i

Kinase Gene Fusions in Melanoma

by

Jacqueline Turner

A thesis submitted to the Faculty of the Undergraduate School of the University of Colorado in partial fulfillment of graduating with Honors from the Department of Chemistry and Biochemistry 2017

i

This thesis entitled: Recurrent Kinase Gene Fusions in Melanoma Written by Jacqueline Turner has been approved for the Department of Chemistry and Biochemistry

______Dr. Joseph Falke Dr. Robert Parson

______Dr. Jennifer Martin Dr. Natalie Ahn

______Dr. John Tentler Dr. William Robinson

Date:______

The final copy of this thesis has been examined by the signatories, and we find that both the content and the form meet acceptable presentation standards of scholarly work in the above mentioned discipline ii

Turner, Jacqueline (B.A., Department of Chemistry and Biochemistry)

Kinase Fusions in Melanoma

Thesis directed by Professor William Robinson

Kinase gene fusions are a mechanism of alternative pathway activation and have been increasingly described in cancer, including malignant melanoma. The prevalence of kinase gene fusions across different subtypes in melanoma has not yet been reported. Additionally, few studies in melanoma have examined the responses of these kinase gene fusions to small molecule inhibitors. We used break-apart fluorescence in situ hybridization (FISH) to identify genomic rearrangements in tissues from 59 patients with various types of malignant melanoma including acral lentiginous, mucosal, superficial spreading, and nodular. We identified four genomic rearrangements involving the genes BRAF, RET, and ROS1. Of these, three were confirmed by immunohistochemistry (IHC) or sequencing. We identified a RET fusion in an acral lentiginous melanoma, an ARMC10-BRAF fusion in an unknown primary melanoma, and an AGK-BRAF fusion in a superficial spreading melanoma. This is the first report of a RET fusion in melanoma and the ARMC10-BRAF fusion has not been previously described in melanoma. These fusions occurred in different subtypes of melanoma but all in tumors lacking known driver mutations.

We went on to generate patient-derived xenograft (PDX) models of both BRAF fusions to show that both the ARMC10-BRAF and AGK-BRAF kinase fusions are sensitive to downstream mitogen-activated protein (MAP) kinase pathway inhibition. We characterized these responses and identified differential responses between the different BRAF gene fusions. Our data suggests gene fusions are more common than previously thought and are actionable therapeutic targets.

Broader screening for kinase fusions in melanomas lacking known driver mutations should become part of routine clinical practice. iii

Dedication This thesis is dedicated to Dr. William Robinson

iv

Acknowledgements I want to first thank Dr. William Robinson who has been a pivotal influence on my life.

His patience, time, and support throughout my entire undergraduate career and throughout the course of this project been made me into the scientist I am today. When I first entered the lab,

Dr. Robinson handed me a paper. It was titled “Kinase gene fusions are frequent in Spitz neoplasms and spitzoid melanomas”. He told me, this is what we want to do. From that point on

I have been completely invested in studying kinase gene fusions and building this project from scratch. Dr. Robinson has always had faith in me. Even when I lost faith in myself, Dr.

Robinson encouraged me to keep moving forward and working hard. He has provided me opportunities, guidance, and support without which I would not be where I am today. Dr.

Robinson is a tremendous mentor and I am incredibly grateful for everything he has done for me.

Thank you.

Another outstanding mentor is Dr. Kasey Couts. Kasey is an amazing teacher and has put in so much time and energy to teaching me new techniques, developing my skills, and furthering my projects. Kasey is a wonderful role model who does her research diligently and with integrity. Kasey is an inspiration to me. She has taught me numerous protocols and procedures. Kasey has given me the tools I need to work independently at the bench. She also challenges me to think about the next step in the experiments, brings me into the creative process while designing projects, and has expanded so many more projects off of my initial data. Kasey and I have worked closely very closely and she even worked with me and taught me how to write an entire paper. I am extremely appreciative for all that she has done and I am very excited for working with her next year. v

Without these incredible mentors providing support and guidance, this work would not be possible.

The entire Robinson Lab including, Judson Bemis, Stacey Bagby, Carol Amato, Allison

Applegate, Rita Gonzalez, Magdelena Glogowska, Steven Robinson, Dr. Jennifer Hintzsche, and

Dr. Keith Wells have been such a big help in moving this project forward. I want to thank them for teaching me in the lab, answering all of my questions, and dealing with my boundless energy day-to-day. I also want to specifically acknowledge Stacey Bagby for her help with the patient- derived xenograft models. Additionally, I want to thank Maren Salzmann-Sullivan and Dr.

Isabel Schlaepfer for teaching me how to run my first-ever western blot and always answering all of my questions. I have enjoyed all of our laughs in the lab and I look forward to working the next year together.

A special thanks goes out to the donors of this research including the Heidi Horner

Foundation, the Amy Davis Foundation, and the Moore Family Foundation. I hope that this work makes you proud. This research would not be where it is at without your generosity.

Thank you to Dr. Marileila Varella-Garcia and her expertise on fluorescence in situ hybridization (FISH). Dr. Garcia allowed me to be extremely involved in the FISH studies and took the time and effort to train me throughout the process. I am so grateful to have worked with such a wonderful investigator.

I wish to thank the entire melanoma Scientific Advisory Board including Dr. John

Tentler, Dr. Yiqun Shellman, Dr. Aik-Choon Tan, Dr. Matthew Rioth, Dr. Theresa Medina, Dr.

Joshua Wisell, and Dr. Neil Box for their guidance and support on this project. vi

I want to thank my thesis committee including Dr. Joe Falke, Dr. Robert Parson, Dr.

Natalie Ahn, Dr. Jennifer Martin, Dr. William Robinson, and Dr. Tentler. I am very excited to present you this research and am very grateful for your service on my committee.

A special thanks to the patients who donated blood and tissue to the International

Melanoma Biorepository and Research Laboratory. I wish to also acknowledge the entire melanoma tissue bank whose efforts make such a unique resource available for biomedical research.

Lastly, I wish to thank my family and Ethan Cabral who have supported me throughout the entire process. I am so appreciative to have them in my life. vii

Contents Chapter 1. Introduction……………………………………………………………………………1

1.1 The Mitogen-Activated Protein Kinase Pathway in Melanoma…………………………..1

1.2 The Etiology of Chromosomal Rearrangement…………………………………………...3

1.3 Kinase Gene Fusions in Melanoma……………………………………………………….5

1.4 Overview…………………………………………………………………………………..8

Chapter 2. Kinase Gene Fusions in Defined Subsets of Melanoma………………………………9

2.1 Introduction………………………………………………………………………………..9

2.1.1 Subsets of Melanoma……………………………………………………………..9

2.1.2 Gene Fusions as an Alternative Kinase Activation Mechanism in Melanoma…..10

2.1.3 Overview…………………………………………………………………………11

2.2 Materials and Methods………………………………………………………...... 12

2.2.1 Melanoma Patient Samples…………………………………………………...….12

2.2.2 Break-apart Fluorescence In-situ Hybridization……………..………………….12

2.2.3 Immunohistochemistry………………………………………………………….13

2.2.4 Next-generation Whole Exome Sequencing…………………….………………13

2.2.5 Targeted RNA Sequencing……………………………………………………...14

2.2.6 RT-PCR, Sanger Sequencing……………………………………………………15

2.2.7 Quantitative Real-Time PCR……………………………………………………15

2.3 Results……………………………………………………………………………………17

2.3.1 Break-apart FISH Identifies Genomic Alterations in RET, ROS1, NTRK1, and

BRAF……………………………………………………………………………………..17 viii

2.3.2 NTRK1, RET, and ROS1 Expression in Tumors with Rearrangements or

Genomic Alterations……………………………………………………………………..21

2.3.3 Characterization of Patient BRAF Rearrangements……………………………..23

2.3.4 Classification of Genomic Alterations…………………………………………..26

2.3.5 Kinase Fusions Occur in Pan-negative Patient Samples………………………...29

2.4 Discussion………………………………………………………………………………..32

2.4.1 Gene Fusions and Rearrangements in Melanoma……………………………….32

2.4.2 Current Methods for Gene Fusion Detection……………………………………33

2.5 Overview…………………………………………………………………………………34

Chapter 3. BRAF Kinase Fusions Exhibit Differential Responses to Targeted Therapy In Vivo..35

3.1 Introduction………………………………………………………………………………35

3.1.1 Targeted Therapy in Melanoma…………………………………………………35

3.1.2 Success in Targeting BRAF Kinase Fusions in Melanoma……………………...36

3.1.3 Overview…………………………………………………………………………37

3.2 Materials and Methods…………………………………………………………………...38

3.2.1 Melanoma Patient Samples……………………………………………………...38

3.2.2 Patient-derived Xenograft (PDX) Treatment Models……………………………38

3.2.3 Immunoblotting………………………………………………………………….38

3.3 Results……………………………………………………………………………………40

3.3.1 BRAF Kinase Fusions are Effective Therapeutic Targets in Melanoma………..40

3.3.2 Differential Responses are Observed in Melanomas with Different BRAF Gene

Fusions…………………………………………………………………………………...43

3.4 Discussion………………………………………………………………………………..46 ix

3.4.1 Targeting BRAF Gene Fusions with Small Molecules…………………………..46

3.4.2 Differential Sensitivity of BRAF Kinase Fusions……………………………….46

3.5 Overview…………………………………………………………………………………47

Chapter 4. Investigating the Role of 5’ Gene Partners in BRAF Kinase Fusions in Melanoma…49

4.1 Introduction………………………………………………………………………………49

4.1.1 Aberrantly Activated BRAF is a Common Therapeutic Target in Melanoma…..49

4.1.2 Overview………………………………………………………………………...50

4.2 Materials and Methods…………………………………………………………………...51

4.2.1 Computational Protein Modeling………………………………………………..51

4.2.2 Plasmids and Stable Gene Expression…………………………………………..51

4.2.3 Flow Cytometry and Florescence-activated Cell Sorting (FACS)……………...51

4.2.4 Quantitative Real-Time PCR…………………………………………………….51

4.2.5 Immunoblotting………………………………………………………………….52

4.2.6 Cell Viability……………………………………………………………………..52

4.3 Results……………………………………………………………………………………53

4.3.1 BRAF Kinase Fusions with Different 5’ Gene Partners Exhibit Different

Functional Characteristics………………………………………………………………..53

4.3.2 Computational Modeling Reveals Variable Kinase Domain Structures…………63

4.4 Discussion………………………………………………………………………………..65

4.4.1 The Impact of the 5’ Gene Partner in BRAF Kinase Fusions……………………65

4.4.2 Chimeric Kinase Structures……………………………………………………...65

4.4.3 Chromosome 7 Instability………………………………………………………..66

4.5 Overview…………………………………………………………………………………66 x

Chapter 5. Conclusion and Future Directions……………………………………………………68

xi

Figures Table of Content

1.1 Structural alterations resulting in chromosomal rearrangement…………………………...…4

1.2 The mitogen-activated protein (MAP) kinase signal transduction pathway…………...……..7

2.1 FISH images show rearrangements and alterations in different melanoma subtypes…………………………………………………………………………………………..19

2.2 Intratumor heterogeneity identified in patient #29………………………………………….20

2.3 Immunohistochemistry analysis of tumor samples with RET, NTRK1, and ROS1 genomic rearrangement and NTRK1 unknown genomic alterations……………………………………...22

2.4 Positive and negative immunohistochemistry control sections……………………………..24

2.5 Characterization and validation of BRAF fusions in rearranged samples…………………...25

2.6 Real-Time PCR for ROS1 rearrangement identified by FISH……………………………...28

2.7 Genomic alterations occur in melanomas without hotspot mutations in common driver genes in all subtypes……………………………………………………………………………………30

3.1 BRAF kinase fusions are sensitive to MAP kinase pathway inhibitors in vivo……………..42

3.2 Differential expression of downstream signaling targets in the AGK-BRAF and ARMC10-

BRAF kinase fusions……………………………………………………………………………..45

4.1 Previously identified BRAF fusion transcripts used in overexpression lentiviral vector system model…………………………………………………………………………………….54

4.2 Stringent gating ensures collection of high expressing ZsGreen1 from stable cell lines…...56

4.3 Expression of fusion constructs in stable cell lines…………………………………………57

4.4 Validation of BRAF fusion breakpoints in established cell lines……………………………58

4.5 mRNA expression of BRAF kinase fusions in stable cell lines……………………………..59 xii

4.6 Differential activity is observed between the BRAF kinase fusions with different 5’ gene partners…………………………………………………………………………………………...60

4.7 Inhibitor response of cell line constructs to MEK inhibitor, trametinib…………………….62

4.8 Protein structures of BRAF fusions with different 5’ gene partners, wild type BRAF, and

BRAFV600E……………………………………………………………………………………….66

xiii

Tables Table of Content

2.1 Clinical characteristics of melanoma patients……………………………………………….18

2.2 Classification of genomic alterations based on findings from FISH, IHC, and targeted RNA sequencing………………………………………………………………………………………..27

2.3 Kinase fusion frequency in melanomas……………………………………………………..31

3.1 Clinical characteristics of patient #9 and patient #56……………………………………….41

4.1 Reference information for each 5’ gene partner…………………………………………….55

1

Chapter 1. Introduction

1.1 The Mitogen-Activated Protein Kinase Pathway in Melanoma

Metastatic melanoma is an aggressive cancer, which can rapidly progress resulting in a low overall survival rate. For many years, the 5-year survival rate for metastatic melanoma was less than 10% until recently when molecularly targeted therapies and various forms of immunotherapy came to the forefront.1,2 In the past 5 years, molecular targeted therapies have begun to change the outlook in melanoma.2 Small molecule inhibitors are an effective treatment in a subset of melanoma patients with activating mutations. Approximately, 50% of cutaneous melanomas harbor an activating BRAFV600E mutation.3 Patients with BRAFV600E mutations often benefit from targeted therapy with inhibitors specific to the mitogen-activated protein (MAP) kinase pathway.4 Vemurafenib, an inhibitor specific to the BRAFV600E mutation, and trametinib, a downstream MEK inhibitor, are now routinely used tyrosine kinase inhibitors in patients with mutations in BRAF. Such inhibitors are widely available to melanoma patients but are effective only in patients whose cancer has activating mutations in the MAP kinase pathway. For melanoma patients whose cancer does not have a BRAFV600E mutation or other activating mutations, molecular therapies are not effective. The melanomas that lack hotspot mutations in common driver genes are known as pan-negative.5 These comprise about 40% of all patients with metastatic melanoma.6 This population of pan-negative melanoma patients have fewer treatment options and a poorer prognosis.7 Therefore, identifying effective treatments for pan- negative melanoma patients is an important and pressing issue.

