVNS1ABP 206245_s_at 221584_s_at potassium large conductance calcium-activated channel, M alpha 1 KCNMA1 204401_at potassium intermediate/small conductance calcium-activated channel N 4 KCNN4 212456_at KIAA0664 protein KIAA0664 213478_at kazrin KIAA1026 212942_s_at KIAA1199 KIAA1199 218651_s_at La ribonucleoprotein domain family, member 6 LARP6 208949_s_at lectin, galactoside-binding, soluble, 3 (galectin 3) LGALS3 221880_s_at hypothetical gene supported by AK075564; BC060873 LOC400451 51158_at 209679_s_at small trans-membrane and glycosylated protein LOC57228 204298_s_at 213640_s_at lysyl oxidase LOX 215446_s_at 202997_s_at lysyl oxidase-like 2 LOXL2 202998_s_at 219042_at leucine zipper, putative tumor suppressor 1 LZTS1 47550_at 221760_at Mannosidase, alpha, class 1A, member 1 MAN1A1 212233_at Microtubule-associated protein 1B MAP1B 207323_s_at myelin basic protein MBP 209072_at 221620_s_at hypothetical protein MGC4825 MGC4825 211026_s_at monoglyceride lipase MGLL 212472_at microtubule associated monoxygenase, calponin and LIM domain containing 2 MICAL2 212473_s_at 207233_s_at microphthalmia-associated transcription factor MITF 206426_at melan-A MLANA 206427_s_at 218211_s_at melanophilin MLPH

155

205413_at metallophosphoesterase domain containing 2 MPPED2 37408_at mannose receptor, C type 2 MRC2 222153_at myelin expression factor 2 MYEF2 204527_at myosin VA (heavy polypeptide 12, myoxin) MYO5A 204114_at nidogen 2 (osteonidogen) NID2 202238_s_at nicotinamide N-methyltransferase NNMT 210510_s_at neuropilin 1 NRP1 212298_at 204589_at NUAK family, SNF1-like kinase, 1 NUAK1 212605_s_at nudix (nucleoside diphosphate linked moiety X)-type motif 3 NUDT3 207303_at phosphodiesterase 1C, calmodulin-dependent 70kDa PDE1C 214582_at phosphodiesterase 3B, cGMP-inhibited PDE3B 218718_at platelet derived growth factor C PDGFC 202273_at platelet-derived growth factor receptor, beta polypeptide PDGFRB 210976_s_at phosphofructokinase, muscle PFKM 204604_at PFTAIRE protein kinase 1 PFTK1 213638_at phosphatase and actin regulator 1 PHACTR1 207938_at peptidase inhibitor 15 PI15 207469_s_at pirin (iron-binding nuclear protein) PIR 219584_at phospholipase A1 member A PLA1A 207943_x_at pleiomorphic adenoma gene-like 1 PLAGL1 209318_x_at 214866_at plasminogen activator, urokinase receptor PLAUR 210198_s_at proteolipid protein 1 PLP1 206470_at plexin C1 PLXNC1 213241_at 201578_at podocalyxin-like PODXL 207808_s_at protein S (alpha) PROS1 204262_s_at presenilin 2 (Alzheimer disease 4) PSEN2 207177_at prostaglandin F receptor (FP) PTGFR 206157_at pentraxin-related gene, rapidly induced by IL-1 beta PTX3 218931_at RAB17, member RAS oncogene family RAB17 209514_s_at 209515_s_at RAB27A, member RAS oncogene family RAB27A 210951_x_at 219412_at RAB38, member RAS oncogene family RAB38 206617_s_at renin binding protein RENBP 210138_at regulator of G-protein signalling 20 RGS20 221127_s_at regulated in glioma RIG 214663_at receptor interacting protein kinase 5 RIPK5 220425_x_at ropporin, rhophilin associated protein 1B ROPN1B 204633_s_at ribosomal protein S6 kinase, 90kDa, polypeptide 5 RPS6KA5 204635_at 221523_s_at Ras-related GTP binding D RRAGD 221524_s_at 212647_at related RAS viral (r-ras) oncogene homolog RRAS 205334_at S100 calcium binding protein A1 S100A1 204268_at S100 calcium binding protein A2 S100A2 218854_at squamous cell carcinoma antigen recognized by T cells 2 SART2 200958_s_at syndecan binding protein (syntenin) SDCBP 202627_s_at serpin peptidase inhibitor, clade E 1 SERPINE1 202628_s_at 209848_s_at silver homolog (mouse) SILV 203123_s_at solute carrier family 11, member 2 SLC11A2 203124_s_at 209610_s_at 212810_s_at solute carrier family 1 (glutamate/neutral amino acid transporter), member 4 SLC1A4 212811_x_at 220245_at solute carrier family 45, member 2 SLC45A2 221644_s_at 216092_s_at solute carrier family 7 (cationic amino acid transporter, y+ system), member 8 SLC7A8

156

209897_s_at slit homolog 2 (Drosophila) SLIT2 204466_s_at 204467_s_at synuclein, alpha (non A4 component of amyloid precursor) SNCA 207827_x_at 211546_x_at 208127_s_at suppressor of cytokine signaling 5 SOCS5 209647_s_at 209843_s_at SRY (sex determining region Y)-box 10 SOX10 38918_at SRY (sex determining region Y)-box 13 SOX13 202935_s_at SRY (sex determining region Y)-box 9 SOX9 210942_s_at ST3 beta-galactoside alpha-2,3-sialyltransferase 6 ST3GAL6 213355_at 203438_at stanniocalcin 2 STC2 203439_s_at 209238_at syntaxin 3A STX3A 205547_s_at transgelin TAGLN 222116_s_at TBC1 domain family, member 16 TBC1D16 205993_s_at T-box 2 TBX2 203753_at 212382_at transcription factor 4 TCF4 212385_at 222146_s_at 212758_s_at transcription factor 8 (represses interleukin 2 expression) TCF8 213361_at tudor domain containing 7 TDRD7 204653_at transcription factor AP-2 alpha (activating enhancer binding protein 2 alpha) TFAP2A 201506_at transforming growth factor, beta-induced, 68kDa TGFBI 201107_s_at 201108_s_at thrombospondin 1 THBS1 201109_s_at 201110_s_at 208850_s_at 208851_s_at Thy-1 cell surface antigen THY1 213869_x_at 201666_at TIMP metallopeptidase inhibitor 1 TIMP1 216997_x_at transducin-like enhancer of split 4 (E(sp1) homolog, Drosophila) TLE4 204137_at transmembrane 7 superfamily member 1 (upregulated in kidney) TM7SF1 213096_at transmembrane and coiled-coil domain family 2 TMCC2 204932_at tumor necrosis factor receptor superfamily, member 11b (osteoprotegerin) TNFRSF11B 204933_s_at 209354_at tumor necrosis factor receptor superfamily, member 14 TNFRSF14 207643_s_at tumor necrosis factor receptor superfamily, member 1A TNFRSF1A 203476_at trophoblast glycoprotein TPBG 206116_s_at 206117_at tropomyosin 1 (alpha) TPM1 210986_s_at 210987_x_at 204083_s_at tropomyosin 2 (beta) TPM2 202369_s_at translocation associated membrane protein 2 TRAM2 213293_s_at tripartite motif-containing 22 TRIM22 206479_at transient receptor potential cation channel, subfamily M, member 1 TRPM1 206630_at tyrosinase (oculocutaneous albinism IA) TYR 205694_at tyrosinase-related protein 1 TYRP1 209946_at vascular endothelial growth factor C VEGFC 221532_s_at WD repeat domain 61 WDR61 205990_s_at wingless-type MMTV integration site family, member 5A WNT5A 221029_s_at wingless-type MMTV integration site family, member 5B WNT5B 217781_s_at zinc finger protein 106 homolog (mouse) ZFP106

