Metastasis Suppressor NME1 Directly Activates Transcription of the ALDOC Gene in Melanoma Cells

Molecular Mechanisms Underlying the Metastasis Suppressor Activity of NME1 in Malignant Melanoma

Item Type dissertation

Authors Pamidimukkala, Nidhi Vela

Publication Date 2019

Abstract Metastatic melanoma is exceedingly lethal and is responsible for the majority of skin cancer related deaths. Impactful research on melanoma metastasis is crucial to improving prognosis. The discovery of metastasis suppressor genes, genes that inhibit...

Keywords ALDOC; NME1; Metastasis; Fructose-Bisphosphate Aldolase; Melanoma; Transcription, Genetic

Download date 26/09/2021 16:56:05

Link to Item http://hdl.handle.net/10713/9611 Curriculum Vitae

Nidhi V. Pamidimukkala

[email protected]

https://www.linkedin.com/in/nidhi-pamidimukkala

Doctor of Philosophy, 2019

EDUCATION

2012-2019 Ph.D., Graduate Program in Molecular Medicine, Cancer Biology Tack University of Maryland School of Medicine – Baltimore, MD

2005-2009 B.S., Biological Sciences, Mellon College of Science Carnegie Mellon University - Pittsburgh, PA

2005-2009 Minor, Chemistry, Mellon College of Science Carnegie Mellon University - Pittsburgh, PA

RESEARCH EXPERIENCE

2012-2019 Graduate Research Assistant, Mentor: David M. Kaetzel, Ph.D University of Maryland School of Medicine-Baltimore, MD

2009-2012 Biochemist-MBV Bioprocessing Group Merck & Co, Inc.-Union, NJ

2008-2008 Future Talent Intern-Drug Metabolism Group Merck & Co., Inc- Rahway, NJ

2007-2009 Undergraduate Research Assistant, Mentor: Gordon S. Rule, Ph.D Carnegie Mellon University-Pittsburgh, PA

RESEARCH PUBLICATIONS

Pamidimukkala, N.V , Leonard, M.K, Snyder, D., McCorkle, J.R., Kaetzel, D.M. (2018). Metastasis Suppressor NME1 Directly Activates Transcription of the ALDOC Gene in Melanoma Cells. Anticancer Research,38(11), 6059-6068. PMID: 30396920

Leonard, M.K, Novak, M., Snyder, D., Snow, G., Pamidimukkala, N.V , McCorkle, J.R., Yang, X.H., Kaetzel, D.M. (2019). The Metastasis Suppressor NME1 Inhibits Melanoma Cell Motility via Direct Transcriptional Induction of the Integrin beta-3 Gene. Experimental Cell Research, 374(1), 86-93. PMID: 30458180

REVIEW PUBLICATIONS

Puts, G.S., Leonard, M.K, Pamidimukkala, N.V, Snyder, D., Kaetzel, D.M. (2018). Nuclear Functions of NME Proteins. Lab Inverstigation,98 (2), 211-218. PMID: 29058704

Leonard, M.K, Pamidimukkala, N.V, Puts, G.S., Snyder, D., Slominski, A.T., Kaetzel, D.M. (2017). The HGF/SF Mouse Model of UV-Induced Melanoma as an In Vivo Sensor for Metastasis-Regulating Gene. International Journal of Molecular Science, 18 (8), E1647 PMID: 28788083

MANUSCRIPTS IN PREPARATION

Leonard, M.K*, Pamidimukkala, N.V*, Puts, G.S*., Manojlovic, Z., Jarrett, S.G., Snyder, D., Novak, M., Dabernat, S., DeFabo, E.C., Noonan, F.P., Slominski, A., Merlino, G., Carpten, J., Kaetzel, D.M., Comprehensive Molecular Profiling of UV- induced Metastatic Melanoma in Nme1/Nme2-deficient Mice Reveals Genomic and Transcriptomic Alterations Associated with Poor Survival in Human Melanoma Patients. In preparation. *Authors contributed equally

Leonard, M.K., Pamidimukkala, N.V ., Novak, M., Snyder, D., Snow, G., McCorkle, J.R., Yang, X.H., Kaetzel, D.M., Impacts of NME1 Expression in the Tumor Microenvironment on Growth and Metastatic Potential of Melanoma Cells. In preparation

Puts, G.C., Leonard, M.K., Jarrett, S.G., Snyder, D., Vincent, R., Yang, Y., Pamidimukkala, N.V. , Portney, B., Rassool, F., Zalzman, M., Kaetzel, D.M., Recruitment of the Metastasis Suppressor of NME1 to Double-Stranded Breaks in DNA Promotes the Non-Homologous End-Joining Pathway of Repair. In Preparation

SELECTED PRESENTATIONS

8th Annual Biochemistry Retreat Baltimore, MD University of Maryland School of Medicine January 2019 Poster Presentation: Metastasis Suppressor NME1 Activates Transcription of the ALDOC Gene in Melanoma

9th Annual Cancer Biology Research Retreat Baltimore, MD University of Maryland School of Medicine May 2018 Oral Presentation: Transcriptional Regulation of the Aldolase C (ALDOC) Gene as a Potential Mechanism for the Metastasis Suppressor Activity of NME1

7th Annual Biochemistry Retreat Baltimore, MD University of Maryland School of Medicine January 2018 Poster Presentation: NME1 Regulation of Aldolase C (ALDOC) as a Potential Mechanism for the Metastasis Suppressor Activity in Melanoma

Molecular Medicine Seminar Baltimore, MD University of Maryland School of Medicine November 2017 Oral Presentation: NME1 Regulation of Aldolase C (ALDOC) as a Potential Mechanism of Metastasis Suppression in Melanoma

Cancer Focus Group Meeting Baltimore, MD University of Maryland School of Medicine June 2017 Oral Presentation: Induction of ALDOC Expression by NME1 May Mediate Metastasis Suppressor Activity in Malignant Melanoma

8th Annual Cancer Biology Research Retreat Baltimore, MD University of Maryland School of Medicine May 2017 Poster Presentation: Induction of ALDOC Expression by NME1 may Mediate Metastasis Suppressor Activity in Malignant Melanoma (Presentation Winner)

AACR Annual Meeting Washington, D.C. Walter E. Washington Convention Center April 2017 Poster Presentation: Induction of Aldolase C by NME1 may Mediate Metastasis Suppressor Activity in Malignant Melanoma

6th Annual Biochemistry Retreat Baltimore, MD University of Maryland School of Medicine January 2017 Poster Presentation: Induction of ALDOC Expression by NME1 may Mediate Metastasis Suppressor Activity in Malignant Melanoma

Pan American Society for Pigment Cell Research Baltimore, MD Royal Sonesta Harbor Court October 2016 Poster Presentation: Identifying Key Effectors in NME1-Mediated Metastasis Suppression

5th Annual Biochemistry Retreat Baltimore, MD University of Maryland School of Medicine January 2016 Poster Presentation: Identifying Key Effectors in NME1-Mediated Metastasis Suppression

6th Annual Cancer Biology Research Retreat Baltimore, MD University of Maryland School of Medicine May 2015 Poster Presentation: Identifying Key Effectors in NME1-Mediated Metastasis Suppression (Presentation Winner)

Cancer Focus Group Meeting Baltimore, MD University of Maryland School of Medicine September 2014 Oral Presentation: The Role of NME1 in Mediating Metastasis Suppression in Malignant Melanoma

Abstract

Title of Dissertation: Molecular Mechanisms Underlying the Metastasis Suppressor Activity of NME1 in Malignant Melanoma

Nidhi V. Pamidimukkala, Doctor of Philosophy, 2019

Dissertation Directed by: David M. Kaetzel, Ph.D, Professor, Biochemistry and Molecular Biology

Metastatic melanoma is exceedingly lethal and is responsible for the majority of skin cancer related deaths. Impactful research on melanoma metastasis is crucial to improving prognosis. The discovery of metastasis suppressor genes, genes that inhibit metastasis but do not affect tumor growth, have advanced the field of metastasis research.

NME1 (also referred to as NDPK-A or NM23-H1) was the first identified metastasis suppressor gene. NME1 is a multifaceted molecule, executing numerous cellular functions which are attributable to the suppressor phenotype. Our lab previously identified a signature of NME1-regualted genes which have prognostic value in human melanoma. As such, we hypothesize that NME1 possess an understudied transcriptional activity to regulate one or more of these signature genes. In the present study, we identified Aldolase C ( ALDOC ) from the gene signature as an ideal candidate to assess

NME1 transcriptional activity. NME1 induced mRNA and protein expression of ALDOC in multiple melanoma cell lines. Transcriptional regulation was evidenced by elevated expression of ALDOC pre-mRNA and activation of the ALDOC promoter in NME1- expressing melanoma cells. Employing chromatin immunoprecipitation, we demonstrated

NME1 localizes to the ALDOC gene and enriches the presence of active transcription indicators at the ALDOC promoter. Together, we provide novel systematic evidence for

NME1 transcriptional activity, which may further be explored as a mechanism for the metastasis suppressor function. Furthermore, we establish unequivocal metastasis suppressor capability of NME1 using a transgenic mouse model susceptible to developing melanoma upon exposure to ultraviolet (UV) radiation. Mice deficient in

NME1 expression experienced significantly increased melanoma metastases to the lung and lymph nodes compared to wild-type mice. Taken together, our studies demonstrate

NME1 as a robust metastasis suppressor of UV-induced melanoma, and provide evidence for novel transcription factor activity of NME1. We aim to enhance our understanding of melanoma metastasis through our continued research of the metastasis suppressor gene,

NME1.

Molecular Mechanisms Underlying the Metastasis Suppressor Activity of NME1 in Malignant Melanoma

by Nidhi V. Pamidimukkala

Dissertation submitted to the Faculty of the Graduate School of the University of Maryland, Baltimore in partial fulfillment of the requirements for the degree of Doctor of Philosophy 2019

Acknowledgments

I sincerely thank my mentor, Dr. David M. Kaetzel. I am so appreciative of everything you have taught me. I know you have mentored several students, but I am happy to have been your first graduate student at UMB. Thank you to my thesis committee, Drs. Thomas Hornyak, Shawn Lupold, Stuart Martin, Tony Passaniti, and Dudley Strickland, for all their valuable advice on my dissertation project. Thank you to all the past and present members of the Kaetzel lab, particularly, Drs. Katie Leonard and Gemma Puts. You both were there to offer support and advice in the lab and outside of it. Thank you for teaching me so much and providing such an open and cooperative learning environment. Thank you to the Molecular Medicine program, and the Biochemistry department for all their support.

To my friends from CMU-Haritha, Anupama, Fateema, Subha and Shephalie, we have grown and accomplished so much since college. I am grateful we have been there to support one another every step of the way.

To my fiancé, David-I am so lucky you are in my life. I couldn’t have gotten through this incredibly challenging time without you. You calm me down when I get stressed, and you encourage me when I want to quit. You make me better each day. Thank you for being the best fiancé, and soon to be husband! To the McGurl/McCarthy families: Dave, Joanne, Julie, Lisa, Brian, Braeden, Mia, and Paige-thank you for welcoming me with open arms and treating me like family since the first day we met.

Lastly, I am so fortunate to have the unconditional support of my immediate and extended family. To my dad-thank you for instilling in me the importance of education and hard work. You grew up in a remote village in Andhra Pradesh, India, earned a Ph.D, and came to America so your children would have better opportunities-I can’t thank you enough for everything you have given me. To my mom-you are a true superwoman. You earned your Masters while taking care of an infant; you worked full time and still were an attentive and loving mother. I don’t know how you do it all-thank you for being the best mom and role model. To my unbelievably amazing sister, Geeta-thank you for being my biggest cheerleader, picking me up when I am down, and always knowing the exact right thing to say. You are more than just my sister, you are my best friend and my soulmate. Thank you dad, mom and Gigi for shaping me in to the person I am today.

iii

Table of Contents

Chapter 1: Introduction ...... 1 1.1 Melanoma Development and Progression...... 1 1.1.1 Melanoma Risk Factors ...... 1 1.1.2 Melanoma Progression...... 2 1.1.3 Melanoma Metastasis...... 4 1.2 NME1 and Proposed Mechanisms of the Metastasis Suppressor Function ...... 5 1.2.1 NME1 Interactions with Cytoplasmic Proteins ...... 6 1.2.2 Nucleoside Diphosphate Kinase Activity of NME1 ...... 8 1.2.3 The 3’ to 5’ Exonuclease Activity of NME1 ...... 10 1.2.4 Putative Roles of NME1 in Transcriptional Regulation ...... 11 1.3 Scope of Work ...... 14 Chapter 2: Identifying Candidate Transcriptional Targets of NME1 ...... 17 2.1 Introduction ...... 17 2.2 Methods ...... 20 2.3 Results ...... 26 2.3.1 NME1 Regulates Expression of Genes That Track with the Metastasis Suppressor Function ...... 26 2.3.2 NME1 is Successfully Over-expressed in the WM1158 Metastatic Melanoma Cell Line ...... 30 2.3.3 ALDOC mRNA Expression is Lower in a Subset of Metastatic Melanoma Human Biospecimen...... 33 2.4 Discussion ...... 37 Chapter 3: Metastasis Suppressor NME1 Directly Activates Transcription of the ALDOC Gene in Melanoma Cells ...... 40 3.1 Introduction ...... 40 3.2 Methods ...... 42 3.3 Results ...... 49 3.3.1 NME1 Regulates Expression of ALDOC in Melanoma Cell Lines ...... 49 3.3.2 NME1 Promotes Transcriptional Activation of the ALDOC Gene ...... 52 3.3.3 NME1 is Recruited to the ALDOC Gene and Enhances Epigenetic Activation Markers and RNA PolII Occupancy ...... 54

3.3.4 NME1 Attenuation of Single-Cell Motility is Not Mediated by ALDOC Expression ...... 59 3.4 Discussion ...... 65 Chapter 4: In vivo Validation of the Metastasis Suppressor Function of NME1 in a Mouse Model of UVR-Induced Cutaneous Melanoma ...... 70 4.1 Introduction ...... 70 4.2 Methods ...... 74 4.3 Results ...... 76 4.3.1 Generating Triple Transgenic Mice for UV Irradiation ...... 76 4.3.2 NME1 Deficiency does not Affect Primary Melanoma Onset, Incidence, or Growth Rate ...... 80 4.3.3 NME1 Deficiency Enhances Lung and Lymph Node Metastases of UVR Initiated Melanoma ...... 85 4.4 Discussion ...... 89 Chapter 5: Summary and Discussion ...... 93 5.1 NME1 Activates Transcription of the ALDOC Gene ...... 93 5.1.1 Potential Mechanisms of NME1 Nuclear Translocation ...... 94 5.1.2 Potential Mechanism of NME1 DNA Binding Activity ...... 97 5.2 NME1 is a Validated Metastasis Suppressor of UV-induced Cutaneous Melanoma ...... 98 5.2.1 Potential Mechanisms of NME1-meidated Inhibition of UVR-initiated Melanoma Metastasis ...... 99 5.3 Conclusion ...... 102 References ...... 103

List of Figures

Figure 1.1: Melanoma Development and Progression ...... 4

Figure 1.2: Metastasis Suppressor Activity of NME1 May Occur Through Protein

Interactions ...... 7

Figure 1.3: NME1 Possess A Nucleoside Diphosphate Kinase Activity (NDPK) Which

Contributes to the Metastasis Suppressor Activity ...... 9

Figure 1.4: The 3’ to 5’ Exonuclease Activity of NME1 May Contribute to the Metastasis

Suppressor Activity Through Maintenance of Genomic Stability ...... 10

Figure 1.5: NME1 May Affect Gene Transcription Through Direct or Indirect

Regulation...... 14

Figure 2.1 ALDOC, NETO2, and LRP1b are Upregulated by NEM1 and Correlate with the Metastasis Suppressor Function ...... 29

Figure 2.2: The Metastatic Melanoma Cell Line WM1158 was Selected for Forced

Expression of NME1...... 31

Figure 2.3: NME1 Upregulates Expression of ALDOC, LRP1b, and NETO2 in a Dose-

Responsive Manner ...... 33

Figure 2.4: ALDOC and NETO2 Expression Drive Hierarchical Clustering in the

GSE8401 Dataset ...... 36

Figure 3.1: NME1 Regulates Expression of ALDOC Specifically ...... 51

Figure 3.2: NME1 Activates Transcription of the ALDOC Gene ...... 54

Figure 3.3: NME1 Localizes to the ALDOC Gene and Enhances Indicators of

Transcriptional Activation ...... 58

Figure 3.4: ALDOC Expression Does not Affect Single Cell Motility ...... 63

Figure 3.5: ALDOC Expression Correlates with Melanoma Stage and Breslow Depth .. 65

Figure 4.1: Breeding Diagram for Generation of a Triple Transgenic Mouse Model ...... 79

Figure 4.2: Study Design for UVR of Transgenic Mice ...... 81

Figure 4.3: Melanoma Locations are not Affected by NME1 Deficiency ...... 82

Figure 4.4: NME1 Deficiency Does Not Affect Primary Melanoma Tumor Development.

...... 84

Figure 4.5: NME1 Deficiency Enhances Lung Metastases ...... 86

Figure 4.6: Loss of NME1 Enhances Lymph Node Metastasis ...... 88

vii

List of Tables

Table 1: Comparison of Complete and Nutrient Deprived DMEM Media ...... 47

viii

List of Abbreviations

5' UTR 5 Prime Untranslated Region

6-4 PP 6-4 pyrimidine-pyrimidone photoproducts

actD Actinomycin D

ALDOA Aldolase A

ALDOB Aldolase B

ALDOC Aldolase C

ApoE Apolipoprotein E

ATP/ADP Adenosine Tri/Di Phosphate

Bcl-2 B-cell Lymphoma 2

ChIP Chromatin Immunoprecipitation

COX-2 Cyclooxygenase 2

CPD Cyclobutane pyrimidine dimers

DHAP Dihydroxyacetone Phosphate

dNTP Dinucleotide Tri-Phosphate

DSBR Double Strand Break Repair

EBNA3C Epstein Barr Virus Nuclear Protein Antigen 3C

EMT Epithelial to Mesenchymal Transition

ER α Estrogen Receptor Alpha

ERE Estrogen Response Elements

ER β Estrogen Receptor Beta

FACS Fluorescent Activated Cell Sorting

G3P Glyceraldehyde 3 Phosphate

GFP Green Fluorescent Protein

H3K27ac Histone 3 Lysine 27 Acetylation

H3K4me3 Histone 3 Lysine 4 trimethylation

HGF Hepatocyte Growth Factor

IRES Internal Ribosomal Entry Sequence

ITGαV Integrin Alpha 5

ITGβ3 Integrin Beta 3

KSR Kinase Suppressor of Ras

Low-density Lipoprotein Receptor-Related LRP1b Protein 1b

MAPK Mitogen-activated protein kinase

Macrophage Migration Inhibitor Factor MIF

MMP-9 Matrix Metallopeptidase 9

MSG Metastasis Suppressor Gene

MSS Metastasis Suppressor Signature

NDPK Nucleoside Diphosphate Kinase

NER Nucleotide Excision Repair

NETO2 Neuropilin and Tolloid-like 2

NF-κ B Nuclear Factor Kappa-Light Chain Enhancer of Activated B-cells

NHEJ Non-Homologous End Joining

NLS Nuclear Localization Signal

NME1 Non-metastatic entity 1

NME2 Non-metastatic entity 2

PAI-I Plasminogen-Activating Inhibitor

PDGF-A Platelet Derived Growth Factor-A

PTEN Phosphate and Tensin Homolog

Quantitative Reverse-Transcription Polymerase qRT-PCR Chain Reaction

RGP Radial Growth Phase

RNApolII RNA polymerase II

Sp1 Specificity Protein 1

STRAP Serine-Threonine Receptor Associated Protein

The Cancer Genome Atlas-Skin Cutaneous TCGA-SKCM Melanoma

U.V Ultraviolet

uPA Urokinase

VGP Vertical Growth Phase

xii

Chapter 1: Introduction

1.1 Melanoma Development and Progression

Melanoma is the deadliest form of skin cancer due to its aggressive nature and increased propensity to metastasize [1, 2]. An estimated 96,480 new cases of melanoma are expected to be diagnosed in 2019, making it the fifth most common cancer type in the

United States [3].

Melanoma is derived from pigment producing melanocytes, which reside at the basal layer of the epidermis of the skin [4]. Progression of melanoma occurs linearly, beginning with the oncogenic transformation and uncontrolled proliferation of melanocytes resulting in a pigmented lesion on the skin, known as a nevus, or more colloquially, as a mole. The ultimate stage of melanoma is the colonization of malignant melanoma cells at a distant organ site (metastasis). Melanoma often develops as a result of heritable mutations or environmental influences.

1.1.1 Melanoma Risk Factors

Heritable mutations in tumor suppressor genes or oncogenes are a significant risk factor of melanoma. Loss of function of the tumor suppressor gene, CDKN2a , and activation of the oncogene, CDK4 , are the most common germline mutations detected in familial melanomas [2, 5-8]. Mutations silencing CDKN2a occur in about 40% of hereditable melanoma cases [5, 9]. The CDKN2a gene encodes for two tumor suppressor proteins, p16 INK4a and p14 ARF , which are necessary for cell cycle arrest. As such, loss of

CDKN2a enhances cell proliferation resulting in increased nevus number and size, a highly likely indicator of melanoma development [5]. Amplification or activating point mutations of the CDK4 gene are responsible for a smaller percentage of familial

1 melanoma cases [2, 5-8]. The CDK4 gene encodes for the cyclin dependent kinase 4 enzyme, necessary for G1 phase cell cycle progression [8]. Constitutive activation of

CDK4 promotes tumorgenicity by enhancing melanoma cell cycle progression [8].

An environmental risk factor of melanoma is exposure to ultraviolet (UV) radiation [6, 10, 11]. The most common source of UV radiation is the sun, which is comprised of varying wavelengths of UV rays (UVA, UVB, and UVC) [6, 10]. UVB is absorbed by the epidermis, and is therefore considered to be the predominant carcinogen for melanoma development [10]. UVB exposure can cause DNA mutations resulting in C to T or CC to TT base transitions. Further, UVB and can link two adjacent DNA pyrimidine bases causing cyclobutane pyrimidine dimers (CPDs) and 6-4 pyrimidine- pyrimidone photoproducts (6-4 PPs) [12, 13]. CPDs and 6-4PPs cause significant alterations in the DNA structure. Nucleotide excision repair (NER) is the primary repair pathway responsible for restoring UV-induced DNA lesions [14]. Accumulation of CPDs and 6-4PPs as a result of extended UVB radiation compromises the efficacy of NER and increases the presence of DNA mutations [15]. UV-induced mutations which activate oncogenes, or silence tumor suppressor genes, can encourage melanoma development and progression [16, 17].

1.1.2 Melanoma Progression

Constitutive activation of the mitogen-activated protein kinase (MAPK) signaling pathway is the most common melanoma initiating event [18-20]. Somatic mutations activating MAPK signaling genes, NRAS and BRAF , occur in 15% and 50% of melanomas respectively [18]. In particular, the point mutation in the BRAF gene causing the substitution of valine to glutamic acid at position 600 (BRAF V600E ) is frequently

2 observed in melanoma [21]. As such, small molecule inhibitors targeted to inactivate

BRAF V600E have been developed to treat melanoma [21]. However, several melanomas are resistant to BRAF V600E targeted treatment, suggesting there may be a compensatory mechanism to activate cell proliferation [21]. Hepatocyte growth factor (HGF) signaling, occurring exclusively through binding to cell surface receptor tyrosine kinase, c-MET, is one such reason for BRAF inhibitor resistance [22-25]. HGF/c-MET signaling activates several cell proliferation signaling pathways, including the MAPK pathway [22]. This activation of the MAPK pathway, occurring independent of BRAF mutations, is proposed as a mechanism for BRAF drug resistance in some melanoma patients [23-25]. The multiple mechanisms for activating the MAPK signaling pathway also highlights the heterogeneity with which melanoma may be initiated.

Activation of the MAPK signaling pathway in melanocytes causes uncontrolled cell proliferation, and results in the phenotypic presentation of a pigmented lesion

(benign nevus) [18]. Most nevi do not progress to malignancy. Secondary genomic events are often required to transform benign nevi to dysplasia. Acquired silencing mutations in tumor suppressor genes CDKN2a , TP53 , or PTEN , cause premalignant lesions which are larger in size and slightly raised (dysplastic nevi) [18-20]. Dysplastic nevi may develop to melanoma in situ (radial growth phase), characterized by superficial spreading of melanoma cells within the epidermis. Through additional genomic events promoting a switch of malignant melanoma cells from epithelial-like to mesenchymal-like phenotype

(EMT), melanoma cells invade through the basal layer of the epidermis in to the dermis

(vertical growth phase) [18]. The final stage of melanoma occurs when malignant melanoma cells colonize at a distant organ site (metastasis). Figure 1.1 diagrams the

stages of melanoma and proposed genomic transformations necessary for melanoma progression.

Figure 1.1: Melanoma Development and Progression : A. Melanoma progresses linearly. Melanocytes are located at epidermal and dermal junctions of normal skin. UV radiation causes the first genomic insult causing activation of the MAPK signaling pathway, resulting in melanocyte transformation and proliferation to a benign nevi . Secondary genomic insults including the loss of tumor suppressors result in the development of a dysplastic nevi . The radial growth phase of melanoma is characterized by superficial spreading of melanoma through the epidermis. Additional genomic alterations cause transformations of malignant melanocytes to a more mesenchymal phenotype, allowing for invasion in to the dermal layer of the skin in the vertical growth phase . Finally, melanoma cells invade vasculature and ultimately colonize at a distant organ in metastatic melanoma . This image was made with images from https://smart.servier.com

1.1.3 Melanoma Metastasis

Metastasis is a multistep process beginning with the dissemination of malignant cells from the tumor bulk, intravasation into capillaries and lymphatic systems, extravasation through endothelial cells, and homing and ultimate proliferation at a distant organ site [26]. A reliable predictor of melanoma metastasis is a measure of melanoma thickness (Breslow depth) [27]. Employed clinically, Breslow depth is a measure of how deep a melanoma has invaded through the epidermis or dermis in the vertical growth phase. Breslow measurements greater than 1mm have required sentinel lymph node biopsies, generally considered to be the first site of melanoma metastasis [28]. In addition

4 to lymph nodes, melanoma often metastasizes to the lung, brain, liver, and bone [26].