Recently, there has been evidence that patients lacking activating mutations in the MAP kinase pathway may still respond to targeted therapy as a result of alternative pathway 2 activation.8-10 One such mechanism of alternative pathway activation is gene rearrangement.

Kinase gene rearrangements have been described as tumorigenic events in both melanoma and other cancers, which may lead to constitutive phosphorylation of the MAP kinase pathway and other pathways.11 These structural aberrations often arise in the absence of a known driver mutation (pan-negative) and offer potential therapeutic targets for patients with no common driver mutations.12 However, the prevalence of these genomic rearrangements has not been thoroughly assessed across melanoma subtypes. As noted above, gene rearrangements have been described in various cancers including, leukemia, thyroid cancer, lung cancer, and skin cancer.11,13-18 Previous findings in other cancers have demonstrated that activating, in-frame gene rearrangements may also serve as therapeutic targets.8,9 Yet there have been few reports of actionable gene rearrangements in melanoma and their documented their sensitivity to inhibitors.10-12 While fusions involving the kinase portion of BRAF with multiple partners has been described in several cancers,19 many fundamental questions about these kinase gene fusions remain unanswered.

The frequency of kinase gene fusions across different melanoma subtypes has yet to be determined. Currently, there is no published data that compares the responses of melanomas driven by BRAF gene fusions with different 5’ partners to inhibitors of the MAP kinase pathway.

Additionally, understanding differential responses between fusion transcripts could separate driver gene fusions from passenger gene fusions. Establishing how these different rearrangements may activate pathways and characterizing their responses to tyrosine kinase inhibitors in melanoma is essential in making such therapies available to patients. To do so, it is important to understand the differences between true driver gene fusions and inactive passenger gene fusions. 3

1.2 The Etiology of Chromosomal Rearrangement

Tumor development and progression in melanoma is often driven by phosphorylation of the MAP kinase pathway though aberrant activation of the BRAF kinase.20 BRAF may be activated by mutations, amplifications, or recently discovered, gene fusions.3 Gene fusions are a result of gene rearrangement, which may be a consequence of intrachromosomal rearrangement or an interchromosomal rearrangement (Figure 1.1). Intrachromosomal rearrangements are the result of gene deletions, duplications, and inversions while interchromosomal rearrangements are resultant from translocations or inverted translocations. Gene fusions have been described in cutaneous melanomas18,19 and some other types of skin cancers16,17 where 3’ kinase genes have been found to be fused to a subset 5’ gene partners.

Chromosomal translocations result from non-homologous end joining (NHEJ),21,22 however, the mechanism of somatic chromosome breakage and subsequent chimeric gene fusion, resulting in oncogenesis, has not been fully investigated. Previous studies have predicted that microhomology-mediated break-induced replication, chromosome shattering (chromothripsis), or mutations leading to genomic instability could drive the formation of these structural alterations.23-25 Select gene fusions identified in leukemia have been determined to be resultant of overactive restriction enzymes such as recombinase activating 1 (RAG1) and activation- induced cytidine deaminase (AID) respectively involved in generating the variable region of lymphocyte receptors and isotype switching.26 In addition, there are few reports of germline mutations in BRCA1 and large NF1 deletions resulting in chromosomal instability.27,28 It has been postulated that germline and somatic chromosomal rearrangements may be derived from similar mechanistic origins.29 However, further research on genomic instability resulting in gene 4

Figure 1.1. Structural alterations resulting in chromosomal rearrangement.

Schematic of gene rearrangements including intrachromosomal rearrangements and interchromosomal rearrangements.

5 fusions is required to understand the exact nature of these structural alterations.

1.3 Kinase Gene Fusions in Melanoma

Genomic rearrangements have been identified across many cancers but have yet to be thoroughly investigated in melanoma. A recent study examined data from 328 cutaneous melanomas from the TCGA database and detected 223 potential rearrangements including 3’ gene partners, BRAF and RAF1.18 Genomic rearrangements incorporating kinase genes ALK,

BRAF, MET, NTRK1, RET, and ROS1 have also been reported in begin skin cancer in Spitz nevi and Spitzoid melanoma.16,17 However, expressed activating, in-frame rearrangements in melanoma have only been described with kinase genes BRAF, MET, and most recently, ROS1. 10-

12,16,30 Determining if these activating and in-frame rearrangements are actionable targets in melanoma patients are essential findings that have been incompletely investigated. However, a recent study from our laboratory identified the first case of an activating GOPC-ROS1 gene rearrangement in a pan-negative acral lentiginous melanoma patient.30 This patient exhibited a durable partial response demonstrating that kinase gene fusions are actionable targets in patients with metastatic melanoma.30 Nevertheless, additional research must be done to understand the entirety of these gene fusion events.

The most commonly identified gene fusion in melanoma are those involving the

BRAF.10,12,19 The BRAF kinase has been well known to play a role in the development of melanoma. BRAF is a component of the MAP kinase pathway (Figure 1.2A) which in its face- to-face dimerized form binds to MEK to form a tetramer RAF-MEK complex to signal for cell growth and proliferation (Figure 1.2B).31 Wild type BRAF signals only through homodimerization or heterodimerization. The only example of BRAF signaling as monomer is when a BRAFV600E mutation is present and is driving signal transduction.32 The 5’ portion of the 6

BRAF transcript contains a RAS auto-inhibitory domain, which is lost during rearrangement.11

As a result, the BRAF kinase may become constitutively phosphorylated when fused to an expressed 5’ gene partner.11,12 These activated pathways may be targeted using signal transduction inhibitors acting downstream in the MEK and ERK pathway, but are resistant to treatment with the BRAFV600E inhibitor, vemurafenib.11,12 BRAF rearrangements may become constitutively activated as a result of losing the RAS auto-inhibitory domain, however, the 3’ region of BRAF often remains wild type.12 It is not known whether BRAF kinase fusions signal as homodimers or heterodimers with wild type BRAF, and whether they might dimerize with other members of the RAF family. There is only one report of a BRAF fusion heterodimerizing with a CRAF protomer.33 These fundamental questions and others must be investigated to fully understand the BRAF fusion mechanism of action and their potential therapeutic targets.

7

Figure 1.2. The mitogen-activated protein (MAP) kinase signal transduction pathway.

Small molecule inhibitors target activated kinases. The BRAFV600E inhibitor, vemurafenib, is specific to cells that harbor the BRAFV600E mutation. Trametinib, a MEK inhibitor, targets phosphorylated MEK kinases. SCH722984 is an ERK downstream inhibitor. These inhibitors prevent signaling phosphorylation which halts cell proliferation. 8

1.4 Overview

Metastatic melanoma survival rates are amongst the lowest of all cancers. Since melanoma rapidly metastasizes and often doesn’t respond to the usual kinds of cancer chemotherapy, patient outcome is poor with stage IV melanoma five-year survival rates under 30% even with the latest molecularly targeted therapies and immunotherapies available.1,2 Identifying and characterizing gene rearrangements in melanoma may provide another avenue of therapeutic approach for patients with advanced melanoma. Understanding how these fusions occur, how they are activated, and can be inhibited provides another unique opportunity in the understanding and treatment of melanoma. To do this we need to understand the oncogenic potential of various

BRAF fusions and to discriminate between driver and passenger gene fusions.

In these studies, we have examined the prevalence of gene fusions in various subtypes of melanoma, assessed the response rates of these fusions to various inhibitors, and are currently characterizing the effect of different 5’ gene partners on BRAF activation. We have shown that gene fusions are enriched in a pan-negative subset of melanomas where these patients have limited treatment options and a worse prognoses. We have also shown that the sensitivity of these gene fusions to small molecule inhibitors is variable, between different gene fusions.

Preliminary studies in this thesis suggest this differential response could be dependent on the 5’ partner. Yet more research must be done to fully understand the magnitude of these findings.

The data presented in this thesis has clinical significance and may provide new treatment options and effective therapy for patients harboring such fusions. These studies could integrate new diagnostic strategies for detecting gene fusions and open a new treatment paradigm in melanoma clinics for patients with tumors driven by BRAF fusions or other kinase fusions. 9

Chapter 2. Kinase Gene Fusions in Defined Subsets of Melanoma

2.1 Introduction

2.1.1 Subsets of Melanoma

Metastatic melanoma is an aggressive cancer with high rates of metastases until recently had a 5-year survival rate of less than 15% when spread to distant sites.2,34 Approximately 90% of all melanomas arise on chronically sun-damaged skin or intermittently sun-exposed skin, traditionally classified into superficial spreading, lentigo maligna, and nodular subtypes.35-37 The second most common type of melanoma are those arising in the pigmented areas in the back of the eye (uveal melanoma) (4%). These melanomas are only partially related to UV exposure.38,39

Although less common, other types can arise on non-sun-exposed areas such as acral lentiginous melanomas and mucosal melanomas. Acral lentiginous melanomas account for 2-3% of all melanomas and occur on non-glabrous skin such as the palms of the hands, soles of the feet, and underneath fingernails and toenails.40,41 Mucosal melanomas (1%) arise in mucosa of the nasopharynx and sinuses, mouth, anorectal, and vulvovaginal regions.42 While these melanoma subtypes differ in their location and molecular characteristics, they are all aggressive once they have metastasized and typically respond poorly to currently available therapy.43

Nearly 50% of cutaneous melanomas arising in sun exposed sites have an activating mutation in BRAF of which V600E is the most common. When present, these patients can be treated with inhibitors specific for the BRAFV600E mutation or its downstream kinase MEK.3,44,45 However,

BRAFV600E mutations are less common in other melanoma subtypes (approximately 10-15% in acral lentiginous and mucosal melanomas).10,46-48 Other oncogenic driver mutations in genes such as NRAS, KIT, GNAQ, and GNA11 have been identified in different melanoma subtypes but 10 at a much lower frequency.20 Overall, approximately 25% of cutaneous melanomas and nearly

40-50% of non-cutaneous melanomas lack known common driver mutations (“pan-negative”) and have no targeted therapy options available.5

2.1.2 Gene Fusions as an Alternative Kinase Activation Mechanism in Melanoma

Gene fusions, have been identified across several cancer types and found in a small percentage of cutaneous melanomas providing an alternate mechanism of BRAF activation.10-

13,19,49 Genomic alterations include structural changes in chromosomes. Gene fusions are the result of genomic rearrangements that join two previously separated genes. Oncogenic fusions in common kinases such as BRAF, ALK, MET, ROS1, RET, and NTRK1 have been found in Spitz nevi and Spitzoid melanomas.16,17 Spitz neoplasms and Spitzoid melanomas are skin growths that exhibit intermediate histopathological features between nevi and melanoma.50 These tumors are rarely diagnosed in melanoma clinics because it is difficult to predict their malignant potential. Thus Spitz neoplasms and Spitzoid melanomas remain quite controversial yet it has been recurrently identified that these tumors are driven by structural rearrangement.16,17,51

Analysis of the TCGA database containing over 320 cutaneous melanomas identified over 220 different fusion events including BRAF and RAF1 kinases, however no fusions in ALK, ROS1,

RET, or NTRK1 were identified in this set of cutaneous melanomas.18 In a recent study, ALK fusions were reported in 4/30 (13%) of acral lentiginous melanomas, and none were identified in the mucosal melanomas.52 These studies suggest that the genes involved in oncogenic fusion events in melanoma may be subtype-specific, however the frequency of BRAF, ALK, ROS1,

RET, and NTRK1 fusions has not been systematically studied across different melanoma subtypes. Characterizing kinase fusions in specific melanoma subtypes is critical in understanding the development of these melanomas and devising new treatment strategies. 11

2.1.3 Overview

In this study, we used break-apart fluorescence in situ hybridization (FISH) to screen for kinase rearrangements in BRAF, ALK, ROS1, RET, and NTRK1 in different subtypes of melanoma from 59 patients. We identified genomic rearrangements in tumors from four patients involving BRAF, RET, and ROS1. These rearrangements occurred in superficial spreading, acral lentiginous, and unknown primary subtypes, and we demonstrated that three out of four rearrangements resulted in gene fusions discovered by IHC or targeted RNA sequencing. These findings highlight the differences in gene rearrangements between specific melanoma subtypes, and prompt further characterization of the gene fusion landscape in all types of melanomas.

12

2.2 Materials and Methods

2.2.1 Melanoma Patient Samples

Ninety-eight tissue samples obtained from 59 melanoma patients in the Melanoma

Research Clinics at the University of Colorado, Anschutz Medical Campus, were included in the analysis. These were obtained between 2008 and 2015 as part of the International Melanoma

Biorepository and Research Laboratory at the University of Colorado Cancer Center, informed after patient consent and approval by the Colorado Institutional Review Board (IRB# 05-0309).

A total of 98 samples, mostly from metastatic sties, were randomly selected from superficial spreading, nodular, acral lentiginous, mucosal, desmoplastic, and unknown primary melanoma subtypes. Formalin-fixed, paraffin-embedded (FFPE) tissue microarrays (TMAs) were made from 1.5 mm punches of melanoma patient tumor FFPE blocks.

2.2.2 Break-apart Fluorescence In-situ Hybridization

This technique uses two different colored fluorescent FISH probes designed in the 5’ and

3’ portion of a gene to identify potential gene fusions. The two signals will overlap in normal cells, but there will be only one signal or split apart signals in cells where a genomic rearrangement in the gene has occurred. Dual-colored FISH assays were performed using break- apart (BA) probes for ALK, BRAF, RET, and ROS1 genes, using commercial reagents (Abbott

Molecular). The NTRK1 probe was developed in our laboratory as previously described.15 The

BRAF BA probe set was developed for this study using BAC clones RP11-837G3 and RP11-

948O19 containing human DNA inserts from regions homologous to the 3’ and 5’ end of BRAF, respectively. The signals from these clones were overlapped or very close in the native copy of the gene. Assays in the tissue microarrays (TMAs) were performed as previously described.53 13

The BA FISH analyses were conducted as previously described with minor modifications.54 Duplicate tissue cores for one or more specimens per patient were analyzed. A total of 50 cells were scored in each core. For each probe set, 3’ and 5’ signals physically separated by ≥1 signal diameter were considered split. Specimens were considered positive for specific gene rearrangement if >15% of the cells showed split signals or single 3’ or 5’ signals.