157

Supplementary Data 2B

Genes with co-regulated cohort specific expression patterns (motifs). Motif 1 (the neural crest signature) Probe Set ID Gene Title Gene Symbol 203300_x_at adaptor-related protein complex 1, sigma 2 subunit AP1S2 204416_x_at apolipoprotein C-I APOC1 203381_s_at apolipoprotein E APOE 203382_s_at 205681_at BCL2-related protein A1 BCL2A1 211890_x_at 214475_x_at calpain 3, (p94) CAPN3 210944_s_at 206837_at cartilage paired-class homeoprotein 1 CART1 204995_at cyclin-dependent kinase 5, regulatory subunit 1 (p35) CDK5R1 207144_s_at Cbp/p300-interacting transactivator with Glu/Asp-rich carboxy-terminal domain 1 CITED1 209235_at chloride channel 7 CLCN7 209570_s_at DNA segment on chromosome 4 (unique) 234 expressed sequence D4S234E 216513_at 216512_s_at dopachrome tautomerase DCT 205337_at 205338_s_at 221031_s_at hypothetical protein DKFZp434F0318 DKFZP434F0318 219648_at likely ortholog of mouse dilute suppressor DSU 204271_s_at endothelin receptor type B EDNRB 206701_x_at 202454_s_at v-erb-b2 erythroblastic leukemia viral oncogene homolog 3 (avian) ERBB3 203397_s_at polypeptide N-acetylgalactosaminyltransferase 3 (GalNAc-T3) GALNT3 217167_x_at glycerol kinase GK 204187_at guanosine monophosphate reductase GMPR 209169_at 209168_at glycoprotein M6B GPM6B 209170_s_at 209167_at 206696_at G protein-coupled receptor 143 GPR143 206582_s_at G protein-coupled receptor 56 GPR56 212070_at 203632_s_at G protein-coupled receptor, family C, group 5, member B GPRC5B 205862_at GREB1 protein GREB1 210964_s_at glycogenin 2 GYG2 205376_at inositol polyphosphate-4-phosphatase, type II, 105kDa INPP4B 204562_at interferon regulatory factor 4 IRF4 203723_at inositol 1,4,5-trisphosphate 3-kinase B ITPKB 47550_at leucine zipper, putative tumor suppressor 1 LZTS1 219042_at 221644_s_at membrane associated transporter MATP 220245_at 207323_s_at myelin basic protein MBP 209072_at 207233_s_at microphthalmia-associated transcription factor MITF 206426_at melan-A MLANA 206427_s_at 213638_at phosphatase and actin regulator 1 PHACTR1 207469_s_at pirin (iron-binding nuclear protein) PIR 219584_at phospholipase A1 member A PLA1A 210198_s_at proteolipid protein 1 PLP1 206470_at plexin C1 PLXNC1 213241_at 219412_at RAB38, member RAS oncogene family RAB38 206617_s_at renin binding protein RENBP 210138_at regulator of G-protein signalling 20 RGS20 221524_s_at Ras-related GTP binding D RRAGD

158

205334_at S100 calcium binding protein A1 S100A1 209848_s_at silver homolog (mouse) SILV 211546_x_at 207827_x_at synuclein, alpha (non A4 component of amyloid precursor) SNCA 204466_s_at 209843_s_at SRY (sex determining region Y)-box 10 SOX10 213355_at ST3 beta-galactoside alpha-2,3-sialyltransferase 6 ST3GAL6 210942_s_at 204653_at transcription factor AP-2 alpha (activating enhancer binding protein 2 alpha) TFAP2A 209354_at tumor necrosis factor receptor superfamily, member 14 TNFRSF14 206479_at transient receptor potential cation channel, subfamily M, member 1 TRPM1 206630_at tyrosinase (oculocutaneous albinism IA) TYR 205694_at tyrosinase-related protein 1 TYRP1

Motif 2 (the TGF-β-like signal signature) Probe Set ID Gene Title Gene Symbol 208161_s_at ATP-binding cassette, sub-family C (CFTR/MRP), member 3 ABCC3 213790_at a disintegrin and metalloproteinase domain 12 (meltrin alpha) ADAM12 202952_s_at 222108_at amphoterin induced gene 2 AMIGO2 204589_at AMP-activated protein kinase family member 5 ARK5 202686_s_at AXL receptor tyrosine kinase AXL 213905_x_at biglycan /// serologically defined colon cancer antigen 33 BGN /// SDCCAG33 204726_at cadherin 13, H-cadherin (heart) CDH13 203440_at cadherin 2, type 1, N-cadherin (neuronal) CDH2 212489_at 203325_s_at collagen, type V, alpha 1 COL5A1 212488_at 209101_at connective tissue growth factor CTGF 209774_x_at chemokine (C-X-C motif) ligand 2 CXCL2 210764_s_at cysteine-rich, angiogenic inducer, 61 CYR61 204602_at dickkopf homolog 1 (Xenopus laevis) DKK1 214247_s_at dickkopf homolog 3 (Xenopus laevis) DKK3 201983_s_at epidermal growth factor receptor EGFR 45297_at EH-domain containing 2 EHD2 221870_at 214446_at elongation factor, RNA polymerase II, 2 ELL2 204363_at coagulation factor III (thromboplastin, tissue factor) F3 202765_s_at fibrillin 1 (Marfan syndrome) FBN1 202766_s_at 204421_s_at fibroblast growth factor 2 (basic) FGF2 204422_s_at 206307_s_at forkhead box D1 FOXD1 222020_s_at neurotrimin HNT 219985_at heparan sulfate (glucosamine) 3-O-sulfotransferase 3A1 HS3ST3A1 205207_at interleukin 6 (interferon, beta 2) IL6 211506_s_at interleukin 8 IL8 202859_x_at 210511_s_at inhibin, beta A (activin A, activin AB alpha polypeptide) INHBA 212942_s_at KIAA1199 KIAA1199 221541_at LCCL domain containing cysteine-rich secretory protein 2 LCRISP2 213640_s_at 215446_s_at lysyl oxidase LOX 204298_s_at 202997_s_at lysyl oxidase-like 2 LOXL2 202998_s_at 212472_at flavoprotein oxidoreductase MICAL2 MICAL2 212473_s_at 204114_at nidogen 2 (osteonidogen) NID2 202238_s_at nicotinamide N-methyltransferase NNMT 210510_s_at neuropilin 1 NRP1 212298_at

159

218718_at platelet derived growth factor C PDGFC 202273_at platelet-derived growth factor receptor, beta polypeptide PDGFRB 209318_x_at pleiomorphic adenoma gene-like 1 PLAGL1 201578_at podocalyxin-like PODXL 207177_at prostaglandin F receptor (FP) PTGFR 206157_at pentaxin-related gene, rapidly induced by IL-1 beta PTX3 218854_at squamous cell carcinoma antigen recognized by T cells 2 SART2 202628_s_at serine (or cysteine) proteinase inhibitor, clade E 1 SERPINE1 202627_s_at 203438_at stanniocalcin 2 STC2 205547_s_at transgelin TAGLN 222146_s_at transcription factor 4 TCF4 212387_at 201506_at transforming growth factor, beta-induced, 68kDa TGFBI 201107_s_at 201110_s_at thrombospondin 1 THBS1 201109_s_at 201108_s_at 213869_x_at 208850_s_at Thy-1 cell surface antigen /// Thy-1 co-transcribed THY1 /// LOC94105 208851_s_at 204932_at tumor necrosis factor receptor superfamily, member 11b (osteoprotegerin) TNFRSF11B 204933_s_at 206117_at 206116_s_at tropomyosin 1 (alpha) TPM1 210986_s_at 210987_x_at 204083_s_at tropomyosin 2 (beta) TPM2 209946_at vascular endothelial growth factor C VEGFC 205990_s_at wingless-type MMTV integration site family, member 5A WNT5A 221029_s_at wingless-type MMTV integration site family, member 5B WNT5B

160

Supplementary Data 3a.

Genes described by others to be downregulated in melanomas with greater metastatic potential and are significantly downregulated from cohort A to cohort C (P<.05). Symbol Gene Affy ID Folda P

(Vlaykova et al., BCL2 B-cell CLL/lymphoma 2 203685_s_at 2.5 0.010 2002)

244035_at 2.8 0.038 (Andersen et al., CDH1 E-cadherin 201131_s_at 3.4 0.019 2004) (Straume and CDKN2A cyclin-dependent kinase inhibitor 2A 211156_at 2.1 0.004 Akslen, 1997) 234583_at 8.2 0.003 CHL1 c cell adhesion molecule with homology to L1CAM (Seftor et al., 2002) 204591_at 18 0.004 216513_at 23 0.002

205338_s_at 37 0.001 (Takeuchi et al., DCT c dopachrome tautomerase 2003) 216512_s_at 95 0.001

205337_at >100 0.002 204271_s_at 22 0.001

EDNRB c endothelin receptor type B 204273_at 24 0.001 (Smith et al., 2002)

206701_x_at 34 0.001 EEF1A2 eukaryotic translation elongation factor 1 alpha 2 204540_at 2.4 0.023 (de Wit et al., 2002) 209168_at 16 0.001

209169_at 52 0.001 GPM6B c glycoprotein M6B (Seftor et al., 2002) 209170_s_at 77 0.001

209167_at 92 0.001 202637_s_at 3.6 0.004 (Anastassiou et al., ICAM1 intercellular adhesion molecule 1 2000) 215485_s_at 2.5 0.005 240555_at 12 0.001

MITF c microphthalmia-associated transcription factor 207233_s_at 19 0.001 (Salti et al., 2000)