Advancements in melanoma therapies, including immunotherapies, have improved melanoma prognosis overall, however, the 5-year survival rate for metastatic melanoma is a dismal 23% in the United States [29].

We may begin to unravel the complexities of melanoma metastasis through the study of metastasis suppressor genes (MSGs), which function to attenuate metastasis, but do not affect primary tumor growth [30]. In this report, we focus our studies on the metastasis suppressor gene, NME1.

1.2 NME1 and Proposed Mechanisms of the Metastasis Suppressor Function

NME1 belongs to a family of 10 known proteins in humans (NME1-10), each possessing a well conserved nucleoside diphosphate kinase (NDPK) domain (described in section 1.2.2). Despite sharing this domain, the NME family of proteins have different catalytic activities. Further, only NME1 and NME2 have exhibited demonstrable metastasis inhibiting capability, and are thus the only NME family members considered as metastasis suppressor genes (MSGs).

NME1 was the first MSG discovered. Through differential hybridization experiments, NME1 mRNA expression was observed to be lowered or absent in highly metastatic derivatives of the mouse melanoma cell line, K-1735, compared to the noninvasive cell counterparts [31]. Since the initial discovery in melanoma, NME1 suppressor activity has been demonstrated across several human cancer types [32-36].

The underlying mechanisms for the metastasis suppressor activity may be explained by the numerous cellular functions performed by NME1.

1.2.1 NME1 Interactions with Cytoplasmic Proteins

NME1 interactions with a diverse array of protein binding partners have been implicated as a mechanism for the metastasis suppressor activity. For example, NME1 was shown to bind to macrophage migration inhibitor factor (MIF) via cysteine-cysteine interactions [37]. NME1 interactions with MIF, relieved MIF suppression of p53- mediated apoptosis, suggesting NME1 may indirectly promote cell cycle arrest [37].

Effects on p53 signaling were also observed through NME1 interactions with the serine- threonine receptor associated protein (STRAP), which promoted p53 activation [38].

Together, the results suggest NME1 may function to inhibit metastasis by interacting with binding partners that activate the tumor suppressor, p53.

NME1 effects on cell motility have been described through its interactions with the actin binding protein, Gelsolin, and with the guanine exchange factor (GEF), Tiam1.

NME1 was shown to interact with Gelsolin through co-immunoprecipitation experiments in murine and human breast cancer cell lines [39]. Further, this interaction inhibited the actin depolymerization activity of Gelsolin, resulting in reduced cell motility. In kidney cells, NME1 interactions with Tiam1, inhibited Tiam1 specific activation of the Rho-

GTPase, Rac1 [40]. Activated Rac1 is necessary for cell growth and motility. As such,

NME1 may inhibit cell motility through interactions and inactivation of Tiam1.

NME1 possess a purported histidine kinase activity, with supposed contribution to the metastasis suppressor function. NME1 self-phosphorylates its histidine residue located at position 118 (H 118 ) through the nucleoside diphosphate kinase activity

(described in detail in the next section). NME1 may transfer this phosphate to receiving protein substrates [41]. NME1 phosphorylation of the kinase suppressor of ras (KSR), is

6 considered to be the primary contributor to the metastasis suppressor activity [42, 43].

KSR is a scaffolding protein and is necessary for propagating MAPK signaling. NME1 phosphorylation of KSR, inhibits its activity, thereby tempering MAPK signaling and tumorigenesis ( Figure 1.2. ) [43].

Although NME1 protein interactions and the histidine kinase activity have once been believed to be primary contributors to the metastasis suppressor function, the activities must be considered cautiously. NME1 is a “sticky” protein, suggesting co-immunoprecipitation experiments used to identify NME1 binding partners, may not in fact reveal a direct protein interaction [44]. Additionally, Levit, et al. were unable to replicate the observed phosphate transfer to protein substrates via the putative histidine kinase activity. The apparent phosphorylation event was instead attributed to residual

ADP present in assay reactions [45]. In light of this, additional mechanisms of the suppressor function of NME1 have been considered.

Figure 1.2: NME1 may Inhibit Metastasis Through Interactions with KSR . A. NME1 transfers a phosphate from the histidine 118 (H 118 ) residue to the kinase suppressor of ras (KSR). KSR activates the MAPK pathway. Phosphorylated KSR inhibits this activity, resulting in decreased MAPK signaling and decreased metastasis. NME1 interactions with KSR is considered to be one mechanism for the metastasis suppressor activity.

1.2.2 Nucleoside Diphosphate Kinase Activity of NME1

NME1 has a well conserved nucleoside diphosphate kinase domain (NDPK). The

NDPK activity functions to transfer a terminal phosphate from a nucleoside triphosphate

((d)NTP) to a nucleoside diphosphate ((d)NDP). The phosphate exchange is reversible and is mediated by an autophosphorylation event on the H 118 residue of NME1 ( Figure

1.3 A). Originally, the NDPK activity was considered unnecessary for the metastasis suppressor function [46]. However, our lab has demonstrated that disrupting the NDPK activity only partially inhibited lung metastases in a spontaneous metastasis experiment performed in athymic nude ( nu/nu ) mice [47]. The result suggested the NDPK activity may in fact contribute to the metastasis suppressor function of NME1.

Nucleoside diphosphate kinases are necessary for maintaining genomic integrity

[48]. Mutations in the bacterial homolog of NDPK, ndk in E.coli , resulted in increased mutation frequency caused by imbalances in dNTP pools [49]. Sufficient supply of dNTPs is necessary for precise synthesis of DNA, and may be one reason for the increased mutational burden observed when NDPK expression is lost. Further, maintenance of dNTP pools is crucial for DNA repair [50]. As such, it was predicted that the NDPK activity of NME1 may function in DNA repair processes.

The role of NME1 in DNA repair was first observed in Saccharomyces cerevisiae . Ablation of the yeast homologue of NME1, YNK1 , delayed the repair of

DNA damage caused by UV radiation [51, 52]. As previously mentioned, UV-radiation causes 6-4 photoproduct (6-4 PP) DNA lesions, which are repaired by the nucleotide excision repair (NER) pathway [53]. Our lab demonstrated NME1 facilitated the removal of 6-4PPs through the NER pathway in a human melanoma cell line [54]. Further, an

NME1 mutant deficient in NDPK activity, slowed repair of 6-4PPs, increasing presentation of these bulky DNA lesions. The results suggest that the NDPK activity of

NME1 may contribute to anti-mutator functions, including removal of 6-4PPs through

NER, and preventing imbalances of dNTP supply to ensure accuracy of DNA synthesis

and repair processes. Collectively, these functions can promote genomic integrity and

inhibit melanoma progression (Figure 1.3 B).

Figure 1.3: NME1 Possess a Nucleoside Diphosphate Kinase Activity (NDPK) Which may Contribute to the Metastasis Suppressor Function . A. The NDPK activity of NME1 transfers a terminal phosphate from ATP to ADP. The phosphate transfer occurs via a self-phosphorylation event on the histidine 118 (H 118 ) residue. The phosphate transfer from (d)NTPs to (d)NDPs is a reversible reaction occurring through a “ping-pong” mechanism. B. The NDPK activity of NME1 may contribute to the metastasis suppressor activity by maintaining genomic stability. UV radiation (UVR) causes DNA damage including bulky DNA lesions known at 6-4 photoproducts (6-4PP). NME1 was shown to aid in the repair of these lesions by enhancing nucleotide excision repair (NER). Further, NME1 may balance dNTP and dNDP pools which are necessary for DNA synthesis and DNA repair processes.

1.2.3 The 3’ to 5’ Exonuclease Activity of NME1

In addition to the NDPK activity, NME1 possesses a 3’ to 5’ exonuclease activity

(EXO). Our lab was the first to identify the EXO activity through in vitro experiments.

NME1 was shown to bind to single-stranded DNA containing a free 3’ hydroxyl group, and subsequently cleave the DNA in a 3’ to 5’ direction [55]. Furthermore, our lab established the EXO activity as a necessary mediator of the metastasis suppressor function of NME1 in vivo [47].

3’ to 5’ exonucleases function in DNA proofreading, DNA synthesis, and double strand break repair (DSBR) processes [56]. The NDPK activity of NME1 contributed to

DNA repair though the NER pathway. As such, the attributed EXO activity of NME1 further suggests a role in DNA repair, likely occurring through DSBR mechanisms. Our lab is currently investigating the role of NME1 in DSBR. We have shown that NME1 is recruited to double strand breaks after DNA induced damage from UV radiation (Puts et al. , manuscript in preparation). Further, NME1 promoted DSBR occurring by nonhomologous end joining (NHEJ). Together, the EXO activity may be contributing to the metastasis suppressor activity of NME1 through the enhancement of DSBR

(Figure 1.4).

Figure 1.4: The 3’ to 5’ Exonuclease Activity of NME1 may Enhance Genomic Stability and Inhibit Metastasis . A. UV radiation (UVR) causes double-strand DNA breaks. NME1 possess a 3’ to 5’ exonuclease activity which may facilitate the removal of 3’ overhanging DNA bases. This activity may help to enhance double strand break repair through the non-homologous end joining (NHEJ) repair pathway. Increasing repair of DNA double strand breaks may prevent metastasis driving mutations, providing a mechanism for the metastasis suppressor activity of NME1.

1.2.4 Putative Roles of NME1 in Transcriptional Regulation 1

NME1 was observed to affect global gene expression through microarray analysis in several cancer types [35, 58-61]. Further, NME1 binding to single stranded DNA, evidenced through the EXO activity, suggest NME1 may be interacting with DNA to affect gene expression. The potential involvement in gene regulation may contribute to the metastasis suppressor activity, as NME1 may regulate expression of cancer associated genes.

Evidence for NME1 Direct Regulation of Gene Expression

Initial evidence for NME1 involvement in gene regulation, was the discovery of

NME1 as a repressor of platelet-derived growth factor A ( PDGF-A) oncogene [62]. The

PDGF-A promoter and upstream gene regions contain G-C rich DNA segments associated with transcriptional silencing of the gene. Screening of a HeLa cDNA expression library for elements binding in vitro to single-stranded segments of these silencing elements, identified NME1 [62]. Forced expression of NME1 in HeLa cells resulted in reduced expression of PDGF-A mRNA, suggesting the binding of NME1 to the silencer elements observed in vitro , may be the reason for the decreased expression

[62]. In a separate study, binding of an NME entity to the PDGF-A promoter region was observed in vivo via chromatin immunoprecipitation (ChIP) [63, 64]. Further, the study identified the promoters of additional cancer associated genes, TP53 and CMYC , as binding targets for the NME entity [64]. While the research did not distinguish between

NME1 and NME2, the group showed that NME1/2 bound to G-C rich segments on the promoters of all 3 genes, consistent with the previous report.

1 57. Puts, G.S., et al., Nuclear functions of NME proteins. Lab Invest, 2018. 98 (2): p. 211-218.

Interestingly, NME1 may also function as a competitive DNA binding factor to regulate gene expression. The rat homolog of NME1, Nm23-β, decreased expression of the pro-migratory gene, gelatinase A [65]. Nm23-β was shown to bind to an enhancer response element on the promoter of gelatinase A, blocking binding of the activating transcription factor, YB-1, to the same region. Nm23-β competitive binding with YB-1, thereby inhibited expression of gelatinase A. Together, these studies suggest NME1 may regulate expression of cancer associated genes through direct binding to gene promoter regions.

Evidence of Indirect Regulation of Gene Expression by NME1

Through interactions with transcription factors, NME1 may indirectly regulate expression of cancer related genes. NME1 interactions with the transcription factor

AP-1, enhanced AP-1 mediated repression of the cyclin D1 gene, resulting in p53- dependent apoptosis in B-cells [34]. In another study, NME1 binding to ERα in a breast cancer cell line, facilitated the interaction of ERα with estrogen response elements (ERE).

The NME1-ERα interaction resulted in decreased expression of Bcl-2, cathepsin D, and cyclin D1 [66]. A further area of research demonstrating NME1 interactions with known transcription factors, involves NME1 interactions with oncogenic viral antigens of the

Epstein-Barr virus.

The Epstein-Barr virus preferentially targets B-cells and epithelial cells, and is associated with Burkitt’s lymphoma, Hodgkin’s disease, and other B-cell lymphomas

[67]. NME1 interactions with the EBNA3C viral antigen of the Epstein-Barr virus, resulted in increased translocation of NME1 from the cytoplasm to the nucleus [68].

NME1 interacted with transcription factors NF-κB and CREB/AP-1 in the nucleus,

12 resulting in increased transcription of the tumor associated COX-2 gene, and the pro- migratory, MMP-9 gene [69, 70]. Further, expression of integrin alpha V (ItgαV), which is expressed on highly migratory cancer cells, was also increased as a result of NME1 and

EBNA3C interactions [67]. NME1 and EBNA3C formed a transcription complex in the nucleus, in which NME1 bound to transcription factor GATA-1, and EBNA3C bound to the transcription factor Sp1. Collectively, this complex induced expression of ItgαV [67].

Interestingly, in the absence of EBNA3C, NME1 was still able to bind to GATA-1, but instead, this interaction resulted in decreased expression of ItgαV.

Contrary to the metastasis suppressor activity of NME1, NME1 interactions with the EBNA3C viral antigen in B-cell lymphoma, enhanced expression of pro-migratory genes. However, in the absence of EBNA3C, NME1 may inhibit expression of pro- migratory genes, as was observed with ItgαV expression. These results emphasize the complexity for the role of NME1 in transcriptional regulation, which may be context and cancer cell type specific. As such, NME1 interactions with transcription factors in melanoma, may result in repression of pro-migratory genes. Figure 1.5 diagrams the proposed transcription activity of NME1 occurring through direct or indirect mechanisms.

A. B.

Figure 1.5: NME1 may Affect Gene Expression Through Direct or Indirect Regulation . A. NME1 was observed to bind to single-stranded G-C rich segments of DNA located on promoter regions of cancer-associated genes. NME1 was able to inhibit or enhance gene expression, demonstrating a potential direct mechanism of gene regulation. B. NME1 was observed to bind to established transcription factors (TF). Interactions with these TFs resulted in enhanced expression of several cancer associated genes. NME1 interactions with TFs demonstrates an indirect mechanism for NME1 regulation of gene expression. Together, NME1 regulation of the expression of genes related to cancer, may be a potential mechanism for the metastasis suppressor activity of NME1.

1.3 Scope of Work

Several molecular mechanism for the metastasis suppressor activity of NME1 have been proposed based on the multiple cellular functions performed by NME1.

However, many of the activities performed by NME1 have yet to be explored in full. In particular, NME1 involvement in gene transcription is an understudied area of research.

In this study, we aim to systematically evaluate NME1 participation in transcriptional regulation.

Prior to the current work, our lab conducted microarray analysis in a human melanoma cell line to observe gene expression profiles in cells with and without forced

NME1 expression [35]. The NME1-differentially expressed genes were further compared to gene profiles from cells with forced expression of NME1 mutants, devoid of metastasis suppressor activity. By selecting genes that are differentially expressed by wildtype

NME1, but unaffected by NME1 mutants, a signature collection of genes regulated by

NME1 and tracking with the metastasis suppressor activity was generated. In the current study, we identified aldolase C ( ALDOC ) from the signature gene list as an ideal candidate to evaluate NME1 transcriptional activity.

ALDOC expression was robustly induced by NME1 in a number of melanoma cell lines, and was thus considered a model gene prototype to study the transcriptional activity of NME1. NME1 induction of ALDOC was analyzed by high-throughput transcriptome methods, qRT-PCR, and immunoblot analyses. NME1 enhanced pre-mRNA expression of ALDOC , and activated the ALDOC promoter, further suggesting NME1 involvement in transcriptional regulation. Finally, we demonstrated that NME1 localizes to the

ALDOC promoter, enhances RNApolII occupancy, and increases epigenetic markers of transcriptional activation. Through comprehensive analyses, we reveal a novel transcription factor function for NME1.

NME1 was first identified over 30 years ago as a metastasis suppressor gene. In melanoma, the metastasis suppressor activity of NME1 has been widely studied in cell culture systems, or in vivo , by tail vein injection and spontaneous metastasis experiments.

Both tail vein injection and spontaneous metastasis experiments require transplantation of melanoma cells to mice, which is an artificial method to initiate melanoma and to study metastasis. However, to date, the metastasis suppressor effects of NME1 on melanomas initiated by UV-radiation, a known risk factor of human melanoma development, had yet to be established in vivo . In this study, we utilized the hepatocyte growth factor (HGF) transgenic mouse model of melanoma, in which melanoma is initiated by UV-irradiation, to mimic human disease. Nme1 gene specific knockout was introduced in to the mouse model to observe NME1 effects on UV-induced melanoma metastasis. NME1 deficiency

15 significantly increased lung metastasis incidence and metastatic lesion number and size.

Further, NME1 knockout mice experienced increased metastases to sentinel lymph nodes.

Together, we demonstrated for the first time, bona fide metastasis suppressor activity of

NME1 using an in vivo mouse model of melanoma. Through continued efforts by our lab to elucidate functions of the metastasis suppressor, NME1, we aim to enhance the ever- evolving paradigm of melanoma metastasis.

Chapter 2: Identifying Candidate Transcriptional Targets of

NME1

2.1 Introduction

Metastasis is a complex process of interacting pathways permitting malignant cells to migrate from the primary tumor, invade vasculature and lymphatic networks, and seed and proliferate at a distant organ [26]. Metastasis suppressor genes (MSG) function to inhibit one or more processes of the metastatic cascade, but do not affect primary tumor formation or growth [33]. The primary mechanism of suppression for most MSGs involves interactions with cytoskeletal elements, or inhibiting signaling pathways that promote metastasis. However, a principal mode of suppression for some MSGs, involves regulation of metastasis associated genes [71]. NME1 is a well-classified metastasis suppressor gene. Like most MSGs, the metastasis suppressor function of NME1 has been attributed to protein interactions which inhibit metastasis signaling pathways. However, evidence has supported NME1 may also function to regulate expression of metastasis- related genes as a potential mechanism for the suppressor activity [55, 72]. Our lab aimed to further investigate NME1 regulation of genes with likely contribution to the metastasis suppressor function.

Previous work from our lab generated NME1 mutants which disrupted the metastasis suppressor function. Compared to wild-type NME1, NME1 mutants with single amino acid substitutions of either glutamic acid to alanine (E 5A), or a substitution of lysine to glutamine (K 12 Q), showed decrease capacity to inhibit both tail vein and spontaneous metastasis in athymic nude ( nu /nu ) [47]. These suppressor-deficient variants were utilized for microarray analysis to identify NME1-regulated genes that track with

17 suppressor activity. This was accomplished by comparing gene profiles from parental

WM793 melanoma cells, with WM793 cells expressing either wild-type or the suppressor-deficient variants of NME1. Genes that were regulated by wild-type NME1, but not the suppressor-deficient mutants, were considered to be genes that correlate with metastasis suppressor function of NME1. We denoted these genes as the NME1-regulated

Metastasis Suppressor Signature (MSS) genes [35, 73]. Further, we selected three genes

(ALDOC , LRP1b and NETO2 ) from the MSS gene signature that were highly probable transcriptional targets of NME1, with increased likelihood of contributing to the metastasis suppressor function.

MSS Gene Candidates

Aldolase C (ALDOC) is a glycolytic enzyme, responsible for catalyzing fructose1,6 bisphosphate to glyceraldehyde-3 phosphate (G3P) and dihydroxyacetone phosphate (DHAP) [74]. The preference of cancer cells to utilize glycolysis as a primary energy source is known as the Warburg effect, and is a well-established hallmark of cancer [75-77]. As such, cancer malignancy is associated with elevated expression of glycolytic enzymes. NME1 was shown to increase expression of ALDOC in our microarray study. The result is surprising given the role of ALDOC in glycolysis.

However, aldolases perform glycolysis-independent functions, which may help to explain the observed NME1-induction of the ALDOC gene. [78-80]. In particular, aldolases bind to actin, independent of their enzymatic activity [81-84]. Although the particular function of the binding is unknown, aldolase interactions with cytoskeletal components is suggestive of a role in cell motility [83, 85]. Interestingly, ALDOC over-expression in oral squamous cell carcinoma cells, did in fact decrease cell motility [86]. ALDOC is a

18 compelling gene target of NME1 due the suggested and observed roles in inhibiting cell motility, a critical component of metastasis.

NME1 also increased expression of the LRP1b gene (low-density lipoprotein receptor-related protein 1b). LRP1b is a member of the LDL receptor family of proteins.

These receptor proteins perform multiple cellular functions, with initial studies focused on the clearance of lipoproteins [87, 88]. In addition to lipoproteins, LRP1b binds to and decreases the extracellular availability of pro-angiogenic and invasion proteins, ApoE, uPA, and PAI-I [89]. Furthermore, loss of expression of LRP1b occurring by hypermethylation at CpG islands, genomic deletion, or proteolytic processing, has been observed across several cancer types [90-93]. Taken together, LRP1b has been ascribed a putative tumor suppressor activity [90]. Interestingly, LRP1b was frequently mutated in melanoma [94]. Although, the operational activity of LRP1b in melanoma metastasis was not determined, NME1 upregulation of LRP1b, a protein frequently lost or mutated in cancer, predicts a functional role of LRP1b in inhibiting melanoma metastasis.

Neuropilin and tolloid-like 2 (NETO2) encodes a transmembrane protein expressed in the retina and brain [95]. Though little is known about the function of

NETO2 in cells, it has been shown to interact with kainate-glutamate receptors, responsible for post-synaptic neural excitation [95, 96]. Despite elusive cellular functions, NETO2 expression has been observed in several cancer types [97, 98].

Interestingly, similar to our own results, NME1 regulated expression of NETO in a microarray study performed in a breast cancer cell line. NME1 reduced expression of

NETO2 in this study [61]. Our microarray studies show NME1 increases expression of

NETO2 in melanoma cells. The contrary results suggest NME1 may differentially

19 regulate expression of NETO2 depending on cancer type. Supporting this idea, is the observation that NETO2 expression also varied based on cancer type. In the study by

Oparina, et al ., NETO2 mRNA expression was assessed in multiple cancers. While some cancer types expressed elevated NETO expression, potentially indicating tumor driving functions, NETO2 expression was lower in stomach cancer [98]. The observed loss of

NETO expression in stomach cancer, suggests NETO may be necessary to inhibit tumor development and progression. NME1 induction of NETO2 in our study, may also indicate

NETO2 performs metastasis inhibiting effects in melanoma. With the limited available literature on NETO2, it is a novel gene to study as a transcriptional target and effector of the suppressor activity of NME1.

2.2 Methods

Microarray Analysis Affymetrix microarray analysis was performed as part of the dissertation work of Dr. Joseph Robert McCorkle in the Kaetzel laboratory at the

University of Kentucky (2003-2010) [35, 73]. Briefly, RNA from WM793 parental,

-NME1 wild-type, -NME1 E 5A mutant, and -NME1 K 12 Q mutant, were isolated using the

Qiagen RNeasy RNA Extraction Kit (Qiagen, Valencia, CA, USA) according to manufacturer instructions. Sample preparation and microarray analysis was performed at the University of Kentucky Microarray Core Facility using the GeneChip®Human Exon

1.0 ST Array (Affymetrix, Santa Clara, CA, USA). Analysis of microarray data was performed by the Informatics Resource Center at the Institute for Genome Sciences,

University of Maryland Baltimore. Data was normalized using the Robust-Multi-Array

Average (RMA) method, and differential gene expression was performed with Linear

Models of Microarray (LIMMA), both methods are part of the R package. P-values less than 0.01 were used to identify significant gene targets. The Nanostring transcriptome assay was performed at the Genomics Research Center at the University of Maryland

Baltimore, with a custom plate consisting of 16 target genes of interest, and 5 housekeeping genes. Total RNA from M14-vector and -NME1 cells used for the analysis was extracted as mentioned above by Dr. Katie Leonard, a post-doctoral fellow in the Kaetzel laboratory at University of Maryland Baltimore (2012-2018). Analysis of the data was performed by Nanostring Technologies using nSolver (v.2.5). Target genes were considered detectable if the mean expression for the target gene was greater than or equal to 2 standard deviations above the mean of negative controls. Analysis of microarray and

Nanostring data were presented in Leonard, et al. [35].

Cell Lines and Cell Culture WM793, WM278, WM1158, 451Lu, WM164, WM9,

WM1232, and WM239 parental cells were gifts from Dr. Meenhard Herlyn (Wistar

Institute, Philadelphia, PA). Melanoma cell lines were cultured in Tu2% media composed of: MCDB:Leibovitz-15 medium (4:1, v/v; Sigma-Aldrich, St. Louis, MO), 2 mM CaCl 2,

o 2.5 µg/ml insulin and 2% fetal bovine serum and maintained at 37 C in 5% CO 2. MDA-

MB-435s/M14 cells (referred to as M14 in this text) were obtained from Dr. Rina Plattner

(University of Kentucky, Lexington, KY) and maintained in complete DMEM media

(Life Technologies, Grand Island, NY) containing 4.5 g/l D-glucose, 4mM glutamine,

o and 1mM sodium pyruvate, supplemented with 10% FBS, and kept at 37 C in 10% CO 2.

WM793 cells stably expressing NME1-wild type, NME1-E5A, or NME1-K12 Q mutants were generated as part of the dissertation work of Dr. Qingbei Zhang in the Kaetzel

21 laboratory at University of Kentucky (2001-2006) [47, 99]. M14 stable NME1 expressing cells were generated by Dr. Marian Novak, a former graduate student in the Kaetzel laboratory at University of Kentucky (2005-2012) [100].

Lentivirus Production Forced expression of NME1 in WM1158 cells was achieved using Life Technologies Gateway Recombination Cloning system (Life Technologies).

NME1 cDNA was cloned in to the Gateway Technologies pENTRA1A plasmid purchased from the University of Maryland Baltimore Viral Core Facility, using restriction enzymes SalI and EcoRI (Invitrogen, Carlsbad, CA, USA). Subcloning of

NME1 from the pENTRA1A plasmid in to the Gateway pSMPUW-CMV-IRES-EGFP lentiviral destination vector, was performed using corresponding recombination sites on the pENTRA1A plasmid and the pSMPUW lentiviral vector. Bulk production of the lentivirus was performed by co-transfecting 293t cells, maintained in viral media

(DMEM, 10% charcoal/dextran stripped FBS, 1mM pyruvate, 1x glutamine, and 24mM

HEPES), with 4µg of pSMPUW-CMV-IRES-EGFP empty-vector or -NME1 lentivirus vector, 0.67µg of pVSVG envelope plasmid, and 4µg of pLenti17531 packaging plasmid using 3:1 Fugene 6 transfection reagent (Promega, Madison, WI, USA). Viral supernatant containing lentiviral particles was collected every 8-16 hours for 3 days. Collected supernatant was centrifuged at 500g for 5min at 4 oC, followed by filtration on a 45µm filter. Lentivirus titer was determined using the Clontech Lenti-X qRT-PCR Titration Kit

(Takara Bio, Kusatu, Shiga Prefecture, Japan).