2.2.3 Immunohistochemistry

Archival FFPE specimens were stained as previously described,55,56 with the following modifications and antibodies: NTRK1 staining was performed with citrate buffer target retrieval solution (TRS pH 6) and a one hour room temperature incubation with primary rabbit monoclonal [EP1058Y] TrkA antibody (ab76291, dilution: 1:500), followed by secondary antibody (Dako Envision+ Anti-Rabbit K4003, pre-prepared standard dilution57). RET staining was performed with citrate buffer target retrieval solution (TRS pH 6) and a one hour room temperature incubation with primary rabbit monoclonal [EPR2871] anti-RET antibody

(ab134100, dilution: 1:50), followed by secondary antibody (Dako Envision+ Anti-Rabbit

K4003, pre-prepared standard dilution57). ROS1 staining was performed using ROS1 monoclonal antibody (Cell Signaling D4D6®, dilution: 1:25) under previously reported conditions.58

2.2.4 Next-generation Whole Exome Sequencing

Whole exome sequencing was done on concordant DNA from tissue and blood samples from 51/59 melanoma patients. Blood samples were collected in PAXGene DNA tubes and stored at 4°C until processed. Depending on material available, tissue genomic DNA was 14 isolated using the DNeasy Blood and Tissue kit or the QiaAmp DNA FFPE kit. DNA concentration and purity was determined using Qubit (Thermo Fisher Scientific) and Agilent

2100 bioanalzyer analysis. 200 ng of genomic DNA was sheared using Covaris S220 at 150bp.

Sheared DNA was end-repaired and used to construct the exome library following Agilent

SureSelect XT Target Enrichment System for Illumina Paired End Multiplexed Sequencing

Library (cat# G9641B). Exome capture was done through hybridization using XT5 probe.

Resulting captured libraries were indexed and purified. The cDNA library was validated on the

Agilent 2100 Bioanalzyer using DNA-1000 chip. Libraries were sequenced on the Illumina

HiSeq 2000 with 125 bp pair-end reads. We obtained an average of 400X and 200X sequencing coverage for the cancer and normal exomes, respectively. Data for blood and tissue samples were analyzed using the data analysis pipeline IMPACT,59 and germ line variants identified in normal blood were removed.

2.2.5 Targeted RNA Sequencing

RNA was isolated from fresh frozen patient tumor tissue using the RNeasy Plus Mini Kit

(Qiagen). Tissues were homogenized with the TissueLyser II (2x 2 minutes at 30 Hz) and on- column DNAse I digest was performed per protocol using the RNAse-free DNAse I set

(Qiagen). Library preparation and sequencing were performed by the University of Colorado

Denver Genomics and Microarray Core. Targeted libraries were generated using the Ovation cDNA module and Ovation Fusion Panel Target Enrichment System (NuGEN, Inc.) with 250 ng

- 750 ng RNA input as previously described.60 Paired-end sequencing (2 x 75 bp) was performed on a MiSeq instrument (Illumina) and approximately 1.5 million total reads were collected for each sample. Fusions were detected by analyzing FASTQ files on the BaseSpace 15 data analysis hosting platform (Illumina) using the NuGEN Ovation Fusion Target App

(NuGEN).

2.2.6 RT-PCR, Sanger Sequencing

RNA was isolated from fresh frozen patient tumor tissue using the RNeasy Plus Mini Kit

(Qiagen). Tissues were homogenized with the TissueLyser II (2x 2 minutes at 30 Hz) and on- column DNAse I digest was performed per protocol using the RNAse-free DNAse I set

(Qiagen). RNA was reverse transcribed into cDNA using the Verso cDNA Synthesis Kit

(Thermo Scientific). PCR was performed using primers specific for the AGK-BRAF fusion

(CGCTTCGAAATCACTGGAAGAA (sense); CCACGAAATCCTTGGTCTCTAATC (anti- sense)) and ARMC10-BRAF fusion (TCGCAGCCTGAAGACTTAAC (sense);

GTGGAATAGCCCATGAAGAGTAG (anti-sense)). PCR products were purified using the

QIAquick PCR purification kit (Qiagen) and submitted to the Barbara Davis Center for sequencing using the BigDye Terminator Cycle Sequencing Ready Reaction kit version 3.1

(Applied Biosystems) with both forward and reverse primers used in the PCR reactions.

2.2.7 Quantitative Real-Time PCR

The real-time PCR reaction was done in triplicate using PowerUp SYBER Green master mix

(Thermo Scientific) and was analyzed on the StepOne Plus real-time PCR system (Applied

Biosystems). Primer sequences for ROS1 are as follow: exons 1-2

(TCCGAAGCTTGTCAATTTTGC (sense); TGTGCCAAGGTCAAGCTG (anti-sense)), exons

3-4 (AGGATGTCACTTTTGGAACTCTG (sense); TCTTCATATGCACCTTCCGC (anti- sense)), exons 36-37 (AAACTGACTCTGCGTCTCTTG (sense);

CTTCTCCTGGTCTGTGGAAC (anti-sense)), exons 40-41 16

(GATGGCTCCAGAAAGTTTGATG (sense); TGCACATAGTTTAACACATCAAGG (anti- sense)).

17

2.3 Results

2.3.1 Break-apart FISH Identifies Genomic Alterations in RET, ROS1, NTRK1, and BRAF

In order to identify genomic rearrangements in BRAF, ALK, RET, ROS1, and NTRK1 kinase genes, we used break-apart FISH to screen 98 samples from 59 patient tumors (27 primary and 32 metastatic) across multiple subtypes including superficial spreading, nodular, acral lentiginous, mucosal, lentigo maligna, desmoplastic, and unknown primary (Table 2.1).

We identified 4 samples that had genomic rearrangements (Figure 2.1A, D-F) and 2 samples that showed an atypical hybridization pattern of NTRK1 (Figure 2.1B-C).

One sample from a patient with acral lentiginous melanoma of the plantar surface of the foot (#30) had a RET hybridization pattern showing single copies of the 3’ probe consistent with a rearrangement (Figure 2.1A). Another sample from a patient with acral lentiginous foot melanoma (#29) had an atypical FISH pattern for NTRK1. In this sample, there was both clustering of the NTRK1 signal and separation of the 3’ and 5’ signals (Figure 2.1B). The distance between them was, however, below the scoring threshold used to indicate the presence of a break in the gene. Multiple samples from this case were examined by FISH and interestingly the different samples had discordant results. Two samples were classified as

NTRK1 atypical while two other samples from the same tumor were classified as negative for any atypical NTRK1 patterns suggesting intra-tumoral heterogeneity. H&E staining and NTRK1

FISH analysis of the original patient specimen showed the negative biopsies were taken from a region which contained only early stage in-situ melanoma cells, whereas the NTRK1 atypical biopsies were taken from a region of the same sample which contained more advanced vertical growth phase tumor cells (Figure 2.2). An atypical FISH pattern for NTRK1 was also observed in a nasal sinusoid mucosal melanoma sample (#41), where single signals of the 5’ portion of 18

Table 2.1. Clinical characteristics of melanoma patients No. Subtype Gender Age Sample Source Tissue Source

1 Superficial spreading M 75 Metastatic Lymph node 2 Superficial spreading M 46 Metastatic GI 3 Superficial spreading M 50 Metastatic Lymph node 4 Superficial spreading M 73 Metastatic Subcutaneous 5 Superficial spreading F 58 Metastatic Brain 6 Superficial spreading F 61 Metastatic Thyroid 7 Superficial spreading M 75 Metastatic Subcutaneous 8 Superficial spreading F 66 Metastatic Lymph node 9 Superficial spreading F 34 Metastatic Subcutaneous 10 Superficial spreading F 54 Primary Right upper chest 11 Superficial spreading F 62 Metastatic Subcutaneous 12 Nodular M 68 Metastatic Bone 13 Nodular M 66 Metastatic Subcutaneous 14 Nodular M 64 Metastatic GI 15 Nodular F 60 Metastatic Subcutaneous 16 Nodular M 88 Metastatic Subcutaneous 17 Nodular M 26 Metastatic Subcutaneous 18 Nodular M 78 Metastatic Subcutaneous 19 Nodular F 69 Metastatic Brain 20 Nodular M 74 Metastatic Subcutaneous 21 Nodular M 78 Primary Left upper arm 22 Nodular M 72 Metastatic Brain 23 Acral lentiginous M 60 Primary Right ankle 24 Acral lentiginous F 64 Primary Bottom of right foot 25 Acral lentiginous F 64 Metastatic Breast 26 Acral lentiginous M 61 Metastatic Subcutaneous 27 Acral lentiginous M 54 Primary Left 4th toe 28 Acral lentiginous M 62 Primary Right lateral foot 29 Acral lentiginous F 39 Primary Left thumbnail 30 Acral lentiginous F 77 Primary Left foot 31 Acral lentiginous M 46 Metastatic Subcutaneous 32 Acral lentiginous F 56 Metastatic Lymph node 33 Acral lentiginous M 38 Primary Left thumb 34 Acral lentiginous M 66 Primary Webspace of 4th and 5th toe 35 Acral lentiginous F 61 Primary Bottom of right foot 36 Acral lentiginous M 87 Primary Left great toe 37 Acral lentiginous M 78 Primary Sole of left foot 38 Acral lentiginous F 66 Primary Right third tow 39 Acral lentiginous F 56 Primary Left heel 40 Mucosal F 35 Metastatic Brain 41 Mucosal M 75 Primary Anorectal 42 Mucosal F 85 Primary Maxillary sinus 43 Mucosal M 78 Primary Pelvic cavity 44 Mucosal F 97 Primary Nasopharyngeal 45 Mucosal F 51 Primary Vaginal 46 Mucosal F 88 Primary Vulvovaginal 47 Mucosal F 61 Metastatic Lymph node 48 Mucosal F 72 Primary Rectum 49 Mucosal F 63 Primary Anal canal 50 Mucosal F 68 Metastatic Pancreas 51 Mucosal F 66 Primary Anorectal 52 Mucosal F 78 Primary Vaginal 53 Mucosal F 71 Primary Vulva 54 Unknown primary F 49 Metastatic Lymph node 55 Unknown primary M 59 Metastatic Subcutaneous 56 Unknown primary M 54 Metastatic Lymph node 57 Lentigo M 68 Metastatic Subcutaneous 58 Lentigo M 46 Primary Mid back 59 Desmoplastic M 57 Metastatic Lung

19

Figure 2.1. FISH images show rearrangements and alterations in different melanoma subtypes.

Representative FISH images from tumors with rearrangement or atypical staining patterns. (A) An acral lentiginous melanoma from patient #30 harboring a RET rearrangement where single copies of the 3’ probes are observed (white arrows). This sample is classified as a rearrangement. (B) An acral lentiginous melanoma from patient #29 shows an NTRK1 atypical FISH pattern including mini clusters of both probe signals (white arrows). This sample is classified as an unknown genomic alteration. (C) A mucosal melanoma with NTRK1 atypical FISH pattern where single copies of the 5’ probes are seen adjacent to intact copies of NTRK1 (white arrows). This sample is classified as an unknown genomic alteration. (D) A superficial spreading melanoma from patient #8 shows single copies of the 3’ ROS1 probe (white arrows). This sample is classified as a rearrangement. (E) An unknown primary melanoma from patient #56 shows single copies of the 3’ BRAF probe (white arrows). This sample is classified as a rearrangement. (F) A superficial spreading melanoma from patient #7 shows single copies of the 3’ BRAF probe. This sample is classified as a rearrangement. 20

Figure 2.2. Intratumor heterogeneity identified in patient #29.

Samples from the same biopsy specimen, taken from different sites, exhibit heterogeneity in the NTRK1 unknown genomic alteration and cellular morphology. The section of the tumor labeled (A) shows invasive vertical growth phase melanoma where tumor cells penetrate the dermoepidermal junction as shown by the H&E and atypical NTRK1 FISH pattern. The section of the tumor labeled (B) shows in situ horizontal growth phase melanoma as shown by the H&E and a normal NTRK1 FISH pattern. 21

NTRK1 were adjacent to an intact copy of the NTRK1 gene (Figure 2.1C). Both NTRK1 atypical hybridization patterns suggest the presence of an unknown genomic alteration but does not indicate a rearrangement or a gene fusion. These three cases are the first demonstrations of RET and NTRK1 genomic alterations in melanoma subtypes other than Spitz nevus and Spitzoid melanoma.

Samples from eleven superficial spreading cutaneous melanomas were examined and two had rearrangements. A ROS1 rearrangement was identified in a sample from patient #8 where single signals were observed for the 3’ probe (Figure 2.1D), and in a sample from patient #9 a

BRAF rearrangement was identified with split 3’ and 5’ signals (Figure 2.1F). Samples from four unknown primary melanomas were included in the analysis. A BRAF rearrangement with extra copies of single 3’ BRAF signals was observed in a metastatic lymph node in one patient with an unknown primary melanoma (Figure 2.1E). Overall, 2 out of 11 superficial spreading melanomas (18.2%), 1 out of 17 acral lentiginous melanomas (5.9%), and 1 out of 3 unknown primary tumors (33%) had genomic rearrangements.

2.3.2 NTRK1, RET, and ROS1 Expression in Tumors with Rearrangements or Genomic

Alterations

Rearranged or atypical hybridization patterns by FISH do not necessarily indicate a functional gene fusion event. We used immunohistochemistry (IHC) with kinase-domain specific antibodies for NTRK1, RET, and ROS1 to determine if protein was expressed in the rearranged or atypical samples. The sample from patient #30 with a RET rearrangement from an early-stage primary in-situ acral lentiginous melanoma showed protein expression suggesting the presence of a functional gene fusion (Figure 2.3A). 22

Figure 2.3. Immunohistochemistry analysis of tumor samples with RET, NTRK1, and ROS1 genomic rearrangements and NTRK1 unknown genomic alterations.