206426_at 27 0.001 206426_at 45 0.003 MLANA c melan-A (Berset et al., 2001) 206427_s_at 77 0.004 (Mao et al., 2001) NME1 non-metastatic cells 1, protein (NM23A) expressed in 201577_at 2.5 0.002 (Sarris et al., 2004) PCNA proliferating cell nuclear antigen 201202_at 2.9 0.012 (Evans et al., 1992) (Juergensen et al., S100A1 c S100 calcium binding protein A1 205334_at >100 0.001 2001) (Korabiowska et al., 1994) 220109_at 5.0 0.029 TF transferrin (Seftor et al., 2002) 203400_s_at 18 0.022 210669_at 4.2 0.019 (Karjalainen et al., TFAP2A c transcription factor AP-2 alpha 2000a)

161

204654_s_at 9.1 0.001

204653_at 20 0.001 TOP2A topoisomerase (DNA) II alpha 170kDa 201291_s_at 7.2 0.049 (Mu et al., 2000) 240386_at 14 0.006

206479_at 32 0.001 transient receptor potential cation channel, subfamily M, (Duncan et al., TRPM1 c member 1 2001) 237070_at 95 0.001

237069_s_at >100 0.001 (Takeuchi et al., TYR c tyrosinase 206630_at 40 0.001 2003) aCalculated using normalized Zürich data. bTwo tailed T-test statistic , assuming variance is equal. c Also present in the 223 gene intersection of the Zürich, Philadelphia and Mannheim cohort-specific expression data sets.

162

Genes described by others to be upregulated in melanomas with greater metastatic potential and are significantly upregulated from cohort A to cohort C (P<.05). Symbol Gene Affy ID Folda P

(Sanders et al., CDH2 c N-cadherin 237305_at 4.9 0.015 1999)

203441_s_at 3.9 0.001 (Maniotis et al., CTGF c connective tissue growth factor 209101_at 6.6 0.033 1999) (Seftor et al., 2002) 201983_s_at 42 0.001

232541_at 22 0.002

201984_s_at 11 0.002 (Hurks et al., EGFR c epidermal growth factor receptor 2000) 224999_at 7.8 0.001

233044_at 3.2 0.013

232120_at 1.8 0.034 200878_at 2.5 0.013 (Giatromanolaki et EPAS1 endothelial PAS domain protein 1 al., 2003) 235963_at 2.5 0.005 (Wang et al., FBLN5 fibulin 5 203088_at 49 0.001 2004) (Giatromanolaki et HIF1A hypoxia-inducible factor 1, alpha subunit 238869_at 2.1 0.016 al., 2003) 211911_x_at 2.2 0.026 HLA-B major histocompatibility complex, class I, B (Real et al., 1999) 209140_x_at 1.6 0.038 202859_x_at 18 0.001 (Nurnberg et al., IL8 c interleukin 8 1999) (Rofstad 211506_s_at 11 0.004 and Halsor, 2000) 215878_at 2.8 0.023 (Nikkola et al., ITGB1 integrin, beta 1 2004) 211945_s_at 1.6 0.047 204679_at 6.9 0.001 (Seftor et al., KCNK1 potassium channel K1 2002) 204678_s_at 5.6 0.018 (Fuchs et al., KRT18 keratin 18 201596_x_at 9.3 0.017 1992) (Wang et al., 2004) 209016_s_at >100 0.001 (Seftor et al., KRT7 keratin 7 2002) 214031_s_at 6.8 0.015 223690_at 22 0.001 (Seftor et al., LTBP2 latent TGFβ binding protein 2 2002) 204682_at 13 0.001 203434_s_at 19 0.010 (Bilalovic et al., MME membrane metallo-endopeptidase 2004) 203435_s_at 8.3 0.012 (Nikkola et al., MMP1 matrix metalloproteinase 1 204475_at 57 0.003 2002; Seftor et al., 2001) (Seftor et al., MMP2 matrix metalloproteinase 2 201069_at 11 0.003 2002) (Vaisanen et al., 1996) (Nikkola et al., MMP3 matrix metalloproteinase 3 205828_at 14 0.007 2002) 212298_at >100 0.001 (Straume and NRP1 c neuropilin 1 Akslen, 2003)

163

210510_s_at 33 0.001

239519_at 17 0.001

233626_at 5.9 0.001

210615_at 3.6 0.034

233701_at 2.8 0.015

242677_at 2.8 0.007 (Seftor et al., PAX8 paired box gene 8 227474_at 6.7 0.021 2002) 211668_s_at 4.5 0.041 (Maniotis et al., PLAU plasminogen activator, urokinase 1999; Seftor et al., 205479_s_at 3.1 0.010 2002) 202619_s_at 8.5 0.010 (Goldberg et al., PLOD2 procollagen-lysine, 2-oxoglutarate 5-dioxygenase 2 2003) 202620_s_at 13 0.018 39729_at 99 0.001

(de Wit et al., PRDX2 peroxiredoxin 2 201006_at 7.6 0.001 2002)

211658_at 7.4 0.001 (Figueiredo et al., PTGS2 prostaglandin-endoperoxide synthase 2 2204748_at 46 0.003 2003) (Seftor et al., RIG c regulated in glioma 221127_s_at 15 0.002 2002) (Massi et al., SPARC osteonectin 212667_at 3.4 0.006 1999) 201107_s_at >100 0.001

201109_s_at 67 0.001

201110_s_at 67 0.001

(Straume and THBS1 c thrombospondin 1 239336_at 41 0.003 Akslen, 2003)

201108_s_at 28 0.001

235086_at 10 0.001

215775_at 1.8 0.004 THBS2 thrombospondin 2 203083_at 23 0.001 (Kunz et al., 2002) TMSB10 thymosin, beta 10 217733_s_at 1.6 0.05 (Liu et al., 2004) (Galvani et al., TNFRSF1A c tumor necrosis factor receptor superfamily 1A 207643_s_at 2.8 0.005 2000) (Bayer-Garner et VEGF vascular endothelial growth factor 212171_x_at 3.2 0.033 al., 1999) (Osella- Abate et al., 2002) (Seftor et al., VEGFC c vascular endothelial growth factor C 209946_at 18 0.001 2002) 213425_s_at 45 0.002

wingless-type MMTV integration site family, (Bittner et al., WNT5A c 231227_at 23 0.001 member 5A 2000)

205990_s_at 20 0.001 aCalculated using normalized Zürich data, bTwo tailed T-test statistic , assuming variance is equal. c Also present in the 223 gene intersection of the Zürich, Philadelphia and Mannheim cohort-specific expression data sets.

164

Supplementary Data 3b

Survey of literature exploring 134 factors of possible prognostic value in melanoma.

Regulationa Reference Factor A2M Down (de Wit et al., 2002) ALCAM Up (van Kempen et al., 2000) AOC3 Down (Forster-Horvath et al., 2004) APAF1 Down (Baldi et al., 2004; Fujimoto et al., 2004) APOC2 Up (de Wit et al., 2002) APOD Up (Miranda et al., 2003) ARF6 Up (Tague et al., 2004) ARHGDIB Up (Seftor et al., 2002) ATOX1 Up (Wang et al., 2004) BCL2 Down/Upb (Hernberg et al., 1998; Vlaykova et al., 2002) BIRC1 Down (Seftor et al., 2002) BIRC5 Up (Gradilone et al., 2003) BNC1 Up (Seftor et al., 2002) CASP6 Up (Woenckhaus et al., 2003) CCL2 Down (Su et al., 2000) CCND1 Up (Errico et al., 2003) CCND3 Up (Florenes et al., 2000) CD44 Down/Up (Dietrich et al., 1997; Karjalainen et al., 2000b) CDH1 Down (Andersen et al., 2004) CDH2 Up (Sanders et al., 1999) CDH3 Down (Sanders et al., 1999) CDH5 Up (Hendrix et al., 2001) CDKN1A Up (Karjalainen et al., 1999) CDKN1B Down (Florenes et al., 1998; Woenckhaus et al., 2004) CDKN2A Down (Straume and Akslen, 1997; Straume et al., 2000) CEACAM1 Up (Thies et al., 2002a) CENPE Up (Goldberg et al., 2003) CHL1 Down (Seftor et al., 2002) CKB Up (de Wit et al., 2002) CPS1 Up (Seftor et al., 2002) CSF2 Up (Ciotti et al., 1999) CTGF Up (Maniotis et al., 1999; Seftor et al., 2002) CTSD Up (Podhajcer et al., 1995) CXCL1 Up (Bordoni et al., 1990) DCT Down (Takeuchi et al., 2003) DKKL1 Up (Goldberg et al., 2003) EDNRB Down (Smith et al., 2002) EEF1A2 Down (de Wit et al., 2002) EGFR Up (Hurks et al., 2000) EPAS1 Up (Giatromanolaki et al., 2003) EPHA1 Up (Easty et al., 1999) EPHA2 Up (Maniotis et al., 1999; Seftor et al., 2002) EPHB6 Down (Hafner et al., 2003) ETS1 Up (Keehn et al., 2003) F2R Up (Tellez and Bar-Eli, 2003) FAP Down (Ramirez-Montagut et al., 2004) FASN Up (Innocenzi et al., 2003) FBLN5 Up (Wang et al., 2004) GJA1 Down (Su et al., 2000) GPM6B Down (Seftor et al., 2002) GSTM1 Up (Depeille et al., 2004) HIF1A Up (Giatromanolaki et al., 2003) HLA-A Up (Blom et al., 1997) HLA-B Up (Blom et al., 1997) HLA-G Up (Real et al., 1999) ICAM1 Down/Up (Anastassiou et al., 2000; Ciotti et al., 1999; Haritopoulos et al., 2003) IGFBP2 Up (Wang et al., 2003)

