Characterization of Growth Hormone Signaling in the NCI60 Cancer Panel

A dissertation presented to

the faculty of

the College of Arts and Sciences of Ohio University

In partial fulfillment

of the requirements for the degree

Doctor of Philosophy

Elahu Gosney Sustarsic

August 2013

This dissertation titled

Characterization of Growth Hormone Signaling in the NCI60 Cancer Panel

ELAHU GOSNEY SUSTARSIC

has been approved for

the Department of Biological Sciences

and the College of Arts and Sciences by

Xiaozhuo Chen

Associate Professor of Microbiology

Robert Frank

Dean, College of Arts and Sciences 3

ABSTRACT

SUSTARSIC, ELAHU GOSNEY, Ph.D., August 2013, Molecular and Cellular Biology

Characterization of Growth Hormone Signaling in the NCI60 Cancer Panel

Director of Dissertation: Xiaozhuo Chen

Growth hormone Receptor (GHR) is a class I cytokine receptor. Under normal conditions, it is expressed in most tissues of the body.In addition to its presence in healthy tissue,

GHR has been detected in cancer cells and tumors. There has been accumulating evidence that implicates GHR activation in carcinogenesis. Most research to date has focused on breast and prostate cancer, with a few studies in other cancer types. A limited number of studies have investigated a direct effect of GH on cancer cells, and even fewer have examined the mechanism of GH action in cancer cells. Our investigations began with a survey of gene expression in the 60 cell lines of the National Cancer Institute’s NCI60 panel. All of the metastatic melanoma lines examined show high expression of GHR. Five of the melanoma cell lines were treated with GH.

In three of the cell lines, GH treatment increases cell growth. In one cell line, GH has a biphasic effect based on dose; a low dose moderately inhibits cell growth, while a high dose induces growth. Cell signaling pathways were analyzed as well. Some activation of mTOR was detected, along with strong STAT5 phosphorylation with GH treatment. A final group of studies examined the effect of GH treatment on cancer metabolism. In these experiments, we discovered a novel action of GH: an ability to increase the bioenergetic potential and potentiate the Warburg effect in cancer cells. These findings may have relevance to growth and invasion of metastatic melanoma cells in vivo. Further studies are warranted to determine if therapeutic interventions aimed at GH signaling might be beneficial in treatment of metastatic melanoma.

PREFACE ON DISSERTATION FORMAT

This dissertation is divided into five chapters. The first one is an introduction to the current body of literature on the role GH may have in cancer. Chapters 2, 3 and 4 are in the form of manuscript drafts that will be submitted for publication in Endocrine-Related Cancer or a similar journal. Therefore, the formatting is set to the author guidelines for this journal.

Importantly, in chapters 2-4, figures are not within the body of the text (as they are for chapters 1 and 5), but are shown at the end of each chapter. The final chapter lists the key findings and provides a roadmap of future research that should be undertaken as follow-up studies. 5

DEDICATION

I dedicate this dissertation to the people who made it possible.

To my mother, brother and sister for their nurturing love and support. Jahiah, your prescient advice to me many years ago put me on the path to this endeavor. Thanks to John for so many

things, most importantly for giving me your passion for science.

And to the love of my life, Riia. Without you, this would not have been possible.

Minun rakkaus pupuni, sinä olet mailman ihana nainen.

E<3R

ACKNOWLEDGMENTS

I would like to acknowledge the many people who played a role in this project. Most importantly, Dr. John Kopchick, for bringing together several world experts who have assisted with this project. We owe a debt of gratitude to Dr. Steven Swanson, who suggested the use of the National Cancer Institute resources, specifically the NCI60 panel, for this investigation. Dr.

Theresa Wood for offering her expertise in the insulin/IGF field, and Dr. Vincent Goffin for his encouragement and guidance in recombinant protein production and gene expression studies.

Many thanks to everyone in the Kopchick lab who have been my friends, mentors and colleagues for many years, including Dr. Darlene Berryman, Dr. Edward List, Dr. Nick Okada and Dr. Bruce

Kelder. Without the training, encouragement and guidance of these scientists, this work would not have been completed. Funding for the project was provided by the Diabetes Institute at Ohio

University, the Molecular and Cellular Biology Program at Ohio University and the State of

Ohio’s Eminent Scholars Program that includes a gift from Milton and Lawrence Goll.