Generation of WM1158-NME1 Expressing Cells Transduction of WM1158 parental cells was achieved by plating cells at 5x10 4cells/well in a 6-well dish. At 60% confluency, cells were infected with pSMPUW-CMV-IRES-EGFP empty-vector or

-NME1 lentivirus at a MOI of 50. Forty-eight hours post infection, virus media was replaced with complete TU2% media. GFP images, collected from an Evos FL fluorescent microscope (Life Technologies), were used to determine transduction efficiency. Transduced cells were expanded and sorted based on GFP expression by fluorescent activated cell sorting (FACS) on an Aria Flow Cytometer (Becton-Dickson,

Mountain View, CA) at the University of Maryland Baltimore Flow Core facility. FACS gating to collect cell populations with various GFP fluorescence intensities (Low GFP,

Medium GFP, and High GFP), was conducted by Dr. Livak at the Flow Core facility.

Immunoblot Analysis Immunoblots of ALDOC, LRP1b, NETO2, and NME1 were conducted as described [35]. Briefly, 50µg of protein from whole cell lysates of M14- vector and M14-NME1 cells were run on a BioRad AnyKD Criterion Precast Midi

Protein gel (Bio-Rad Laboratories, Hercules, CA, USA), followed by transfer to a nitrocellulose membrane. Membranes were incubated with anti-ALDOC (SC-12065,

Santa Cruz Biotechnologies, Santa Cruz, USA), anti-LRP1b (K-20, SC-49229, Santa

Cruz), anti-NETO2 (AF3859, R&D Systems, Minneapolis, MN, USA ), anti-NME1

(610247, BD Biosciences, San Jose, CA, USA), and anti-β-tubulin (9F3, Cell Signaling

Technologies, Danvers, MA, USA) antibodies.

To observe NME1 expression in the melanoma cell panel, WM793, 451Lu,

WM164, WM1158, WM9, WM1232, and WM239, cells were plated in 10 cm dishes,

23 and harvested when 80-90% confluent. Lysates were generated using 1x RIPA lysis buffer (10mM Tris-HCl pH 7.5, 1mM EDTA, 0.5mM EGTA, 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% sodium dodecyl sulfate, 140mM NaCl) supplemented with

Halt Protease Inhibitor Cocktail (Thermo Scientific, Rockford, IL, USA). SDS- polyacrylamide gel electrophoresis using a BioRad AnyKD Criterion Precast Midi

Protein gel (Bio-Rad Laboratories), was performed on 20μg of protein lysate as determined by the BCA assay (Thermo Scientific). Following transfer to a nitrocellulose membrane, blots were incubated with anti-NME1 antibody (D14H1, Cell Signaling) for

16 hours at 4 oC. NME1 protein was detected by chemiluminescence, following addition of Amersham ECL Prime Western Blot Detection Reagent (GE Healthcare, Chicago, IL,

USA). To ensure equal loading of protein, Coomassie stain of the blot was performed.

Immunoblot analysis of WM1158 transduced cells was performed in a similar manner to the procedure above, with some notable exceptions. Following transfer to nitrocellulose membranes, blots were incubated with anti-NME1 antibody from BD

Biosciences (610247, BD Biosciences). Further, Histone 3 (D1H12, Cell Signaling) was used as a loading control.

Quantitative Real-time Polymerase Chain Reaction (qRT-PCR) Total RNA was isolated based on manufacturer instructions using the Qiagen RNeasy Mini Kit (Qiagen).

A total of 1µg of isolated RNA was converted to cDNA using the High Capacity cDNA

Reverse Transcription Kit (Applied Biosystems, Carlsbad, CA, USA). qRT-PCR was performed in triplicate or quadruplicate wells on a CFX Real-Time PCR Detection

System (Bio-Rad Laboratories). Relative fold changes of genes of interest were

24 determined by the 2 -ΔΔCT method using SYBR green qPCR master mix (Applied

Biosystems) with normalization to the internal control, RPL13a. Specified primer sequences, NME1 sense: CAGAGAAGGAGATCGGCTTGT, NME1 antisense:

GCACAGCTCGTGTAATCTACCA, ALDOC sense:

GCTGTCCCAGGAGTGACCTT, ALDOC antisense: CATTCACCTCAGCCCGCTT,

LRP1b sense: CAGCGCAGGCTTCCTTAC, LRPP1b antisense:

TGAATCATCAGGGCAGTCAG, NETO2 sense: CTTATAGATCGTTACTGTGGC,

NETO2 antisense: ACTGACAATCTCCTAGGTAAG, RPL13a sense:

CATAGGAAGCTGGGAGCAAG, RPL13a antisense:

GCCCTCCAATCAGTCTTCTG , were used to perform the analysis. Statistical analysis was performed by paired, two-tailed t-test using Microsoft Excel v 1809.

Unsupervised Hierarchical Clustering and Genome Expression Analysis Melanoma dataset GSE8401 was obtained from the Genome Expression Omnibus [101].

Unsupervised hierarchal clustering of the dataset was performed by Drs. Amol Shetty and

Anup Mahurkar from the Informatics Resource Center at the Institute for Genome

Sciences, University of Maryland Baltimore. Briefly, hierarchical clustering was performed based on Pearson correlation distance matrix, followed by the average linkage clustering algorithm, using log2 transformed data of specified genes. A heatmap of gene expression values was performed using MeV software. The heatmap illustrates clustering of samples based on the selected genes.

For expression analysis, pre-normalized mRNA expression values were downloaded from the GEO repository. ALDOC and NETO2 log2 mRNA expression

25 values were matched to corresponding clusters, or to primary and metastatic clinical patient categories. Statistical differences in mRNA expression of ALDOC or NETO2 between hierarchical clusters was determined by one-way-analysis of variance (ANOVA) using the Dunn method, performed with the SigmaPlot software (Sysat Software, San

Jose, CA, USA). Statistical differences in mRNA expression of ALDOC or NETO2 between primary and metastatic groups was determined by two-tailed t-test using

Microsoft Excel v. 1809.

2.3 Results

2.3.1 NME1 Regulates Expression of Genes That Track with the Metastasis

Suppressor Function

We employed multiple transcriptome methods to identify NME1-regualted genes which correlate with the metastasis suppressor function of NME1. Figure 2.1 A outlines the workflow and stringent validation criteria used to identify these target

NME1-regualted genes. For the initial step, Affymetrix Human Exon 1.0 ST microarray analysis was performed on the WM793 melanoma cell line. Although not a traditional metastatic melanoma cell line, parental WM793 cells were able to colonize the lung after tail vein injection in nude mice [35]. In the same experiment, WM793 cells expressing wild-type NME1, significantly reduced lung colonization compared to the parental cells.

Furthermore, WM793 cells expressing the suppressor-deficient NME1 mutants, -E5A and

-K12 Q, enhanced lung colonization to incidences similarly observed in the parental condition, confirming these mutants exhibit compromised metastasis suppressor activity

[35].

To first identify genes which are regulated by NME1, gene profiles from parental cells were compared with the profiles from NME1-wildtype (NME1-wt) expressing cells.

Next, NME1-wt expression profiles were compared with expression profiles from either

NME1-E5A or NME1-K12 Q mutants ( Figure 2.1 B). Twenty genes were observed to be expressed in either of the comparisons made (WT to parent, WT to E 5A, WT to K 12 Q).

Six genes, identified in bold in Figure 2.1 B, were common to all comparisons performed. Further, these 6 genes were differentially expressed by wild-type NME1, but the expression was unaffected in one or both of the NME1 suppressor-deficient mutants.

This collection of 6 genes, which are regulated by NME1, and track with the metastasis suppressor function of NME1, were denoted by our lab as the Metastasis Suppressor

Signature (MSS) genes. We have reported that the MSS genes collectively have prognostic value in melanoma, further demonstrating these genes have increased likelihood of contributing to the suppressor function of NME1 [35].

Results from the microarray were validated by Nanostring, a reliable, high throughput method to observe differential expression of target genes [35]. Sixteen

NME1-regulated gene targets from the microarray were selected for validation by

Nanostring. Nanostring analysis was performed using the M14 metastatic melanoma cell line with forced expression of wild-type NME1. Of the 16 genes tested, 9 genes were validated to be differentially expressed by NME1 in the M14 cell line

(Figure 2.1 C). We chose to focus our future analyses on ALDOC , NETO2 , and LRP1b

(marked in a red box in Figure 2.1 C), as these genes were part of the initial 6-gene MSS gene list, and therefore have increased probability of affecting metastasis. Further,

ALDOC, NETO2, and LRP1b expression have all been associated with cancer in the literature [86, 88, 89, 97, 98].

Protein induction of ALDOC, NETO2, and LRP1b by NME1 was confirmed in the M14 cell line by immunoblot analysis ( Figure 2.1 D). qRT-PCR analysis was the final step in the validation process for the selection of target NME1-regualted genes

(Figure 2.1 A). The WM793 cell line was selected for qRT-PCR validation analysis as this cell line was used for the microarray study. Upregulation of ALDOC and LRP1b mRNA was confirmed to be induced by NME1 in the WM793 cell line ( Figure 2.1 E).

Taken together, 6 NME1-regulted genes were identified through microarray analysis as candidate effectors of the metastasis suppressor function (MSS genes). Of these MSS genes, 3 target genes ( ALDOC , NETO2 , and LRP1b ) were validated to be regulated by

NME1 via Nanostring, immunoblot, and qRT-PCR analyses. These genes serve as ideal transcriptional targets of NME1, with high potential of contributing to the metastasis suppressor function.

B. C.

D. E.

Figure 2.1 ALDOC, NETO2, and LRP1b are Upregulated by NME1 and Correlate with the Metastasis Suppressor Function. A. A schematic workflow for identifying and validating NME1 regulated genes with high probability of contributing to metastasis suppressor function. B. A heatmap display of microarray analysis performed in WM793 cells expressing NME1-wt, and NME1 suppressor deficient mutants (E 5A and K 12 Q). Gene expression comparison of specified groups was performed to identify genes that track with the metastasis suppressor function of NME1. Six genes identified in bold were common to all comparisons made. The figure was adapted from Leonard, et al. [35]. C. Nanostring validation of target genes resulted in 9 confirmed NME1-regulated genes in the M14 cell line. ALDOC, NETO2, and LRP1b (highlighted with a red box) were selected as NME1-regulated candidate transcriptional targets. The figure was adapted from Leonard, et al. [35] D. Increased protein expression of ALDOC, NETO2 and LRP1b was confirmed in NME1 overexpressing M14 cells. The figure was adapted from Leonard, et al. [35]. E. ALDOC and LRP1b were induced by NME1 in the WM793 cell line by qRT-PCR analysis. **p≤0.01, and *p≤0.05.

2.3.2 NME1 is Successfully Over-expressed in the WM1158 Metastatic Melanoma

Cell Line

Microarray analysis was conducted in the WM793 cell line. WM793 cells were derived from an invasive melanoma stage, known as the vertical growth phase (VGP), which occurs prior to metastasis. Nanostring transcriptome analysis to validate results from the microarray was performed in a metastatic melanoma cell line (M14). We aimed to identify an additional metastatic melanoma cell line to confirm NME1 regulation of the 3 target MSS genes. Immunoblot analysis was performed on whole cell lysates from 6 metastatic melanoma cell lines to identify cells with endogenously low expression of

NME1. Although NME1 expression is variable across this panel of cells, WM1158 cells express low NME1, similar to levels observed in the WM793 cells ( Figure 2.2 A). The presence of 2 bands on the immunoblot in Figure 2.2A was due to the use of an antibody targeted to the N-terminal region of NME1. The higher molecular weight band was identified as a splice variant of NME1, containing 25 additional amino acid residues at the N-terminus compared to the predominant NME1 isoform [102]. Forced expression of

NME1 in melanoma cell lines was performed with cDNA coding for the predominant

NME1 isoform. To ensure specificity of detection of this predominant isoform, an antibody targeted to the C-terminal region of NME1was used for all future analyses.

We selected the WM1158 cell line to over-express NME1. The Life Technologies

Gateway Recombination Cloning system was used to clone NME1 cDNA in to a lentiviral vector. The Gateway Cloning system provided a more reliable method of cloning compared to restriction enzyme cloning methods, due to the use of recombination sites present on donor and receiver plasmids. NME1 cDNA from the donor pENTR1A

30 plasmid was cloned in to a pSMPUW lentiviral receiver vector via the attR and attL recombination sites ( Figure 2.2 B). The lentiviral vector expresses the constitutively active CMV promoter, and an IRES-EGFP reporter ( Figure 2.2 B). The GFP reporter allowed for visualization of transduction efficiency by fluorescent microscopy ( Figure

2.2 C). Approximately 50% of cells were GFP positive, with cells expressing varying intensities of GFP. NME1 over-expression in the WM1158 cell line was confirmed by immunoblot analysis ( Figure 2.2 C). Further, qRT-PCR analysis was performed to observe NME1 , ALDOC , LRP1b , and NETO2 transcript levels. NME1 mRNA over- expression was confirmed in the WM1158 cells. Further, of the MSS genes tested, only

ALDOC expression was induced by NME1 with a 2-fold increase in transcript levels

(Figure 2.2 D) .

A. B.

C. D.

Figure 2.2: The Metastatic Melanoma Cell Line WM1158 was Selected for Forced Expression of NME1 . A. Immunoblot analysis was performed on a panel of metastatic melanoma cell lines (MET) to observed NME1 protein expression. WM1158 cells express endogenously low levels of NME1 protein. B. NME1 was cloned in to the pSMPUW lentiviral vector using Gateway Recombination Cloning method via recombination sites attL and attR on the pENTR1A and pSMPUW vectors C. Post infection, transduction efficiency was evaluated in WM1158 cells by fluorescence microscopy using the GFP reporter. NME1 over-expression was confirmed by immunoblot analysis D. qRT-PCR analysis confirmed increased NME1 mRNA expression post transduction. Further, ALDOC expression was induced 2 fold by NME1. **p≤0.01, and *p≤0.05.

2.3.3: MSS Genes are Dose-Responsive to NME1 Expression

Fluorescent microscopy images of the WM1158 transduced cells revealed a mixed population of cells expressing varying intensities of GFP expression ( Figure 2.2 C). The internal

ribosomal entry sequence (IRES) on the pSMPUW lentiviral vector results in efficient translation

of both NME1 and GFP proteins, permitting GFP to serve as a surrogate for NME1 expression.

WM1158 cells were sorted by fluorescent activated cell sorting (FACS) with gating performed to

collect pure cell populations expressing low, medium, and high GFP (Figure 2.3 A).

Approximately 50% of cells were negative for GFP, which corresponded to the transduction

efficiency previously observed. Equal cell populations for the low, medium, and high GFP

conditions were collected ( Figure 2.3 A). As expected, NME1 mRNA expression correlated with

GFP expression, with low, medium, and high GFP pooled cells also having low, medium, and

high NME1 expression ( Figure 2.3 B). ALDOC , LRP1b , and NETO2 expression was observed in

each of these cell populations. Interestingly, all 3 genes were dose-responsive to NME1

expression. Both ALDOC and LRP1b expression were significantly increased in a step-wise

manner in each of the low, medium and high NME1 expressing cells. NETO2 expression was not

significantly enhanced in the low or medium cell populations. However, an increasing trend in

NETO2 mRNA expression was observed, with significant induction obtained in the high-NME1

expressing cell population ( Figure 2.3 C). Neither LRP1b or NETO2 were observed to be

upregulated by NME1 prior to sorting. Post-sorted, pure populations of NME1 expressing cells,

appeared to reveal the NME1 dependent induction. This result may suggest that non-NME1

expressing cells may be masking induction of these genes, potentially through intercellular

communication between non-NME1 and NME1 expressing cells.

A. B.

Figure 2.3: NME1 Upregulates Expression of ALDOC, LRP1b, and NETO2 in a Dose- Responsive Manner . A. WM1158 cells were sorted by FACS in to low, medium, and high GFP expressing populations. B. NME1 mRNA expression corresponds to GFP expression. C. ALDOC, LRP1b, and NETO2 expression is induced in a step-wise manner with increasing expression of NME1. **p≤ 0.01

2.3.3 ALDOC mRNA Expression is Lower in a Subset of Metastatic Melanoma

Human Biospecimen

While the MSS genes collectively were observed to have prognostic value in melanoma, we hypothesized that singular MSS genes may also demonstrate clinical significance in melanoma [35]. We utilized the Genome Expression Omnibus (GEO) repository to aid in the identification of NME1-regulated MSS genes which demonstrated clinical relevance. The GSE8401 dataset was preferentially selected as it is comprised of primary and metastatic human melanoma biospecimen [101]. Although not all the MSS

33 genes were represented in this dataset, the top 3 gene candidates, ALDOC , LRP1b , and

NETO2 were presented. Unsupervised hierarchical clustering was performed by the

Informatics Resource Center to observe correlational relationships between MSS genes and primary and metastatic clinical samples. A heatmap displaying transformed log2 expression of the MSS genes revealed 2 primary clusters which appeared to be driven by differences in ALDOC and NETO2 expression ( Figure 2.4 A). Cluster 1 was comprised of both primary and metastatic samples, while cluster 2 was comprised of only metastatic samples. We further observed sub-clustering of cluster 2 in to clusters 2a and 2b, once again driven by differences in ALDOC and NETO2 expression. To visually observe expression differences in each of the clusters and sub-clusters, log2 mRNA expression of

ALDOC and NETO2 were graphed on a box plot ( Figure 2.4 A, bottom). Primary and metastatic samples were separated in to two distinct groups within cluster 1. Within cluster 1, no expression differences were observed between primary and metastatic samples for either ALDOC or NETO2 transcripts. NETO2 expression was significantly elevated in both sub-cluster 2a and 2b comprised solely of metastatic samples, when compared to primary or metastatic samples from cluster 1. Although no difference in

ALDOC expression was observed in sub-cluster 2a, a significant decrease in expression was observed in cluster 2b compared to primary samples in cluster 1 and the metastatic samples in cluster 1 and 2a.

NME1 increased expression of ALDOC and NETO2 in melanoma cell lines, suggesting these genes may aide in inhibiting melanoma progression. Providing this hypothesis, we expected to observe decreased expression of ALDOC and NETO2 in metastatic melanoma samples compared to primary melanomas. Although we did not

34 observe a difference in expression of ALDOC when the GSE8401 dataset was stratified by sample type (primary vs metastatic) (Figure 2.4 B ), we did observe significantly lower expression of ALDOC in metastatic samples from cluster 2b compared to the primary melanoma samples from cluster 1. The result suggests that in a subset of metastatic melanoma samples, ALDOC may be acting as a metastasis inhibiting agent, supporting our initial hypothesis. Interestingly, NETO2 expression was observed not only to be elevated in both sub-cluster 2a and 2b, but also significantly elevated in the collection of metastatic specimen compared to primary melanomas ( Figure 2.4 B) . The result is in contrast to our hypothesis that NETO2 functions to inhibit metastasis, and instead supports a role for NETO2 in cancer progression.

ALDOC expression was consistently upregulated by NME1 in a panel of melanoma cell lines by microarray, qRT-PCR, and immunoblot analyses. Further,

ALDOC mRNA expression was observed to be lower in a collection of human metastatic melanoma specimen compared to primary melanomas, supporting a putative role as an inhibitor of metastasis. Taken together, we have identified ALDOC as a candidate target of NME1-transcriptional regulation, with high probability of executing the metastasis suppressor function of NME1.

Figure 2.4: ALDOC and NETO2 Expression Drive Hierarchical Clustering in the GSE8401 Dataset. A. A heatmap display (top) of MSS gene expression in GSE8401 shows clustering of genes in to 2 predominant clusters (Cluster 1 and Cluster 2). Cluster 2 is subclustered in to 2a and 2b. Clustering is driven by ALDOC and NETO2 expression in this dataset. Clustering and heatmap generation was performed by Drs. Amol Shetty and Anup Mahurkar from the Institute for Genome Sciences (UMB). ALDOC is lower in a subset of metastatic samples (bottom). NETO2 expression is higher in both 2a and 2b clusters comprised of metastatic melanoma samples. Plots not sharing a symbol are significant (p ≤ 0.05) as determined by ANOVA using the Dunn method. B. No changes in ALDOC expression is observed between primary and metastatic samples. NETO2 expression is higher in metastatic melanoma samples compared to primary melanomas. Significance was determined by two-tailed t-test ** p ≤ 0.01.

2.4 Discussion

Several molecular mechanisms have been attributed to the metastasis suppressor function of NME1, including a potential ability to regulate expression of genes which may be functional in suppressing metastasis [52, 72, 99, 103]. In breast cancer, NME1 regulated expression of several genes in a microarray study. Knockdown of the EDG2 gene, which was downregulated in response to NME1 expression, resulted in decreased in vivo metastasis in breast cancer [61]. The result not only demonstrates that NME1 regulates global gene expression, but that there may be a predominant NME1-regulated gene that contributes to the metastasis suppressor activity.

Through multiple transcriptome methods and employing stringent filtering criteria, we identified 3 NME1-upregualted genes ( ALDOC , NETO2 , LRP1b) with high probability of affecting metastasis. Amongst the 3 genes, only ALDOC mRNA expression was lower in a subset of human metastatic melanoma biospecimen compared to primary melanoma samples in the GSE8401 dataset [101]. The result suggests that loss of ALDOC expression may contribute to melanoma progression. Further compelling is the result by Li, et al ., in which elevated ALDOC expression was associated with increased probability of survival in oral squamous cell carcinoma [86]. As such, we hypothesize that ALDOC is a transcriptional target of NME1, functioning to mediate the metastasis suppressor function. We cannot eliminate the possibility that interactions amongst several of the NME1-regualted genes may contribute to the suppressor function.

However, our results indicate that ALDOC may be a dominant effector of the metastasis suppressor function.

Despite evidence demonstrating substantial metastasis suppressor activity in melanoma cell lines, mRNA expression of NME1 in human melanoma specimen has not been a reliable prognostic indicator [104, 105]. NME1 mRNA expression in the

GSE8401 dataset, and the larger data collection from The Cancer Genome Atlas, was higher in metastatic melanoma samples compared to primary melanomas (data not shown). This discrepancy may be explained by the finding that the decrease in NME1 expression observed in metastatic melanoma cell lines is a result of lysosome-mediated degradation of NME1 protein, and not due to decreases in NME1 mRNA expression

[106]. As such, NME1 protein expression was observed to be a more reliable biomarker in breast cancer, pancreatic ductal carcinoma, and melanoma [104, 107, 108]. Publicly available clinical melanoma datasets have limited or no proteomics data, making it difficult to ascertain meaningful conclusions for prognosis from NME1 expression. Gene expression of the aldolase family members, including ALODC , have been shown to be reliable predictors of human outcomes in cancers and other diseases [109]. ALDOC may serve as a functional proxy for NME1 as a prognostic predictor in human melanomas.

Like many cancers, melanoma displays a high degree of heterogeneity [110].

Melanoma patients have variable oncogenic mutations and other variances across the genomic landscape, making targeted therapies a challenge [111]. Even with targeted treatments (eg. BRAF inhibitors), melanoma patients demonstrate either initial or acquired resistance to the therapies [21]. Additional genomic and proteomic approaches to stratify patients in to subgroups may provide more impactful therapeutic strategies. For example, patients exhibiting decreased NME1 protein expression with decreased ALDOC expression, may represent subsets of patients that would be more responsive to a

38 particular treatment. Furthermore, elucidating the correlational relationship between

NME1 and ALDOC expression by understanding a mechanism with which NME1 regulates ALDOC gene expression, can provide valuable insight in to approaches to develop more personalized targeted therapies.

Chapter 3: Metastasis Suppressor NME1 Directly Activates

Transcription of the ALDOC Gene in Melanoma Cells 1

3.1 Introduction

We have identified a cohort of NME1-regualted genes that track with the metastasis suppressor function. Collectively, the genes are prognostic predictors of survival in melanoma [35]. Amongst these genes, aldolase C ( ALDOC ) was upregulated by NME in melanoma cell lines. The robust correlation of NME1 and ALDOC transcripts, suggest NME1 might regulate transcription of ALDOC .

The NME family of proteins have been implicated in gene transcription through both direct and indirect modes of regulation [57]. To date, transcriptional activity has mostly been studied and ascribed to NME2, a close homolog of NME1. Through screening of a HeLa cDNA expression library, NME2 was identified as a binding factor of PuF consensus DNA segments located on the promoter of the CMYC gene [113]. Since this initial finding, NME2 transcriptional regulation of the CMYC gene has been widely studied [114]. NME2 binds to G-rich segments on the CMYC promoter which arrange in to a non-B form DNA structure, identified as the G-quadraplex (G-4 ) [115].

Interestingly, binding of the G-4 complex was specific to NME2, as NME1 did not bind to the particular DNA motif [113-115]. Although NME1 and NME2 share strong sequence homology, they do not share the same mode of DNA binding, emphasizing distinct properties of these proteins.

1 112. Pamidimukkala, N.V., et al., Metastasis Suppressor NME1 Directly Activates Transcription of the ALDOC Gene in Melanoma Cells. Anticancer Res, 2018. 38 (11): p. 6059-6068.

In spite of the inability to bind to the G-4 complex, NME1 does exhibit DNA binding activity. Previous work from our lab, attributed a 3’ to 5’ exonuclease (EXO) activity to NME1. NME1 binds to single stranded segments of DNA, with preferential cleavage of the DNA in a 3’ to 5’ directionality [55, 62, 116]. The discovery of the EXO activity came at an initial finding by our lab, demonstrating in vitro binding of NME1 to single stranded DNA segments located on the PDGF-A gene [55, 62, 116]. Binding to

DNA segments present on gene loci is highly suggestive for a role of NME1 as a regulator of gene transcription. While a few studies have alluded to NME1 as a direct transcriptional regulator, no study has demonstrated rigorous transcription factor functionality of NME1 [63, 64]. The ALDOC gene serves as an excellent prototype to elucidate NME1 mechanisms of transcription.