165

IL8 Up (Rofstad and Halsor, 2000) (Nurnberg et al., 1999) IL24 Down (Ellerhorst et al., 2002) ILK Up (Dai et al., 2003) IRF1 Down (Lowney et al., 1999) ITGA4 Up (Schadendorf et al., 1993) ITGA6 Down (Schadendorf et al., 1993) ITGAV Down (Nikkola et al., 2004) ITGB1 Down/Up (Nikkola et al., 2004; Vihinen et al., 2000) ITGB3 Up (Hieken et al., 1996) KCNK1 Up (Seftor et al., 2002) KRT7 Up (Seftor et al., 2002) KRT18 Up (Fuchs et al., 1992; Wang et al., 2004) L1CAM Up (Fogel et al., 2003; Thies et al., 2002b) LAMC1 Up (Seftor et al., 2001) LCP1 Up (de Wit et al., 2002) LTBP2 Up (Seftor et al., 2002) MDM2 Down (Polsky et al., 2002) MET Up (Cruz et al., 2003) MIA Up (Juergensen et al., 2001) MITF Down (Salti et al., 2000) MKI67 Up (Hazan et al., 2002) MLANA Down (Berset et al., 2001) MME Up (Bilalovic et al., 2004; Seftor et al., 2002) MMP1 Up (Nikkola et al., 2002; Seftor et al., 2001) MMP2 Up (Seftor et al., 2001; Vaisanen et al., 1998; Vaisanen et al., 1999; Vaisanen et al., 1996) MMP3 Up (Nikkola et al., 2002) MXI1 Down (Ariyanayagam-Baksh et al., 2003) MYLK Down (Seftor et al., 2002) NME1 Down (Mao et al., 2001; Sarris et al., 2004) NOS2A Down/Up (Ekmekcioglu et al., 2000; Tschugguel et al., 1999) NRP1 Up (Straume and Akslen, 2003) NTRK1 Up (Florenes et al., 2004) NUP88 Up (Zhang et al., 2002) PAWR Down (Lucas et al., 2001) PAX8 Up (Seftor et al., 2002) PCNA Down (Evans et al., 1992) PECAM1 Down (Hendrix et al., 2001) PLAT Down (Ferrier et al., 2000) PLAU Up (Maniotis et al., 1999; Seftor et al., 2002) PLK1 Up (Kneisel et al., 2002) PLOD2 Up (Goldberg et al., 2003) POU3F2 Up (Thomson et al., 1995) PRDX2 Up (de Wit et al., 2002) PTEN Down (Stahl et al., 2003; Tsao et al., 2003) PTGS2 Up (Figueiredo et al., 2003) RB1 Down (Korabiowska et al., 2001) RELA Up (Kashani-Sabet et al., 2004) RIG Up (Seftor et al., 2002) S100A1 Down (Juergensen et al., 2001; Korabiowska et al., 1994) S100A4 Up (Andersen et al., 2004) S100B Up (Banfalvi et al., 2002; Banfalvi et al., 2003) SEMA4D Up (Seftor et al., 2002) SERPINE1 Down (Su et al., 2000) SPARC Up (Massi et al., 1999) STHM Down (Seftor et al., 2002) TAP1 Down (Kamarashev et al., 2001) TCEB3 Up (Seftor et al., 2002) TF Down (Seftor et al., 2002) TFAP2A Down (Karjalainen et al., 2000a) TGM1 Up (Seftor et al., 2002) THBS1 Down/Up (Grant et al., 1998; Straume and Akslen, 2003) THBS2 Up (Kunz et al., 2002) TIE Up (Hendrix et al., 2001; Maniotis et al., 1999; Seftor et al., 2002)

166

TM4SF1 Up (Seftor et al., 2002) TMSB10 Up (Liu et al., 2004) TMSB4X Up (Cha et al., 2003) TNFRSF1A Up (Galvani et al., 2000; Ocvirk et al., 2000; Redondo et al., 2002) TOP2A Down (Mu et al., 2000) TP53 Up (McGregor et al., 1993) TRIP Down (Wang et al., 2004) TRPM1 Down (Duncan et al., 2001) TXNIP Down (Goldberg et al., 2003) TYR Down (Takeuchi et al., 2003) VEGF Up (Bayer-Garner et al., 1999; Osella-Abate et al., 2002) VEGFC Up (Seftor et al., 2002) WNT5A Up (Bittner et al., 2000) WWP2 Up (Seftor et al., 2002)

aRegulation associated with poor prognosis or metastatic aggressiveness. bDifferent researchers reporting opposing results.

167

Supplementary Data 4.

Total RNA extracted from samples was subjected to reverse transcription PCR using the 1st Strand cDNA Synthesis Kit for PCR (Roche) and amplifications were perfomed using the LightCycler FastStart DNA Master SYBR Green I (Roche) on a LightCycler 2.0 instrument (Roche) according to manufacturers instructions.

Primers and conditions. Gene Primers (sense & antisense) Denaturation Annealing Elongation Cycles MgCl2 CGGGACCTTGCTGTCTTCTC BGN 95°C / 15s 55°C / 5s 72°C / 17s 50 5 mM CCCGGCAAGAACCTGAAAG GCGTCTGCGCCAAGCA CTGF 95°C / 1s 56°C / 2s 72°C / 17s 55 3 mM TGGACCAGGCAGTTGGCT ACGGATTTGGTCGTATTGGG GAPDH 95°C / 10s 56°C / 2s 72°C / 17s 45 4 mM CGCTCCTGGAAGATGGTGAT ATGCCAAGAGAAGATGCT MLANA 95°C / 15s 55°C / 5s 72°C / 17s 50 4 mM GGAGAACATTAGATGTCTG TGCTGGTGAATGCCCTCTACT SERPINE1 95°C / 15s 60°C / 5s 72°C / 18s 35 3 mM CGGTCATTCCCAGGTTCTCTA GGCGGCGGCCGGGGGCGA SOX10 95°C / 10s 66°C / 2s 72°C / 14s 35 2 mM TCAGGGCAGGAGCCAGACAGAAA ACAACAGCCATCAGTCT TYR 95°C / 0s 60°C / 15s 72°C / 13s 40 4 mM CCTGTACCTGGGACATT