IHC analysis of tumors with rearrangement or atypical FISH hybridization pattern. (A) Tumor from patient #30, identified with a RET rearrangement by FISH, shows staining with the brown DAB chromagen (left panel) indicating expression of the RET protein. The H&E section of this tumor (right panel) shows large, pleomorphic tumor cells with some blood. (B) NTRK1 staining (black arrows) of a red chromagen in of an acral lentiginous melanoma from patient #29 with an unknown genomic alteration is shown (left panel) and the H&E stain of malignant tumor cells is shown (right panel). (C) Lower protein expression for NTRK1 is seen in scattered tumor cells precipitating the red chromagen (black arrow) in a sample from patient #41 (left panel) that was identified with an unknown genomic alteration by FISH. The matched H&E section (right panel) shows small amounts of melanin (brown pigment) are scattered throughout the section. (D) ROS1 IHC (left panel) of a superficial spreading melanoma from patient #8 found to have a ROS1 genomic rearrangement by FISH. The ROS1 IHC does not show any staining of the brown DAB chromagen (left panel) and thus is negative for ROS1 protein expression. The H&E stain of the tumor is shown (right panel). 23

The sample from patient #29 with an atypical NTRK1 FISH hybridization pattern showed scattered expression of the NTRK1 protein suggesting the atypical genomic alteration results in protein expression (Figure 2.3B). Expression of NTRK1 was observed only in scattered malignant melanocytes in the sample from patient #41 (Figure 2.3C) and was substantially different from patient #29. The conclusion from IHC is consistent with the FISH analysis which found different genomic events between the two tumors.

No IHC expression of ROS1 was demonstrated for patient #8 (Figure 2.3D). Real-time qPCR was also performed on tumor tissue and ROS1 transcript expression was undetectable, confirming the negative IHC result (Figure 2.4). These data suggest the genomic rearrangement did not result in a functional gene fusion.

2.3.3 Characterization of Patient BRAF Rearrangements

BRAF fusions in melanoma have been reported to occur frequently in tumor samples with a Spitz-like morphology.10,11 Both of our samples with BRAF rearrangements were examined by an experienced dermatopathologist, and did not show Spitz cytomorphology (Figure 2.5A). We performed targeted RNA sequencing for each sample positive for a BRAF rearrangement and identified the BRAF fusion transcripts and 5’ gene partners.

In the metastatic sample from patient #9 with superficial spreading melanoma, an AGK-

BRAF fusion transcript was found that fused exon 2 of AGK with exons 8 through 18 of BRAF

(Figure 2.5B). The sample from patient #56, a melanoma of unknown primary, was found to have an ARMC10-BRAF transcript with a fusion of exons 1-4 of ARMC10 and exons 11-18 of

24

Figure 2.4. Positive and negative immunohistochemistry control sections.

(A) RET IHC of positive control human colon stains with the brown DAB chromagen which is seen on the left panel and negative control colon is seen on the right panel. (B) NTRK1 IHC brain positive control is seen on the left panel with the scattered positive neurons staining with the red chromagen (black arrow). The negative control skin with melanin (brown pigment) in the sample is seen on the left panel. (C) ROS1 IHC HCC78 cell line positive control is seen on the left panel where cells stain for the brown DAB chromagen and negative control skin is seen on the right panel. 25

Figure 2.5. Characterization and validation of BRAF fusions in rearranged samples. (A) Representative H&E stained images from each BRAF fusion tumor show lack of Spitz associated morphology in these tumors. (B) Schematic of exonic breakpoints in the BRAF fusions. Targeted RNA sequencing was used to identify the breakpoints and 5’ partners for each BRAF fusion. (C) Sanger sequencing confirmation of BRAF fusion transcripts. RT-PCR products for each fusion were sequenced and chromatograms confirm in-frame transcripts for ARMC10-BRAF and AGK-BRAF fusions with the expected breakpoints.

26

BRAF (Figure 2.5B). The fusion and breakpoint were confirmed for both patient samples using

RT-PCR and Sanger sequencing (Figure 2.5C). AGK-BRAF fusions have been previously identified in melanomas,10,11 however, this is the first report of an ARMC10-BRAF fusion occurring in a melanoma.

2.3.4 Classification of Genomic Alterations

Break-apart FISH analysis identified genomic rearrangements in tumors from four melanoma patients. Tumor from patient #30 with a RET rearrangement also had positive RET

IHC staining. Because RET is normally hardly detectable, positive RET staining suggests upregulation due to a gene fusion. There was, however, not sufficient tumor material for confirmation by targeted RNA sequencing (Table 2.2). Tumor from patient #8 with a ROS1 rearrangement was negative for ROS1 protein expression by IHC, therefore a functional kinase fusion did not occur. Targeted RNA sequencing was performed on RNA from this tumor, and as expected, no ROS1 fusion transcripts were identified (Table 2.2). This result is consistent with our real-time qPCR data (Figure 2.7). BRAF rearrangements in tumors from patients #56 and #7 were confirmed using targeted RNA sequencing. Melanomas express endogenous BRAF protein so IHC cannot be used for validation (Table 2.2). Therefore, we identified a total of three kinase fusions occurring in BRAF and RET genes.

Tumors from two patients, #29 and #41, were negative for rearrangements in the break- apart FISH analysis but had atypical hybridization patterns for NTRK1. Although these tumors did not have rearrangements, they were IHC positive for NTRK1 suggesting unknown genomic alterations led to expression of NTRK1 protein. 27

Table 2.2. Classification of genomic alterations based on findings from FISH, IHC, and targeted RNA sequencing Technique Classification Unknown Targeted Kinase genomic No. Gene FISH IHC RNA seq Rearrangement fusion alteration

Patient #30 RET Positive Positive -* Yes Yes No Patient #29 NTRK1 Negative, atypical Positive - No No Yes Patient #41 NTRK1 Negative, atypical Positive - No No Yes Patient #8 ROS1 Positive Negative Negative Yes No No Patient #56 BRAF Positive - Positive Yes Yes No Patient #9 BRAF Positive - Positive Yes Yes No

“-“ indicates experiment was not performed * Attempted but unsuccessful due to lack of material 28

Figure 2.6. Real-Time PCR for ROS1 rearrangement identified by FISH.

(A) Schematic of primer design where two early primers are specific for the 5’ portion of ROS1 and two late primers are designed within the kinase domain of ROS1. (B) Analysis of ROS1 expression relative to GAPDH for patient #8 and HCC28. (C) ΔCT values of control GAPDH plotted for patient #8 and the control HCC28 cell line that is previously known to harbor a ROS1 fusion. 29

2.3.5 Kinase Fusions Occur in Pan-negative Patient Samples

Using next generation whole exome sequencing, we examined all samples for the presence of mutations in seven common melanoma driver genes (BRAF, NRAS, KRAS, HRAS, KIT, GNAQ, and GNA11) as well as NF1, which has been recently described as a major subgroup of melanoma.61 Consistent with previous observations, all 6 samples with genomic alterations detected by FISH, across different subtypes, occurred in pan-negative melanoma patients whose tumor lacked any known hotspot mutations in common driver genes (Figure 2.6).11,12 We observed an NF1 mutation in the ROS1 rearranged superficial spreading melanoma tumor, however this rearrangement did not result in a functional gene fusion. The frequency of kinase fusions within the subset of pan-negative melanomas showed strong enrichment in cutaneous, acral lentiginous, and unknown primary subtypes compared to the overall frequency (Table 2.3).

These findings suggest kinase fusions are more frequent across multiple subtypes of pan- negative melanomas.

30

Figure 2.6. Genomic alterations occur in melanomas without hotspot mutations in common driver genes in all subtypes.

Results from the FISH analysis and whole exome sequencing are presented for each melanoma patient sample. Samples were classified positive if > 15% of cells exhibited rearrangement patterns. Samples were classified as atypical if they had a consistent atypical FISH hybridization pattern and were not positive for rearrangement. Patients within each subtype are grouped by the presence of driver mutations, followed by the presence of genomic rearrangements or alterations. Dark grey bars indicate missing data for the FISH analysis (failure of hybridization for that specific gene) and the whole exome sequencing analysis (failure of library preparation for sequencing or insufficient material to use for sequencing analysis). Clinical information was analyzed for mutation testing in patients where samples were unable to be sequenced, and resulted in only partially missing driver gene mutation status for some patients. 31

Table 2.3. Kinase fusion frequency in melanomas

Number of kinase fusions

Subtype All patients Pan-negative patients

Superficial spreading 1/11 (9.1%) 1/6 (16.7%) Nodular 0/11 (0%) 0/7 (0%) Acral lentiginous 1/17 (5.9%) 1/5 (20%) Mucosal 0/14 (0%) 0/12 (0%) Unknown primary 1/3 (33.3%) 1/1 (100%) Lentigo 0/2 (0%) 0/1 (0%) Desmoplastic 0/1 (0%) 0/1 (0%) Total: 3/59 (5.1%) 3/33 (9.1%)

32

2.4 Discussion

2.4.1 Gene Fusions and Rearrangements in Melanoma

In this study we report a high frequency of kinase fusions in pan-negative melanoma samples across different subtypes of melanoma, particularly in those that lack common driver gene mutations. We identified a RET fusion in an acral lentiginous melanoma and unknown genomic alterations in NTRK1 in acral lentiginous and mucosal melanomas. These genomic alterations have not been previously described in melanoma. We did not find RET and NTRK1 alterations in cutaneous melanoma subtypes. Conversely, we observed BRAF fusions in cutaneous melanoma subtypes but not in acral lentiginous or mucosal melanomas. Finally, we discovered a ROS1 rearrangement in a cutaneous superficial spreading melanoma although lack of expression suggests this may not be a functional gene fusion. Our patient numbers are not large enough to conclude that these events are mutually exclusive.

Kinase fusions in NTRK1 and RET have been previously reported in Spitz neoplasms17 but have yet to be described in malignant melanoma. Here, we identified the first case of a RET fusion in a metastatic acral lentiginous melanoma. Acral lentiginous melanomas are known to have a lower mutational burden, but higher levels of genomic instability and more DNA copy number alterations62. We also found an atypical FISH pattern in an acral lentiginous melanoma which corresponded with NTRK1 protein expression. Interestingly, the atypical NTRK1 FISH pattern was observed in more advanced vertical growth phase cells but not early stage melanoma in-situ tumor cells, suggesting the development of the atypical NTRK1 genomic alteration may be associated with melanoma progression. Further investigation of the genomic NTRK1 region is required to fully understand our observations and the tumorigenic potential and significance of

NTRK1 alterations in melanoma. 33

In our cohort, we also observed a ROS1 rearrangement by FISH which did not express protein by IHC. To rule out the possibility of a false negative IHC result, we examined the samples with positive and negative controls that were stained within two hours of being cut

(Figure 2.4). The results were unchanged. Hence, the probability for a false negative outcome by IHC is unlikely. We also performed real-time qPCR and did not detect any ROS1 transcript expression (Figure 2.6). Alternate explanations for the inconsistency between the ROS1 rearrangement FISH and IHC results include a 5’ gene partner that is not expressed or an intergenic rearrangement where ROS1 does not have a 5’ gene partner.

2.4.2 Current Methods for Gene Fusion Detection

Detection of activating rearrangements in the clinical arena can be complex and time- consuming. Structural genomic alterations such as potential gene fusions can be identified using

FISH, but their expression must be confirmed by other methods since these alterations are not always functional. Using break-apart FISH as the only molecular diagnostic technique may also result in false negatives when the target genes are located on the same chromosome and are less than 10 Mbp apart, unless the fusion is generated by a chromosomal deletion which encompasses the sequences of one of the probes. Examples of this have been observed with the EML4-ALK rearrangements in lung adenocarcinoma.63 Likewise, IHC can generate false positives when protein expression is a consequence of pathway crosstalk.64 Multiplex analysis, such as combination of FISH and IHC, increases the likelihood of detection and has been shown to increase efficacy of detection in contrast to single screening techniques. Nonetheless, there may still be discordant outcomes, which could be a causal result of tumor heterogeneity, out-of-frame rearrangements, or minimally expressed fusion transcripts. Lambros et al. suggested using a massive parallel sequencing technique (MassARRAY) in conjunction with FISH and IHC to 34 decrease the possibility of false positive or false negative results.65 Overall there are methodological advantages and drawbacks to both FISH and IHC. In addition, FISH is of limited value in detecting new and previously undescribed rearrangements. This is best done using RNA techniques as have been previously described.65 We think multiplex analysis using a combination of FISH, RNA analysis, IHC, and other expression techniques will be needed to uncover a multitude of gene rearrangements important in the pathophysiology and treatment of melanoma and other cancers.

2.5 Overview

Altogether, our findings from FISH and IHC analyses may have important clinical significance. We observed a high frequency of kinase fusions in pan-negative melanomas across most clinical subtypes. Kinase fusions are actionable therapeutic targets in other cancers and the same may be true for melanoma.9,13,14,19 The demonstration here of activating kinase fusions in melanoma, particularly in pan-negative patient tumors, suggests expanded screening for fusions in patients with metastatic malignant melanoma should be considered. The data presented likely provides only a limited scope of the frequency of fusions in melanoma. Further work in this area is warranted and could uncover other gene fusions that are important in the development, growth and treatment of melanoma.

35

Chapter 3. BRAF Kinase Fusions as Therapeutic Targets

3.1 Introduction

3.1.1 Targeted Therapy in Melanoma

Small molecules have been effective in suppressing tumor growth across many cancers, including melanoma. Often, these small molecules are target specific with some small molecule inhibitors targeting only one mutation. These target specific inhibitors have fewer side effects in practice.66 However, some small molecules are multikinase inhibitors that are more promiscuous and will target entire kinase families, such as the RAF inhibitor, sorafenib.67 As a result, these promiscuous inhibitors have more side effects which may dramatically decrease quality of life and in certain cases may even be fatal.