NME1 has previously displayed preferential binding to G-C rich segments of

DNA [55, 62, 116]. The ALDOC promoter is comprised of several G-C rich regulatory regions, which may serve as potential locales for NME1 binding [117-119]. Further, a G-

C rich region located near the ALDOC transcription start site was identified as a transcriptional enhancer element [119].

The current work demonstrates in vivo binding of NME1 to the ALDOC gene.

Further, NME1 expression enhanced epigenetic activation markers H3K27ac and

H3K4me3 and increased RNApolII occupancy at the ALDOC proximal promoter regions.

To our knowledge, we are the first lab to comprehensively illustrate NME1 regulation of a single gene occurring through transcriptional mechanisms.

3.2 Methods

Cell lines and Cell Culture Cell culture conditions and NME1-forced expression in M14 and WM1158 cell lines can be found in the methods section of Chapter 2. Additionally, lentiviral production for shRNA viral vectors can be found in Chapter 2. Knockdown of

NME1 in the WM278 cell line was achieved by transient lentiviral transduction with

Mission Sigma (SHC002, Sigma-Aldrich, St. Louis, MO, USA) shNME1

TRCN0000010062 or pLKO.1 non-target shRNA at a MOI of 1000. For stable knockdown of ALDOC in the M14 cell line, cells were infected with Mission Sigma lentivirus shRNA targeted to the coding sequence of ALDOC. ALDOC was effectively knocked-down using shALDOC#1 (TRCN0000052513), shALDOC#2

(TRCN0000052414), or a pLKO.1 non-target shRNA at a MOI of 1000. Transduced

M14 cells were maintained in puromycin (1μg/mL) for 7 days to obtain stable ALDOC knock-down cells.

Quantitative Real-time PCR (qRT-PCR) Detailed methodology for qRT-PCR is described in Chapter 2. Analysis was performed on 3 independent biological replicates for M14 and WM1158 cells, with each replicate performed in quadruplet replicate wells.

In addition to NME1, ALDOC, and the RPL13a control primer (specific sequences found in Chapter 2), ALDOA sense primer: CGTCCACGGACTCTCCGTTA and antisense primer: GCGATGTCAGACAGCTCCTT, ALDOB sense primer:

CGGCAGTCCCGAGAAATCCT and antisense primer:

CATCAAGCCCTTGAATGGTGG, and pre-mRNA ALDOC sequence primers targeted to intron 1; sense primer: CTGAGAGAAGAGGGTGGCAA and antisense primer:

CCAATCAAGGATGCCAACCC were used. Statistical analysis was performed by paired, two-tailed t-test using Microsoft Excel v 1809.

Immunoblot Complete details for immunoblot methodology can be found in Chapter 2.

Briefly, whole cell lysates of M14, WM1158, and WM278 cells were generated using 1x

RIPA lysis buffer. SDS-polyacrylamide gel electrophoresis was performed on 5-20 μg of protein lysate as quantified by the BCA assay (Thermo Scientific), followed by transfer to nitrocellulose membranes. Membranes were incubated with specific antibodies against

NME1 (610247, BD Biosciences), ALDOC (SC-12065, Santa Cruz Biotechnologies), or

Sp1 (CS200631, Millipore). An antibody directed to Histone 3 (D1H12, Cell Signaling) was used to verify equal loading of protein. Three washes with TBST buffer were performed, followed by incubation with one of the following isotype specific-HRP- conjugated secondary antibodies: ECL-conjugated anti-mouse IgG for anti-NME1 blots

(GE Healthcare), anti-goat-IgG-HRP for anti-ALDOC blots (Santa Cruz) and ECL- conjugated anti-rabbit IgG for anti-Histone 3 and anti-Sp1 blots (GE Healthcare).

Proteins were detected on immunoblots by chemiluminescence after addition of

Amersham ECL Prime Western Blot Detection Reagent (GE Healthcare) and exposure to

Amersham Hyperfilm ECL (GE Healthcare).

mRNA Decay Kinects of ALDOC and CMYC Cultured M14 cells were incubated with actinomycin D at 5μg/mL (Sigma Aldrich) to inhibit transcription. Cell pellets were collected at 1 hour, 2 hour, 3 hour, and 4 hour timepoints. Timepoint measurements were limited to 4 hours to prevent confounding effects occurring by actinomycin D-mediated

43 apoptosis [120]. A 0 hour timepoint in which cells were treated with DMSO was used for baseline measurements of ALDOC and CMYC mRNA. Total RNA was isolated based on manufacturer instructions using the Qiagen RNeasy Mini Kit (Qiagen). mRNA quantification was performed by Taqman probe qPCR method using the GoTaq Probe 1-

Step RT-qPCR System (Promega). ALDOC probe primer set: sense: GATGCCTCTTCTTCGCTCTG, antisense: CCTGTCCCATCAAGTATACCC with probe: /56-FAM/CAGCTGTCC/ZEN/CAGGAGTGACCTTC/3IABkFQ and CMYC probe primer set: sense: CAGTAGAAATACGGCTGCAC, antisense: TTCGGGTAGTGGAAAACCAG with probe: /56-FAM/CCGCGACGA/ZEN/TGCCCCTCAA/3IABkFQ was purchased from Integrated DNA Technologies (IDT, Coralville, IA, USA). A GAPDH probe primer set was used as an internal normalization control and was kindly provided by Dr. Gerald

Wilson; sense: GACAGTCAGCCGCATCTT, antisense: ACTCCGACCTTCACCTTCC with probe: /5Cy5/CGCCAGCCG/TAO/AGCCACATCGC/3IAbRQSp. Each single reaction was performed with: 10ng/μL of RNA, 500nM of ALDOC or CMYC primers with 250nM of respective probe, 100nM of GAPDH primers, with 50nM of GAPDH probe. GoTaq Probe qPCR master mix and GoScript RT Mix for 1-step were added to reactions according to manufacturer instructions. Reverse transcription and qPCR one- step analysis was performed on a BioRad CFX96 Touch Real-Time PCR Detection

System courtesy of Dr. Wilson using dual fluorescent channels FAM (detection at 510-

530nm) and Cy5 (detection at 675-690nm) for quantification. mRNA measurements from single reactions were normalized to the corresponding internal GAPDH control. The average quantified mRNA amounts of 4 single reactions for each timepoint was

44 determined for ALDOC and CMYC genes. Percent mRNA remaining was calculated relative to the 0 hour timepoint. A total of 3 independent experiments were performed, with a representative experiment displayed. mRNA decay curves and t1/2 measurements were calculated based on t1/2 =ln2/k using Prism 5.0 software (GraphPad, San Diego, CA,

USA). Significance of t1/2 measurements between vector control and NME1-forced expressing cells was determined by paired, two-tailed t-test using Microsoft Excel v

1809.

Luciferase Promoter Activity Assay M14 and WM1158 vector control and NME1- forced expressing cells were transiently transfected with 100ng of a Renilla luciferase pLightSwitch promoter plasmid containing the transcriptional regulatory region of

ALDOC (-880 to +118) (SwitchGear Genomics, Menlo Park, CA, USA). Additionally,

50ng of a pcDNA3.1-based Firefly luciferase plasmid courtesy of Dr. William Kaelin

[121] (Addgene, Cambridge, MA, USA), was co-transfected to serve as a transfection efficiency control. Cells were transfected using Fugene 6 (Promega) as previously described in Chapter 2. Luciferase activity was measured in cell lysates 48 hours post- transfection using the Dual Luciferase Reporter Assay Kit (Promega). The Renilla luciferase promoter signal was normalized to the Firefly luciferase signal, and relative fold-change in relative luminescence units (RLU) was quantified. The average of 2 independent experiments is displayed for M14 cells. A representative experiment from two independent experiments is displayed for WM1158 cells. Statistical analysis was performed by paired, two-tailed t-test using Microsoft Excel v 1809.

Chromatin Immunoprecipitation (ChIP) Sheared chromatin for ChIP analysis was generated using the ChIP-IT Express Kit (Active Motif, Carlsbad, CA, USA). Fifteen micrograms of chromatin was pre-cleared on Protein G magnetic beads (Cell Signaling) with isotype control antibodies, mouse IgG (G3A1, Cell Signaling) or rabbit IgG (DA1E,

Cell Signaling), for 1 hour at 4 oC. Immunoprecipitation with 5 μg of specified antibodies to NME1 (SC-NM301, Santa Cruz Biotechnologies), Histone 3 lysine 27 acetylation

(H3K27ac, clone D5E4, Cell Signaling), Histone 3 lysine 4 trimethylation (H3K4me3,

Cell Signaling), RNA Polymerase II (Active Motif) , and Sp1 (CS200631, Millipore) was performed overnight at 4 oC on an end-to-end rotator. Eluted DNA was purified using the

QiaQuick PCR Purification Kit (Qiagen) prior to qPCR analysis. qPCR was performed using primer sequences targeted to specified locations on the ALDOC gene: -542/-385 sense: AAACTCTGGGCACTGGACTG, antisense: TTCAGGGCAGGAAGTAATGAA;

-188/-4 sense: GAGGGCAGTCCCTAACAGC, antisense:

GTAAATGAGGCTGCGGATGT; and +31/+422 sense:

GGATCAGAACCCGAGCTGT, antisense: GAGCCATGGGTTTCAAGTA.

Quantification was achieved by normalization to input DNA, and expressed as % input.

A representative experiment is displayed from 2 independent experiments performed.

Statistical analysis was performed by paired, two-tailed t-test using Microsoft Excel v

1809.

Cell Proliferation Assay Cell proliferation of M14 stable ALDOC knockdown cells was determined by plating equal cell numbers in 24-well culture plates in complete DMEM media. Cells were detached from wells using 0.05% Trypsin-EDTA solution (Thermo

Scientific) and counted on a hematocytometer using trypan blue solution at 24, 48, 72, and 96 hours post plating. A total of 4 counts for each timepoint was performed.

Statistical analysis for significant changes in cell proliferation was performed by two-way analysis of variance (ANOVA) using Shapiro-Wilk normality test performed on

SigmaPlot software (Sysat Software).

Time-lapse Microscopy M14 cells with stable ALDOC knockdown were cultured in complete DMEM media or nutrient deprived DMEM media for 2 days, followed by serum-starvation for 24 hours. Specific components of complete DMEM or nutrient deprived DMEM media are listed in the table below. After serum-starvation, cells were detached from culture dishes using PBS-based Enzyme-free Disassociation buffer

(Thermo Scientific), and plated in 3.5cm culture dishes at 6x10 4 cells in complete

DMEM or nutrient deprived media supplemented with 1% FBS. Four hours post plating, time-lapse microscopy was performed using the Viva View FL Incubator microscope

(Olympus Life Science, Waltham, MA, USA) supplied by the Department of

Biochemistry and Molecular Biology Shared Instrumentation Service. Differential interference contrast (DIC) images were collected at 10X magnification every 20 minutes for 24 hours. Five different image fields per cell condition were obtained. Cell velocity measurements from a total of 100-130 cells for each cell condition were calculated using the Manual Tracking plugin for ImageJ. Significance of single cell velocity measurements were determined by two-way analysis of variance (ANOVA) using

Shapiro-Wilk normality test performed with SigmaPlot software (Sysat Software).

Table 1: Comparison of Complete and Nutrient Deprived DMEM Media Component Complete DMEM Nutrient Deprived DMEM

Glucose 25mM 5.5mM

Glutamine 4mM 0mM

Sodium Pyruvate 1mM 0mM

Gene Expression Profiling Gene expression data from the GSE46517 dataset published by Kabbarah, et al . was obtained from the Gene Expression Omnibus [122]. Normalized log2 expression for ALDOC was evaluated in 17 normal skin and benign nevi, 31 primary melanomas, and 73 metastatic melanoma specimens. Statistical differences for

ALDOC mRNA expression between each of the samples types was performed by one- way-analysis of variance (ANOVA) using post-hoc:Holm-Sidak method using SigmaPlot software (Sysat Software). Genomic and clinical data of the Skin Cutaneous Melanoma collection from The Cancer Genome Atlas (TCGA-SKCM) was obtained via cBioPortal

[123, 124]. Patient identifiers, and normalized log2 mRNA expression of ALDOC were downloaded from cBioPortal and aligned with corresponding clinical attributes

(e.g . Breslow depth). For Breslow depth correlations to ALDOC expression, primary melanoma samples were assigned in to one of two groups: Breslow depth measurements of less than or equal to 4 mm, or measurements greater than 4 mm. Differences in

ALDOC log2 expression was determined by unpaired, two-tailed t-test Microsoft Excel v

1809.

3.3 Results

3.3.1 NME1 Regulates Expression of ALDOC in Melanoma Cell Lines

ALDOC expression was induced by NME1 as measured by microarray analysis in the invasive WM793 melanoma cell line (Figure 2.1 B, Figure 2.1 D). ALDOC upregulation by NME1 was confirmed by Nanostring analysis, a transcriptome method performed to validate our microarray results (Figure 2.1 C). In the present report, we observed robust upregulation of ALDOC mRNA in the M14-NME1 over-expressing condition through qRT-PCR analysis (Figure 3.1 A). Immunoblots performed with vector and NME1 expressing M14 lysates, also demonstrated increased steady-state protein levels of ALDOC in the NME1 expressing condition ( Figure 3.1 A). We have previously demonstrated a dose-dependent increase in ALDOC mRNA expression in response to increasing amounts of NME1 in the WM1158 cell line (Figure 2.3 B). We collected additional replicates (n=3) of total RNA from the WM1158 cell condition expressing high levels of NME1. A significant increase in ALDOC mRNA expression was confirmed in the WM1158-high NME1 expressing cells compared to the control condition (Figure 3.1 B). Furthermore, ALDOC protein expression was also induced by

NME1 in the WM1158 cell line (Figure 3.1 B). To affirm the NME1-ALDOC regulatory axis, the impact of NME1 knockdown on ALDOC expression was assessed. WM278 cells, which express endogenously high levels of NME1, were infected with a lentiviral shRNA construct targeted to NME1. NME1 was effectively silenced in this cell line, resulting in reduced expression of ALDOC protein ( Figure 3.1 C). In sum, the acquired results corroborate the robust regulation of ALDOC by NME1 in multiple melanoma cell lines.

ALDOC is a member of the aldolase family of enzymes comprised of ALDOA,

ALDOB, and ALDOC [125, 126]. The aldolase isotypes are expressed in a tissue specific manner. ALDOA is the most ubiquitously expressed isozyme, with principal expression occurring in the muscle, while ALDOB is localized to the liver and kidney cortex [127].

ALDOC is expressed in the brain and other neural crest derived cells, including melanocytes [127, 128]. To determine the specificity with which NME1 upregulates expression of ALDOC compared to the other aldolase isozymes, ALDOA and ALDOB mRNA expression was measured in the metastatic melanoma cell lines M14 and

WM1158. No changes in ALDOA expression were observed in either of the cell lines

(Figure 3.1 D). ALDOB mRNA expression was undetected in WM1158 cells, while expression was induced by NME1 in the M14 cell line (Figure 3.1 D, right). Despite the apparent induction, ALDOB mRNA transcripts were detected at very low levels as observe by cycle time quantification values from qRT-PCR analysis. The average Ct for

ALDOC in the M14-vector condition, which expresses relatively low amounts of

ALDOC, was 31.0+/- 0.1. The average Ct for ALDOB in the M14-NME1 condition, which expresses relatively high amounts of ALDOB, was only 37.4+/-0.2. As such,

ALDOC was the more readily detected and expressed isoform in the M14 cell line. The result further highlights the tissue specificity with which the isozymes are expressed.

C. D.

Figure 3.1: NME1 Regulates Expression of ALDOC Specifically . A. NME1 upregulates mRNA and protein expression of ALDOC in the M14 cell line. B. NME1 further upregulates expression of ALDOC in the WM1158 cell line. C. Silencing NME1 expression in the shNME1 condition in WM278 cells, results in decreased ALDOC protein expression D. mRNA expression from additional aldolase isoforms (ALDOA and ALDOB) was analyzed by qRT-PCR analysis. No significant changes in ALDOA expression is observed between vector and NME1 conditions in both the M14 and WM1158 cell lines. ALDOB expression is not detected (nd) in the WM1158 cells, while NME1 appears to induce ALDOB expression in the M14 cells. **p≤0.01.

3.3.2 NME1 Promotes Transcriptional Activation of the ALDOC Gene

Several examples from the literature implicate NME1 regulation of target genes occurring through a suggested transcriptional mechanism. However, observed increases in mRNA expression may not solely be attributable to transcription, and may instead be a result of post-transcriptional regulation inhibiting degradation of mRNA transcripts

[129]. We have observed a very strong induction of ALDOC mRNA in NME1 overexpressing M14 cells compared to the vector condition ( Figure 3.1 A). The significant increase in ALDOC transcript may be due to NME1-enhancing stability of the mRNA. To test this hypothesis, M14 cells were treated with actinomycin D (actD), a DNA intercalating agent, which inhibits transcription. In the absence of new transcripts, the rate of mRNA decay of existing endogenous transcripts can be evaluated [129]. The percent of ALDOC mRNA remaining after actD treatment was not significantly different between the vector control and NME1 over-expressing cells ( Figure 3.2 A, left panel).

Interestingly, ALDOC mRNA remained stable, displaying no signs of decay even after 4 hours of actD treatment. To eliminate the possibility that actD treatment was ineffective in the M14 cells, mRNA stability of CMYC , a gene known to have a short mRNA half- life, was measured [130]. CMYC mRNA was rapidly degraded in both M14-vector and

M14-NME1 conditions ( Figure 3.2 A, right panel), demonstrating efficient inhibition of transcription by actD treatment. mRNA decay kinetic measurements further showed that

NME1 over-expression did not impact the rate of CMYC mRNA decay (t1/2=1.0 in vector condition, and t/12=0.8 in the NME1 condition, p=0.2, t-test). The decidedly stable nature of the ALDOC mRNA even after actD treatment suggests it is not regulated through post-transcriptional mechanisms affecting mRNA stability. The results further

52 support NME1 acting to enhance transcription of ALDOC , rather than stabilizing mRNA transcripts.

An initiating event of gene expression involves transcription of DNA to pre- cursor RNA (pre-mRNA) occurring in the nucleus [131]. Prior to being processed and exported to the cytoplasm as mRNA, pre-mRNA reflects nascent RNA transcribed directly from the DNA template, and is considered a preliminary determinant of transcription. A primer set directed to the intronic region of ALDOC was used to detect pre-mRNA expression. Significant increases in ALDOC pre-mRNA were observed in

M14 and WM1158-NME1 expressing cells compared to the control, strongly suggesting

NME1 regulates transcription of ALDOC (Figure 3.2 B).

We next determined if NME1 activates the ALDOC promoter, an additional indicator of transcriptional regulation. M14 and WM1158 cells were transfected with a plasmid containing a portion of the ALDOC promoter (-800 bp to +118bp) located upstream a Renilla luciferase reporter. An observed increase in luciferase activity was observed in NME1-overexpressing M14 cells compared to the vector control condition, indicating ALDOC promoter activation ( Figure 3.2 C). The result is consistent with

NME1 upregulation of ALDOC mRNA and protein expression in this cell line.

Interestingly, an apparent decrease in luciferase activity was observed in NME1- overexpressing WM1158 cells. This result was unexpected as we have previously observed significant increases in ALDOC mRNA and protein levels in this cell line.

However, subsequent qRT-PCR and immunoblot analyses revealed loss of NME1-forced expression in the WM1158 cells used for this study. The loss of NME1 expression corresponded to blunted ALDOC mRNA induction by NME1, providing an explanation

53 for the discrepancy observed in the promoter activity assay. Unstable expression of

NME1 was specific to the WM1158 cell line, as M14 cells maintained forced-expression of NME1. M14 cells were selected for future analyses to minimize confounding results caused by instability of NME1-forced expression.

B. C.

Figure 3.2: NME1 Activates Transcription of the ALDOC Gene. A. ALDOC mRNA is stable after 4 hours of actinomycin D (actD) treatment in both the NME1 and vector condition in the M14 cell line (left). Significant decay of CMYC mRNA was observed (right) confirming effectiveness of the actD treatment. B. NME1 increased unspliced pre-mRNA levels of ALDOC in both the M14 and WM1158 cell lines. C. Promoter activity assays performed in the M14 and WM1158 cells reveal that NME1 activates the ALDOC promoter in the M14 cells, but not in WM1158 cells. **p≤0.01, and *p≤0.05 .

3.3.3 NME1 is Recruited to the ALDOC Gene and Enhances Epigenetic Activation

Markers and RNA PolII Occupancy

To further assess the impact of NME1 transcriptional regulation of ALDOC , chromatin immunoprecipitation (ChIP) analysis was performed using 3 different DNA amplicons (-542 to -398, -188 to -4, and +31 to +244) within the regulatory region of the

ALDOC gene. Prior to performing ChIP analysis, we consulted data from the

Encyclopedia of DNA Elements (ENCODE) consortium using the University of

California, Santa Cruz (USCS) browser interface [132, 133]. The ENCODE project involved conducting ChIP-seq analysis on several different cell types in order to map genome wide regulation occurring by transcription factor binding and histone modification presentation [133]. Using this data, we were able to predict the location of likely epigenetic targets on the ALDOC gene. The activating H3K27ac epigenetic marker was observed in the upstream region of the ALDOC promoter at -630 to -200 base pairs from the transcription start site. Further, ENCODE data predicted the H3K4me3 activation maker to be localized to the 5’UTR of the ALDOC gene (Figure 3.3 A).

Through ChIP analysis, we observed significant increases in H3K27ac presentation at all tested locales in the NME1-overexpressing condition compared to the control ( Figure 3.3

B, upper panel). Predominant enrichment of H3K27ac occurred at the site predicted by

ENCODE. Similarly, the H3K4me3 epigenetic mark was solely detected in NME1 expressing cells at the predicted ENCODE site ( Figure 3.3 B, upper panel). Histone modifications have been shown to correlate with gene expression [134]. Further,

H3K27ac and H3K4me3 modifications are associated with transcriptional activation

[135]. The observed increase of these epigenetic marks on the ALDOC gene in NME1 expressing cells, may demonstrate NME1 promotes an open chromatin state permissive for initiation of transcription.

An additional feature of transcriptional activation is the presence of the RNA synthesis enzyme, RNA polII, poised at gene promoter regions. Indeed, NME1 increased recruitment of RNA polII to the promoter and 5’UTR of the ALDOC gene ( Figure 3.3 B,

55 lower panel). The presence of RNA polII at locales adjacent to the transcription start site reflect the bimodal binding property of RNA polII [136]. The observed increase in

RNA polII recruitment to the ALDOC gene in NME1 expressing cells, further supports

NME1-mediated upregulation of ALDOC occurring by transcriptional mechanisms. We next aimed to determine if NME1 also localizes to the ALDOC promoter. While NME1 was undetected at ALDOC regulatory regions in the vector control condition, NME1 occupancy was observed at all surveyed regions in the NME1 over-expressing cells

(Figure 3.3 B, lower panel). Elevated NME1 expression and subsequent localization to the ALDOC gene, ascribes a potential transcription factor function to NME1.

Interestingly, NME1 may also increase recruitment of the transcription factor Sp1 to the ALDOC gene. Sp1, or a Sp1 like factor, was observed to bind to the G-C rich segment located at the ALDOC proximal promoter region [119]. Further, consensus Sp1

DNA-binding sequences were observed at -648bp and -72bp from the ALDOC transcription start site. As a result, preliminary ChIP experiments were performed using

Sp1 as a positive control for optimizing experimental conditions. We anticipated Sp1 to be present at the -188/+4 region of the ALDOC gene harboring the consensus binding site. Surprisingly, Sp1 was localized to regions on the ALDOC gene not containing the

Sp1 binding sequence in both the NME1 and vector control cells ( Figure 3.3 C). We did observe Sp1 binding at the predicted proximal ALDOC promoter region (-188/+4), with additional enhanced recruitment of Sp1 to this region in the NME1-expressing cells

(Figure 3.3 C). Sp1 recruitment to the consensus binding site on the ALDOC gene, may be due to the increase in Sp1 protein expression observed in the NME1-expressing cells compared to the vector control cells (Figure 3.3 D). Further, NME1 enhanced both

56 phosphorylated (~105 kDa) and unphosphorylated (~95 kDa) Sp1 protein ( Figure 3.3 D)

[137]. Unphosphorylated Sp1 was shown to exhibit increased DNA binding specificity and affinity [138, 139]. The observed Sp1 localization to DNA segments on the ALDOC gene not harboring the consensus binding sequence in both vector and NME1 expressing cells, may be due to the presence of phosphorylated Sp1, which displays more non- discriminant binding. Likewise, the presence of unphosphorylated Sp1 in the NME1 expressing cells may explain the increased recruitment of Sp1 to its specified DNA binding location.

M14 cells expressing elevated levels of NME1 expression promoted chromatin remodeling and enhanced RNApolII and Sp1 recruitment to the ALDOC gene. Further,

NME1 was observed to also localize to the ALDOC gene. Taken together, the ChIP results strongly imply NME1 displays transcription factor activity to upregulate expression of ALDOC through transcriptional mechanism.

A. Predicted Predicted H3K27ac site H3K4me3 site

-800 -630 -200 +1 +85 +295 x Exon1 Exon2

B. rIgG H3K27ac H3K4me3 )

-4 4.5 ** 3.0 * ** ** 1.5 0.0 NME1

Input % (x10

vec ) mIgG NME1 ) RNAPolII -4 1.5 * -3 1.5 ** ** 1.0 ** 1.0 ** ** 0.5 0.5 0.0 0.0 % Input % Input (x10 % Input % Input (x10

C. D. rIgG Sp1 )

-4 6.0 NME1 4.0 * 2.0

0.0 vec % Input (x10

Figure 3.3: NME1 Localizes to the ALDOC Gene and Enhances Indicators of Transcriptional Activation. A. A schematic representation of epigenetic marker presentation from the ENCODE consortium. B. NME1 enhances epigenetic activation markers H3K27ac and H3K4me3 (upper panel). NME1 also localizes to the ALDOC gene at all tested locales (lower panel). NME1 further increases occupancy of RNApolII (lower panel), and the transcription factor Sp1 ( C. ) to the ALDOC gene. D. NME1 enhances both phosphorylated (105kDa) and unphosphorylated (95kDa) protein expression of Sp1. **p≤0.01, and *p≤0.05.

3.3.4 NME1 Attenuation of Single-Cell Motility is Not Mediated by ALDOC

Expression

Our previous microarray results established a correlational relationship between

ALDOC expression and the metastasis suppressor function of NME1. Further, strong experimental evidence supports NME1 transcriptional regulation of ALDOC . Taken together, we hypothesize that NME1 upregulates expression of ALDOC to facilitate the metastasis suppressor activity.