168

Supplementary References

Anastassiou, G., Schilling, H., Stang, A., Djakovic, S., Heiligenhaus, A. and Bornfeld, N. (2000) Expression of the cell adhesion molecules ICAM-1, VCAM-1 and NCAM in uveal melanoma: a clinicopathological study. Oncology, 58, 83-88. Andersen, K., Nesland, J.M., Holm, R., Florenes, V.A., Fodstad and Maelandsmo, G.M. (2004) Expression of S100A4 combined with reduced E-cadherin expression predicts patient outcome in malignant melanoma. Mod Pathol, 17, 990-997. Ariyanayagam-Baksh, S.M., Baksh, F.K., Swalsky, P.A. and Finkelstein, S.D. (2003) Loss of heterozygosity in the MXI1 gene is a frequent occurrence in melanoma. Mod Pathol, 16, 992-995. Baldi, A., Santini, D., Russo, P., Catricala, C., Amantea, A., Picardo, M., Tatangelo, F., Botti, G., Dragonetti, E., Murace, R., Tonini, G., Natali, P.G., Baldi, F. and Paggi, M.G. (2004) Analysis of APAF-1 expression in human cutaneous melanoma progression. Exp Dermatol, 13, 93-97. Banfalvi, T., Boldizsar, M., Gergye, M., Gilde, K., Kremmer, T. and Otto, S. (2002) Comparison of prognostic significance of serum 5-S-Cysteinyldopa, LDH and S-100B protein in Stage III-IV malignant melanoma. Pathol Oncol Res, 8, 183-187. Banfalvi, T., Udvarhelyi, N., Orosz, Z., Gergye, M., Gilde, K. and Timar, J. (2003) Heterogenous S-100B protein expression patterns in malignant melanoma and association with serum protein levels. Oncology, 64, 374-379. Bayer-Garner, I.B., Hough, A.J., Jr. and Smoller, B.R. (1999) Vascular endothelial growth factor expression in malignant melanoma: prognostic versus diagnostic usefulness. Mod Pathol, 12, 770-774. Berset, M., Cerottini, J.P., Guggisberg, D., Romero, P., Burri, F., Rimoldi, D. and Panizzon, R.G. (2001) Expression of Melan-A/MART-1 antigen as a prognostic factor in primary cutaneous melanoma. Int J Cancer, 95, 73-77. Bilalovic, N., Sandstad, B., Golouh, R., Nesland, J.M., Selak, I. and Torlakovic, E.E. (2004) CD10 protein expression in tumor and stromal cells of malignant melanoma is associated with tumor progression. Mod Pathol, 17, 1251-1258. Bittner, M., Meltzer, P., Chen, Y., Jiang, Y., Seftor, E., Hendrix, M., Radmacher, M., Simon, R., Yakhini, Z., Ben-Dor, A., Sampas, N., Dougherty, E., Wang, E., Marincola, F., Gooden, C., Lueders, J., Glatfelter, A., Pollock, P., Carpten, J., Gillanders, E., Leja, D., Dietrich, K., Beaudry, C., Berens, M., Alberts, D. and Sondak, V. (2000) Molecular classification of cutaneous malignant melanoma by gene expression profiling. Nature, 406, 536-540. Blom, D.J., Luyten, G.P., Mooy, C., Kerkvliet, S., Zwinderman, A.H. and Jager, M.J. (1997) Human leukocyte antigen class I expression. Marker of poor prognosis in uveal melanoma. Invest Ophthalmol Vis Sci, 38, 1865-1872. Bordoni, R., Fine, R., Murray, D. and Richmond, A. (1990) Characterization of the role of melanoma growth stimulatory activity (MGSA) in the growth of normal melanocytes, nevocytes, and malignant melanocytes. J Cell Biochem, 44, 207-219. Cha, H.J., Jeong, M.J. and Kleinman, H.K. (2003) Role of thymosin beta4 in tumor metastasis and angiogenesis. J Natl Cancer Inst, 95, 1674-1680. Ciotti, P., Pesce, G.P., Cafiero, F., Rainero, M.L., Sementa, A., Nicolo, G., Mela, G.S., Bagnasco, M., Santi, P.L. and Bianchi-Scarra, G. (1999) Intercellular adhesion molecule-1 (ICAM-1) and granulocyte-macrophage colony stimulating factor (GM-CSF) co- expression in cutaneous malignant melanoma lesions. Melanoma Res, 9, 253-260.

169

Cruz, J., Reis-Filho, J.S., Silva, P. and Lopes, J.M. (2003) Expression of c-met tyrosine kinase receptor is biologically and prognostically relevant for primary cutaneous malignant melanomas. Oncology, 65, 72-82. Dai, D.L., Makretsov, N., Campos, E.I., Huang, C., Zhou, Y., Huntsman, D., Martinka, M. and Li, G. (2003) Increased expression of integrin-linked kinase is correlated with melanoma progression and poor patient survival. Clin Cancer Res, 9, 4409-4414. de Wit, N.J., Burtscher, H.J., Weidle, U.H., Ruiter, D.J. and van Muijen, G.N. (2002) Differentially expressed genes identified in human melanoma cell lines with different metastatic behaviour using high density oligonucleotide arrays. Melanoma Res, 12, 57-69. Depeille, P., Cuq, P., Mary, S., Passagne, I., Evrard, A., Cupissol, D. and Vian, L. (2004) Glutathione S-transferase M1 and multidrug resistance protein 1 act in synergy to protect melanoma cells from vincristine effects. Mol Pharmacol, 65, 897-905. Dietrich, A., Tanczos, E., Vanscheidt, W., Schopf, E. and Simon, J.C. (1997) High CD44 surface expression on primary tumours of malignant melanoma correlates with increased metastatic risk and reduced survival. Eur J Cancer, 33, 926-930. Duncan, L.M., Deeds, J., Cronin, F.E., Donovan, M., Sober, A.J., Kauffman, M. and McCarthy, J.J. (2001) Melastatin expression and prognosis in cutaneous malignant melanoma. J Clin Oncol, 19, 568-576. Easty, D.J., Hill, S.P., Hsu, M.Y., Fallowfield, M.E., Florenes, V.A., Herlyn, M. and Bennett, D.C. (1999) Up-regulation of ephrin-A1 during melanoma progression. Int J Cancer, 84, 494-501. Ekmekcioglu, S., Ellerhorst, J., Smid, C.M., Prieto, V.G., Munsell, M., Buzaid, A.C. and Grimm, E.A. (2000) Inducible nitric oxide synthase and nitrotyrosine in human metastatic melanoma tumors correlate with poor survival. Clin Cancer Res, 6, 4768-4775. Ellerhorst, J.A., Prieto, V.G., Ekmekcioglu, S., Broemeling, L., Yekell, S., Chada, S. and Grimm, E.A. (2002) Loss of MDA-7 expression with progression of melanoma. J Clin Oncol, 20, 1069-1074. Errico, M.E., Staibano, S., Tranfa, F., Bonavolonta, G., Lo Muzio, L., Somma, P., Lucariello, A., Mansueto, G., D'Aponte, A., Ferrara, G. and De Rosa, G. (2003) Expression of cyclin- D1 in uveal malignant melanoma. Anticancer Res, 23, 2701-2706. Evans, A.T., Blessing, K., Orrell, J.M. and Grant, A. (1992) Mitotic indices, anti-PCNA immunostaining, and AgNORs in thick cutaneous melanomas displaying paradoxical behaviour. J Pathol, 168, 15-22. Ferrier, C.M., Suciu, S., van Geloof, W.L., Straatman, H., Eggermont, A.M., Koops, H.S., Kroon, B.B., Lejeune, F.J., Kleeberg, U.R., van Muijen, G.N. and Ruiter, D.J. (2000) High tPA-expression in primary melanoma of the limb correlates with good prognosis. Br J Cancer, 83, 1351-1359. Figueiredo, A., Caissie, A.L., Callejo, S.A., McLean, I.W., Gold, P. and Burnier, M.N., Jr. (2003) Cyclooxygenase-2 expression in uveal melanoma: novel classification of mixed- cell-type tumours. Can J Ophthalmol, 38, 352-356. Florenes, V.A., Faye, R.S., Maelandsmo, G.M., Nesland, J.M. and Holm, R. (2000) Levels of cyclin D1 and D3 in malignant melanoma: deregulated cyclin D3 expression is associated with poor clinical outcome in superficial melanoma. Clin Cancer Res, 6, 3614-3620. Florenes, V.A., Maelandsmo, G.M., Holm, R., Reich, R., Lazarovici, P. and Davidson, B. (2004) Expression of activated TrkA protein in melanocytic tumors: relationship to cell proliferation and clinical outcome. Am J Clin Pathol, 122, 412-420. Florenes, V.A., Maelandsmo, G.M., Kerbel, R.S., Slingerland, J.M., Nesland, J.M. and Holm, R. (1998) Protein expression of the cell-cycle inhibitor p27Kip1 in malignant melanoma: inverse correlation with disease-free survival. Am J Pathol, 153, 305-312.