In 2011, vemurafenib was introduced for the treatment of metastatic malignant melanoma in which BRAF was mutated. This was considered a major breakthrough in the treatment of melanoma and led the way for kinase inhibitors in other cancers.68 Vemurafenib, a nucleoside analog, contains nitrogenous heterocyclic rings which mimic purine rings.69 Vemurafenib is a type 1 inhibitor that competes for ATP in the ATP-binding site yet it confers specificity for the

V600E mutation by binding the active kinase where the Asp-Phe-Gly (DFG) regulator motif is flipped in.70,71 This conformation of the DFG motif is referred to as ‘DFG-in’. This DFG-in active conformation is stabilized by the 600E residue.70,71 Consequently, vemurafenib is only mildly specific for the V600E mutation since it can also bind non-V600E mutated active

BRAF.71

Since vemurafenib is a nucleoside analog and competes for ATP, there is a selective pressure to mutate.72 As a result, despite rapid tumor regression, prompt tumor resistance often 36 develops resulting in the short lived responses.73 One of the most common resistance mechanisms observed in BRAFV600E mutated melanomas is the acquisition of an NRAS mutation, specifically NRASQ61R.74,75 This mutation leads to the constitutive activation of CRAF thus bypassing the BRAF inhibitor.74,75 To combat this, most patients are now given a combination of vemurafenib and trametinib, a MEK inhibitor. Trametinib was FDA-approved in 2014 for the treatment of BRAF mutated melanomas.76 Trametinib, a third class of inhibitor, functions as an allosertic modulator to favor the inactive conformation of MEK leading to phosphoinhibition.76,77

Since MEK does not compete for ATP, there is less of a selective pressure to mutate and as a result, has longer-lasting effects than vemurafenib.78 When these two kinase inhibitors are used together, the response rates are considerably higher than when either is used alone.79 Currently, these therapies are available only to patients with activating mutations and are not FDA approved to treat melanoma patients with BRAF kinase fusions although our data and data from others suggests that MEK inhibitors alone may be effective in treating patients with activating BRAF fusions.

3.1.2 Success in Targeting BRAF Kinase Fusions

While the mutated BRAF response to targeted therapy has been well characterized, BRAF kinase fusions are relatively unexplored as therapeutic targets. There have been a few studies examining the responses of activating BRAF kinase fusions to treatment in vitro.11,12 These studies found that BRAF fusions are targetable in vitro. A previous study also identified a

PPFIBP2-BRAF kinase fusion in a superficial spreading melanoma patient.10 This patient was treated with trametinib and exhibited a dramatic response of 90% reduction of extracranial metastases and a 19% reduction of intracranial metastases.10 Altogether these findings suggest 37 that BRAF kinase fusions could be potential therapeutic targets but have yet to be fully characterized as actionable therapeutic targets.

Gene fusions are only actionable targets if the fusion is in-frame and activating.80 These genomic alterations may be out-of-frame, fused to an intergenic region of the genome, or have a low expression of their 5’ gene partner. Therefore, these gene rearrangements may be present but not targetable. Of the gene rearrangements that are in-frame and activating, it has not been determined which fusions are more sensitive or less sensitive to inhibition with small molecule inhibitors. To use these gene fusions as clinical therapeutic targets, it is important to understand the response rates of these fusions in a patient background.

3.1.3 Overview

In this study we generated patient derived xenograft models from two previously identified

BRAF kinase fusions, AGK-BRAF and ARMC10-BRAF81 and examined their responses in vivo to

MAP kinase pathway inhibitors specific to MEK1/2 and ERK 1/2. We observed differential responses and signaling patterns between the AGK-BRAF and ARMC10-BRAF kinase fusions.

These findings better characterize fusion responses to targeted therapy in a patient-specific background. Outlining therapeutic response rates may affect treatment regimens and provide a more personalized approach to medicine.

38

3.2 Materials and Methods

3.2.1 Melanoma Patient Samples

Melanoma patient samples were collected from the International Melanoma

Biorepository and Research Laboratory at the University of Colorado Cancer Center at Anschutz

Medical Campus. Patients were consented under the approval from the Institutional Review

Board (IRB# 05-0309). Viable and flash frozen tissue samples were obtained and used for experiments.

3.2.2 Patient-derived Xenograft (PDX) Treatment Models

Patient tumor from patient #32 and patient #6 were injected subcutaneously (SQ) into athymic nude mice after surgery at the Universityof Colorado Hospital. Growth and passage into

5 new athymic nude mice was continued 5 times until establishing the F5 generation. At this point tumors were implanted SQ into both flanks in >5 mice per treatment group. Tumors were permitted to grow until reaching an average of 150 – 300 mm3. The treatment groups include vehicle, vemurafenib (BRAFi), trametinib (MEKi), and SCH772948 (ERKi) and mice were treated for at least 21 days. Vemurafenib was administered at 50 mg/kg oral gavage daily, the trametinib treatment group was dosed by oral gavage daily at 1mg/kg, and the ERK inhibitor,

SCH772984, cohort was treated twice a day Intraperitoneal at 25mg/kg. The vehicle control group had no drug given.

3.2.3 Immunoblotting

Protein concentration was determined and normalized using the DCTM Protein Assay (Bio-

Rad). Samples were run on precast BoltTM 4-12% Bis-Tris Plus gels (Invitrogen) and transferred using the iBlot2 (Life Technologies) on iBlot®2 NC Mini Stacks (Invitrogen). Immunoblotting 39 was done with the following primary antibodies: phospho-MEK (Ser217/221) (dilution: 1:1000)

(9121, Cell Signaling), MEK (Ser217/221) (dilution: 1:1000) (9122, Cell Signaling), phospho- p44/42 MAPK (ERK1/2) (Thr202/Tyr204) (dilution: 1:1000) (4370, Cell Signaling), p44/42

MAPK (ERK1/2) (dilution 1:1000) (9102, Cell Signaling), and β-Actin (13E5) (dilution:

1:10000) (4970, Cell Signaling). Immunoblotting was done with the following secondary antibodies: IRDye® 680RD Goat (polyclonal) anti-Mouse IgG (H+L) Highly Cross-Adsorbed

(dilution: 1:15000) (925-68070, LI-COR) and IRDye® 800CW Goat (polyclonal) anti-Rabbit

IgG (H+L), Highly Cross-Adsorbed (dilution: 1:15000) (925-32211, LI-COR). Western blots were then quantified using the ImageJ software.

40

3.3 Results

3.3.1 BRAF Kinase Fusions are Effective Therapeutic Targets in Melanoma

We generated PDX mouse models with the previously identified AGK-BRAF and

ARMC10-BRAF from patient #9 and patient #56 in the Turner et al. study81. The AGK-BRAF gene fusion was originally identified in a tumor from a female patient with stage IV metastatic melanoma arising from a primary superficial spreading melanoma (Table 3.1). Patient #9 had a history of multiple pre-existing of dysplastic nevi (dysplastic nevus syndrome) and has generally fair skin and a history of multiple sunburns. Yet the melanoma tumors from this patient did not have a BRAFV600E mutation or any other hotspot mutations of common driver genes and was classified as pan-negative. The ARMC10-BRAF gene fusion was found in a tumor sample from a deceased male with stage IV metastatic melanoma from an unknown primary melanoma (Table

3.1). Patient #56 previously had vitiligo and extremely fair skin but had no hotspot mutations in common driver genes and thus was pan-negative.

We examined the response of tumors in PDX mice containing the AGK-BRAF and

ARMC10-BRAF fusions to treatment with small molecules inhibitors vemurafenib (BRAFi), trametinib (MEKi), and SCH722984 (ERKi) (Figure 3.1). Both models responded to downstream inhibition with trametinib and SCH722984. Trametinib, allosteric modulator, acts downstream of the BRAF kinase by inhibiting constitutive MAP kinase phosphorylation resulting in suppression of tumor proliferation. Similarly, SCH772984, a small molecule inhibits

ERK through allosteric modulation. The SCH772984 ERKi halts tumor growth by allosteric modulation downstream of the aberrantly activated BRAF kinase. Both PDX models were resistant to treatment with the BRAFV600E inhibitor, vemurafenib, since neither had this mutation.

This confirmed our hypothesis that the BRAF kinase fusions have different mechanisms of 41

Table 3.1. Clinical characteristics of patient #9 and patient #56 Patient Subtype Gender Age at Ethnicity Skin Type Stage Primary Sample Tissue Number Diagnosis Location Source Source

9 SSM F 36 Caucasian Fair IV Right Knee Metastatic Subcutaneous

56 Unknown M 56 Caucasian Extremely IV Unknown Metastatic Lymph Node Primary Fair ‘SSM’ denotes superficial spreading melanoma

42

Figure 3.1. BRAF kinase fusions are sensitive to MAP kinase pathway inhibitors in vivo.

(A) Growth curves of the BRAF fusion PDX models from patient #56 and patient #9. (B) Waterfall plots show tumor regression in the ARMC10-BRAF PDX model and displayed by percent of change in tumor volume (Δ = (V -V )/V ). ED B B 43

BRAF activation than stabilizing the active DFG-in conformation which is seen in canonically mutated BRAFV600E melanomas.

Although both of these fusion genes activated the MAP kinase pathway, their response to inhibition by MEK and ERK inhibition was significantly different. Tumors with the ARMC10-

BRAF fusion from patient #56 clearly were more sensitive to treatment with trametinib and

SCH722984 (Figure 3.1A). The ARMC10-BRAF model even showed tumor regression when treated with trametinib or SCH722984. In comparison, tumors with the AGK-BRAF fusion in a tumor from patient #9 were more resistant to treatment with no tumor regression after treatment with trametinib or SCH722984 (Figure 3.1B). These findings suggest that activating and in- frame BRAF kinase fusions are actionable therapeutic targets but BRAF kinase fusions may respond quite differently.

3.3.2 Differential Responses are Observed in Melanomas with Different BRAF Gene

Fusions

To further characterize these two BRAF fusions and better understand the differential response observed to targeted therapy, MAP kinase signaling was assessed (Figure 3.2).

Differences in pathway phosphorylation was observed between these two fusions. The tumor from patient #56 with an ARMC10-BRAF fusion showed stronger expression of phospho-MEK and phospho-ERK than the tumor from patient #9 (Figure 3.2A). Notably, the most significant difference was observed in the phospho-MEK expression between patient #56 and patient #9

(Figure 3.2B) where there was more phospho-MEK expression in patient #56 than there was in patient #9 suggesting there are differences in signaling of the MAP kinase pathway between these different BRAF fusions. 44

Phospho-ERK1 and phospho-ERK2 were then quantified and we observed that there was significant difference between phospho-ERK1 and phospho-ERK2 expression in a tumor with the AGK-BRAF kinase fusion from patient #9 where phospho-ERK2 was in greater abundance than phospho-ERK1 (Figure 3.2C). This difference in ERK isoform expression was not observed to the same extent in a tumor with the ARMC10-BRAF fusion from patient #56.

Additionally, we demonstrated a significant difference in abundance of phospho-ERK between patient #56 and patient #9. These results prompted us to question whether there is differential signaling and sensitivity of different BRAF gene fusions in different generic backgrounds and exactly what role the 5’ gene partner plays in chimeric fusions.

45

Figure 3.2. Differential expression of downstream signaling targets in the AGK-BRAF and ARMC10-BRAF kinase fusions.

(A) Immunoblotting of the mitogen-activated protein (MAP) kinase pathway of the patient #56 and patient #9. (B) Quantification of the ratio of phosphorylated MEK:total MEK in patient #56 and patient #9. (C) Quantification of phosphorylated ERK isoforms 1/2 in patient #56 and patient #9. (D) Sum of total phosphorylated ERK in patient #56 and patient #9. 46

3.4 Discussion

3.4.1 Targeting BRAF Gene Fusions with Small Molecules

In this study, we treated PDX models from two different tumors with BRAF kinase fusions, AGK-BRAF and ARMC10-BRAF, with inhibitors specific to the MAP kinase pathway.

Both BRAF fusions were sensitive to the downstream inhibition suggesting that BRAF kinase fusions are actionable therapeutic targets in melanoma patients. We also observed different responses in the PDX models derived from patient #9 and patient #56 where the tumors from patient #56, with an ARMC10-BRAF fusion, were more sensitive to treatment. This finding suggests that BRAF kinase fusions may respond differently which could be a consequence of patient background or a result of the gene fusion event. This the first study to examine treatment response of the ARMC10-BRAF fusion and report significant differential responses between

BRAF fusions in melanoma.

3.4.2 Differential Sensitivity of BRAF Kinase Fusions

In addition, the expression of the phosphorylated MEK and ERK isoforms was significantly different between the AGK-BRAF and the ARMC10-BRAF fusions. The differential expression of phosphorylated MEK may suggest the chimeric ARMC10-BRAF in a tumor from patient #56 may be activating downstream MEK more than the AGK-BRAF in a tumor from patient #9. Moreover, the tumor from patient #9 exhibits stronger expression of phospho-ERK relative to the expression of upstream phospho-MEK suggesting other pathways may be converging onto the downstream ERK kinase. These observations could be the result of patient- specific background, however, both melanomas are pan-negative and do not have any other known driver events. Another possibility for this observation is that the differential pathway 47 activation is a consequence of the intrinsic signaling capability and sensitivity of individual

BRAF gene fusions.

These BRAF gene fusions both had different exonic breakpoints and different 5’ gene partners which may be contributing to the differential signaling and responses found with two

BRAF fusions in the patient samples. We postulated that the 5’ partners in these fusions may play a significant role in downstream activation and response to inhibitors. The 5’gene partner

ARMC10 is an armadillo repeat containing protein 10 (also known as the splicing variant involved in hepatocarcinogenesis protein, SVH) which negatively affects the activity of TP5382 and thus is oncogenic when wild type ARMC10 is amplified as has been noted in hepatocellular carcinomas.83 The AGK gene codes for the acylglycerol kinase and phosphorylates acylglycerols in lipid and glycolipid metabolism.84 The products of this phosphorylation reaction are phosphatidic acid and lysophopatidic acid (LPA).85 An increase in LPA secretion has been shown to transactivate EGFR and ERK1/2.85 Our data suggests that the action of gene fusions may result not only from the 3’ gene partner, in this case BRAF, but there may also be effects from the 5’ partner. Further studies are needed to fully clarify this.