A vital component of metastasis is the ability of malignant cells to migrate from the primary tumor bulk to lymphatic and vasculature systems [140]. Changes in cytoskeletal dynamics, in particular actin assembly and interactions with scaffolding proteins, are crucial for cell movement [140-142]. The aldolase isozymes have been observed to bind to actin filaments in a number of cell types, including melanoma cells

[143]. Aldolases have further been shown to affect actin polymerization, with strong implications in maintenance of cytoskeletal structure and cell motility [84, 85].

Correspondingly, NME1 expression triggered increased actin stress fiber and focal adhesion formation, resulting in increased cell spreading and decreased cell motility

[100]. Together, we predicted that NME1 upregulation of ALDOC would have predominant impacts on cell motility.

To evaluate effects on cell motility, ALDOC expression was silenced in both vector and NME1-expressing M14 cells using two different shRNA constructs targeted to

ALDOC . While ALDOC expression was modestly silenced in the vector control cells, significant reductions in ALDOC protein expression were observed in the NME1- overexpressing condition (Figure 3.4 A). Prior to performing motility experiments, we

59 examined ALDOC knockdown effects on cell proliferation as impacts to cell proliferation could confound effects on cell motility. We previously observed that NME1 expression does not affect cell proliferation [99]. Similarly, we did not observe any significant changes to cell proliferation in the ALDOC knockdown conditions ( Figure

3.4 B).

Two primary modes of migration are employed by metastatic cells: individual

(single cell migration) or collective cell migration [144]. It is proposed that single cells may disseminate from the tumor bulk to lay a “track” for additional transformed cells to follow [145]. Single cell migration assessments can help to capture this putative initiating event of metastasis [145]. Time-lapse microscopy was utilized to measure single cell motility. Cells were plated at a low density to measure unimpeded cell movement, however, cells were not so sparsely distributed so as to permit intercellular crosstalk, a critical component of cell motility [146].

As a candidate effector of the suppressor activity of NME1, we predicted that silencing ALDOC would result in increased cell motility. We did not observe any effects on cell motility in the vector cells in the ALDOC knockdown conditions (Figure 3.4 C).

This result was anticipated as ALDOC is expressed at low levels in the vector cells, and expression was only minimally diminished in the knockdown conditions. As predicted,

NME1 significantly reduced single cell velocity when compared to the vector control cells (NME1-NTC vs. vec-NTC). Decreased cell velocity was also observed in the NME1 over-expressing cells in which ALDOC expression was silenced. The result highlights the robust impact of NME1 on impeding cell motility. Surprisingly, ALDOC-knockdown did not significantly affect cell motility in the NME1 expressing cells. We expected that

60 silencing ALDOC expression in the NME1 cells would significantly increase cell motility. However, it has been previously demonstrated that cell migration is highly dependent on the extracellular environment, including the availability of growth factors

[147]. Furthermore, it was suggested that a cell culture environment deprived of glucose may expose the non-metabolic activities of glycolytic enzymes, such as ALDOC

(communication with Dr. Kavita Bhalla, UMB).

We performed our initial time-lapse experiments in a fetal bovine serum (FBS) depleted environment (1% FBS) to minimize growth factor stimuli, however, additional cell culture conditions were maintained. To fully limit growth factor stimuli which may trigger glycolysis, we performed time-lapse microscopy under nutrient deprived conditions. A comparison of the different cell culture conditions used to perform the time-lapse microscopy experiments is shown in Table 1 in the methods section. Nutrient deprivation did not affect protein expression of NME1 and ALDOC, and ALDOC induction by NME1 was maintained ( Figure 3.4 D, left). Further, ALDOC expression continued to be effectively silenced in the knockdown conditions.

Interestingly, nutrient deprivation reduced the overall mean cell velocity of vector cells compared to vector cells maintained in complete media (0.03μm/min vs.

0.06μm/min) ( Figure 3.4 D, right). Furthermore, differences in cell motility between vector and NME1 over-expressing cells were no longer observed as a result of this slowed motility phenotype adopted by nutrient deprived vector cells. Cell motility in

NME1 expressing cells was not significantly impacted by the nutrient deprived media, providing an explanation for the lack of an observed NME1-effect on cell motility.

Although a decrease in cell motility as a result of glucose deprivation was demonstrated

61 in a previous study, the effects of glucose deprivation were shown to be cell-type specific

[148, 149]. This could explain why the low-NME1 expressing vector cells displayed increased sensitivity to nutrient deprivation compared to the NME1-overexpressing cells, and could suggest that NME1 has protective effects from changes to the microenvironment. While nutrient deprivation effects on the vector control cells was unexpected, we anticipated that nutrient deprivation would reveal ALDOC impacts on cell motility previously unseen in the complete media condition. However, silencing

ALDOC expression in either the vector or the NME1 over-expressing condition did not show significant changes in cell motility ( Figure 3.4 D, right). Although ALDOC was not observed to impact single cell motility in the nutrient-deprived condition, it may be possible that optimal cell culture conditions to expose ALDOC moonlighting functions were still not obtained. For example, any combination of glutamine, sodium pyruvate, or glucose elimination may be key to unveiling ALDOC-mediated effects on cell motility.

Furthermore, the unintended effect of nutrient deprivation which masked the robust effect of NME1 on cell motility, highlights the importance of the cell culture environment for in vitro metastasis assessments. While it may be difficult to recapitulate what occurs in human disease in vitro , observations of ALDOC expression in human melanoma biospecimen may help to overcome these specific challenges.

M14 Complete Media A. B.

M14 Nutrient Deprived Media D.

Figure 3.4: ALDOC Expression Does not Affect Single Cell Motility. A. ALDOC protein expression is effectively silenced using 2 different shRNAs. B. Knockdown of ALDOC expression did not significantly affect M14 cell proliferation. C. Time-lapse microscopy was performed to measure the distance cells move over time (cell velocity). NME1 significantly decreases single cell velocity compared to vector conditions, however, silencing ALDOC expression does not affect single cell velocity. D. NME1 and ALDOC protein expression is not impacted by media stripped of glucose, glutamine, and sodium pyruvate additives (nutrient deprived). A nutrient deprived environment does not unmask moonlighting functions of ALDOC as an inhibitor of single cell motility (right). *p≤0.05.

3.3.5 Decreased ALDOC Expression Correlates With Metastatic and Invasive

Melanomas in Human Biospecimens

Clinical data from the Genome Expression Omnibus (GEO) was previously utilized by our lab to perform unsupervised hierarchical clustering of NME1-regulated genes. By performing this analysis, we aimed to examine correlational relationships between the cohort of NME1-regualted genes with melanoma sample type (primary vs. metastatic) (Figure 2.4 A). The results demonstrated ALDOC expression to be lower in a subset of metastatic samples compared to primary melanomas in the GSE8401 dataset

(Figure 2.4 A) [101]. We employed the GEO database repository once more to identify an additional dataset to determine if ALDOC expression is associated with melanoma progression.

Normalized log2 mRNA expression data were downloaded from the GSE46517 dataset comprised of samples of progressive stages of melanoma: normal skin, nevi, primary melanoma, and metastatic melanoma [122]. ALDOC mRNA expression was significantly lower in metastatic melanoma samples compared to normal skin and early stages of melanoma ( Figure 3.5 A). This result is consistent with our hypothesis that

ALDOC expression may function to inhibit metastasis. In addition to the GEO dataset, we analyzed data collected from The Cancer Genome Atlas-Skin Cutaneous Melanoma

(TCGA-SKCM) consortium [18]. The TCGA-SKCM dataset is comprised of primary and metastatic melanoma samples with corresponding clinical attributes from the samples collected. Normalized log2 mRNA values were downloaded from the c-BioPortal website

[123, 124]. While no differences in ALDOC expression were observed between primary and metastatic melanoma groups ( Figure 3.5 B, left )), ALDOC expression was markedly

64 different in samples with a higher Breslow depth measurement (Figure 3.5 B, right).

Breslow depth is an index of melanoma invasion; an increased Breslow thickness measurement (>4mm) was determined to be a predictor of metastasis [150]. The observed decrease in ALDOC expression correlating with increased melanoma progression, strongly suggests ALDOC may be a functional effector to inhibit metastasis in human disease.

A. B.

Figure 3.5: ALDOC Expression Correlates with Melanoma Stage and Breslow Depth. A. Normalized ALDOC mRNA expression is significantly lower in metastatic melanoma samples compared to normal skin, or early stages of melanoma in the GEO dataset GSE46517. B. No differences in ALDOC mRNA expression is observed in primary and metastatic melanoma biospecimen in the skin cutaneous melanoma dataset from the cancer genome atlas (TCGA- SKCM). Decreased ALDOC mRNA expression is observed in clinical samples with a higher Breslow depth measurement, an index of melanoma invasion (right). *p≤0.05

3.4 Discussion

NME1 is a well-established metastasis suppressor, however, mechanisms of the suppressor activity and additional cellular functions of NME1, remain to be elucidated

[31]. A proposed mechanism of suppression involves NME1 regulation of genes which execute the suppressor function. While studies have demonstrated co-transcriptional functions of NME1 through interactions with transcription factors and viral oncogenic

65 antigens, no study to date has assessed robust direct transcriptional mechanisms for

NME1 regulation of gene expression [68, 69, 151]. We confirmed a strong correlational relationship between NME1 expression and ALDOC transcript levels in several melanoma cell lines. Accordingly, ALDOC served as an excellent prototype for assessing

NME1 transcriptional activity.

We concluded that NME1 is a transcriptional regulator of ALDOC . We observed increased ALDOC mRNA and unspliced pre-mRNA levels in NME1-forced expressing

M14 and WM1158 cells, and NME1 activated the ALDOC promoter in the M14 cell line.

Further evidence of NME1 promoting transcriptional activation of ALDOC , was the increased presentation of H3K27ac and H3K4me3 epigenetic modifications on the

ALDOC gene in NME1 expressing cells. H3K27ac and H3K4me3 markers are associated with enhanced transcriptional activation. Interestingly, it has been shown that NME1 affects expression of several histone modification enzymes including HDACs and

DNMTs [152, 153]. This result may suggest that NME1 may be encouraging histone alterations on the ALDOC gene through mediating expression of HDACs and DNMTs.

However, it is possible NME1 may exert more direct effects on chromatin remodeling. It was shown that binding of NME2 to various genes increased nucleosome repositioning at proximal promoter regions, resulting in transcriptional activation [154]. Through ChIP analysis, we demonstrated that NME1 binds to the ALDOC gene. Further, NME1 expression enhanced RNApolII occupancy at the ALDOC gene. NME1 may perform similar functions observed by NME2, and may be promoting nucleosome shifting at the

ALDOC promoter, allowing for RNApolII to bind to the promoter and activate transcription. NME1 directly binding to the ALDOC gene and facilitating chromatin

66 remodeling would suggest that NME1 exhibits transcription factor-like activity to enhance ALDOC expression.

The formaldehyde cross-linking technique used for ChIP analyses favors detection of direct protein-DNA interactions [155]. However, a co-transcriptional role for

NME1, in which NME1 interacts with transcription factors, must also be considered.

NME1 increased protein expression of the transcription factor Sp1, a result similarly observed in lung cancer cells by Hung, et al. [156]. Increased recruitment of Sp1 to the

ALDOC proximal promoter region harboring a known Sp1 binding site, was also observed in NME1 expressing cells. Unlike Sp1, NME1 does not bind to a consensus

DNA sequence. Both NME1 and Sp1 bind to G-C rich motifs, suggesting NME1 and Sp1 could be acting cooperatively to bind to these DNA segments, a phenomenon observed with other transcription factors [62, 113, 157, 158]. Interestingly, prior studies examining the ALDOC promoter revealed a G-C rich transcriptional activation box at the proximal promoter that could bind to Sp1, or a Sp1-like factor [119]. It remains to be elucidated if

Sp1 and NME1 co-localize to the ALDOC gene in a regulatory complex, or if NME1 binding to the ALDOC gene promotes Sp1 recruitment to its cognate binding site.

ALDOC is a member of the aldolase family of enzymes (ALDOA, ALDOB, and

ALDOC), predominantly functioning in glycolysis. No changes in ALDOA or ALDOB mRNA expression were observed in response to NME1 expression, suggesting ALDOC is a specific target of NME1 transcription. Specific upregulation of ALDOC compared to the other isotypes may be biologically significant. Although the aldolase isotypes are structurally similar, they are catalytically distinct. ALDOA exhibits greater glycolytic capability in fructose 1,6 bisphosphate turnover to glyceraldehyde 3 phosphate (G3P) and

67 dihydroxyacetone phosphate (DHAP) compared to ALDOB and ALDOC [125, 126].

Increased ALDOA expression has been associated with decreased probability of survival in colorectal cancer, pancreatic cancer, and lung squamous carcinoma, with these studies implicating the increased rate of glycolysis as the principal means for poor clinical outcome [159-162]. In contrast, ALDOB and ALDOC isozymes inhibited metastasis in hepatocellular carcinoma and oral squamous cell carcinoma (OSCC) respectively, with the latter study suggesting a decrease in glycolysis as a potential mechanism of suppression [86, 163]. Decreased lactate production was observed in OSCC cells overexpressing ALDOC compared to control cells with low ALDOC expression [86].

Elevated glycolysis favoring a less efficient metabolic pathway that produces lactate as a biproduct, is a hallmark of cancer cells (Warburg Effect) [75]. The observed decrease in lactate production in OSCC cells suggests ALDOC may be reducing glycolytic flux.

NME1 may show preference for regulating ALDOC expression in melanoma cells as a means to regulate glycolysis. We did not measure lactate production in the current study.

However, future experiments to measure lactate and other glycolysis biproducts may be helpful in providing an explanation for NME1 specific induction of ALDOC.

Consistent with the metastasis suppressor phenotype of NME1, we observed significant decreases in single cell motility in M14 cells stably expressing NME1.

Surprisingly, shRNA targeted knockdown of ALDOC in this cell line did not affect cell motility. The result suggests ALDOC expression alone may not be sufficient to execute the suppressor function of NME1. ALDOC was selected from a collection of NME1- regulated genes which strongly correlated with the metastasis suppressor function. It may

68 be likely that ALDOC operates cooperatively with other NME1-regulated genes to inhibit metastatic activity.

In the present study, we confirm NME1 upregulates expression of ALDOC in melanoma cell lines. NME1 directly binds to the ALDOC gene and enhances markers of active transcription. As such, we a ascribe a putative transcription factor activity to

NME1, which may be explored as a mechanism for the metastasis suppressor activity.

Chapter 4: In vivo Validation of the Metastasis Suppressor

Function of NME1 in a Mouse Model of UVR-Induced

Cutaneous Melanoma

4.1 Introduction

The metastasis suppressor activity of NME1 has been demonstrated across several cancer types primarily by in vitro bioassays using human cancer cell lines, or by in vivo metastasis experiments that utilize immunocompromised mice [61, 103, 164-166]. While these methods provide a low-cost, rapid technique for studying metastasis, they lack critical features involved in cancer initiation and progression. Next-generation hallmarks of cancer emphasize the importance of the microenvironment in tumor progression [77].

For example, the immune system has been shown to be an active participant in tumor malignancy as evidenced by the detection of infiltrating immune cells in the tumor bulk

[167].

The hepatocyte growth factor/scatter factor (HGF/SF) transgenic mice are immunocompetent mice and serve as an ideal model for studying melanoma. Under the control of the metallothionein promoter, HGF/SF mice robustly express hepatocyte growth factor/scatter factor, a known driver of cancer progression, in both mesenchymal and epithelial cells [168]. Elevated expression of HGF/SF results in increased melanocyte presentation at the superficial epidermal and dermal layers of the skin in mice

[168]. Mice characteristically express melanocytes at the hair follicles, while humans express melanocytes at the skin surface [169]. As such, few genetically engineered mouse models display the more humanized distribution of melanocytes that HGF/SF mice

70 present [170]. This mimicry of human skin displayed by HGF/SF mice allow for the study of melanoma with real-world applicability.

A known risk-factor for human cutaneous melanoma is exposure to ultraviolet radiation (UVR) [10]. A primary source of UVR comes from the sun which is comprised of UVA (320-400nm) UVB (280-320nm), and UVC (<280nm) wavelengths. While UVC is absorbed by the ozone, UVA and UVB radiation can be absorbed by the skin, resulting in insults to the genome. Both UVA and UVB radiation have implications in melanoma initiation and progression [10]. Due to the shorter wavelength of UVB radiation, it is typically absorbed by the epidermis. UVB radiation has been shown to be the primary cause of DNA damages resulting in cyclobutene pyrimidine dimers (CPD), and 6-4 pyrimidine-pyrimidone photoproducts (6-4PPs) causing C to T and CC to TT transformations of the DNA [10]. Though less established in melanoma genesis, UVA demonstrates more indirect impacts on DNA damage. UVA was shown to increase reactive oxygen species (ROS) in HGF/SF mice, implicating ROS-driving mutations and damages to DNA as contributors to melanoma formation and progression [171].

Together, it is likely that both UVA and UVB exposure contribute to melanoma formation. Exposure of HGF/SF mice to a combination of UVA and UVB irradiation resulted in the development of UV-induced melanomas with histological features similar to human melanomas [171, 172]. The results further demonstrate the applicability of the

HGF/SF transgenic mouse model to human disease.

Providing the aforementioned benefits of HGF/SF mice (referred to as [ HGF +] in the text), the model is ideal for in vivo validation of the metastasis suppressor activity of

NME1. UVR initiated melanomas of [HGF +] mice are poorly metastatic, allowing for the

71 singular assessment of NME1 as a metastasis suppressor in the absence of spontaneous metastasis occurrences [173]. While NME1 has considerable impacts on inhibiting metastasis, the close homolog, NME2, also exhibits suppressor activity [174-177].

Further, endogenous expression of NME1 and NME2 have been shown to form hetero- hexamers in vivo , suggesting the two family members may collectively conduct physiological functions [178]. As such, we previously assessed the effects of deletion of

NME1 and NME2, termed Nme1 and Nme2 respectively in mice, using the HGF/SF mouse model in the C57/BL6 background. Homozygous deletion of both Nme1 and

Nme2 genes results in midgestational death and disruptions in erythropoiesis, but a transgenic mouse model with hemizygous deletion of Nme1 and Nme2 (referred to as

[Nme1 +/-:Nme2 +/-]) is viable with no phenotypic abnormalities [179]. Tandem hemizygous deletion of [ Nme1+/-:Nme2 +/-] in the [ HGF +] mouse model (referred to as

[HGF +: Nme1 +/-:Nme2 +/-]) did not affect primary tumor incidence or primary tumor growth rate when compared to wild-type [HGF +] control mice after exposure to UVR

[180]. However, melanoma metastasis incidence to lymph nodes, lung, and liver was significantly enhanced in the [HGF +: Nme1 +/-:Nme2 +/-] mice. The study was important in establishing concomitant effects of NME1 and NME2 deficiency on melanoma metastasis, however, the individual contributions of either protein to preventing metastasis remained to be elucidated. In the present study, we assess the singular roles of the metastasis suppressor activity of NME1 and NME2. Observations of melanoma initiation, growth, and metastasis were evaluated for both genes, however, the report herein assesses NME1 specific metastasis suppressor activity in the [HGF +] model.

Homozygous-null Nme1 (referred to as [ Nme1 -/-]) mice were generated by the

Arnaud-Dabernat lab at the University of Bordeaux in Bordeaux, France. Nme1 null mice did not display embryonic lethality, and exhibited normal Mendelian genetic distribution

[181]. The impact of Nme1 deficiency was first evaluated in a mouse model which spontaneously develop hepatocellular carcinoma [182]. Crossing [Nme1 -/-] mice to this mouse model resulted in significant lung metastasis occurrence, demonstrating in vivo metastasis suppressor functionality of NME1 in hepatocellular carcinoma. Similarly, we aim to assess for the first time, NME1 suppressor activity of UVR initiated melanomas.

[Nme1 -/-] mice were bred with [ HGF +] mice deficient of the Ink4a gene. Deletion of the

Ink4a gene was introduced to our mouse model as loss of this gene is frequently observed in human melanomas.

The INK4a/ARF gene locus, also known as the CDKN2a gene, encodes tumor suppressor proteins p16 INK4a and p14 ARF in humans [183]. Germline mutations and somatic deletion of CDKN2a have been attributed to familial and spontaneous melanomas [9, 183, 184]. CDKN2a gene deletion occurs in 50% of human melanomas, and is the most common genomic deletion observed [185]. Furthermore, deletion of

Ink4a/Arf in [HGF +] mice of the albino FVB/N background, accelerated melanoma initiation post UVR exposure [186]. Together, the results establish a significant role of

INK4a/ARF in melanoma genesis. Interestingly, several studies demonstrated loss of p16 INK4a was a more prominent driver of melanoma initiation and progression than p14 ARF [187, 188]. As such, we introduced the Ink4a specific deletion (herein referred to as [ p16 -/-]) in to our transgenic mouse model to decrease tumor latency and generate melanomas more closely resembling humans. The present study demonstrates NME1 as a

73 bona fide metastasis suppressor of UVR initiated melanomas in the [HGF +:p16 -/-] transgenic mouse model.

4.2 Methods

*All humane handling of mice for experimental procedures were in accordance with

University of Maryland Baltimore IAUCUC approved protocols: 0612013 and 0515008

Genotyping Transgenic mice used for breeding and generation of experimental mice were confirmed by genotyping methods. [ HGF +] mice have phenotypic features including pigmented tail, ears, and paws, distinct from C57/BL6 mice, which permit a visual approach for identifying mice harboring this genotype. Polymerase-chain reaction

(PCR) based methods were conducted to confirm Nme1 or p16 knockout genotypes.

Ear-punch tissue samples taken from mice at the time of weaning were collected.

Genomic DNA was extracted from ear tissue according to manufacturer instructions using the DNeasy Blood and Tissues Extraction Kit (Qiagen). PCR was performed on an

Eppendorf Master Cycler Nexus Gradient thermocycler (Eppendorf, Hauppauge, NY,

USA) to confirm effective knockout of respective Nme1 and p16 genes. A PCR reaction using the Taq-based Master Mix (Promega) with the primers: ex2S:

AGATCATCAAGCGGTTCGAG, int2AS2 : CAACCTACTTCGAGGGTGGA, and

KONeo2: CGTTGGCTACCCGTGATATT were used to determine the genotypes for

[Nme1 -/-] mice. The 5 Prime HotMasterMix (Eppendorf), also a Taq-based method, with primer sets: 5P3: GGCAAATAGCGCCACCTAT,

3P4: GACTCCATGCTGCTCCAGAT, and LP: GCCGCTGGACCTAATAACTTC were used to confirm the genotype of [ p16 -/-] mice. DNA amplicons were visualized after gel- electrophoresis on a 2-3% agarose gel, stained with ethidium bromide using a UVP

BioDoc-IT Imaging System.

UVA/UVB Irradiation To induce melanoma in mice with the [ HGF +] transgene, postnatal day 4 mice were subject to a single erythematous dose of UVA/UVB irradiation at 9 kJ/m 2. Mice were placed dorsal-side up in a 6-well plastic dish for maximum exposure of the back to irradiation. Plastic dishes containing the mice were placed on a

UVP-XX Series UV bench lamp platform (UVP, Upland, CA, USA) encapsulated by a protective box to limit UV exposure to researchers. A UV bulb (Ushio Light Edge

Technologies, Tokyo, Japan) emitting a spectral output of 61% UVB, 28.3% UVA, and

2.09% UVC was used to administer UV irradiation. Appropriate UV dosage was monitored using a radiometric/photometric light meter (International Light Technologies,

Peabody, MA). Mice experienced minor skin reddening and desquamation, as previously observed, but no effects on mice development or mortality were detected [180].

Melanoma Surveillance and Mouse Necropsy At time of weaning (postnatal day 30),

UV-irradiated mice were first inspected for upraised pigmented melanoma lesions.

Subsequently, mice were surveyed once weekly for melanoma lesions. Melanoma incidence was recorded based on first observation of a melanoma lesion. Tumor volume and growth rates were determined on a weekly basis by caliper measurements as previously described [180, 189]. Experimental endpoints were determined when one of

75 the following events occurred, a single tumor reached 500mm 3 in size, a cumulative tumor burden on a single mouse reached 2000mm 3, or a mouse reached 9.5 months of age. At the specified endpoint, mice were humanely sacrificed by CO 2 asphyxiation and cervical dislocation. Mouse necropsies were performed after depilation to excise and image primary tumors, lymph nodes, lungs, liver, spleen, and any additional afflicted organs. Excised primary tumors and distal organs with melanoma metastases were sectioned and maintained in RNALater for future analyses of whole-genome and RNA- seq transcriptome methods to be conducted by our collaborator, Dr. John Carpten, at the

Translational Genomics Research Institute in Phoenix, Arizona. Primary tumor and metastasis sections were also maintained in 10% Formalin solution for future histopathological confirmation of melanomas by our collaborator Dr. Andrezj Slominski at the University of Tennessee Health Science Center.

Statistical Analysis Significant comparisons between control, wild-type Nme1 mice and

Nme1 knockout mice was determined by Mann-Whitney Rank Sum test using SigmaPlot software (Sysat Software).

4.3 Results

4.3.1 Generating Triple Transgenic Mice for UV Irradiation

C57/BL6 mice with germline homozygous deletion of Nme1 (referred to as

[Nme1 -/-]) were obtained from Dr. Sandrine Arnaud-Dabernat at the University of

Bordeaux in Bordeaux, France. To ensure genetic homogeneity, six backcrosses in to our own C57/BL6 colony were performed. The Ink4a homozygous-null mice, referred to as

[p16 -/-], were provided by Dr. Glenn Merlino at the National Cancer Institute in Bethesda,

Maryland, USA. The acquired [p16 -/-] mice were in the albino C57/BL6 background, and were further backcrossed in to our black C57/BL6 colony for six generations. No backcrosses for mice expressing HGF were necessary, as these mice were already fully- integrated in to our C57/BL6 colony from previously performed experiments [54, 180].

Once backcrosses were complete, homozygous-null [ Nme1 -/-] and [ p16 -/-] mice were crossed with one another to generate mice heterozygous for both Nme1 and p16 genes, [ Nme1 +/-:p16 +/-]. Heterozygous pairs were further crossed with one another and mice homozygous-null for both genes [ Nme1 -/-: p16 -/-], as determined by PCR genotyping, were preferentially selected. Congruently, [ p16 -/-] female mice were crossed with male mice expressing the HGF transgene to generate mice heterozygous for both

HGF and p16 [HGF +/-:p16 +/-]. Breeding pairs with mice expressing HGF were limited to the use of males due to apparent infertility of HGF expressing female mice [190].