170

Fogel, M., Mechtersheimer, S., Huszar, M., Smirnov, A., Abu-Dahi, A., Tilgen, W., Reichrath, J., Georg, T., Altevogt, P. and Gutwein, P. (2003) L1 adhesion molecule (CD 171) in development and progression of human malignant melanoma. Cancer Lett, 189, 237-247. Forster-Horvath, C., Dome, B., Paku, S., Ladanyi, A., Somlai, B., Jalkanen, S. and Timar, J. (2004) Loss of vascular adhesion protein-1 expression in intratumoral microvessels of human skin melanoma. Melanoma Res, 14, 135-140. Fuchs, U., Kivela, T., Summanen, P., Immonen, I. and Tarkkanen, A. (1992) An immunohistochemical and prognostic analysis of cytokeratin expression in malignant uveal melanoma. Am J Pathol, 141, 169-181. Fujimoto, A., Takeuchi, H., Taback, B., Hsueh, E.C., Elashoff, D., Morton, D.L. and Hoon, D.S. (2004) Allelic imbalance of 12q22-23 associated with APAF-1 locus correlates with poor disease outcome in cutaneous melanoma. Cancer Res, 64, 2245-2250. Galvani, V., Pretnar Hartman, K., Rupreht, R.R., Novakovic, S., Stabuc, B., Ocvirk, J., Menart, V., Gaberc Porekar, V., Stalc, A., Rozman, P. and Curin Serbec, V. (2000) Soluble tumor necrosis factor receptor I (sTNFRI) as a prognostic factor in melanoma patients in Slovene population. Pflugers Arch, 440, R61-63. Giatromanolaki, A., Sivridis, E., Kouskoukis, C., Gatter, K.C., Harris, A.L. and Koukourakis, M.I. (2003) Hypoxia-inducible factors 1alpha and 2alpha are related to vascular endothelial growth factor expression and a poorer prognosis in nodular malignant melanomas of the skin. Melanoma Res, 13, 493-501. Goldberg, S.F., Miele, M.E., Hatta, N., Takata, M., Paquette-Straub, C., Freedman, L.P. and Welch, D.R. (2003) Melanoma metastasis suppression by chromosome 6: evidence for a pathway regulated by CRSP3 and TXNIP. Cancer Res, 63, 432-440. Gradilone, A., Gazzaniga, P., Ribuffo, D., Scarpa, S., Cigna, E., Vasaturo, F., Bottoni, U., Innocenzi, D., Calvieri, S., Scuderi, N., Frati, L. and Agliano, A.M. (2003) Survivin, bcl-2, bax, and bcl-X gene expression in sentinel lymph nodes from melanoma patients. J Clin Oncol, 21, 306-312. Grant, S.W., Kyshtoobayeva, A.S., Kurosaki, T., Jakowatz, J. and Fruehauf, J.P. (1998) Mutant p53 correlates with reduced expression of thrombospondin-1, increased angiogenesis, and metastatic progression in melanoma. Cancer Detect Prev, 22, 185-194. Hafner, C., Bataille, F., Meyer, S., Becker, B., Roesch, A., Landthaler, M. and Vogt, T. (2003) Loss of EphB6 expression in metastatic melanoma. Int J Oncol, 23, 1553-1559. Haritopoulos, K.N., Lazaris, A.C., Kavantzas, N., Tseleni-Balafouta, S., Thomopoulou, G. and Aroni, K. (2003) ICAM-1 and beta(3) integrin immunoexpression in malignant melanoma cells: can they be used as additional predictors? Apmis, 111, 421-429. Hazan, C., Melzer, K., Panageas, K.S., Li, E., Kamino, H., Kopf, A., Cordon-Cardo, C., Osman, I. and Polsky, D. (2002) Evaluation of the proliferation marker MIB-1 in the prognosis of cutaneous malignant melanoma. Cancer, 95, 634-640. Hendrix, M.J., Seftor, E.A., Meltzer, P.S., Gardner, L.M., Hess, A.R., Kirschmann, D.A., Schatteman, G.C. and Seftor, R.E. (2001) Expression and functional significance of VE- cadherin in aggressive human melanoma cells: role in vasculogenic mimicry. Proc Natl Acad Sci U S A, 98, 8018-8023. Hernberg, M., Turunen, J.P., von Boguslawsky, K., Muhonen, T. and Pyrhonen, S. (1998) Prognostic value of biomarkers in malignant melanoma. Melanoma Res, 8, 283-291. Hieken, T.J., Farolan, M., Ronan, S.G., Shilkaitis, A., Wild, L. and Das Gupta, T.K. (1996) Beta3 integrin expression in melanoma predicts subsequent metastasis. J Surg Res, 63, 169-173.

171

Hurks, H.M., Metzelaar-Blok, J.A., Barthen, E.R., Zwinderman, A.H., De Wolff-Rouendaal, D., Keunen, J.E. and Jager, M.J. (2000) Expression of epidermal growth factor receptor: risk factor in uveal melanoma. Invest Ophthalmol Vis Sci, 41, 2023-2027. Innocenzi, D., Alo, P.L., Balzani, A., Sebastiani, V., Silipo, V., La Torre, G., Ricciardi, G., Bosman, C. and Calvieri, S. (2003) Fatty acid synthase expression in melanoma. J Cutan Pathol, 30, 23-28. Juergensen, A., Holzapfel, U., Hein, R., Stolz, W., Buettner, R. and Bosserhoff, A. (2001) Comparison of two prognostic markers for malignant melanoma: MIA and S100 beta. Tumour Biol, 22, 54-58. Kamarashev, J., Ferrone, S., Seifert, B., Boni, R., Nestle, F.O., Burg, G. and Dummer, R. (2001) TAP1 down-regulation in primary melanoma lesions: an independent marker of poor prognosis. Int J Cancer, 95, 23-28. Karjalainen, J.M., Eskelinen, M.J., Kellokoski, J.K., Reinikainen, M., Alhava, E.M. and Kosma, V.M. (1999) p21(WAF1/CIP1) expression in stage I cutaneous malignant melanoma: its relationship with p53, cell proliferation and survival. Br J Cancer, 79, 895- 902. Karjalainen, J.M., Kellokoski, J.K., Mannermaa, A.J., Kujala, H.E., Moisio, K.I., Mitchell, P.J., Eskelinen, M.J., Alhava, E.M. and Kosma, V.M. (2000a) Failure in post- transcriptional processing is a possible inactivation mechanism of AP-2alpha in cutaneous melanoma. Br J Cancer, 82, 2015-2021. Karjalainen, J.M., Tammi, R.H., Tammi, M.I., Eskelinen, M.J., Agren, U.M., Parkkinen, J.J., Alhava, E.M. and Kosma, V.M. (2000b) Reduced level of CD44 and hyaluronan associated with unfavorable prognosis in clinical stage I cutaneous melanoma. Am J Pathol, 157, 957-965. Kashani-Sabet, M., Shaikh, L., Miller, J.R., 3rd, Nosrati, M., Ferreira, C.M., Debs, R.J. and Sagebiel, R.W. (2004) NF-kappa B in the vascular progression of melanoma. J Clin Oncol, 22, 617-623. 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. Kneisel, L., Strebhardt, K., Bernd, A., Wolter, M., Binder, A. and Kaufmann, R. (2002) Expression of polo-like kinase (PLK1) in thin melanomas: a novel marker of metastatic disease. J Cutan Pathol, 29, 354-358. Korabiowska, M., Mirecka, J., Brinck, U., Szuta, M., Stypulkowska, J., Wiese, G., Bartkowski, S. and Schauer, A. (1994) Immunohistochemical demonstration of S100 protein in malignant melanomas of the facial skin and oral cavity. J Nihon Univ Sch Dent, 36, 117-121. Korabiowska, M., Ruschenburg, I., Betke, H., Stachura, J., Schlott, T., Cardo, C.C. and Brinck, U. (2001) Downregulation of the retinoblastoma gene expression in the progression of malignant melanoma. Pathobiology, 69, 274-280. Kunz, M., Koczan, D., Ibrahim, S.M., Gillitzer, R., Gross, G. and Thiesen, H.J. (2002) Differential expression of thrombospondin 2 in primary and metastatic malignant melanoma. Acta Derm Venereol, 82, 163-169. Liu, C.R., Ma, C.S., Ning, J.Y., You, J.F., Liao, S.L. and Zheng, J. (2004) Differential thymosin beta 10 expression levels and actin filament organization in tumor cell lines with different metastatic potential. Chin Med J (Engl), 117, 213-218. Lowney, J.K., Boucher, L.D., Swanson, P.E. and Doherty, G.M. (1999) Interferon regulatory factor-1 and -2 expression in human melanoma specimens. Ann Surg Oncol, 6, 604-608. Lucas, T., Pratscher, B., Krishnan, S., Fink, D., Gunsberg, P., Wolschek, M., Wacheck, V., Muster, T., Romirer, I., Wolff, K., Pehamberger, H., Eichler, H.G., Rangnekar, V.M. and