3.5 Overview

Our data supports previous findings that BRAF kinase fusions are actionable therapeutic targets which should be routinely screened for the in the clinical setting in melanoma patients with pan-negative tumors. In addition, we observed differential sensitivity to MAP kinase inhibition with different BRAF fusions, AGK-BRAF and ARMC10-BRAF. This could be a reflection of a patient-specific background and tumor heterogeneity or result from differences in the 5’ gene partners. Recognizing and characterizing different activating BRAF fusions and their 48 response rates to various inhibitors adds an entirely new dimension to individualized treatment and personalized medicine. 49

Chapter 4. Investigating the Role of 5’ Gene Partners in BRAF

Kinase Fusions in Melanoma

4.1 Introduction

4.1.1 Aberrantly Activated BRAF is a Common Therapeutic Target in Melanoma

Recently, suppression of noncanonical BRAF activation has been recognized as a potential therapeutic avenue in melanoma. A number of activating mutations other than

BRAFV600E have been described and new ones are being discovered. BRAF mutations such as

K601E and L597Q, which are found in 1-2% of melanomas, have been recognized as activating mutations that can be treated with targeted therapy using BRAFi/MEKi in combination.86,87

Additionally, less common subtypes of melanomas including, acral lentiginous and mucosal melanoma, have been found to have different genomic landscapes than that of cutaneous melanomas.48,88,89 Studies of non-mutational oncogenic events is also beginning to expand with

BRAF amplifications and aberrant BRAF splice forms being investigated as potential therapeutic targets,90,91 along with BRAF fusions as described here and elsewhere.10,11,33,81

To date, there is little known about comparative responses among different BRAF fusions in melanoma or other cancers.92 One study has shown a paradoxical action of different transcripts of the KIAA1549-BRAF kinase fusion in pediatric astrocytoma.92 However, there has yet to be any studies comparative studies assessing oncogenic potential between kinase fusions with different 3’ and 5’ gene partners in melanoma. Examining fundamental mechanisms of

BRAF kinase activation in chimeric BRAF fusion proteins may provide insight into new novel targets for BRAF fusion driven melanomas.

4.1.2 Overview 50

In this study have examined the behavior of six different BRAF kinase fusions. This data represents only preliminary studies in this work which is ongoing. Currently we have identified fundamental differences between six different BRAF kinase fusions. The conclusions and findings that will come from this study are important in understanding the treatment of patients with different BRAF fusions. We believe that this information will lead to new treatments and improve the outcome of all patients with BRAF gene fusions.

51

4.2 Materials and Methods

4.2.1 Computational Protein Modeling

Computational modeling was done using the Phyre tertiary protein modeling software methods from Kelley LA et al.93 which uses sequence homology detection to construct 3D models of from the amino acid sequence.

4.2.2 Plasmids and Stable Gene Expression

cDNA sequences for wild type BRAF, AGK-BRAF, ARMC10-BRAF, KIAA1549-BRAF,

PPFIBP2-BRAF, TRIM24-BRAF, and ZKSCAN1-BRAF were synthesized by GenScript and were cloned into the pLVX-EF1a-IRES-ZsGreen1 vector. The BRAFV600E mutation was generated from the BRAF sequence using site directed mutagenesis. Vectors were introduced and grown in

One Shot® Mach1TM T1 Phage-Resistant Chemically Competent E. coli (Invitrogen). DNA was extracted using the HiSpeed® Plasmid Midi Kit (Qiagen). Vectors were then transfected into human embryonic kidney (HEK) 293T cells (ATCC) using the Lenti-XTM Packaging Single

Shots (VSV-G) (Clonetch) and 8 μg/mL of polybrene (Santa Cruz Biotechnology). Virus was harvested and filtered at 48 hours and used to infect NIH3T3 cells (ATCC).

4.2.3 Flow Cytometry and Florescence-activated Cell Sorting (FACS)

Infected NIH3T3 cells were sorted using flow cytometry analysis on a MoFfo XDP 100 instrument (Beckman Coulter). ZsGreen1-positive cells were sorted using the FITC (blue laser) channel.

4.2.4 Quantitative Real-Time PCR 52

The real-time PCR reaction was done in triplicate using PowerUp SYBER Green master mix (Thermo Scientific) and was analyzed on the StepOne Plus real-time PCR system (Applied

Biosystems). Primer sequences are as follow: BRAF exons 3-4

(AGTGCTACCTTCATCTCTTTCAG (sense); TGTAACTCCACACCTTGCAG (anti-sense)),

BRAF exons 12-13 (TGATGTGGCAGTGAAAATGTTG (sense);

GTTGTGGCTTTGTGGAATAGC (anti-sense)).

4.2.5 Immunoblotting

Protein concentration was determined and normalized using the DCTM Protein Assay

(Bio-Rad). Samples were run on precast BoltTM 4-12% Bis-Tris Plus gels (Invitrogen) and transferred using the iBlot2 (Life Technologies) on iBlot®2 NC Mini Stacks (Invitrogen).

Immunoblotting was done with the following primary antibodies: phospho-BRAF (Ser445)

(dilution: 1:1000) (2696, Cell Signaling), Raf-B (C-19) (dilution: 1:250) (sc-166, Santa Cruz

Biotechnology), and Raf-B (F7) (dilution: 1:200) (sc-5284, Santa Cruz Biotechnology).

Immunoblotting was done with the following secondary antibodies: IRDye® 680RD Goat

(polyclonal) anti-Mouse IgG (H+L) Highly Cross-Adsorbed (dilution: 1:15000) (925-68070, LI-

COR) and IRDye® 800CW Goat (polyclonal) anti-Rabbit IgG (H+L), Highly Cross-Adsorbed

(dilution: 1:15000) (925-32211, LI-COR).

4.2.6 Cell Viability

Cell viability was determined using the Cell Titer-Glo® Luminescent Cell Viability Assay

(Promega). Cells were plated at 750 cells per well and treated the following day. Luminescence was measured 72 hours after treatment.

53

4.3 Results

4.3.1 BRAF Kinase Fusions with Different 5’ Gene Partners Exhibit Different Functional

Characteristics

We cloned six previously identified BRAF gene fusions with different 5’ gene partners

AGK, ARMC10, KIAA1549, PPFIBP2, TRIM24, and ZKSCAN1 (Figure 4.1). All of these BRAF kinase fusions were identified in melanoma or Spitzoid melanoma tumors12,19,81 except for the

KIAA1549-BRAF which was detected in spindle cell tumors of indeterminate histogenesis10,94

(Table 4.1). The KIAA1549-BRAF fusion was included in this study since it has been found to be an important diagnostic, prognostic, and potential therapeutic target in central nervous system associated tumors95,96. These BRAF kinase fusions were cloned into a lentiviral vector, which uses an internal ribosome entry site (IRES) to express ZsGreen1 and our BRAF kinase fusion constructs. Only clones overexpressing ZsGreen1 were collected using a stringent gating strategy (Figure 4.2). All transformed cells were expressing ZsGreen1 at high levels (Figure

4.3). We validated the breakpoint sequence, or in the case of wild type and V600E mutated

BRAF we sequenced across the 600 residue, in each stable cell line (Figure 4.4). We then examined the expression of BRAF by running quantitative real-time PCR across the BRAF gene in the early 5’ portion and in the late 3’ portion, specifically the kinase domain, of BRAF (Figure

4.5). All clones exhibited similar expression of the fusion constructs, however, TRIM24-BRAF, expression was significantly higher than the other BRAF kinase fusions.

Immunoblotting of the six different BRAF kinase fusions was done using C-terminus specific antibodies to analyze differential BRAF expression and activation of the MAP kinase pathway (Figure 4.6). Chimeric total and phospho-BRAF protein was detected in all fusions except the AGK-BRAF and KIAA1549-BRAF (Figure 4.6A-B). This could be a consequence of 54

Figure 4.1. Previously identified BRAF fusion transcripts used in overexpressed lentiviral vector system model.

AGK-BRAF (1,422 bps and ~60 kDa), ARMC10- BRAF (1,515 bps and ~53 kDa), KIAA1549-BRAF (6,090 bps and ~219 kDa), PPFIBP2- BRAF (1,266 bps and ~45 kDa), TRIM24-BRAF (2,691 bps and ~61 kDa), and ZKSCAN1-BRAF (1,923 bps and ~70 kDa) were all overexpressed in a generic back to examine differential sensitivity and signaling. 55

Table 4.1. Reference information for each 5’ gene partner 5’ Gene Parter AGK ARMC10 KIAA1549 PPFIBP2 TRIM24 ZKSCAN1

5’ Exonic 2 4 15 3 9 5 Breakpoint

3’ Exonic 8 11 9 11 9 10 Breakpoint

Total Exons in 16 7 20 24 19 6 5’ Gene Parter

Chromosome 7 7 7 11 7 7 Location

Cytogenetic 141,551,189- 103,074,881- 138,831,381- 7,513,770- 138,460,334- 100,015,581- Location of the 141,655,244 103,099,764 138,981,318 7,653,756 138,589,993 100,041,689 5’ Gene Partner

Melanoma Spindle Cell Subtype/Tissue Superifical Unknown Tumor of Superficial Not Otherwise Spitzoid Type from Spreading Primary indeterminate Spreading Specified Melnaoma Original Sample histogenesis

Reference Turner et al. Turner et al. Ross et al. Menzies et al. Hutchinson et al. Ross et al. (2017) (2017) (2016) (2015) (2013) (2016)

56

Figure 4.2. Stringent gating ensures collection of high expressing ZsGreen1 from stable cell lines.

Example of gating strategy from ZKSCAN1-BRAF construct used on all transformed cells to generate a homogenous high expressing population of cells. 57

Figure 4.3. Expression of fusion constructs in stable cell lines.

Micrographs of each transformed and sorted cell line where expression of fusion construct correlates to ZsGreen1 expression. Bright field images also highlight morphological characteristics each fusion cell line. 58

Figure 4.4. Validation of BRAF fusion breakpoints in established cell lines

Chromatograms from Sanger sequences across the breakpoints of the different BRAF fusion clones. 59

Figure 4.5. mRNA expression of BRAF kinase fusions in stable cell lines.

Analysis of exon imbalance using quantitative real-time PCR of each fusion construct using primers from the early portion of BRAF in exons 3-4 and primers from the late portion of BRAF in the kinase domain in exons 12-13. 60

Figure 4.6. Differential activity is observed between the BRAF kinase fusions with different 5’ gene partners.

(A) Immunoblot of C-terminus specific total BRAF and loading control β-Actin. (B) Immunoblot of C-terminus specific phospho- BRAF. (C) Immunoblot of N-terminus specific total BRAF. (D) Quantification of phosphorylated chimeric, wild type, or mutated BRAF:total BRAF protein in the fusion constructs identified to be expressing BRAF protein by western blotting. 61 expressing these kinase fusions in a fibroblast background and not in a melanoma or melanocyte specific background. Notably, many of the BRAF kinase fusions had bands at unexpected sizes of both phospho- and total BRAF this was not observed in the total BRAF using an N-terminus specific antibody (Figure 4.6C) suggesting that these are intrinsic features of the BRAF kinase fusions. Such bands were seen in KIAA1549-BRAF at approximately 100 kDa and in ZKSCAN1-

BRAF at approximately 110 kDa. Currently, the functional significance of these unexpected bands is undetermined. Of the expressing BRAF fusions, differential fusion phosphorylation was observed where both PPFIBP2-BRAF and ARMC10-BRAF exhibited less phosphorylation of the fusion protein than TRIM24-BRAF, ZKSACN1-BRAF, wild type BRAF, and mutated BRAFV600E

(Figure 4.6D).

Cell viability assays were used to examine treatment response between the different

BRAF kinase fusions (Figure 4.7) to downstream MAP kinase inhibition. All of the BRAF kinase fusions were more sensitive to treatment with the MEK inhibitor, trametinib, than wild type BRAF and the empty vector controls. However all fusions except the KIAA1549-BRAF were less sensitive to treatment with trametinib than the BRAFV600E cell line. Between the BRAF fusions there was some variability in sensitivity. The AGK-BRAF was found to be the least sensitive to treatment and interestingly, the KIAA1549-BRAF was the most sensitive. The IC50 values from the fusion constructs ranged from 9.39 nM to 2.33 nM respectively. Notably, this is a narrow range for IC50 values. However, since NIH3T3 have basal activation of the MAP kinase pathway these results become more difficult to interpret and will be repeated from a transient transfection of HEK293T cells where there is no basal phosphorylation in the MAP kinase pathway to produce more robust and reliable data. Our preliminary findings do suggest

62

Figure 4.7. Inhibitor response of cell line constructs to MEK inhibitor, trametinib.

Viability from luminescence based assay after 72 hour treatment with MEK inhibitor, trametinib. IC50 values of transformed BRAF fusions including AGK-BRAF, ARMC10-BRAF, KIAA1549-BRAF, PPFIBP2-BRAF, TRIM24-BRAF, and ZKSCAN1-BRAF are compared to the IC50 values of the empty vector, wild type BRAF, and BRAFV600E. 63 that many BRAF gene fusions exhibit intermediate responses where they are not as sensitive as

BRAFV600E mutated cells but are more sensitive than normal wild type BRAF to downstream treatment. Ultimately, these conclusions are only preliminary and more studies must be done to fully understand the sensitivity of these BRAF fusions to MAP kinase pathway inhibition.

4.3.2 Computational Modeling Reveals Variable Kinase Domain Structures

To analyze whether structural differences between different BRAF fusion transcripts, we computationally modeled the tertiary protein structure of the kinase domain of the six different

BRAF kinase fusions using the Phyre software2 (Figure 4.8).93 We compared these computational models to the protein structures of wild type BRAF, and mutated BRAFV600E. The modeled protein structures showed predicted the catalytic sites between in the BRAF kinase fusions. The only kinase fusion that exhibited an alternative catalytic site is the KIAA1549-

BRAF. These differences in the catalytic site may potentially impact the local and regional phosphorylation of the downstream kinase, MEK. Additionally, the software predicted distinct binding pockets for each fusion transcript. These findings have implications for binding efficacy where size and shape of the binding pocket may affect BRAF dimerization and the ability of the

BRAF kinase to bind MEK and other scaffolding proteins. Interestingly, we observed large differences in mutational sensitivity in BRAF wild type and BRAFV600E compared to the fusion proteins. The mutational sensitivity is a computational prediction of the susceptibility for an amino acid residue to be mutated. Our results suggest that BRAF wild type and BRAFV600E are more susceptible to mutation than any of the BRAF fusion proteins. This computational prediction supports previous evidence that BRAF gene fusions are enriched in tumors that lack driver common mutations.12 64

Figure 4.8. Protein structures of BRAF fusions with different 5’ gene partners, wild type BRAF, and BRAFV600E.

Computational modeling was done using the Phyre tertiary protein modeling software methods from Kelley LA et al. Nature Protocols 10, 845-858 (2015) and shows the prediction for the catalytic site, binding pocket, and mutational sensitivity in the kinase domain. 65

4.4 Discussion

4.4.2 The Impact of the 5’ Gene Partner in BRAF Kinase Fusions

To characterize BRAF kinase fusions with different 5’ gene partners we cloned different

BRAF fusion transcripts along with wild type BRAF and BRAFV600E into a generic background.