Therefore, males heterozygous for HGF expression were crossed with wildtype C57/BL6 female mice, resulting in 50% HGF penetrance in mice of this breeding pair. Single allelic expression of HGF is sufficient for distributing melanocytes from mouse hair follicles to epidermal and dermal layers of mouse skin. This dispersal of melanocyte presentation in HGF expressing C57/BL6 mice results in mice exhibiting hyperpigmented tails, ears, and paws, distinguishing them from wildtype C57/BL6 mice.

For simplicity in this text, HGF heterozygous mice will be referred to as HGF positive

([ HGF +]), symbolizing expression of the HGF gene.

Crossing of [ HGF +:p16 +/-] mice with [ p16 -/-] female mice was performed to preferentially select for male [ HGF +:p16 -/-] mice. A breeding pair with [HGF +:p16 -/]

77 male mice with [ Nme1 -/-:p16 -/-] female mice generated mice homozygous-null for p16 , and heterozygous for Nme1 . Of these mice, [ HGF +] male mice were preferentially selected. A final crossing of [ HGF +:p16 -/-:Nme1 +/-] male mice with [ Nme1 -/-:p16 -/-] female mice, generated triple transgenic [ HGF +:p16 -/-:Nme1 -/-] males that were used as breeders for the ultimate generation of the experimental mice exposed to UVR.

Numerous breeding pairs with [ HGF +:p16 -/-:Nme1 -/-] male mice with [Nme1 -/-:p16 -/-] females, resulted in a total of 22 triple transgenic female, and 34 triple transgenic male mice for the UVR experiments. Additionally, [ HGF +:p16 -/-] double transgenic mice, wild-type for the Nme1 gene, were generated as controls for the UVR experiments. A detailed procedure of the multiple breeding pairs is diagramed in Figure 4.1.

Both male and female mice were utilized for the UVR experimental procedure given the arduous breeding process to obtain the necessary triple transgenic [HGF +:p16 -/-

:Nme1 -/-] mice. The use of both male and female mice allowed for a larger experimental population of mice to increase statistical power. Further, the use of both male and female mice may provide insight in to gender differences in melanoma presentation, growth rate, and metastasis

Figure 4.1: Breeding Diagram for Generation of a Triple Transgenic Mouse Model . A. Nme1 and p16 knockout mice ( Nme1 -/-, and p16 -/- respectively) required six backcrosses in to our C57BL/6 colony to ensure genetic homogeneity. Multiple breeding pairs were necessary to generate a triple transgenic mouse expressing the HGF transgene, and homozygous-null for Nme1 and p16 expression referred to as [ HGF +: p16 -/-: Nme1 -/-] used for the UVR initiated melanoma study.

4.3.2 NME1 Deficiency does not Affect Primary Melanoma Onset, Incidence, or

Growth Rate

At postnatal day 4, mice from breeding pairs generating the experimental triple transgenic mice were subject to a single 9kJ/m 2 dose of UVA/UVB irradiation. Day 4 pups were placed dorsal side up in a plastic 6-well dish to maximize UVR exposure to the backs of mice ( Figure 4.2 A). At time of weaning (postnatal day 30), [ HGF +] mice were preferentially selected. Mice expressing HGF were distinguished from their non-HGF counterparts by the presentation of pigmented ears, tails and paws ( Figure 4.2 B).

Furthermore, at time of weaning, [ HGF +] mice began the surveillance regimen for primary melanoma presentation ( Figure 4.2 C). Adult mice were surveyed once weekly for upraised pigmented lesions to determine melanoma onset time. The experimental endpoint was determined when a single tumor on a mouse reached 500mm 3, total tumor burden on a single mouse reached 2000mm 3, or a mouse reached 300 days of age ( Figure

4.2 C). The current study is ongoing, with some [ HGF +:p16 -/-:Nme1 -/-] mice not reaching the endpoint at the time of writing the dissertation. However, several mice which have been analyzed, contribute significant statistical power to establish observable trends in the data.

A. B.

Figure 4.2: Study Design for UVR of Transgenic Mice. A. Postnatal day 4 pups are exposed to a single dose of UVA/UVB irradiation. B. HGF + mice are identified by their pigmented tails, ears, and paws. C. A schematic representation of the experimental timeline is displayed.

As mentioned, UVR was primarily directed to the backs of mice. Both the control

[HGF +:p16 -/-] mice and the Nme1 deficient triple mice had majority of primary tumor presentation localized to the back ( Figure 4.3 A,B). However, melanomas were also detected on the sides, head, and mouse extremities (ears, legs, tail) as these sites were also exposed to UVR. There were no differences in the location of tumor presentation between control [ HGF +:p16 -/-] and the [ HGF +:p16 -/-:Nme1 -/-] mice ( Figure 4.3 B).

Additionally, multiple tumors were detected on both [ HGF +:p16 -/-] and

[HGF +:p16 -/-:Nme1 -/-] mice. Control [ HGF +:p16 -/-] mice had an average of 3.2 melanomas per mouse (range: 1-8), while [HGF +:p16 -/-:Nme1 -/-] mice had an average of

2.8 melanomas per mouse (range: 1-6). However, no significant differences in the final cumulative tumor volumes were detected at the time of necropsy. Together, the results indicate that NME1 deficiency does not affect primary tumor location or burden.

Figure 4.3: Melanoma Locations are not Affected by NME1 Deficiency . A. UVR exposure to [HGF +] mice results in melanoma presentation at multiple locations. B. Majority of melanomas are localized to the backs of mice. Both control mice and NME1 deficient mice show equal presentation of melanomas at all locations.

NME1 deficiency did not affect melanoma onset time or incidence. Melanoma onset was determined at the first detection of a primary melanoma lesion. Melanoma onset for [ HGF +:p16 -/-] mice ranged from 57 days to 215 days post UVR, with a median onset time of 143 days post UVR. The median onset time for [HGF +:p16 -/-:Nme1 -/-] mice was 125 days post UVR, with a range of 40 to 242 days. No statistical differences in melanoma onset were determined ( Figure 4.4 A ). Similarly, we did not observe any noticeable differences in melanoma occurrence or incidence ( Figure 4.4 B) . A total of 24 control mice were evaluated, of which, 21 mice developed melanoma (84% occurrence).

The occurrence rate for [ HGF +:p16 -/-:Nme1 -/-] mice was 80%, with 12 mice developing melanomas from a total of 15 mice analyzed. Collectively, the results are consistent with our previous report, in which concomitant hemizygous deletion of Nme1 and Nme2 did not impact melanoma onset or incidence [180].

Metastasis suppressor genes inhibit metastasis, but do not affect primary tumor growth [30]. We confirmed that hemizygous deletion of Nme1 and Nme2 did not impact primary melanoma growth [180]. To evaluate the singular effects of Nme1 deficiency on tumor growth, caliper measurements on detected melanomas were taken once weekly until experimental endpoints were met. Growth rate curves for each mouse were generated from the cumulative tumor volume calculated at the specified timepoint

(Figure 4.4 C ). To determine the tumor growth rate, the specific growth rate formula was applied using the initial tumor volume at time of onset, and the last measured tumor volume [191]. The average specific growth rate for both [ HGF +:p16 -/-] control mice

(n=20) and [ HGF +:p16 -/-:Nme1 -/-] mice (n=11) was 0.04% per day. As expected, NME1 loss did not affect primary melanoma growth rate. The absence of significant differences

in UVR initiated primary melanoma development or growth rate between control

[HGF +:p16 -/-] and [ HGF +:p16 -/-:Nme1 -/-] mice, allow for determinations on metastasis more specifically attributable to NME1 deficiency.

A. B.

Figure 4.4: NME1 Deficiency does not Affect Primary Melanoma Tumor Development. A,B. Both [ HGF +:p16 -/-] control mice and [ HGF +:p16 -/-:Nme1 -/- ] mice display no significant differences in melanoma onset time (p=0.6 Mann Whitney Rank test) or tumor incidence (p=0.8 Survival Log-Rank test). C. Cumulative tumor growth curves for each of the mice analyzed were generated until endpoints were met. Each line represents the cumulative tumor growth for a single mouse. No differences in the specific growth rate for primary melanomas was detected (p=0.9 Mann Whitney Rank test).

4.3.3 NME1 Deficiency Enhances Lung and Lymph Node Metastases of UVR

Initiated Melanoma

At the experimental endpoint, necropsies on mice were performed to measure the effects of NME1 deficiency on metastasis. Lymph nodes and visceral organs were examined for evidence of metastasis including enlargement and pigmented lesions.

NME1 deficiency did indeed increase metastases to the lung and lymph nodes. Compared to the control [ HGF +:p16 -/-] mice, a greater percentage of [ HGF +:p16 -/-:Nme1 -/-] mice experienced lung (58% vs. 29%) and lymph node (75% vs. 52%) metastases (Figure 4.5

A, 4.6 A).

Lung metastases were evaluated based on the size and number of lesions a tumor bearing mice presented. To normalize comparisons of lung metastases between the control and NME1 deficient mice, a weighted composite score was implemented.

Metastatic lesions were quantified as well as stratified by the diameter of the lesion:

≤ 0.5mm, 0.5-1mm, 1-5mm, and ≥ 5mm. A metastatic lesion with a larger diameter was weighted more than smaller lesions to generate a composite score. While most lesions were ≤ 0.5mm for both [ HGF +:p16 -/-] and [ HGF +:p16 -/-:Nme1 -/-], the NME1 deficient mice had a greater number of lesions, and larger lesions compared to control mice

(Figure 4.5 B). Representative lung images in Figure 4.5 B were obtained from mice with the 2 highest composite scores for the control and NME1 deficient conditions. Lung composite scores observed in the images for [ HGF +:p16 -/-] were 18 (left image) and 3

(right image), while lung composite scores for the [ HGF +:p16 -/-:Nme1 -/-] mice represented were 180 (left image) and 87 (right image). Composite lung metastasis scores for [ HGF +:p16 -/-] mice ranged from 0-18, while composite score range for

[HGF +:p16 -/-:Nme1 -/-] mice was 0-180. Composite scores were statistically different between [ HGF +:p16 -/-] and [ HGF +:p16 -/-:Nme1 -/-] mice ( Figure 4.5 C).

Figure 4.5: NME1 Deficiency Enhances Lung Metastases. A. 58% of [ HGF +:p16 -/-:Nme1 -/-] tumor bearing mice experienced metastasis to the lung compared to 29% of [ HGF +:p16 -/-] mice B. [HGF +:p16 -/-: Nme1 -/-] mice had more metastatic lesions and larger lesions compared to [HGF +:p16 -/-]mice. C. Composite scores for lung metastases were determined based on lesion number and size. Each bar represents a single mouse. [ HGF +:p16 -/-:Nme1 -/-] mice experienced increased lung metastases compared to control mice (p=0.03, Mann-Whitney Rank test).

Control [ HGF +:p16 -/-] and [ HGF +:p16 -/-:Nme1 -/-] experimental mice also displayed metastasis to lymph nodes. Axillary, Brachial, Cervical, and Inguinal lymph nodes were the most afflicted, exhibiting enlargement as an indicator of metastasis. Given the symmetry of lymph node presentation, both left and right lymph nodes were examined singularly for evidence of metastasis. As observed in Figure 4.6 B , representative images from a single [ HGF +:p16 -/-:Nme1 -/-] mouse exhibited differences in lymph node metastasis between left and right lymph nodes, with the left lymph nodes significantly enlarged compared to normal lymph node size (right). Composite scoring for lymph node metastases for each tumor bearing mouse was determined by presence of metastasis enlargement for each lymph node, as well as the volume of the enlarged lymph node. Lymph nodes with increased volumes, indicative of a more aggressive metastasis, were weighted more in composite scoring. A final sum of composite scores from each of the left and right lymph nodes examined was used to compare differences in lymph node metastasis presentation between [ HGF +:p16 -/-] and [ HGF +:p16 -/-:Nme1 -/-] mice. NME1 deficient mice exhibited significantly more lymph node metastases (composite score range 0-66) compared to control mice (composite score range 0-33) (Figure 4.6 C ).

Figure 4.6: Loss of NME1 Enhances Lymph Node Metastasis . A. A greater percentage of NME1 deficient mice experienced lymph node metastases. B. A representative image of axially, brachial, and inguinal lymph nodes from a single [ HGF +:p16 -/-:Nme1 -/-] mouse is displayed. Enhanced metastasis (enlargement) of left lymph nodes are observed compared to the right lymph nodes. C. [HGF +:p16 -/-:Nme1 -/- ] mice experienced more lymph node metastases presentation compared to control mice. Each bar represents a single mouse (p=0.03, Mann-Whitney Rank test).

The increased presentation of both lung and lymph node metastases in NME1 deficient mice compared to control mice, demonstrate for the first time, NME1 suppressor activity of UV initiated primary melanomas.

4.4 Discussion

We previously reported that tandem hemizygous deletion of Nme1 and Nme2 genes enhanced metastasis of UVR initiated melanomas in the [ HGF +] mouse model

[180]. However, the singular contribution of NME1 to inhibiting metastasis remained to be elucidated. The present study demonstrated for the first time, the impact of NME1 deficiency on UVR initiated melanoma metastasis in the [ HGF +:p16 -/-] transgenic mouse model.

As mentioned, there are several benefits to using the [ HGF +] model, including

UV initiated melanomas with histological resemblance to humans [171, 180].

Interestingly, [ HGF +] mice rarely harbor mutations in the BRAF gene [192, 193]. The absence of BRAF mutations is likely due to the compensatory mechanism of constitutive activation of the MAPK pathway occurring from HGF binding and stimulation of the cell-surface receptor tyrosine kinase, c-Met [169, 192, 194]. As such, the metastasis inhibiting effects of NME1 observed in the [ HGF +:p16 -/-] model may not be attributable to melanomas with BRAF mutations. Although BRAF mutations commonly occur in melanomas, signaling through the HGF/c-MET pathway is also active in several melanomas with and without BRAF mutations [22]. In fact, HGF/c-MET signaling is thought to be one such reason for the innate and acquired resistance to BRAF targeted therapies [22-25]. Taken together, the study of melanomas with focus on particular

89 genetic profiles may aide in improved therapeutic options. As such, our study concentrates on melanomas initiated by UV irradiation with active HGF/c-MET signaling, and also accounts for additional risk factors associated with melanoma development.

In the current study, neonatal mice were exposed to a single erythemal UV dosage of 9kJ/m 2. The dosage provided is equivalent to UV exposure in the summer months in midlatitude regions of the Northern Hemisphere [195-197]. Neonatal mice were specifically exposed to irradiation to experimentally mimic a significant risk factor of melanoma involving the age at which sun damage occurs [196, 198-200]. Sunburns occurring in childhood are considered at greater risk of melanoma development due to the predicted increased susceptibility of young melanocytes to the carcinogenic effects of

UV radiation [199, 200]. Neonatal mice have increased melanocyte progenitor cells than adult mice, also making them more susceptible to DNA damage caused by UV irradiation

[201]. Although neonatal mice skin and human childhood skin are not completely equatable, exposing neonatal mice to UV radiation simulates the melanoma risk of childhood sunburn [196].

UV initiated melanomas in [ HGF +:p16 -/-] and [ HGF +:p16 -/-:Nme1 -/-] mice developed at different anatomical locations, a phenomenon similarly observed in humans.

Although NME1 deficiency did not affect where melanomas developed, it is possible that gender may yield differences in melanoma localization. Our previous UVR experiment using [ HGF +:Nme1 +/-:Nme2 +/-] mice was performed on male mice only. In the current study, UV irradiation was performed on both male and female mice. Gender differences in melanoma development and metastasis could not be accurately determined at the time

90 of this report as the study is ongoing. However, epidemiological studies have demonstrated significant gender differences in the anatomical location of melanoma and in melanoma survival rate [202-205]. Females are more likely to present melanomas on the arms and legs, whereas males are more likely to develop melanoma on the trunk, back, neck, and head [203, 204]. Interestingly, females have a decreased risk of metastasis, and thus an overall better probability of survival than males [202, 205].

Although anatomical location of melanoma affects overall survival, another proposed explanation for the gender differences in survival rates may be due to differences in estrogen levels between males and females [202, 204, 205]. Both NME1 and NME2 have been shown to interact with estrogen receptors, suggesting the proteins may possibly contribute to the gender differences of melanoma prognosis [66, 206].

NME1 has demonstrated ability to bind to ERα, enhancing ERα and Estrogen

Response Element (ERE) complex formation in breast cancer cells [66]. The study also showed decreased expression of NME1 resulted in increased expression of pro-migratory proteins [66]. The results suggest that NME1 interactions with ERα may contribute to the metastasis suppressor effects of NME1. Although ERα has been shown to affect skin thickness and collagen content, ERβ is the predominant receptor expressed in melanoma

[207, 208]. Furthermore, elevated ERβ expression has been correlated with decreased melanoma progression [207]. Interestingly, through yeast-two hybrid screening, NME2 co-localized with ERβ in breast cancer cells. NME2 over-expression, along with estrogen treatment, was shown to synergistically inhibit cell migration [206]. Although NME1 deficiency may not disclose enhanced metastasis incidence in females, our study may reveal gender differences in metastasis incidence in mice deficient in NME2 expression.

We have demonstrated robust metastasis suppressor activity of NME1 in the

[HGF +:p16 -/-] transgenic mouse model. NME1 deficiency significantly increased lung and lymph node metastases from UVR initiated melanomas. To enhance our study, we will perform RNA-seq on primary melanomas and matched metastatic lesions, through our collaborator, Dr. John Carpten (Translational Genomics Research Institute, Phoenix,

AZ). Our lab has previously demonstrated that NME1 functions in nucleotide excision repair (NER), specifically mediating removal of UV initiated 6-4 photoproducts [54]. We hypothesize that NME1 deficiency causes increased mutational burden, thereby resulting in the enhanced metastasis observed. Performing whole-genome sequencing on primary melanomas and matched metastatic lesions, we aim to identify metastasis driving mutations as a result of NME1 deficiency. Together, our study substantiates NME1 as a bona fide metastasis suppressor of UVR induced melanomas. Further, we aim to elucidate a potential mechanism of NME1 suppressor activity occurring through protective effects from UV initiated genomic insults. The detection of metastasis driving mutations occurring as a result of NME1 deficiency will aide in the discovery of diagnostic and prognostic factors of melanoma, and help facilitate more targeted therapies to prevent metastasis initiation and development.

Chapter 5: Summary and Discussion

5.1 NME1 Activates Transcription of the ALDOC Gene

To date, much of the literature demonstrating NME involvement in direct gene regulation has revolved around NME2, a close homolog of NME1. However, an emerging field of study is beginning to focus on NME1 as a key component in gene regulation. NME1 regulation on global gene expression is evidenced across multiple cancer types [35, 58, 59, 61]. By in vitro analyses, NME1 attached to single stranded segments of DNA belonging to the PDGF-A and CMYC promoters [62, 114]. NME binding to DNA regions on the promoters of the PDGF-A, CMYC , TP53 , and other genes was demonstrated in vivo [63, 64]. Together, the results are highly suggestive of a role for

NME1 in transcriptional gene regulation. However, the latter study did not differentiate between NME1 and NME2 proteins, and neither study examines a mechanism of functional transcriptional regulation of gene expression by NME1.

Several studies have suggested NME1 may be a co-regulator of gene expression, interacting with known transcription factors to affect gene transcription. However, the

DNA binding capacity of NME1 suggest the molecule may be playing a more direct role in transcription. We determine for the first time, a comprehensive examination of NME1 as a transcriptional regulator of the aldolase C ( ALDOC ) gene. Applying strict filtering criteria from multiple transcriptome analyses, ALDOC was selected as an ideal gene to study NME1 transcription activity.

We observed NME1 induction of ALDOC mRNA and protein in the WM793,

WM1158, and M14 melanoma cell lines. Performing an actinomycin D (actD) time course experiment, we demonstrated that NME1 does not stabilize ALDOC mRNA,

93 establishing the strong correlation of elevated ALDOC mRNA in response to NME1 expression is likely due to transcription. We confirmed a putative involvement of NME1 in ALDOC transcript regulation by the observed increase in unspliced, pre-mRNA of

ALDOC , and activation of the ALDOC promoter by NME1. To elucidate direct contribution of NME1 to regulation of the ALDOC gene, chromatin immunoprecipitation

(ChIP) was performed. We observed NME1 localization to G-C rich segments at upstream regions and proximal promoter regions of the ALDOC gene, highly suggestive that NME1 may exhibit transcription factor functionality. NME1 transcription factor functionality was further highlighted by the increased recruitment of the transcriptional initiator, RNA polymerase II (RNApolII) to the ALDOC gene in NME1 over-expressing cells. NME1 may also be functional in chromatin remodeling, as NME1 expression enhanced the presentation of epigenetic activation markers H3K4me3 and H3K27ac.

Together, these results provide strong evidence for NME1 as a direct transcriptional regulator of genes, specifically regulating and inducing expression of ALDOC . However, absent of a nuclear localization signal or consensus DNA binding sequence, the question remains as to how NME1 enters the nucleus and binds to DNA segments on the ALDOC gene. We discuss these open ended questions below.

5.1.1 Potential Mechanisms of NME1 Nuclear Translocation

Initially believed to reside only in the cytoplasm, our lab and others have demonstrated NME1 is in fact present in the nucleus [54, 209-216]. Despite strong evidence of NME1 presence in the nucleus and various nuclear functions performed by

NME1, translocation of NME1 to the nucleus remains elusive. The lack of a nuclear

94 localization signal (NLS) would suggest NME1 is transported to the nucleus via nuclear shuttle proteins.

It is widely reported that NME1 translocates to the nucleus upon DNA damage

[54, 212-215]. Through immunoprecipitation assays, NME1 was shown to interact with the double-strand break repair complex of proteins Mre11, Rad50, and Nbs1 (MRN) in the cytoplasm after X-ray induced DNA damage [213]. The MRN complex is rapidly recruited from the cytoplasm to the nucleus after DNA damage. NME1 was also present in the nucleus after X-ray irradiation. The results suggest a mechanism of NME1 nuclear translocation occurring by cytoplasmic interactions with the MRN complex [213].

Supporting this mechanism, is the observed interaction of NME1 with another DNA repair complex, the SET complex [214, 215]. Along with NME1, the SET complex contains the Ape1 protein, necessary for base excision repair. In the absence of DNA damage, the SET complex associates with the endoplasmic reticulum [215]. Granzyme-A mediated DNA damage resulted in localization of the SET complex, including NME1, to the nucleus [214, 215]. Since DNA damage frequently occurs in melanoma, NME1 may be recruited to the nucleus through one of these repair complexes. In addition to interactions with DNA repair complexes, NME1 translocation to the nucleus was observed through interactions with oncogenic viruses [217]. The most well studied of these interactions is NME1 binding to the nuclear antigen of the Epstein-Barr virus,

EBNA3C [151].

NME1 physically interacted with EBNA3C, resulting in localization of NME1 from the cytoplasm to the nucleus [68, 218]. Furthermore, nuclear localization of the

EBNA3C/NME1 complex resulted in activation of several genes [68, 69, 151, 217, 218].

Epstein-Barr virus is linked to several cancers, including Burkitt’s lymphoma, but is not observed in melanoma [219]. Viral-mediated translocation of NME1 to the nucleus in melanoma cells may require examination of oncogenic viruses linked to melanoma.

Human endogenous retrovirus K (HERV-K) is expressed in human melanoma clinical samples and cell lines [220-222]. Furthermore, HERV-K proteins were absent in melanocytes, suggesting that activation of HERV-K genes and retroviral assembly occurs exclusively in melanoma [220]. It is possible NME1 may interact with HERV-K proteins, providing a mechanism with which NME1 enters the nucleus in melanoma. Examining the melanoma cell lines utilized in the current study (WM793, WM1158, and M14) for the presence of HERV-K mRNA and proteins through qRT-PCR and immunoblot analyses may provide some insight as to how NME1 enters the nucleus to alter gene expression.

Another means by which NME1 may be shuttled in to the nucleus is through interactions with transcription factors. We have shown that NME1 increases protein expression of the transcription factor Sp1. Interestingly, NME1 interacted with Sp1 in the nucleus in a complex with EBNA3C and transcription factor GATA-1 [67]. Our current study may be enhanced by showing that NME1 not only increases expression of Sp1, but also interacts with Sp1 in the cytoplasm. Co-immunoprecipitation and yeast-two hybrid analyses may help to determine if NME1 and Sp1 interact. Further, co- immunofluorescence may determine if NME1 and Sp1 interactions occur in the cytoplasm and the nucleus, suggesting NME1 may localize to the nucleus through interactions with Sp1.

5.1.2 Potential Mechanism of NME1 DNA Binding Activity

X-ray crystallography determining the 3-dimensional structure of NME1, did not reveal any DNA binding domains or pockets, generating skepticism as to the DNA binding ability of NME1 [223, 224]. However, most structural analysis has been performed on the hexameric NME1 conformation. Although mostly hexameric, NME1 has been shown to exist in both a hexamer and dimer conformation [225]. It was previously demonstrated that the oligomeric structure of the NME proteins dictates the

DNA binding capacity, with dimers demonstrating greater affinity for DNA binding

[226]. NME1 dimers were shown to be present in the cytoplasm, however, the exact conformation of NME1 in the nucleus has yet to be determined [227].

Through in vitro experiments, our lab observed NME1 binding indiscriminately to single stranded G-C rich segments on the PDGF-A promoter [55, 62, 228]. The study suggested that although NME1 does not bind to specific sequences of DNA, NME1 may activate transcription by binding to single stranded DNA segments at promoter regions.

Transcription factor functionality of proteins exclusively binding to single stranded DNA have previously been described [229, 230]. In the current study, we have demonstrated

NME1 localization to the ALDOC promoter in vivo , particularly at regions with high G-C content [112]. Our lab has also shown through ChIP analysis, NME1 presence at G-C rich segments on the integrin β3 ( ITGβ3 ) promoter [231]. Additional structural analysis of these particular segments of DNA may determine if the segments are prone to forming single stranded DNA, as is typically observed at transcription “bubbles” [232].

Furthermore, single-stranded DNA associates with histones [233]. We have shown that

NME1 enhances modifications of histones both on the ALDOC and the ITGβ3 gene [112,

231]. Additional analysis to determine if these modifications expose single-stranded

DNA segments which NME1 may also be binding to could provide meaningful insight in to NME1 DNA binding mechanisms.