172

Jansen, B. (2001) Differential expression levels of Par-4 in melanoma. Melanoma Res, 11, 379-383. Maniotis, A.J., Folberg, R., Hess, A., Seftor, E.A., Gardner, L.M., Pe'er, J., Trent, J.M., Meltzer, P.S. and Hendrix, M.J. (1999) Vascular channel formation by human melanoma cells in vivo and in vitro: vasculogenic mimicry. Am J Pathol, 155, 739-752. Mao, H., Liu, H., Fu, X., Fang, Z., Abrams, J. and Worsham, M.J. (2001) Loss of nm23 expression predicts distal metastases and poorer survival for breast cancer. Int J Oncol, 18, 587-591. Massi, D., Franchi, A., Borgognoni, L., Reali, U.M. and Santucci, M. (1999) Osteonectin expression correlates with clinical outcome in thin cutaneous malignant melanomas. Hum Pathol, 30, 339-344. McGregor, J.M., Yu, C.C., Dublin, E.A., Barnes, D.M., Levison, D.A. and MacDonald, D.M. (1993) p53 immunoreactivity in human malignant melanoma and dysplastic naevi. Br J Dermatol, 128, 606-611. Miranda, E., Vizoso, F., Martin, A., Quintela, I., Corte, M.D., Segui, M.E., Ordiz, I. and Merino, A.M. (2003) Apolipoprotein D expression in cutaneous malignant melanoma. J Surg Oncol, 83, 99-105. Mu, X.C., Tran, T.A., Ross, J.S. and Carlson, J.A. (2000) Topoisomerase II-alpha expression in melanocytic nevi and malignant melanoma. J Cutan Pathol, 27, 242-248. Nikkola, J., Vihinen, P., Vlaykova, T., Hahka-Kemppinen, M., Heino, J. and Pyrhonen, S. (2004) Integrin chains beta1 and alphav as prognostic factors in human metastatic melanoma. Melanoma Res, 14, 29-37. Nikkola, J., Vihinen, P., Vlaykova, T., Hahka-Kemppinen, M., Kahari, V.M. and Pyrhonen, S. (2002) High expression levels of collagenase-1 and stromelysin-1 correlate with shorter disease-free survival in human metastatic melanoma. Int J Cancer, 97, 432-438. Nurnberg, W., Tobias, D., Otto, F., Henz, B.M. and Schadendorf, D. (1999) Expression of interleukin-8 detected by in situ hybridization correlates with worse prognosis in primary cutaneous melanoma. J Pathol, 189, 546-551. Ocvirk, J., Stabuc, B., Rudolf, Z., Galvani, V. and Curin-Serbec, V. (2000) Serum values of tumour necrosis factor-alpha and of soluble tumour necrosis factor-R55 in melanoma patients. Melanoma Res, 10, 253-258. Osella-Abate, S., Quaglino, P., Savoia, P., Leporati, C., Comessatti, A. and Bernengo, M.G. (2002) VEGF-165 serum levels and tyrosinase expression in melanoma patients: correlation with the clinical course. Melanoma Res, 12, 325-334. Podhajcer, O.L., Bover, L., Bravo, A.I., Ledda, M.F., Kairiyama, C., Calb, I., Guerra, L., Capony, F. and Mordoh, J. (1995) Expression of cathepsin D in primary and metastatic human melanoma and dysplastic nevi. J Invest Dermatol, 104, 340-344. Polsky, D., Melzer, K., Hazan, C., Panageas, K.S., Busam, K., Drobnjak, M., Kamino, H., Spira, J.G., Kopf, A.W., Houghton, A., Cordon-Cardo, C. and Osman, I. (2002) HDM2 protein overexpression and prognosis in primary malignant melanoma. J Natl Cancer Inst, 94, 1803-1806. Ramirez-Montagut, T., Blachere, N.E., Sviderskaya, E.V., Bennett, D.C., Rettig, W.J., Garin- Chesa, P. and Houghton, A.N. (2004) FAPalpha, a surface peptidase expressed during wound healing, is a tumor suppressor. Oncogene, 23, 5435-5446. Real, L.M., Cabrera, T., Collado, A., Jimenez, P., Garcia, A., Ruiz-Cabello, F. and Garrido, F. (1999) Expression of HLA G in human tumors is not a frequent event. Int J Cancer, 81, 512-518. Redondo, P., Solano, T., B, V.A., Bauza, A. and Idoate, M. (2002) Fas and Fas ligand: expression and soluble circulating levels in cutaneous malignant melanoma. Br J Dermatol, 147, 80-86.

173

Rofstad, E.K. and Halsor, E.F. (2000) Vascular endothelial growth factor, interleukin 8, platelet-derived endothelial cell growth factor, and basic fibroblast growth factor promote angiogenesis and metastasis in human melanoma xenografts. Cancer Res, 60, 4932-4938. Salti, G.I., Manougian, T., Farolan, M., Shilkaitis, A., Majumdar, D. and Das Gupta, T.K. (2000) Micropthalmia transcription factor: a new prognostic marker in intermediate- thickness cutaneous malignant melanoma. Cancer Res, 60, 5012-5016. Sanders, D.S., Blessing, K., Hassan, G.A., Bruton, R., Marsden, J.R. and Jankowski, J. (1999) Alterations in cadherin and catenin expression during the biological progression of melanocytic tumours. Mol Pathol, 52, 151-157. Sarris, M., Scolyer, R.A., Konopka, M., Thompson, J.F., Harper, C.G. and Lee, S.C. (2004) Cytoplasmic expression of nm23 predicts the potential for cerebral metastasis in patients with primary cutaneous melanoma. Melanoma Res, 14, 23-27. Schadendorf, D., Gawlik, C., Haney, U., Ostmeier, H., Suter, L. and Czarnetzki, B.M. (1993) Tumour progression and metastatic behaviour in vivo correlates with integrin expression on melanocytic tumours. J Pathol, 170, 429-434. Seftor, E.A., Meltzer, P.S., Kirschmann, D.A., Pe'er, J., Maniotis, A.J., Trent, J.M., Folberg, R. and Hendrix, M.J. (2002) Molecular determinants of human uveal melanoma invasion and metastasis. Clin Exp Metastasis, 19, 233-246. Seftor, R.E., Seftor, E.A., Koshikawa, N., Meltzer, P.S., Gardner, L.M., Bilban, M., Stetler- Stevenson, W.G., Quaranta, V. and Hendrix, M.J. (2001) Cooperative interactions of laminin 5 gamma2 chain, matrix metalloproteinase-2, and membrane type-1- matrix/metalloproteinase are required for mimicry of embryonic vasculogenesis by aggressive melanoma. Cancer Res, 61, 6322-6327. Smith, S.L., Damato, B.E., Scholes, A.G., Nunn, J., Field, J.K. and Heighway, J. (2002) Decreased endothelin receptor B expression in large primary uveal melanomas is associated with early clinical metastasis and short survival. Br J Cancer, 87, 1308-1313. Stahl, J.M., Cheung, M., Sharma, A., Trivedi, N.R., Shanmugam, S. and Robertson, G.P. (2003) Loss of PTEN promotes tumor development in malignant melanoma. Cancer Res, 63, 2881-2890. Straume, O. and Akslen, L.A. (1997) Alterations and prognostic significance of p16 and p53 protein expression in subgroups of cutaneous melanoma. Int J Cancer, 74, 535-539. Straume, O. and Akslen, L.A. (2003) Increased Expression of VEGF-Receptors (FLT-1, KDR, NRP-1) and Thrombospondin-1 is Associated with Glomeruloid Microvascular Proliferation, an Aggressive Angiogenic Phenotype, in Malignant Melanoma. Angiogenesis, 6, 295-301. Straume, O., Sviland, L. and Akslen, L.A. (2000) Loss of nuclear p16 protein expression correlates with increased tumor cell proliferation (Ki-67) and poor prognosis in patients with vertical growth phase melanoma. Clin Cancer Res, 6, 1845-1853. Su, Y.A., Bittner, M.L., Chen, Y., Tao, L., Jiang, Y., Zhang, Y., Stephan, D.A. and Trent, J.M. (2000) Identification of tumor-suppressor genes using human melanoma cell lines UACC903, UACC903(+6), and SRS3 by comparison of expression profiles. Mol Carcinog, 28, 119-127. Tague, S.E., Muralidharan, V. and D'Souza-Schorey, C. (2004) ADP-ribosylation factor 6 regulates tumor cell invasion through the activation of the MEK/ERK signaling pathway. Proc Natl Acad Sci U S A, 101, 9671-9676. Takeuchi, H., Kuo, C., Morton, D.L., Wang, H.J. and Hoon, D.S. (2003) Expression of differentiation melanoma-associated antigen genes is associated with favorable disease outcome in advanced-stage melanomas. Cancer Res, 63, 441-448. Tellez, C. and Bar-Eli, M. (2003) Role and regulation of the thrombin receptor (PAR-1) in human melanoma. Oncogene, 22, 3130-3137.