We demonstrated the expression of the BRAF gene fusions and show there are some similarities and differences in BRAF activation and inhibitor sensitivity between different BRAF gene fusions. We are currently working on better characterizing these constructs. Additional immunoblotting from transiently transfected HEK293T will done to clarify the differential MAP kinase signaling. This transient transfection into HEK293T cells will be a better model since

HEK293T cells have no baseline expression of the MAP kinase pathway while there is basal expression of the MAP kinase pathway in NIH3T3 cells. Sensitivity will also carried out using a transient transfection of HEK293T cell which produce more robust viability for similar reasons.

We also identified structural differences in theoretical models of BRAF gene fusions with different 5’ gene partners including AGK-BRAF, ARMC10-BRAF, KIAA1549-BRAF, PPFIBP2-

BRAF, TRIM24-BRAF, and ZKSCAN1-BRAF. In total, the finalized functional studies will be compared to our computational structures of the BRAF gene fusions to better understand the mechanism of aberrant BRAF activation as a result of gene fusion.

4.4.2 Chimeric Kinase Structures

To date, the effects of gene fusion on kinase function have not been evaluated and are poorly understood. Kinases require various proteomic interactions to function and are highly regulated proteins.97,98 Kinases are structured into two lobes, the N-lobe and the C-lobe, which upon activation by binding its substrate, will interact with one another.99,100 A bridge will be formed between the N-lobe and the αC-helix of the C-lobe to flex the hinge region and form a 66 closed structure by undergoing a conformational change when bound to its substrate.100-102 Upon rearrangement, the N-lobe of a kinase is lost or replaced with another protein. The impacts of this on kinase activity and function are unknown. Additionally, how gene fusions affect kinase dimerization, regulation, cellular localization, and active site accessibility are all also unknown.

Understanding how gene fusions effect the fundamental functioning of kinases are important questions that must be addressed to completely understand the consequences of gene tumor cells.

4.4.3 Chromosome 7 Instability

The mechanism of generating somatic gene fusions in melanoma is also very poorly understood. There has yet to be a study documenting the origin of a gene fusion resulting in melanoma development and progression. There have been postulations about aberrant DNA breaks resulting from hypermutation leading to chromosome rearrangement103 and dysfunction of the DNA repair system or chromothripsis resulting in overall chromosomal instability.25

Notably, five out the six BRAF gene fusions were from loci clustered in the same region on the same arm of chromosome 7 (Table 4.1). This region of chromosome 7 is recurrently rearranged not just in melanoma but across cancers, including but not limited, to colorectal, lung, and thyroid carcinoma.19,104 To date, how these gene fusions manifest and arise in recurrent hotspot locations of the genome with recurrent gene partners has not been evaluated and is an uninvestigated frontier in cancer research.

4.5 Overview

While this study is still in progress, these preliminary results may have clinical implications in that BRAF gene fusions are therapeutic targets. Our findings suggest there may be some variability between different BRAF gene fusions and thus it should be considered for patients to 67 have personalized treatment specific to the BRAF fusion transcript. Evaluating oncogenic potential between different BRAF gene fusions will discriminate passenger gene fusions from driver gene fusions. Thus these studies may provide a more advanced and personalized approach for targeted therapeutics.

68

Chapter 5. Conclusion and Future Directions

In this study, we identified kinase gene fusions are as enriched in the pan-negative melanomas where they are present in 9.1% of these melanomas. We have also examined BRAF gene fusions as possible actionable therapeutic targets and additionally have characterized their differential responses to MAP kinase inhibitors in vitro and in vivo to inhibitors of the MAP kinase pathway. More studies are required to fully understand the full scope and oncogenic potential BRAF gene fusions in melanoma. Future studies will include those to understand and establish the role of the 5’ gene partner. These studies will include transient transfections of

HEK293T cells to produce more robust signaling and sensitivity data. In addition, we will examine the invasive, clonogenic, and migratory properties of each of the BRAF gene fusions with 5’ gene partners AGK, ARMC10, KIAA1549, PPFIBP2, TRIM24, and ZKSCAN1. We will also examine the resistance mechanisms of the AGK-BRAF PDX model from patient #9 and the

ARMC10-BRAF PDX model from patient #56. Together these studies will bring the role of gene fusions in melanoma to the forefront and improve patient care.

Research on gene fusions has been a rapidly expanding field within the past five years.

Identification of kinase gene fusions across many cancers types including prostate, lung, thyroid, astrocytoma, gastric, breast, and melanoma have been described and treatments for these activating targeted fusions developed and tested.13,19,105-107 To date, however, no agent specific for activating fusion genes have been FDA approved other than the previously described kinase inhibitors for the BCR-ABL fusion in chronic myeloid leukemia, ALK fusions in non-small-cell lung cancer, and PDGFR fusions for myelodysplastic syndrome and chronic eosinophilic leukemia.108,109 69

In melanoma, there are only a few case reports of patients with activating fusion genes treated with a targeted agent.10,11 The research presented in thesis, in conjunction with existing data on gene fusions in melanoma, has prompted initiation of a new clinical trial at Anschutz

Medical Campus for melanoma patients identified with activating, in-frame BRAF gene fusions.

The ultimate goal of this research is to bring forth new knowledge that will benefit patients diagnosed with challenging and complex illnesses.

70

References 1. Balch CM, Buzaid AC, Soong SJ, et al. Final version of the American Joint Committee

on Cancer staging system for cutaneous melanoma. Journal of clinical oncology : official

journal of the American Society of Clinical Oncology 2001;19:3635-48.

2. Leung AM, Hari DM, Morton DL. Surgery for distant melanoma metastasis. Cancer

journal 2012;18:176-84.

3. Davies H, Bignell GR, Cox C, et al. Mutations of the BRAF gene in human cancer.

Nature 2002;417:949-54.

4. Chapman PB, Hauschild A, Robert C, et al. Improved survival with vemurafenib in

melanoma with BRAF V600E mutation. The New England journal of medicine

2011;364:2507-16.

5. Xia J, Jia P, Hutchinson KE, et al. A meta-analysis of somatic mutations from next

generation sequencing of 241 melanomas: a road map for the study of genes with

potential clinical relevance. Molecular cancer therapeutics 2014;13:1918-28.

6. Lovly CM, Dahlman KB, Fohn LE, et al. Routine multiplex mutational profiling of

melanomas enables enrollment in genotype-driven therapeutic trials. PloS one

2012;7:e35309.

7. Wang R, Zhang Y, Pan Y, et al. Comprehensive investigation of oncogenic driver

mutations in Chinese non-small cell lung cancer patients. Oncotarget 2015;6:34300-8.

8. Shaw AT, Engelman JA. Ceritinib in ALK-rearranged non-small-cell lung cancer. The

New England journal of medicine 2014;370:2537-9.

9. Shaw AT, Ou SH, Bang YJ, et al. Crizotinib in ROS1-rearranged non-small-cell lung

cancer. The New England journal of medicine 2014;371:1963-71. 71

10. Menzies AM, Yeh I, Botton T, Bastian BC, Scolyer RA, Long GV. Clinical activity of

the MEK inhibitor trametinib in metastatic melanoma containing BRAF kinase fusion.

Pigment cell & melanoma research 2015;28:607-10.

11. Botton T, Yeh I, Nelson T, et al. Recurrent BRAF kinase fusions in melanocytic tumors

offer an opportunity for targeted therapy. Pigment cell & melanoma research

2013;26:845-51.

12. Hutchinson KE, Lipson D, Stephens PJ, et al. BRAF fusions define a distinct molecular

subset of melanomas with potential sensitivity to MEK inhibition. Clinical cancer

research : an official journal of the American Association for Cancer Research

2013;19:6696-702.

13. Palanisamy N, Ateeq B, Kalyana-Sundaram S, et al. Rearrangements of the RAF kinase

pathway in prostate cancer, gastric cancer and melanoma. Nature medicine 2010;16:793-

8.

14. Downing JR. Targeted therapy in leukemia. Modern pathology : an official journal of the

United States and Canadian Academy of Pathology, Inc 2008;21 Suppl 2:S2-7.

15. Vaishnavi A, Capelletti M, Le AT, et al. Oncogenic and drug-sensitive NTRK1

rearrangements in lung cancer. Nature medicine 2013;19:1469-72.

16. Yeh I, Botton T, Talevich E, et al. Activating MET kinase rearrangements in melanoma

and Spitz tumours. Nature communications 2015;6:7174.

17. Wiesner T, He J, Yelensky R, et al. Kinase fusions are frequent in Spitz tumours and

spitzoid melanomas. Nature communications 2014;5:3116.

18. Cancer Genome Atlas N. Genomic Classification of Cutaneous Melanoma. Cell

2015;161:1681-96. 72

19. Ross JS, Wang K, Chmielecki J, et al. The distribution of BRAF gene fusions in solid

tumors and response to targeted therapy. International journal of cancer Journal

international du cancer 2015.

20. Hodis E, Watson IR, Kryukov GV, et al. A landscape of driver mutations in melanoma.

Cell 2012;150:251-63.

21. Lieber MR. NHEJ and its backup pathways in chromosomal translocations. Nature

structural & molecular biology 2010;17:393-5.

22. Lieber MR, Gu J, Lu H, Shimazaki N, Tsai AG. Nonhomologous DNA end joining

(NHEJ) and chromosomal translocations in humans. Sub-cellular biochemistry

2010;50:279-96.

23. Javadekar SM, Raghavan SC. Snaps and mends: DNA breaks and chromosomal

translocations. The FEBS journal 2015;282:2627-45.

24. Chiang C, Jacobsen JC, Ernst C, et al. Complex reorganization and predominant non-

homologous repair following chromosomal breakage in karyotypically balanced germline

rearrangements and transgenic integration. Nature genetics 2012;44:390-7, S1.

25. Weckselblatt B, Hermetz KE, Rudd MK. Unbalanced translocations arise from diverse

mutational mechanisms including chromothripsis. Genome research 2015;25:937-47.

26. Tsai AG, Lu H, Raghavan SC, Muschen M, Hsieh CL, Lieber MR. Human chromosomal

translocations at CpG sites and a theoretical basis for their lineage and stage specificity.

Cell 2008;135:1130-42.

27. Bengesser K, Vogt J, Mussotter T, et al. Analysis of crossover breakpoints yields new

insights into the nature of the gene conversion events associated with large NF1 deletions

mediated by nonallelic homologous recombination. Human mutation 2014;35:215-26. 73

28. Yu VP, Koehler M, Steinlein C, et al. Gross chromosomal rearrangements and genetic

exchange between nonhomologous chromosomes following BRCA2 inactivation. Genes

& development 2000;14:1400-6.

29. van Heesch S, Simonis M, van Roosmalen MJ, et al. Genomic and functional overlap

between somatic and germline chromosomal rearrangements. Cell reports 2014;9:2001-

10.

30. Couts KL, McCoach CE, Murphy D, Christiansen J, Turner J, Lewis KD, Robinsion WA,

Doebele RC. Acral lentiginous melanoma harboring a ROS1 fusion with clinical response

to entrectinib. Journal of Clinical OncologyTM Precision Oncology 2017;107.17.101.186.

31. Haling JR, Sudhamsu J, Yen I, et al. Structure of the BRAF-MEK complex reveals a

kinase activity independent role for BRAF in MAPK signaling. Cancer cell 2014;26:402-

13.

32. Gowrishankar K, Snoyman S, Pupo GM, Becker TM, Kefford RF, Rizos H. Acquired

resistance to BRAF inhibition can confer cross-resistance to combined BRAF/MEK

inhibition. The Journal of investigative dermatology 2012;132:1850-9.

33. Kim HS, Jung M, Kang HN, et al. Oncogenic BRAF fusions in mucosal melanomas

activate the MAPK pathway and are sensitive to MEK/PI3K inhibition or MEK/CDK4/6

inhibition. Oncogene 2017.

34. Chang AE, Karnell LH, Menck HR. The National Cancer Data Base report on cutaneous

and noncutaneous melanoma: a summary of 84,836 cases from the past decade. The

American College of Surgeons Commission on Cancer and the American Cancer Society.

Cancer 1998;83:1664-78. 74

35. Bulliard JL, De Weck D, Fisch T, Bordoni A, Levi F. Detailed site distribution of

melanoma and sunlight exposure: aetiological patterns from a Swiss series. Annals of

oncology : official journal of the European Society for Medical Oncology / ESMO

2007;18:789-94.

36. Whiteman DC, Stickley M, Watt P, Hughes MC, Davis MB, Green AC. Anatomic site,

sun exposure, and risk of cutaneous melanoma. Journal of clinical oncology : official

journal of the American Society of Clinical Oncology 2006;24:3172-7.

37. Shain AH, Bastian BC. From melanocytes to melanomas. Nature reviews Cancer

2016;16:345-58.

38. Singh AD, Turell ME, Topham AK. Uveal melanoma: trends in incidence, treatment, and

survival. Ophthalmology 2011;118:1881-5.

39. Pane AR, Hirst LW. Ultraviolet light exposure as a risk factor for ocular melanoma in

Queensland, Australia. Ophthalmic epidemiology 2000;7:159-67.

40. Phan A, Touzet S, Dalle S, Ronger-Savle S, Balme B, Thomas L. Acral lentiginous

melanoma: histopathological prognostic features of 121 cases. The British journal of

dermatology 2007;157:311-8.

41. Bradford PT, Goldstein AM, McMaster ML, Tucker MA. Acral lentiginous melanoma:

incidence and survival patterns in the United States, 1986-2005. Archives of dermatology

2009;145:427-34.

42. Chang CL, Schabert VF, Munakata J, et al. Comparative healthcare costs in patients with

metastatic melanoma in the USA. Melanoma research 2015;25:312-20.

43. Bhatia S, Tykodi SS, Thompson JA. Treatment of metastatic melanoma: an overview.

Oncology 2009;23:488-96. 75

44. McArthur GA, Chapman PB, Robert C, et al. Safety and efficacy of vemurafenib in

BRAF(V600E) and BRAF(V600K) mutation-positive melanoma (BRIM-3): extended

follow-up of a phase 3, randomised, open-label study. The Lancet Oncology

2014;15:323-32.

45. Holderfield M, Deuker MM, McCormick F, McMahon M. Targeting RAF kinases for

cancer therapy: BRAF-mutated melanoma and beyond. Nature reviews Cancer

2014;14:455-67.