5.2 NME1 is a Validated Metastasis Suppressor of UV-induced Cutaneous

Melanoma

NME1 is an eminent metastasis suppressor gene. However, the suppressor activity has almost exclusively been demonstrated in cancer cell lines. To date, no lab has demonstrated substantial metastasis inhibiting effects for NME1 in a melanoma model mimicking human disease. The hepatocyte growth factor/scatter factor, [ HGF +] mice, express melanocytes at epidermal and dermal junctions, similar to humans, and are thus susceptible to melanomas initiated by UV irradiation. As UV irradiation is a predominant risk factor for melanoma genesis, [ HGF +] mice serve as an ideal model to study melanoma with real-world applicability. To this effect, our lab introduced tandem hemizygous deletion of NME1 and NME2 in to [ HGF +] mice, followed by initiating melanoma by exposing the transgenic mice to UV irradiation. Mice with hemizygous deletion of NME1 and NME2 showed enhanced metastasis to draining lymph nodes and distal orangs [180]. The study was important in establishing robust metastasis suppressor activity of both NME1 and NME2, but the individual contributions to suppressor activity had yet to be determined.

For the first time, we establish unequivocal in vivo metastasis suppressor activity of NME1 of UV-initiated melanomas in [ HGF +:p16 -/-] mice. The particular [ HGF +] model in which mice are deficient in p16 expression, was used for the present study as

98 this model demonstrated increased efficiency in developing melanoma initiated by UV irradiation [186]. Mice deficient in NME1 expression did not demonstrate significant differences in primary melanoma onset, incidence, or growth rate. However, significant differences in metastasis presentation to sentinel lymph nodes and the lung were observed. Mice lacking the NME1 transgene had an increased number of metastatic lesions in the lung, as well as larger lesions in the lymph nodes and the lung. Together, the results validate NME1 as a metastasis suppressor of UV-initiated melanoma. We further aim to elucidate a mechanism of metastasis attenuation by NME1 expression in our transgenic mouse model.

5.2.1 Potential Mechanisms of NME1-meidated Inhibition of UVR-initiated

Melanoma Metastasis

Exposure to UV irradiation (UVR) results in significant insults to the genome. In particular, UV exposure causes the dimerization of adjacent DNA pyrimidine bases causing cyclobutane pyrimidine dimers (CPDs) and 6-4 -pyrimidone photoproducts (6-4

PPs) [13]. These bulky DNA lesions can produce activating mutations in oncogenes encouraging melanoma development. Nucleotide excision repair (NER) is a predominant repair pathway for UV induced DNA lesions [14]. Accumulation of CPDs and 6-4PPs as a result of UVR can compromise the efficacy of NER [14, 15]. We have demonstrated

NME1 may contribute to NER, particularly by removing 6-4PPs [54]. We therefore predict that NME1 may function to inhibit metastasis by preventing metastasis driving mutations.

With the aid of our collaborator (Dr. John Carpten), we aim the test the hypothesis that NME1 guards the genome of metastasis driving mutations, by performing RNA-seq on primary melanomas and matched metastatic lesions from our transgenic mouse model.

Whole-exome sequencing and RNA-seq analyses have been performed in human clinical samples and melanoma cell lines of matched primary tumor and metastatic lesions [122,

234]. While informative in identification of melanoma copy number alterations and mutations differentially expressed between primary and metastatic melanomas, the studies were confounded by genome heterogeneity from biopsies taken within the tumor bulk. Examining genome profiles from our melanoma model offers several advantages.

The transgenic mice are genetically homogenous, eliminated confounding variables of whole genome heterogeneity. Furthermore, we will evaluate melanoma content of excised primary tumors and metastatic lesions, ensuring purity of the samples collected, and eliminating the genome profiles from infiltrating stromal cells. Differences in genetic profiles in human melanomas have also been attributed to anatomical location of melanoma presentation [185]. Our study can mitigate these specific genomic differences by preferentially selecting melanoma lesions that occur on the backs of the transgenic mice. Eliminating these outstanding variables, we anticipate to observe recurrent de novo mutation(s) occurring in metastatic lesions that are not present in the primary tumor.

Further, these mutation(s) should be enhanced in mice deficient of NME1 expression. In this way, we may establish a mechanism with which NME1 is inhibiting metastasis, occurring by guarding the genome from metastasis driving mutations.

Another posited mechanism of metastasis suppressor activity of NME1 may occur through inhibiting infiltrating immunosuppressive cells to the tumor bulk [235]. The

100

[HGF +] mouse model with a fully intact immune system, offers an additional advantage for evaluating NME1 deficiency on the tumor microenvironment. Interestingly, NME1 was observed to bind to the macrophage migration inhibitory factor (MIF), an inflammatory cytokine associated with increased angiogenesis and cellular migration [37,

236]. NME1 binding to MIF inhibited MIF suppression of p53-mediated cell cycle arrest

[37]. It is also possible that NME1 interaction with MIF may sequester MIF to the intracellular compartment, preventing the extracellular release. Similarly, NME1 may be inhibiting the release of additional cytokines and pro-angiogenic factors from the primary tumor and the microenvironment. We can isolate cells from excised primary tumors from both control and NME1 deficient mice. Staining for markers of infiltrating immune cells followed by flow cytometry may help to quantify the presence of immunosuppressive cells. Furthermore, qRT-PCR analysis to observe expression of MIF, VEGF, CXCL-2, and other cytokines associated with tumor migration and angiogenesis can be performed.

We hypothesize that NME1 deficiency will increase infiltrating immunosuppressive cells to the tumor bulk, and enhance expression of tumorigenic cytokines.

While melanoma initiating mutations such as BRAF and NRAS are well established, metastasis initiating mutations are less studied. We predict NME1 may prevent metastasis driving mutations. Further, we can identify these novel genes susceptible to mutational events, which may serve as useful biomarkers for melanoma progression.

101

5.3 Conclusion

NME1 is a multifaceted gene, predominantly recognized for the metastasis suppressor activity. Despite extensive research, NME1 cellular functions, and mechanisms of metastasis suppression are continually being unveiled. We add to an emerging body of research examining putative transcription factor activity for NME1.

We are the first lab to systematically demonstrate transcription factor functionality of

NME1 resulting in transcriptional activation of the ALDOC gene. Further, we establish

NME1 as a bona fide melanoma metastasis suppressor. Our future work to examine metastasis driving mutations as a result of NME1 deficiency, may help to identify patients more susceptible to the mutations, aiding in the development of targeted therapies, and enhancing an understanding of melanoma metastasis.

102

References

1. Maio, M., Melanoma as a model tumour for immuno-oncology. Ann Oncol, 2012. 23 Suppl 8 : p. viii10-4.

2. Paluncic, J., et al., Roads to melanoma: Key pathways and emerging players in melanoma progression and oncogenic signaling. Biochim Biophys Acta, 2016. 1863 (4): p. 770-84.

3. Siegel, R.L., K.D. Miller, and A. Jemal, Cancer Statistics, 2019. CA: A Cancer Journal for Clinicians, 2019. 69 : p. 7-34.

4. Dupin, E. and L. Sommer, Neural crest progenitors and stem cells: from early development to adulthood. Dev Biol, 2012. 366 (1): p. 83-95.

5. Damsky, W.E. and M. Bosenberg, Melanocytic nevi and melanoma: unraveling a complex relationship. Oncogene, 2017. 36 (42): p. 5771-5792.

6. Bertolotto, C., Melanoma: from melanocyte to genetic alterations and clinical options. Scientifica (Cairo), 2013. 2013 : p. 635203.

7. Tucker, M.A., Melanoma epidemiology. Hematol Oncol Clin North Am, 2009. 23 (3): p. 383-95, vii.

8. Udayakumar, D., et al., Genetic determinants of cutaneous melanoma predisposition. Semin Cutan Med Surg, 2010. 29 (3): p. 190-5.

9. Hussussian, C.J., et al., Germline p16 mutations in familial melanoma. Nat Genet, 1994. 8(1): p. 15-21.

10. von Thaler, A.K., Y. Kamenisch, and M. Berneburg, The role of ultraviolet radiation in melanomagenesis. Exp Dermatol, 2010. 19 (2): p. 81-8.

11. Sample, A. and Y.Y. He, Mechanisms and prevention of UV-induced melanoma. Photodermatol Photoimmunol Photomed, 2018. 34 (1): p. 13-24.

12. Goto, N., et al., Detection of UV-induced cyclobutane pyrimidine dimers by near- infrared spectroscopy and aquaphotomics. Sci Rep, 2015. 5: p. 11808.

13. Douki, T., et al., Formation of the main UV-induced thymine dimeric lesions within isolated and cellular DNA as measured by high performance liquid chromatography-tandem mass spectrometry. J Biol Chem, 2000. 275 (16): p. 11678-85.

103

14. Budden, T. and N.A. Bowden, The role of altered nucleotide excision repair and UVB-induced DNA damage in melanomagenesis. Int J Mol Sci, 2013. 14 (1): p. 1132-51.

15. Budden, T., et al., Repair of UVB-induced DNA damage is reduced in melanoma due to low XPC and global genome repair. Oncotarget, 2016. 7(38): p. 60940- 60953.

16. Tornaletti, S. and G.P. Pfeifer, Slow repair of pyrimidine dimers at p53 mutation hotspots in skin cancer. Science, 1994. 263 (5152): p. 1436-8.

17. Viros, A., et al., Ultraviolet radiation accelerates BRAF-driven melanomagenesis by targeting TP53. Nature, 2014. 511 (7510): p. 478-482.

18. Miller, A.J. and M.C. Mihm, Jr., Melanoma. N Engl J Med, 2006. 355 (1): p. 51- 65.

19. Bastian, B.C., The molecular pathology of melanoma: an integrated taxonomy of melanocytic neoplasia. Annu Rev Pathol, 2014. 9: p. 239-71.

20. Testa, U., G. Castelli, and E. Pelosi, Melanoma: Genetic Abnormalities, Tumor Progression, Clonal Evolution and Tumor Initiating Cells. Med Sci (Basel), 2017. 5(4).

21. Cantwell-Dorris, E.R., J.J. O'Leary, and O.M. Sheils, BRAFV600E: implications for carcinogenesis and molecular therapy. Mol Cancer Ther, 2011. 10 (3): p. 385- 94.

22. Czyz, M., HGF/c-MET Signaling in Melanocytes and Melanoma. Int J Mol Sci, 2018. 19 (12).

23. Straussman, R., et al., Tumour micro-environment elicits innate resistance to RAF inhibitors through HGF secretion. Nature, 2012. 487 (7408): p. 500-4.

24. Wilson, T.R., et al., Widespread potential for growth-factor-driven resistance to anticancer kinase inhibitors. Nature, 2012. 487 (7408): p. 505-9.

25. Vergani, E., et al., Identification of MET and SRC activation in melanoma cell lines showing primary resistance to PLX4032. Neoplasia, 2011. 13 (12): p. 1132- 42.

26. Damsky, W.E., N. Theodosakis, and M. Bosenberg, Melanoma metastasis: new concepts and evolving paradigms. Oncogene, 2014. 33 (19): p. 2413-22.

104

27. Balch, C.M., et al., Prognostic factors analysis of 17,600 melanoma patients: validation of the American Joint Committee on Cancer melanoma staging system. J Clin Oncol, 2001. 19 (16): p. 3622-34.

28. Gonzalez, A., Sentinel Lymph Node Biopsy: Past and Present Implications for the Management of Cutaneous Melanoma with Nodal Metastasis. Am J Clin Dermatol, 2018. 19 (Suppl 1): p. 24-30.

29. Melanoma Survival Rates . 2018 [cited 2019.

30. Khan, I. and P.S. Steeg, Metastasis suppressors: functional pathways. Lab Invest, 2018. 98 (2): p. 198-210.

31. Steeg, P.S., et al., Evidence for a novel gene associated with low tumor metastatic potential. J Natl Cancer Inst, 1988. 80 (3): p. 200-4.

32. Scambia, G., et al., nm23 in ovarian cancer: correlation with clinical outcome and other clinicopathologic and biochemical prognostic parameters. J Clin Oncol, 1996. 14 (2): p. 334-42.

33. Yan, J., Q. Yang, and Q. Huang, Metastasis suppressor genes. Histol Histopathol, 2013. 28 (3): p. 285-92.

34. Choudhuri, T., et al., Nm23-H1 can induce cell cycle arrest and apoptosis in B cells. Cancer Biol Ther, 2010. 9(12): p. 1065-78.

35. Leonard, M.K., et al., Identification of a gene expression signature associated with the metastasis suppressor function of NME1: prognostic value in human melanoma. Lab Invest, 2018. 98 (3): p. 327-338.

36. McCorkle, J.R., et al., The metastasis suppressor NME1 regulates expression of genes linked to metastasis and patient outcome in melanoma and breast carcinoma. Cancer Genomics Proteomics, 2014. 11 (4): p. 175-94.

37. Jung, H., H.A. Seong, and H. Ha, Direct interaction between NM23-H1 and macrophage migration inhibitory factor (MIF) is critical for alleviation of MIF- mediated suppression of p53 activity. J Biol Chem, 2008. 283 (47): p. 32669-79.

38. Jung, H., H.A. Seong, and H. Ha, NM23-H1 tumor suppressor and its interacting partner STRAP activate p53 function. J Biol Chem, 2007. 282 (48): p. 35293-307.

39. Marino, N., et al., Nm23-h1 binds to gelsolin and inactivates its actin-severing capacity to promote tumor cell motility and metastasis. Cancer Res, 2013. 73 (19): p. 5949-62.

105

40. Otsuki, Y., et al., Tumor metastasis suppressor nm23H1 regulates Rac1 GTPase by interaction with Tiam1. Proc Natl Acad Sci U S A, 2001. 98 (8): p. 4385-90.

41. Wagner, P.D., P.S. Steeg, and N.D. Vu, Two-component kinase-like activity of nm23 correlates with its motility-suppressing activity. Proc Natl Acad Sci U S A, 1997. 94 (17): p. 9000-5.

42. Hartsough, M.T., et al., Nm23-H1 metastasis suppressor phosphorylation of kinase suppressor of Ras via a histidine protein kinase pathway. J Biol Chem, 2002. 277 (35): p. 32389-99.

43. Takacs-Vellai, K., The metastasis suppressor Nm23 as a modulator of Ras/ERK signaling. J Mol Signal, 2014. 9: p. 4.

44. Salerno, M., et al., Inhibition of signal transduction by the nm23 metastasis suppressor: possible mechanisms. Clin Exp Metastasis, 2003. 20 (1): p. 3-10. 45. Levit, M.N., et al., Interactions between Escherichia coli nucleoside-diphosphate kinase and DNA. J Biol Chem, 2002. 277 (7): p. 5163-7.

46. MacDonald, N.J., et al., A serine phosphorylation of Nm23, and not its nucleoside diphosphate kinase activity, correlates with suppression of tumor metastatic potential. J Biol Chem, 1993. 268 (34): p. 25780-9.

47. Zhang, Q., et al., Metastasis suppressor function of NM23-H1 requires its 3'-5' exonuclease activity. Int J Cancer, 2011. 128 (1): p. 40-50.

48. Ji, J.P. and C.K. Mathews, Analysis of mutagenesis induced by a thermolabile T4 phage deoxycytidylate hydroxymethylase suggests localized deoxyribonucleotide pool imbalance. Mol Gen Genet, 1991. 226 (1-2): p. 257-64.

49. Lu, Q., et al., The gene for nucleoside diphosphate kinase functions as a mutator gene in Escherichia coli. J Mol Biol, 1995. 254 (3): p. 337-41.

50. Niida, H., et al., Mechanisms of dNTP supply that play an essential role in maintaining genome integrity in eukaryotic cells. Cancer Sci, 2010. 101 (12): p. 2505-9.

51. Yang, M., et al., YNK1, the yeast homolog of human metastasis suppressor NM23, is required for repair of UV radiation- and etoposide-induced DNA damage. Mutat Res, 2009. 660 (1-2): p. 74-8.

52. Kaetzel, D.M., et al., Potential contributions of antimutator activity to the metastasis suppressor function of NM23-H1. Mol Cell Biochem, 2009. 329 (1-2): p. 161-5.

106

53. Hanawalt, P.C., J.M. Ford, and D.R. Lloyd, Functional characterization of global genomic DNA repair and its implications for cancer. Mutat Res, 2003. 544 (2-3): p. 107-14.

54. Jarrett, S.G., et al., Metastasis suppressor NM23-H1 promotes repair of UV- induced DNA damage and suppresses UV-induced melanomagenesis. Cancer Res, 2012. 72 (1): p. 133-43.

55. Ma, D., J.R. McCorkle, and D.M. Kaetzel, The metastasis suppressor NM23-H1 possesses 3'-5' exonuclease activity. J Biol Chem, 2004. 279 (17): p. 18073-84. 56. Shevelev, I.V. and U. Hubscher, The 3' 5' exonucleases. Nat Rev Mol Cell Biol, 2002. 3(5): p. 364-76.

57. Puts, G.S., et al., Nuclear functions of NME proteins. Lab Invest, 2018. 98 (2): p. 211-218.

58. Zhao, H., et al., Expression profile of genes associated with antimetastatic gene: nm23-mediated metastasis inhibition in breast carcinoma cells. Int J Cancer, 2004. 109 (1): p. 65-70.

59. Bosnar, M.H., et al., Downstream targets of Nm23-H1: gene expression profiling of CAL 27 cells using DNA microarray. Mol Carcinog, 2006. 45 (8): p. 627-33.

60. Che, G., et al., Transfection of nm23-H1 increased expression of beta-Catenin, E- Cadherin and TIMP-1 and decreased the expression of MMP-2, CD44v6 and VEGF and inhibited the metastatic potential of human non-small cell lung cancer cell line L9981. Neoplasma, 2006. 53 (6): p. 530-7.

61. Horak, C.E., et al., Nm23-H1 suppresses tumor cell motility by down-regulating the lysophosphatidic acid receptor EDG2. Cancer Res, 2007. 67 (15): p. 7238-46.

62. Ma, D., et al., NM23-H1 and NM23-H2 repress transcriptional activities of nuclease-hypersensitive elements in the platelet-derived growth factor-A promoter. J Biol Chem, 2002. 277 (2): p. 1560-7.

63. Cervoni, L., et al., In vivo cross-linking of nm23/nucleoside diphosphate kinase to the PDGF-A gene promoter. Mol Biol Rep, 2003. 30 (1): p. 33-40.

64. Cervoni, L., et al., DNA sequences acting as binding sites for NM23/NDPK proteins in melanoma M14 cells. J Cell Biochem, 2006. 98 (2): p. 421-8.

65. Cheng, S., et al., Tumour metastasis suppressor, nm23-beta, inhibits gelatinase A transcription by interference with transactivator Y-box protein-1 (YB-1). Biochem J, 2002. 366 (Pt 3): p. 807-16.

107

66. Curtis, C.D., et al., Interaction of the tumor metastasis suppressor nonmetastatic protein 23 homologue H1 and estrogen receptor alpha alters estrogen-responsive gene expression. Cancer Res, 2007. 67 (21): p. 10600-7.

67. Choudhuri, T., et al., Expression of alpha V integrin is modulated by Epstein-Barr virus nuclear antigen 3C and the metastasis suppressor Nm23-H1 through interaction with the GATA-1 and Sp1 transcription factors. Virology, 2006. 351 (1): p. 58-72.

68. Subramanian, C. and E.S. Robertson, The metastatic suppressor Nm23-H1 interacts with EBNA3C at sequences located between the glutamine- and proline- rich domains and can cooperate in activation of transcription. J Virol, 2002. 76 (17): p. 8702-9.

69. Kaul, R., et al., Epstein-Barr virus protein can upregulate cyclo-oxygenase-2 expression through association with the suppressor of metastasis Nm23-H1. J Virol, 2006. 80 (3): p. 1321-31.

70. Kuppers, D.A., et al., Regulation of matrix metalloproteinase 9 expression by Epstein-Barr virus nuclear antigen 3C and the suppressor of metastasis Nm23- H1. J Virol, 2005. 79 (15): p. 9714-24.

71. Cook, L.M., D.R. Hurst, and D.R. Welch, Metastasis suppressors and the tumor microenvironment. Semin Cancer Biol, 2011. 21 (2): p. 113-22.

72. Marino, N., J.C. Marshall, and P.S. Steeg, Protein-protein interactions: a mechanism regulating the anti-metastatic properties of Nm23-H1. Naunyn Schmiedebergs Arch Pharmacol, 2011. 384 (4-5): p. 351-62.

73. McCorkle, J.R., NM23-H1 Blocks Cell Motility Independent of its Known Enzymatic Activities in a Cohort of Human Melanoma Cells , in Molecular and Biomedical Pharmacology . 2010, University of Kentucky: Lexington, KY. p. 107.

74. Marsh, J.J. and H.G. Lebherz, Fructose-bisphosphate aldolases: an evolutionary history. Trends Biochem Sci, 1992. 17 (3): p. 110-3.

75. Schwartz, L., C.T. Supuran, and K.O. Alfarouk, The Warburg Effect and the Hallmarks of Cancer. Anticancer Agents Med Chem, 2017. 17 (2): p. 164-170.

76. Xu, X.D., et al., Warburg effect or reverse Warburg effect? A review of cancer metabolism. Oncol Res Treat, 2015. 38 (3): p. 117-22.

77. Hanahan, D. and R.A. Weinberg, Hallmarks of cancer: the next generation. Cell, 2011. 144 (5): p. 646-74.

108

78. Erez, A. and R.J. DeBerardinis, Metabolic dysregulation in monogenic disorders and cancer - finding method in madness. Nat Rev Cancer, 2015. 15 (7): p. 440-8.

79. Huberts, D.H. and I.J. van der Klei, Moonlighting proteins: an intriguing mode of multitasking. Biochim Biophys Acta, 2010. 1803 (4): p. 520-5.

80. Langellotti, S., Unraveling Aldolase C Moonlighting Functions: A Multitask Approach , in Molecular Medicine . 2012, European School of Molecular Medicine: Naples, Italy. p. 124.

81. Hu, H., et al., Phosphoinositide 3-Kinase Regulates Glycolysis through Mobilization of Aldolase from the Actin Cytoskeleton. Cell, 2016. 164 (3): p. 433- 46.

82. Kusakabe, T., K. Motoki, and K. Hori, Mode of interactions of human aldolase isozymes with cytoskeletons. Arch Biochem Biophys, 1997. 344 (1): p. 184-93.

83. Ritterson Lew, C. and D.R. Tolan, Aldolase sequesters WASP and affects WASP/Arp2/3-stimulated actin dynamics. J Cell Biochem, 2013. 114 (8): p. 1928- 39.

84. Wang, J., et al., The molecular nature of the F-actin binding activity of aldolase revealed with site-directed mutants. J Biol Chem, 1996. 271 (12): p. 6861-5.

85. Schindler, R., et al., Aldolase-localization in cultured cells: cell-type and substrate-specific regulation of cytoskeletal associations. Biochem Cell Biol, 2001. 79 (6): p. 719-28.

86. Li, Y.J., et al., Suppression of fructose-bisphosphate aldolase C expression as a predictor of advanced oral squamous cell carcinoma. Head Neck, 2016. 38 Suppl 1: p. E1075-85.

87. Beisiegel, U., W. Weber, and G. Bengtsson-Olivecrona, Lipoprotein lipase enhances the binding of chylomicrons to low density lipoprotein receptor-related protein. Proc Natl Acad Sci U S A, 1991. 88 (19): p. 8342-6.

88. Herz, J. and D.K. Strickland, LRP: a multifunctional scavenger and signaling receptor. J Clin Invest, 2001. 108 (6): p. 779-84.

89. Liu, C.X., et al., The putative tumor suppressor LRP1B, a novel member of the low density lipoprotein (LDL) receptor family, exhibits both overlapping and distinct properties with the LDL receptor-related protein. J Biol Chem, 2001. 276 (31): p. 28889-96.

109

90. Liu, C.X., et al., LRP-DIT, a putative endocytic receptor gene, is frequently inactivated in non-small cell lung cancer cell lines. Cancer Res, 2000. 60 (7): p. 1961-7.

91. Lu, Y.J., et al., Aberrant methylation impairs low density lipoprotein receptor- related protein 1B tumor suppressor function in gastric cancer. Genes Chromosomes Cancer, 2010. 49 (5): p. 412-24.

92. Nakagawa, T., et al., Genetic or epigenetic silencing of low density lipoprotein receptor-related protein 1B expression in oral squamous cell carcinoma. Cancer Sci, 2006. 97 (10): p. 1070-4.

93. Sonoda, I., et al., Frequent silencing of low density lipoprotein receptor-related protein 1B (LRP1B) expression by genetic and epigenetic mechanisms in esophageal squamous cell carcinoma. Cancer Res, 2004. 64 (11): p. 3741-7.

94. Nikolaev, S.I., et al., Exome sequencing identifies recurrent somatic MAP2K1 and MAP2K2 mutations in melanoma. Nat Genet, 2011. 44 (2): p. 133-9.

95. Stohr, H., et al., A novel gene encoding a putative transmembrane protein with two extracellular CUB domains and a low-density lipoprotein class A module: isolation of alternatively spliced isoforms in retina and brain. Gene, 2002. 286 (2): p. 223-31.

96. Zhang, W., et al., A transmembrane accessory subunit that modulates kainate- type glutamate receptors. Neuron, 2009. 61 (3): p. 385-96.

97. Hu, L., et al., Upregulation of NETO2 expression correlates with tumor progression and poor prognosis in colorectal carcinoma. BMC Cancer, 2015. 15 : p. 1006.

98. Oparina, N., et al., [Increase in NETO2 gene expression is a potential molecular genetic marker in renal and lung cancers]. Genetika, 2012. 48 (5): p. 599-607.

99. Zhang, Q., Molecular Mechanisms that Mediate Metastasis Suppressor Activity of NM23-H1 , in Molecular and Biomedical Pharmacology . 2006, University of Kentucky: Lexington, KY. p. 162.

100. Novak, M., et al., Metastasis suppressor NME1 regulates melanoma cell morphology, self-adhesion and motility via induction of fibronectin expression. Exp Dermatol, 2015. 24 (6): p. 455-61.

101. Xu, L., et al., Gene expression changes in an animal melanoma model correlate with aggressiveness of human melanoma metastases. Mol Cancer Res, 2008. 6(5): p. 760-9.