174

Thies, A., Moll, I., Berger, J., Wagener, C., Brummer, J., Schulze, H.J., Brunner, G. and Schumacher, U. (2002a) CEACAM1 expression in cutaneous malignant melanoma predicts the development of metastatic disease. J Clin Oncol, 20, 2530-2536. Thies, A., Schachner, M., Moll, I., Berger, J., Schulze, H.J., Brunner, G. and Schumacher, U. (2002b) Overexpression of the cell adhesion molecule L1 is associated with metastasis in cutaneous malignant melanoma. Eur J Cancer, 38, 1708-1716. Thomson, J.A., Murphy, K., Baker, E., Sutherland, G.R., Parsons, P.G., Sturm, R.A. and Thomson, F. (1995) The brn-2 gene regulates the melanocytic phenotype and tumorigenic potential of human melanoma cells. Oncogene, 11, 691-700. Tsao, H., Mihm, M.C., Jr. and Sheehan, C. (2003) PTEN expression in normal skin, acquired melanocytic nevi, and cutaneous melanoma. J Am Acad Dermatol, 49, 865-872. Tschugguel, W., Pustelnik, T., Lass, H., Mildner, M., Weninger, W., Schneeberger, C., Jansen, B., Tschachler, E., Waldhor, T., Huber, J.C. and Pehamberger, H. (1999) Inducible nitric oxide synthase (iNOS) expression may predict distant metastasis in human melanoma. Br J Cancer, 79, 1609-1612. Vaisanen, A., Kallioinen, M., Taskinen, P.J. and Turpeenniemi-Hujanen, T. (1998) Prognostic value of MMP-2 immunoreactive protein (72 kD type IV collagenase) in primary skin melanoma. J Pathol, 186, 51-58. Vaisanen, A., Kallioinen, M., von Dickhoff, K., Laatikainen, L., Hoyhtya, M. and Turpeenniemi-Hujanen, T. (1999) Matrix metalloproteinase-2 (MMP-2) immunoreactive protein--a new prognostic marker in uveal melanoma? J Pathol, 188, 56-62. Vaisanen, A., Tuominen, H., Kallioinen, M. and Turpeenniemi-Hujanen, T. (1996) Matrix metalloproteinase-2 (72 kD type IV collagenase) expression occurs in the early stage of human melanocytic tumour progression and may have prognostic value. J Pathol, 180, 283-289. van Kempen, L.C., van den Oord, J.J., van Muijen, G.N., Weidle, U.H., Bloemers, H.P. and Swart, G.W. (2000) Activated leukocyte cell adhesion molecule/CD166, a marker of tumor progression in primary malignant melanoma of the skin. Am J Pathol, 156, 769-774. Vihinen, P., Nikkola, J., Vlaykova, T., Hahka-Kemppinen, M., Talve, L., Heino, J. and Pyrhonen, S. (2000) Prognostic value of beta1 integrin expression in metastatic melanoma. Melanoma Res, 10, 243-251. Vlaykova, T., Talve, L., Hahka-Kemppinen, M., Hernberg, M., Muhonen, T., Collan, Y. and Pyrhonen, S. (2002) Immunohistochemically detectable bcl-2 expression in metastatic melanoma: association with survival and treatment response. Oncology, 62, 259-268. Wang, H., Shen, S.S., Diwan, A.H., Zhang, W., Fuller, G.N. and Prieto, V.G. (2003) Expression of insulin-like growth factor-binding protein 2 in melanocytic lesions. J Cutan Pathol, 30, 599-605. Wang, Z., Dooley, T.P., Curto, E.V., Davis, R.L. and VandeBerg, J.L. (2004) Cross-species application of cDNA microarrays to profile gene expression using UV-induced melanoma in Monodelphis domestica as the model system. Genomics, 83, 588-599. Woenckhaus, C., Fenic, I., Giebel, J., Hauser, S., Failing, K., Woenckhaus, J., Dittberner, T. and Poetsch, M. (2004) Loss of heterozygosity at 12p13 and loss of p27KIP1 protein expression contribute to melanoma progression. Virchows Arch, 445, 491-497. Woenckhaus, C., Giebel, J., Failing, K., Fenic, I., Dittberner, T. and Poetsch, M. (2003) Expression of AP-2alpha, c-kit, and cleaved caspase-6 and -3 in naevi and malignant melanomas of the skin. A possible role for caspases in melanoma progression? J Pathol, 201, 278-287. Zhang, H., Schneider, J. and Rosdahl, I. (2002) Expression of p16, p27, p53, p73 and Nup88 proteins in matched primary and metastatic melanoma cells. Int J Oncol, 21, 43-48.

175

176

Acknowledgments

I wish to thank Professor Reinhard Dummer for giving me the chance to conduct my research in a clinical environment I was not familiar with. Thank you for all the freedom in research you have given me….. this freedom has surely contributed to making me a better researcher. Also, many thanks to Professor Sabine Werner, without whom, the quality of this thesis would not have been the same. Thank you for being always so involved even though we met only a few times per year. Thank you to Keith who introduced me to melanoma research. Thank you for sharing your ideas on which the new melanoma model could be built. I have to also say thank you for putting up with me when stress overwhelmed me…. it was only the two of us for two years….. it’s a chance we are still talking to each other! And I therefore have to thank Ossia who joined our group in 2005, bringing fresh ideas and plenty of energy. Your ideas and technical expertise was a real gift to our small group. To the girls in F14, Giulia, Julia and Ossia, thank you for the continuous support… we had laughs and we had tears… thank you for always being there. Moving to F14 was a great idea! I would also like to thank Niki for all the great work he has done in the lab and for listening to all our small requests….. F14 has been a quiet place to write and it could not have been without your help. Thank you to all the technicians and the members of the Department of Dermatology for their support. Un énorme merci à ma famille qui depuis le Québec m’a toujours soutenue. Merci de n’avoir jamais perdu patience lors de mes crises d’angoisse…..je crois vraiment que cette thèse sera la dernière…… Merci aussi à mes amies qui ont toujours gardé contact…merci d’avoir cru en la “nerd” que je suis! I also send a warm thank you to Patrick’s family in New Zealand… thank you for caring so much. And finally, I would like to thank Patrick for having been so present and patient….. thank you for all the love, understanding and support you have given me...thank you…

177

Curriculum vitae

Natalie Schlegel Kirchenfeld 37 8052 Zürich, Switzerland Telephone: +41 (0)43 268 9804 E-mail: [email protected]

Date of Birth: 20.01.1976 Nationalities: Swiss and Canadian

Academic Profile

2004-2008 Doctorate of Philosophy in Biology Subject: Cell signalling in melanoma University Hospital Zürich, Zürich, Switzerland Swiss Federal Institute of Technology (ETH) Zürich, Switzerland

2002-2003 Doctorate of Philosophy in Microbiology (interrupted) Subject: Legionella pneumophila pathogenesis Swiss Federal Institute of Technology (ETH), Zürich

1999-2001 Master of Science in Pharmacology with distinction Project conducted in Molecular Pathology Thesis entitled "Genetic susceptibility to migraine in families with bipolar disorder" University of Otago, Dunedin, New Zealand

1996-1999 Bachelor of Science in Biochemistry McGill University, Montreal, Canada

Professional Experience

2001- 2002 Junior Research Fellow Microbiology Department University of Otago, Dunedin, New Zealand

2000-2001 Medical testing supevisor School of Medicine University of Otago, Dunedin, New Zealand

178

Natalie Schlegel Page 2

2000 Demonstrator (teaching assistant) Pharmacology Department University of Otago, Dunedin, New Zealand

1995-1999 Volunteer work towards patients Montreal Children’s Hospital Montreal, Canada

Scientific Presentations

4th International Melanoma Congress, 1-4 Nov 2007, New York, USA

37th Annual European Society for Dermatological Research (ESDR) Meeting, 6-8 Sept 2007, Zürich, Switzerland

ISREC Conference on Cancer Research, 11-13 Oct 2006, Lausanne, Switzerland

36th Annual ESDR Meeting, 7-9 Sept 2006, Paris, France

2nd Cancer research retreat, Cancer Network Zürich, 1-3 Sept 2006, Ascona, Switzerland

35th Annual ESDR Meeting, 22-24 Sept 2005, Tübingen, Germany

Swiss Molecular Microbiology Workshop 2003, June 23-25 2003, Cartigny, Switzerland

Scientific Publications

Hoek, K.S., Schlegel, N.C., Lin, W.M., Mnich, C., Zipser, M., Kobert, N., Storz, M., Mihic, D., Moch, H., Garraway, L.A., and Dummer, R.. Prognostic loss of heterozygosity patterns in melanoma and the effect of MITF amplification in vivo. Manuscript.

Schlegel, N.C., Eichhoff, O.M., Hemmi, S., Mihic, D., Werner, S., Dummer, R., Hoek, K.S.. (2008) Id2 suppression of p15Ink4b in melanoma abrogates TGF-β-mediated anti- proliferation. Manuscript.

Hoek, K.S., Eichhoff, O.M., Schlegel, N.C., Döbbeling, U., Kobert, N., Schaerer, L., Hemmi, S. and Dummer, R.. (2008). In vivo switching of human melanoma cells between proliferative and invasive states. Cancer Res in press.

Hoek, K.S., Schlegel, N.C., Brafford, P., Sucker, A., Ugurel, S., Kumar, R., Weber, B.L., Nathanson, K.L., Phillips, D.J., Herlyn, M., Schadendorf D. and Dummer R.. (2006). Metastatic potential of melanomas defined by specific gene expression profiles with no BRAF signature. Pigment Cell Res 19, 290-302.

179