46. Yamazaki N, Tanaka R, Tsutsumida A, et al. BRAF V600 mutations and pathological

features in Japanese melanoma patients. Melanoma research 2015;25:9-14.

47. Lee JH, Choi JW, Kim YS. Frequencies of BRAF and NRAS mutations are different in

histological types and sites of origin of cutaneous melanoma: a meta-analysis. The British

journal of dermatology 2011;164:776-84.

48. Sasaki Y, Niu C, Makino R, et al. BRAF point mutations in primary melanoma show

different prevalences by subtype. The Journal of investigative dermatology

2004;123:177-83.

49. Stransky N, Cerami E, Schalm S, Kim JL, Lengauer C. The landscape of kinase fusions

in cancer. Nature communications 2014;5:4846.

50. Stefanaki C, Stefanaki K, Chardalias L, Soura E, Stratigos A. Differential diagnosis of

Spitzoid melanocytic neoplasms. Journal of the European Academy of Dermatology and

Venereology : JEADV 2016;30:1269-77.

51. Busam KJ, Kutzner H, Cerroni L, Wiesner T. Clinical and pathologic findings of Spitz

nevi and atypical Spitz tumors with ALK fusions. The American journal of surgical

pathology 2014;38:925-33. 76

52. Niu HT, Zhou QM, Wang F, et al. Identification of anaplastic lymphoma kinase break

points and oncogenic mutation profiles in acral/mucosal melanomas. Pigment cell &

melanoma research 2013;26:646-53.

53. Toschi L, Finocchiaro G, Nguyen TT, et al. Increased SOX2 gene copy number is

associated with FGFR1 and PIK3CA gene gain in non-small cell lung cancer and predicts

improved survival in early stage disease. PloS one 2014;9:e95303.

54. Davies KD, Le AT, Theodoro MF, et al. Identifying and targeting ROS1 gene fusions in

non-small cell lung cancer. Clinical cancer research : an official journal of the American

Association for Cancer Research 2012;18:4570-9.

55. Narita M, Nagumo Y, Miyatake M, Ikegami D, Kurahashi K, Suzuki T. Implication of

protein kinase C in the orexin-induced elevation of extracellular dopamine levels and its

rewarding effect. The European journal of neuroscience 2007;25:1537-45.

56. Zhang FX, Lai CH, Tse YC, Shum DK, Chan YS. Expression of Trk receptors in otolith-

related neurons in the vestibular nucleus of rats. Brain research 2005;1062:92-100.

57. Elias JM, Gown AM, Nakamura RM, et al. Quality control in immunohistochemistry.

Report of a workshop sponsored by the Biological Stain Commission. American journal

of clinical pathology 1989;92:836-43.

58. Boyle TA, Masago K, Ellison KE, Yatabe Y, Hirsch FR. ROS1 immunohistochemistry

among major genotypes of non-small-cell lung cancer. Clinical lung cancer 2015;16:106-

11.

59. Hintzsche J, Kim J, Yadav V, et al. IMPACT: a whole-exome sequencing analysis

pipeline for integrating molecular profiles with actionable therapeutics in clinical

samples. Journal of the American Medical Informatics Association : JAMIA 2016. 77

60. Scolnick JA, Dimon M, Wang IC, Huelga SC, Amorese DA. An Efficient Method for

Identifying Gene Fusions by Targeted RNA Sequencing from Fresh Frozen and FFPE

Samples. PloS one 2015;10:e0128916.

61. Krauthammer M, Kong Y, Bacchiocchi A, et al. Exome sequencing identifies recurrent

mutations in NF1 and RASopathy genes in sun-exposed melanomas. Nature genetics

2015;47:996-1002.

62. Curtin JA, Fridlyand J, Kageshita T, et al. Distinct sets of genetic alterations in

melanoma. The New England journal of medicine 2005;353:2135-47.

63. Ying J, Guo L, Qiu T, et al. Diagnostic value of a novel fully automated

immunochemistry assay for detection of ALK rearrangement in primary lung

adenocarcinoma. Annals of oncology : official journal of the European Society for

Medical Oncology / ESMO 2013;24:2589-93.

64. Sullivan HC, Fisher KE, Hoffa AL, et al. The role of immunohistochemical analysis in

the evaluation of EML4-ALK gene rearrangement in lung cancer. Applied

immunohistochemistry & molecular morphology : AIMM / official publication of the

Society for Applied Immunohistochemistry 2015;23:239-44.

65. Lambros MB, Wilkerson PM, Natrajan R, et al. High-throughput detection of fusion

genes in cancer using the Sequenom MassARRAY platform. Laboratory investigation; a

journal of technical methods and pathology 2011;91:1491-501.

66. Widakowich C, de Castro G, Jr., de Azambuja E, Dinh P, Awada A. Review: side effects

of approved molecular targeted therapies in solid cancers. The oncologist 2007;12:1443-

55. 78

67. Tang N, Ratner D. Managing Cutaneous Side Effects From Targeted Molecular

Inhibitors for Melanoma and Nonmelanoma Skin Cancer. Dermatologic surgery : official

publication for American Society for Dermatologic Surgery [et al] 2016;42 Suppl 1:S40-

8.

68. Kim A, Cohen MS. The discovery of vemurafenib for the treatment of BRAF-mutated

metastatic melanoma. Expert opinion on drug discovery 2016;11:907-16.

69. Zhang J, Yang PL, Gray NS. Targeting cancer with small molecule kinase inhibitors.

Nature reviews Cancer 2009;9:28-39.

70. Tsai J, Lee JT, Wang W, et al. Discovery of a selective inhibitor of oncogenic B-Raf

kinase with potent antimelanoma activity. Proceedings of the National Academy of

Sciences of the United States of America 2008;105:3041-6.

71. Bollag G, Hirth P, Tsai J, et al. Clinical efficacy of a RAF inhibitor needs broad target

blockade in BRAF-mutant melanoma. Nature 2010;467:596-9.

72. Spagnolo F, Ghiorzo P, Queirolo P. Overcoming resistance to BRAF inhibition in BRAF-

mutated metastatic melanoma. Oncotarget 2014;5:10206-21.

73. Swaika A, Crozier JA, Joseph RW. Vemurafenib: an evidence-based review of its clinical

utility in the treatment of metastatic melanoma. Drug design, development and therapy

2014;8:775-87.

74. Van Allen EM, Wagle N, Sucker A, et al. The genetic landscape of clinical resistance to

RAF inhibition in metastatic melanoma. Cancer discovery 2014;4:94-109.

75. Romano E, Pradervand S, Paillusson A, et al. Identification of multiple mechanisms of

resistance to vemurafenib in a patient with BRAFV600E-mutated cutaneous melanoma 79

successfully rechallenged after progression. Clinical cancer research : an official journal

of the American Association for Cancer Research 2013;19:5749-57.

76. Abe H, Kikuchi S, Hayakawa K, et al. Discovery of a Highly Potent and Selective MEK

Inhibitor: GSK1120212 (JTP-74057 DMSO Solvate). ACS medicinal chemistry letters

2011;2:320-4.

77. Lugowska I, Kosela-Paterczyk H, Kozak K, Rutkowski P. Trametinib: a MEK inhibitor

for management of metastatic melanoma. OncoTargets and therapy 2015;8:2251-9.

78. Tran KA, Cheng MY, Mitra A, et al. MEK inhibitors and their potential in the treatment

of advanced melanoma: the advantages of combination therapy. Drug design,

development and therapy 2016;10:43-52.

79. Abdel-Rahman O, ElHalawani H, Ahmed H. Doublet BRAF/MEK inhibition versus

single-agent BRAF inhibition in the management of BRAF-mutant advanced melanoma,

biological rationale and meta-analysis of published data. Clinical & translational

oncology : official publication of the Federation of Spanish Oncology Societies and of

the National Cancer Institute of Mexico 2016;18:848-58.

80. Lee YJ, Kim JH, Bae S, Rho SK, Choe SY. Mechanism of transcriptional repression by

TEL/RUNX1 fusion protein. Molecules and cells 2004;17:217-22.

81. Turner J, Couts K, Sheren J, et al. Kinase gene fusions in defined subsets of melanoma.

Pigment cell & melanoma research 2017;30:53-62.

82. Zhou X, Yang G, Huang R, Chen X, Hu G. SVH-B interacts directly with p53 and

suppresses the transcriptional activity of p53. FEBS letters 2007;581:4943-8. 80

83. Huang R, Xing Z, Luan Z, Wu T, Wu X, Hu G. A specific splicing variant of SVH, a

novel human armadillo repeat protein, is up-regulated in hepatocellular carcinomas.

Cancer research 2003;63:3775-82.

84. Waggoner DW, Johnson LB, Mann PC, Morris V, Guastella J, Bajjalieh SM. MuLK, a

eukaryotic multi-substrate lipid kinase. The Journal of biological chemistry

2004;279:38228-35.

85. Bektas M, Payne SG, Liu H, Goparaju S, Milstien S, Spiegel S. A novel acylglycerol

kinase that produces lysophosphatidic acid modulates cross talk with EGFR in prostate

cancer cells. The Journal of cell biology 2005;169:801-11.

86. Mikhailenko DS, Efremov GD, Safronova NY, Strelnikov VV, Alekseev BY. Detection

of Rare Mutations by Routine Analysis of KRAS, NRAS, and BRAF Oncogenes.

Bulletin of experimental biology and medicine 2017;162:375-8.

87. Bowyer SE, Rao AD, Lyle M, et al. Activity of trametinib in K601E and L597Q BRAF

mutation-positive metastatic melanoma. Melanoma research 2014;24:504-8.

88. Hsieh R, Nico MM, Camillo CM, Oliveira KK, Sangueza M, Lourenco SV. Mutational

Status of NRAS and BRAF Genes and Protein Expression Analysis in a Series of Primary

Oral Mucosal Melanoma. The American Journal of dermatopathology 2017;39:104-10.

89. Rawson RV, Johansson PA, Hayward NK, et al. Unexpected UVR and non-UVR

mutation burden in some acral and cutaneous melanomas. Laboratory investigation; a

journal of technical methods and pathology 2017;97:130-45.

90. Corcoran RB, Dias-Santagata D, Bergethon K, Iafrate AJ, Settleman J, Engelman JA.

BRAF gene amplification can promote acquired resistance to MEK inhibitors in cancer

cells harboring the BRAF V600E mutation. Science signaling 2010;3:ra84. 81

91. Salton M, Kasprzak WK, Voss T, Shapiro BA, Poulikakos PI, Misteli T. Inhibition of

vemurafenib-resistant melanoma by interference with pre-mRNA splicing. Nature

communications 2015;6:7103.

92. Sievert AJ, Lang SS, Boucher KL, et al. Paradoxical activation and RAF inhibitor

resistance of BRAF protein kinase fusions characterizing pediatric astrocytomas.

Proceedings of the National Academy of Sciences of the United States of America

2013;110:5957-62.

93. Kelley LA, Mezulis S, Yates CM, Wass MN, Sternberg MJ. The Phyre2 web portal for

protein modeling, prediction and analysis. Nature protocols 2015;10:845-58.

94. Subbiah V, Westin SN, Wang K, et al. Targeted therapy by combined inhibition of the

RAF and mTOR kinases in malignant spindle cell neoplasm harboring the KIAA1549-

BRAF fusion protein. Journal of hematology & oncology 2014;7:8.

95. Faulkner C, Ellis HP, Shaw A, et al. BRAF Fusion Analysis in Pilocytic Astrocytomas:

KIAA1549-BRAF 15-9 Fusions Are More Frequent in the Midline Than Within the

Cerebellum. Journal of neuropathology and experimental neurology 2015;74:867-72.

96. Kumar A, Pathak P, Purkait S, et al. Oncogenic KIAA1549-BRAF fusion with activation

of the MAPK/ERK pathway in pediatric oligodendrogliomas. Cancer genetics

2015;208:91-5.

97. Perander M, Al-Mahdi R, Jensen TC, et al. Regulation of atypical MAP kinases ERK3

and ERK4 by the phosphatase DUSP2. Scientific reports 2017;7:43471.

98. Kaur A, Westermarck J. Regulation of protein phosphatase 2A (PP2A) tumor suppressor

function by PME-1. Biochemical Society transactions 2016;44:1683-93. 82

99. Bayliss R, Sardon T, Vernos I, Conti E. Structural basis of Aurora-A activation by TPX2

at the mitotic spindle. Molecular cell 2003;12:851-62.

100. Johnson LN, Noble ME, Owen DJ. Active and inactive protein kinases: structural basis

for regulation. Cell 1996;85:149-58.

101. Bjarnadottir U, Nielsen JE. Predicting the open conformations of protein kinases using

molecular dynamics simulations. Biopolymers 2012;97:65-72.

102. McClendon CL, Kornev AP, Gilson MK, Taylor SS. Dynamic architecture of a protein

kinase. Proceedings of the National Academy of Sciences of the United States of

America 2014;111:E4623-31.

103. Drier Y, Lawrence MS, Carter SL, et al. Somatic rearrangements across cancer reveal

classes of samples with distinct patterns of DNA breakage and rearrangement-induced

hypermutability. Genome research 2013;23:228-35.

104. Mitelman F, Johansson B, Mertens F. Fusion genes and rearranged genes as a linear

function of chromosome aberrations in cancer. Nature genetics 2004;36:331-4.

105. Kelly LM, Barila G, Liu P, et al. Identification of the transforming STRN-ALK fusion as

a potential therapeutic target in the aggressive forms of thyroid cancer. Proceedings of the

National Academy of Sciences of the United States of America 2014;111:4233-8.

106. Giacomini CP, Sun S, Varma S, et al. Breakpoint analysis of transcriptional and genomic

profiles uncovers novel gene fusions spanning multiple human cancer types. PLoS

genetics 2013;9:e1003464.

107. Veeraraghavan J, Ma J, Hu Y, Wang XS. Recurrent and pathological gene fusions in

breast cancer: current advances in genomic discovery and clinical implications. Breast

cancer research and treatment 2016;158:219-32. 83

108. Meric-Bernstam F, Johnson A, Holla V, et al. A decision support framework for

genomically informed investigational cancer therapy. Journal of the National Cancer

Institute 2015;107.

109. Mertens F, Johansson B, Fioretos T, Mitelman F. The emerging complexity of gene

fusions in cancer. Nature reviews Cancer 2015;15:371-81.