110

102. You, D.J., et al., A splicing variant of NME1 negatively regulates NF-kappaB signaling and inhibits cancer metastasis by interacting with IKKbeta. J Biol Chem, 2014. 289 (25): p. 17709-20.

103. Novak, M., et al., Multiple mechanisms underlie metastasis suppressor function of NM23-H1 in melanoma. Naunyn Schmiedebergs Arch Pharmacol, 2011. 384 (4- 5): p. 433-8.

104. Pacifico, M.D., et al., nm23 as a prognostic marker in primary cutaneous melanoma: evaluation using tissue microarray in a patient group with long-term follow-up. Melanoma Res, 2005. 15 (5): p. 435-40.

105. Xerri, L., et al., NM23 expression in metastasis of malignant melanoma is a predictive prognostic parameter correlated with survival. Br J Cancer, 1994. 70 (6): p. 1224-8.

106. Fiore, L.S., et al., c-Abl and Arg induce cathepsin-mediated lysosomal degradation of the NM23-H1 metastasis suppressor in invasive cancer. Oncogene, 2014. 33 (36): p. 4508-4520.

107. Cropp, C.S., et al., NME1 protein expression and loss of heterozygosity mutations in primary human breast tumors. J Natl Cancer Inst, 1994. 86 (15): p. 1167-9.

108. Takadate, T., et al., Nm23/nucleoside diphosphate kinase-A as a potent prognostic marker in invasive pancreatic ductal carcinoma identified by proteomic analysis of laser micro-dissected formalin-fixed paraffin-embedded tissue. Clin Proteomics, 2012. 9(1): p. 8.

109. Chang, Y.C., et al., Roles of Aldolase Family Genes in Human Cancers and Diseases. Trends Endocrinol Metab, 2018. 29 (8): p. 549-559.

110. Grzywa, T.M., W. Paskal, and P.K. Wlodarski, Intratumor and Intertumor Heterogeneity in Melanoma. Transl Oncol, 2017. 10 (6): p. 956-975.

111. Hayward, N.K., et al., Whole-genome landscapes of major melanoma subtypes. Nature, 2017. 545 (7653): p. 175-180.

112. Pamidimukkala, N.V., et al., Metastasis Suppressor NME1 Directly Activates Transcription of the ALDOC Gene in Melanoma Cells. Anticancer Res, 2018. 38 (11): p. 6059-6068.

113. Postel, E.H., et al., Human c-myc transcription factor PuF identified as nm23-H2 nucleoside diphosphate kinase, a candidate suppressor of tumor metastasis. Science, 1993. 261 (5120): p. 478-80.

111

114. Berberich, S.J. and E.H. Postel, PuF/NM23-H2/NDPK-B transactivates a human c-myc promoter-CAT gene via a functional nuclease hypersensitive element. Oncogene, 1995. 10 (12): p. 2343-7.

115. Thakur, R.K., et al., Metastases suppressor NM23-H2 interaction with G- quadruplex DNA within c-MYC promoter nuclease hypersensitive element induces c-MYC expression. Nucleic Acids Res, 2009. 37 (1): p. 172-83.

116. Liu, B., R.S. Maul, and D.M. Kaetzel, Jr., Repression of platelet-derived growth factor A-chain gene transcription by an upstream silencer element. Participation by sequence-specific single-stranded DNA-binding proteins. J Biol Chem, 1996. 271 (42): p. 26281-90.

117. Vibert, M., et al., The brain-specific gene for rat aldolase C possesses an unusual housekeeping-type promoter. Eur J Biochem, 1989. 181 (1): p. 33-9.

118. Buono, P., et al., Cis-acting elements in the promoter region of the human aldolase C gene. FEBS Lett, 1993. 328 (3): p. 243-9.

119. Cibelli, G., et al., A (G+C)-rich motif in the aldolase C promoter functions as a constitutive transcriptional enhancer element. Eur J Biochem, 1996. 237 (1): p. 311-7.

120. Suzuki, A., et al., Resistance to Fas-mediated apoptosis: activation of caspase 3 is regulated by cell cycle regulator p21WAF1 and IAP gene family ILP. Oncogene, 1998. 17 (8): p. 931-9.

121. Safran, M., et al., Mouse model for noninvasive imaging of HIF prolyl hydroxylase activity: assessment of an oral agent that stimulates erythropoietin production. Proc Natl Acad Sci U S A, 2006. 103 (1): p. 105-10.

122. Kabbarah, O., et al., Integrative genome comparison of primary and metastatic melanomas. PLoS One, 2010. 5(5): p. e10770.

123. Cerami, E., et al., The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov, 2012. 2(5): p. 401-4.

124. Gao, J., et al., Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci Signal, 2013. 6(269): p. pl1.

125. Penhoet, E.E., M. Kochman, and W.J. Rutter, Ioslation of fructose diphosphate aldolases A, B, and C. Biochemistry, 1969. 8(11): p. 4391-5.

126. Penhoet, E.E., M. Kochman, and W.J. Rutter, Molecular and catalytic properties of aldolase C. Biochemistry, 1969. 8(11): p. 4396-402.

112

127. Lebherz, H.G. and W.J. Rutter, Distribution of fructose diphosphate aldolase variants in biological systems. Biochemistry, 1969. 8(1): p. 109-21.

128. Mort, R.L., I.J. Jackson, and E.E. Patton, The melanocyte lineage in development and disease. Development, 2015. 142 (4): p. 620-32.

129. Cheneval, D., et al., A review of methods to monitor the modulation of mRNA stability: a novel approach to drug discovery and therapeutic intervention. J Biomol Screen, 2010. 15 (6): p. 609-22.

130. Bernstein, P.L., et al., Control of c-myc mRNA half-life in vitro by a protein capable of binding to a coding region stability determinant. Genes Dev, 1992. 6(4): p. 642-54.

131. Moore, M.J. and N.J. Proudfoot, Pre-mRNA processing reaches back to transcription and ahead to translation. Cell, 2009. 136 (4): p. 688-700.

132. Kent, W.J., et al., The human genome browser at UCSC. Genome Res, 2002. 12 (6): p. 996-1006.

133. An integrated encyclopedia of DNA elements in the human genome. Nature, 2012. 489 (7414): p. 57-74.

134. Karlic, R., et al., Histone modification levels are predictive for gene expression. Proc Natl Acad Sci U S A, 2010. 107 (7): p. 2926-31.

135. Guenther, M.G., et al., A chromatin landmark and transcription initiation at most promoters in human cells. Cell, 2007. 130 (1): p. 77-88.

136. Quinodoz, M., et al., Characteristic bimodal profiles of RNA polymerase II at thousands of active mammalian promoters. Genome Biol, 2014. 15 (6): p. R85.

137. Spengler, M.L., L.W. Guo, and M.G. Brattain, Phosphorylation mediates Sp1 coupled activities of proteolytic processing, desumoylation and degradation. Cell Cycle, 2008. 7(5): p. 623-30.

138. Chuang, J.Y., et al., Sp1 phosphorylation by cyclin-dependent kinase 1/cyclin B1 represses its DNA-binding activity during mitosis in cancer cells. Oncogene, 2012. 31 (47): p. 4946-59.

139. Leggett, R.W., et al., Sp1 is phosphorylated and its DNA binding activity downregulated upon terminal differentiation of the liver. J Biol Chem, 1995. 270 (43): p. 25879-84.

140. Fife, C.M., J.A. McCarroll, and M. Kavallaris, Movers and shakers: cell cytoskeleton in cancer metastasis. Br J Pharmacol, 2014. 171 (24): p. 5507-23.

113

141. Rottner, K., et al., Actin assembly mechanisms at a glance. J Cell Sci, 2017. 130 (20): p. 3427-3435.

142. Olson, M.F. and E. Sahai, The actin cytoskeleton in cancer cell motility. Clin Exp Metastasis, 2009. 26 (4): p. 273-87.

143. Glass-Marmor, L. and R. Beitner, Detachment of glycolytic enzymes from cytoskeleton of melanoma cells induced by calmodulin antagonists. Eur J Pharmacol, 1997. 328 (2-3): p. 241-8.

144. Lintz, M., A. Munoz, and C.A. Reinhart-King, The Mechanics of Single Cell and Collective Migration of Tumor Cells. J Biomech Eng, 2017. 139 (2).

145. Lawson, D.A., et al., Tumour heterogeneity and metastasis at single-cell resolution. Nat Cell Biol, 2018. 20 (12): p. 1349-1360.

146. Friedl, P. and K. Wolf, Plasticity of cell migration: a multiscale tuning model. J Cell Biol, 2010. 188 (1): p. 11-9.

147. Charras, G. and E. Sahai, Physical influences of the extracellular environment on cell migration. Nat Rev Mol Cell Biol, 2014. 15 (12): p. 813-24.

148. Liu, D., et al., HSF1 promotes the inhibition of EMT-associated migration by low glucose via directly regulating Snail1 expression in HCC cells. Discov Med, 2016. 22 (120): p. 87-96.

149. Wu, S., et al., c-MYC responds to glucose deprivation in a cell-type-dependent manner. Cell Death Discov, 2015. 1: p. 15057.

150. Maurichi, A., et al., Prediction of survival in patients with thin melanoma: results from a multi-institution study. J Clin Oncol, 2014. 32 (23): p. 2479-85.

151. Murakami, M., et al., Nucleoside diphosphate kinase/Nm23 and Epstein-Barr virus. Mol Cell Biochem, 2009. 329 (1-2): p. 131-9.

152. Pandey, S. and E.S. Robertson, Oncogenic Epstein-Barr virus recruits Nm23-H1 to regulate chromatin modifiers. Lab Invest, 2018. 98 (2): p. 258-268.

153. Pal, S., et al., Human SWI/SNF-associated PRMT5 methylates histone H3 arginine 8 and negatively regulates expression of ST7 and NM23 tumor suppressor genes. Mol Cell Biol, 2004. 24 (21): p. 9630-45.

154. Yadav, V.K., et al., Promoter-proximal transcription factor binding is transcriptionally active when coupled with nucleosome repositioning in immediate vicinity. Nucleic Acids Res, 2014. 42 (15): p. 9602-11.

114

155. Hoffman, E.A., et al., Formaldehyde crosslinking: a tool for the study of chromatin complexes. J Biol Chem, 2015. 290 (44): p. 26404-11.

156. Hung, C.Y., et al., Nm23-H1-stabilized hnRNPA2/B1 promotes internal ribosomal entry site (IRES)-mediated translation of Sp1 in the lung cancer progression. Sci Rep, 2017. 7(1): p. 9166.

157. Tamaki, T., et al., Characterization of a GC-rich region containing Sp1 binding site(s) as a constitutive responsive element of the alpha 2(I) collagen gene in human fibroblasts. J Biol Chem, 1995. 270 (9): p. 4299-304.

158. Hernandez-Munain, C., J.L. Roberts, and M.S. Krangel, Cooperation among multiple transcription factors is required for access to minimal T-cell receptor alpha-enhancer chromatin in vivo. Mol Cell Biol, 1998. 18 (6): p. 3223-33.

159. Du, S., et al., Fructose-bisphosphate aldolase a is a potential metastasis- associated marker of lung squamous cell carcinoma and promotes lung cell tumorigenesis and migration. PLoS One, 2014. 9(1): p. e85804.

160. Kawai, K., et al., Fructose-bisphosphate aldolase A is a key regulator of hypoxic adaptation in colorectal cancer cells and involved in treatment resistance and poor prognosis. Int J Oncol, 2017. 50 (2): p. 525-534.

161. Yamamoto, T., et al., Identification of aldolase A as a potential diagnostic biomarker for colorectal cancer based on proteomic analysis using formalin-fixed paraffin-embedded tissue. Tumour Biol, 2016. 37 (10): p. 13595-13606.

162. Grandjean, G., et al., Definition of a Novel Feed-Forward Mechanism for Glycolysis-HIF1alpha Signaling in Hypoxic Tumors Highlights Aldolase A as a Therapeutic Target. Cancer Res, 2016. 76 (14): p. 4259-4269.

163. Tao, Q.F., et al., Aldolase B inhibits metastasis through Ten-Eleven Translocation 1 and serves as a prognostic biomarker in hepatocellular carcinoma. Mol Cancer, 2015. 14 : p. 170.

164. Qu, L., et al., Inhibitory effect of upregulated DR-nm23 expression on invasion and metastasis in colorectal cancer. Eur J Cancer Prev, 2013. 22 (6): p. 512-22.

165. Jung, S., et al., Expression of Nm23 in gliomas and its effect on migration and invasion in vitro. Anticancer Res, 2006. 26 (1a): p. 249-58.

166. Liu, S.J., et al., Downregulated NM23-H1 expression is associated with intracranial invasion of nasopharyngeal carcinoma. Br J Cancer, 2008. 98 (2): p. 363-9.

115

167. Pages, F., et al., Immune infiltration in human tumors: a prognostic factor that should not be ignored. Oncogene, 2010. 29 (8): p. 1093-102.

168. Takayama, H., et al., Diverse tumorigenesis associated with aberrant development in mice overexpressing hepatocyte growth factor/scatter factor. Proc Natl Acad Sci U S A, 1997. 94 (2): p. 701-6.

169. Noonan, F.P., et al., Animal models of melanoma: an HGF/SF transgenic mouse model may facilitate experimental access to UV initiating events. Pigment Cell Res, 2003. 16 (1): p. 16-25.

170. Day, C.P., et al., Mouse models of UV-induced melanoma: genetics, pathology, and clinical relevance. Lab Invest, 2017. 97 (6): p. 698-705.

171. Noonan, F.P., et al., Accelerated ultraviolet radiation-induced carcinogenesis in hepatocyte growth factor/scatter factor transgenic mice. Cancer Res, 2000. 60 (14): p. 3738-43.

172. Wolnicka-Glubisz, A. and F.P. Noonan, Neonatal susceptibility to UV induced cutaneous malignant melanoma in a mouse model. Photochem Photobiol Sci, 2006. 5(2): p. 254-60.

173. Otsuka, T., et al., c-Met autocrine activation induces development of malignant melanoma and acquisition of the metastatic phenotype. Cancer Res, 1998. 58 (22): p. 5157-67.

174. Saha, D., et al., Epigenetic suppression of human telomerase (hTERT) is mediated by the metastasis suppressor NME2 in a G-quadruplex-dependent fashion. J Biol Chem, 2017. 292 (37): p. 15205-15215.

175. Liu, Y.F., et al., NME2 reduces proliferation, migration and invasion of gastric cancer cells to limit metastasis. PLoS One, 2015. 10 (2): p. e0115968.

176. Thakur, R.K., et al., Non-metastatic 2 (NME2)-mediated suppression of lung cancer metastasis involves transcriptional regulation of key cell adhesion factor vinculin. Nucleic Acids Res, 2014. 42 (18): p. 11589-600.

177. Thakur, R.K., et al., Mechanisms of non-metastatic 2 (NME2)-mediated control of metastasis across tumor types. Naunyn Schmiedebergs Arch Pharmacol, 2011. 384 (4-5): p. 397-406.

178. Potel, C.M., D. Fasci, and A.J.R. Heck, Mix and match of the tumor metastasis suppressor Nm23 protein isoforms in vitro and in vivo. Febs j, 2018. 285 (15): p. 2856-2868.

116

179. Postel, E.H., et al., Targeted deletion of Nm23/nucleoside diphosphate kinase A and B reveals their requirement for definitive erythropoiesis in the mouse embryo. Dev Dyn, 2009. 238 (3): p. 775-87.

180. Jarrett, S.G., et al., NM23 deficiency promotes metastasis in a UV radiation- induced mouse model of human melanoma. Clin Exp Metastasis, 2013. 30 (1): p. 25-36.

181. Arnaud-Dabernat, S., et al., Knockout mice as model systems for studying nm23/NDP kinase gene functions. Application to the nm23-M1 gene. J Bioenerg Biomembr, 2003. 35 (1): p. 19-30.

182. Boissan, M., et al., Increased lung metastasis in transgenic NM23-Null/SV40 mice with hepatocellular carcinoma. J Natl Cancer Inst, 2005. 97 (11): p. 836-45.

183. Sharpless, E. and L. Chin, The INK4a/ARF locus and melanoma. Oncogene, 2003. 22 (20): p. 3092-8.

184. FitzGerald, M.G., et al., Prevalence of germ-line mutations in p16, p19ARF, and CDK4 in familial melanoma: analysis of a clinic-based population. Proc Natl Acad Sci U S A, 1996. 93 (16): p. 8541-5.

185. Curtin, J.A., et al., Distinct sets of genetic alterations in melanoma. N Engl J Med, 2005. 353 (20): p. 2135-47.

186. Recio, J.A., et al., Ink4a/arf deficiency promotes ultraviolet radiation-induced melanomagenesis. Cancer Res, 2002. 62 (22): p. 6724-30.

187. Monahan, K.B., et al., Somatic p16(INK4a) loss accelerates melanomagenesis. Oncogene, 2010. 29 (43): p. 5809-17.

188. Krimpenfort, P., et al., Loss of p16Ink4a confers susceptibility to metastatic melanoma in mice. Nature, 2001. 413 (6851): p. 83-6.

189. Tomayko, M.M. and C.P. Reynolds, Determination of subcutaneous tumor size in athymic (nude) mice. Cancer Chemother Pharmacol, 1989. 24 (3): p. 148-54.

190. Tormo, D., et al., Rapid growth of invasive metastatic melanoma in carcinogen- treated hepatocyte growth factor/scatter factor-transgenic mice carrying an oncogenic CDK4 mutation. Am J Pathol, 2006. 169 (2): p. 665-72.

191. Mehrara, E., et al., Specific growth rate versus doubling time for quantitative characterization of tumor growth rate. Cancer Res, 2007. 67 (8): p. 3970-5.

117

192. Gaffal, E., et al., Neonatal UVB exposure accelerates melanoma growth and enhances distant metastases in Hgf-Cdk4(R24C) C57BL/6 mice. Int J Cancer, 2011. 129 (2): p. 285-94.

193. Cheng, L., et al., Molecular testing for BRAF mutations to inform melanoma treatment decisions: a move toward precision medicine. Mod Pathol, 2018. 31 (1): p. 24-38.

194. Organ, S.L. and M.S. Tsao, An overview of the c-MET signaling pathway. Ther Adv Med Oncol, 2011. 3(1 Suppl): p. S7-s19.

195. De Fabo, E.C., Initial studies on an in vivo action spectrum for melanoma induction. Prog Biophys Mol Biol, 2006. 92 (1): p. 97-104.

196. Noonan, F.P., et al., Neonatal sunburn and melanoma in mice. Nature, 2001. 413 (6853): p. 271-2.

197. van Schanke, A., et al., Induction of nevi and skin tumors in Ink4a/Arf Xpa knockout mice by neonatal, intermittent, or chronic UVB exposures. Cancer Res, 2006. 66 (5): p. 2608-15.

198. Oliveria, S.A., et al., Sun exposure and risk of melanoma. Arch Dis Child, 2006. 91 (2): p. 131-8.

199. Balk, S.J., Ultraviolet radiation: a hazard to children and adolescents. Pediatrics, 2011. 127 (3): p. e791-817.

200. Green, A.C., S.C. Wallingford, and P. McBride, Childhood exposure to ultraviolet radiation and harmful skin effects: epidemiological evidence. Prog Biophys Mol Biol, 2011. 107 (3): p. 349-55.

201. Hirobe, T., Developmental changes of the proliferative response of mouse epidermal melanocytes to skin wounding. Development, 1988. 102 (3): p. 567-74.

202. Roh, M.R., et al., Cutaneous melanoma in women. Int J Womens Dermatol, 2017. 3(1 Suppl): p. S11-s15.

203. Chevalier, V., et al., Comparison of anatomical locations of cutaneous melanoma in men and women: a population-based study in France. Br J Dermatol, 2014. 171 (3): p. 595-601.

204. Voinea, S., et al., Impact of gender and primary tumor location on outcome of patients with cutaneous melanoma. J Med Life, 2016. 9(4): p. 444-448.

205. Joosse, A., et al., Gender differences in melanoma survival: female patients have a decreased risk of metastasis. J Invest Dermatol, 2011. 131 (3): p. 719-26.

118

206. Rayner, K., et al., Discovery of NM23-H2 as an estrogen receptor beta-associated protein: role in estrogen-induced gene transcription and cell migration. J Steroid Biochem Mol Biol, 2008. 108 (1-2): p. 72-81.

207. Markiewicz, M., et al., A role for estrogen receptor-alpha and estrogen receptor- beta in collagen biosynthesis in mouse skin. J Invest Dermatol, 2013. 133 (1): p. 120-7.

208. Ohata, C., T. Tadokoro, and S. Itami, Expression of estrogen receptor beta in normal skin, melanocytic nevi and malignant melanomas. J Dermatol, 2008. 35 (4): p. 215-21.

209. Bosnar, M.H., et al., Subcellular localization of A and B Nm23/NDPK subunits. Exp Cell Res, 2004. 298 (1): p. 275-84.

210. Ismail, N.I., et al., Nuclear localization and intensity of staining of nm23 protein is useful marker for breast cancer progression. Cancer Cell Int, 2008. 8: p. 6.

211. Marioni, G., et al., Nm23-H1 nuclear expression is associated with a more favourable prognosis in laryngeal carcinoma: univariate and multivariate analysis. Histopathology, 2012. 61 (6): p. 1057-64.

212. Yoon, J.H., et al., Characterization of the 3' --> 5' exonuclease activity found in human nucleoside diphosphate kinase 1 (NDK1) and several of its homologues. Biochemistry, 2005. 44 (48): p. 15774-86.

213. Sheng, Y., et al., Nm23-H1 is involved in the repair of ionizing radiation-induced DNA double-strand breaks in the A549 lung cancer cell line. BMC Cancer, 2018. 18 (1): p. 710.

214. Fan, Z., et al., Tumor suppressor NM23-H1 is a granzyme A-activated DNase during CTL-mediated apoptosis, and the nucleosome assembly protein SET is its inhibitor. Cell, 2003. 112 (5): p. 659-72.

215. Chowdhury, D., et al., The exonuclease TREX1 is in the SET complex and acts in concert with NM23-H1 to degrade DNA during granzyme A-mediated cell death. Mol Cell, 2006. 23 (1): p. 133-42.

216. Kim, S.H., et al., Nuclear localization of Nm23-H1 in head and neck squamous cell carcinoma is associated with radiation resistance. Cancer, 2011. 117 (9): p. 1864-73.

217. Banerjee, S., H.C. Jha, and E.S. Robertson, Regulation of the metastasis suppressor Nm23-H1 by tumor viruses. Naunyn Schmiedebergs Arch Pharmacol, 2015. 388 (2): p. 207-24.

119

218. Subramanian, C., M.A. Cotter, 2nd, and E.S. Robertson, Epstein-Barr virus nuclear protein EBNA-3C interacts with the human metastatic suppressor Nm23- H1: a molecular link to cancer metastasis. Nat Med, 2001. 7(3): p. 350-5.

219. McLaughlin-Drubin, M.E. and K. Munger, Viruses associated with human cancer. Biochim Biophys Acta, 2008. 1782 (3): p. 127-50.

220. Muster, T., et al., An endogenous retrovirus derived from human melanoma cells. Cancer Res, 2003. 63 (24): p. 8735-41.

221. Buscher, K., et al., Expression of human endogenous retrovirus K in melanomas and melanoma cell lines. Cancer Res, 2005. 65 (10): p. 4172-80.

222. Schiavetti, F., et al., A human endogenous retroviral sequence encoding an antigen recognized on melanoma by cytolytic T lymphocytes. Cancer Res, 2002. 62 (19): p. 5510-6.

223. Kim, M.S., et al., A preliminary X-ray study of human nucleoside diphosphate kinase A under oxidative conditions. Acta Crystallogr Sect F Struct Biol Cryst Commun, 2010. 66 (Pt 11): p. 1490-2.

224. Dumas, C., et al., X-ray structure of nucleoside diphosphate kinase. Embo j, 1992. 11 (9): p. 3203-8.

225. Kim, Y.I., et al., Point mutations affecting the oligomeric structure of Nm23-H1 abrogates its inhibitory activity on colonization and invasion of prostate cancer cells. Biochem Biophys Res Commun, 2003. 307 (2): p. 281-9.

226. Mesnildrey, S., F. Agou, and M. Veron, The in vitro DNA binding properties of NDP kinase are related to its oligomeric state. FEBS Lett, 1997. 418 (1-2): p. 53- 7.

227. Engel, M., et al., Glyceraldehyde-3-phosphate dehydrogenase and Nm23- H1/nucleoside diphosphate kinase A. Two old enzymes combine for the novel Nm23 protein phosphotransferase function. J Biol Chem, 1998. 273 (32): p. 20058-65.

228. Kimura, K., et al., c-Myc gene single-strand binding protein-1, MSSP-1, suppresses transcription of alpha-smooth muscle actin gene in chicken visceral smooth muscle cells. Nucleic Acids Res, 1998. 26 (10): p. 2420-5.

229. Tomonaga, T. and D. Levens, Activating transcription from single stranded DNA. Proc Natl Acad Sci U S A, 1996. 93 (12): p. 5830-5.

120

230. Di Matteo, G., et al., Interactions with single-stranded and double-stranded DNA- binding factors and alternative promoter conformation upon transcriptional activation of the Htf9-a/RanBP1 and Htf9-c genes. J Biol Chem, 1998. 273 (1): p. 495-505.

231. Leonard, M.K., et al., The metastasis suppressor NME1 inhibits melanoma cell motility via direct transcriptional induction of the integrin beta-3 gene. Exp Cell Res, 2019. 374 (1): p. 85-93.

232. Larijani, M. and A. Martin, Single-stranded DNA structure and positional context of the target cytidine determine the enzymatic efficiency of AID. Mol Cell Biol, 2007. 27 (23): p. 8038-48.

233. Wang, Y., et al., Biophysical characterization of the association of histones with single-stranded DNA. Biochim Biophys Acta Gen Subj, 2017. 1861 (11 Pt A): p. 2739-2749.

234. Berger, M.F., et al., Integrative analysis of the melanoma transcriptome. Genome Res, 2010. 20 (4): p. 413-27.

235. Kitamura, T., B.Z. Qian, and J.W. Pollard, Immune cell promotion of metastasis. Nat Rev Immunol, 2015. 15 (2): p. 73-86.

236. Balogh, K.N., D.J. Templeton, and J.V. Cross, Macrophage Migration Inhibitory Factor protects cancer cells from immunogenic cell death and impairs anti-tumor immune responses. PLoS One, 2018. 13 (6): p. e0197702.

121