TABLE OF CONTENTS

Page

Abstract ...... 3 Preface on Dissertation Format ...... 4 Dedication ...... 5 Acknowledgments...... 6 List of Tables ...... 11 List of Figures ...... 12 Chapter 1: Introduction ...... 14 General overview ...... 14 GH background ...... 19 The growth hormone gene family ...... 19 Regulation of pituitary GH ...... 20 GH activation of signaling via cell surface receptors ...... 20 Signaling pathways affected by GH ...... 22 Endocrine GH Action: the somatomedin hypothesis ...... 23 Autocrine/paracrine GH ...... 25 GH and cancer literature review ...... 26 Evidence from cell culture and human biopsies ...... 26 Evidence from animal studies ...... 36 Evidence from human studies (epidemiological and clinical) ...... 43 Contradictory data ...... 46 Summary ...... 48 Comment on PRLR signaling and cancer ...... 50 Hypotheses and research objectives ...... 51 Research approach ...... 52 Limitations of the study ...... 54 Value of the research ...... 55 Publication of data and dissertation outline ...... 55 References ...... 57 Chapter 2: Expression of growth hormone, prolactin, and cognate receptors in the NCI60 cancer cell line panel ...... 67 Abstract ...... 67 Introduction ...... 68 Materials and methods ...... 71 NCI60 cancer cell line RNA samples ...... 71 Real-time RT qPCR ...... 71 Primer design and validation ...... 72 Human melanoma biopsy cDNA arrays ...... 73 8

Broad-Novartis Cancer Cell Line Encyclopedia ...... 73 COMPARE correlations with existing NCI60 data ...... 74 Data analysis ...... 74 Results ...... 75 GHR, PRLR, GH and PRL expression across the NCI60 panel ...... 75 GHR, PRLR, GH and PRL expression between 9 human cancer types ...... 76 GHR, PRLR, GH and PRL expression within cancer types: differences between the cell lines ...... 77 Expression of GHR in metastatic melanoma samples ...... 79 GHR, IGF1 and IGF1R mRNA levels in stage III and IV metastatic melanoma samples by sex ...... 79 NCI60 COMPARE: correlations with microarray and GI50 data ...... 79 Comparison to CCLE Data ...... 81 Discussion ...... 81 GHR is highly expressed in melanoma cell lines ...... 82 PRLR is highly expressed in breast cancer cell lines ...... 83 GHR and IGF1 mRNA is higher in tumors from males while GHR and IGF1R mRNA is elevated in advanced stage IV metastatic tumors ...... 83 Low GHR expression levels in colon cancer cell lines is in contrast with previous data from biopsies...... 84 Limited evidence for autocrine and/or paracrine GH and PRL action...... 85 Cell lines with high GHR and/or PRLR expression tend to be resistant to tubulin inhibiting drugs ...... 86 GHR is positively correlated with rapamycin sensitivity and expression of MAPK-related genes ...... 86 PRL expression is positively associated with effectiveness of estrogen suppression ...... 87 Declaration of interest ...... 88 Funding ...... 88 Author contributions ...... 88 Acknowledgements ...... 89 References ...... 89 Tables ...... 96 Figures ...... 97 Supplementary figures ...... 103 Chapter 3: Growth hormone modulates proliferation and intracellular signaling in melanoma cell lines ...... 107 Abstract ...... 107 Introduction ...... 108 Materials and methods ...... 111 NCI60 melanoma cell lines, culture ...... 111 Proliferation assays ...... 111 9

Analysis of MAPK/Erk, PI3K/Akt and mTOR signaling ...... 112 In-cell STAT signaling assay ...... 113 Data analysis ...... 114 Results ...... 114 Dose Response of five melanoma cell lines to GH ...... 114 Effect of low and high GH treatment on proliferation of SK-MEL-5, UACC-62 and MDA-MB-435 ...... 115 Basal activation of Erk, Akt and mTOR in SK-MEL-5, UACC-62 and MDA-MB-435 116 MAPK/Erk ...... 116 PI3K/Akt ...... 117 mTOR ...... 117 STAT1, STAT3 and STAT5 ...... 118 Discussion ...... 118 GH exerts a biphasic effect on proliferation in melanoma cells ...... 119 The effect of low and high GH on MAPK/Erk, PI3K/Akt and mTOR signaling ...... 120 The effect of low and high GH on STAT1, STAT3 and STAT5 signaling ...... 121 Other factors that may contribute to a biphasic effect of GH in melanoma ...... 122 Conclusions ...... 123 Declaration of interest ...... 124 Funding ...... 124 Author contributions ...... 124 Acknowledgements ...... 125 References ...... 125 Tables ...... 131 Figures ...... 132 Supplementary figures ...... 138 Chapter 4: Growth hormone regulates metabolic programming and the Warburg effect in metastatic melanoma cells ...... 140 Abstract ...... 140 Introduction ...... 141 Materials and methods ...... 144 Cell culture ...... 144 GH treatment ...... 145 Mitochondrial stress test ...... 145 Mitochondrial stress test calculations ...... 146 Glycolysis stress test ...... 147 Glycolysis stress test calculations ...... 147 Lactate assay ...... 148 Results ...... 149 Mitochondrial stress test ...... 149 Glycolysis stress test ...... 151 10

GH-induced lactate production ...... 152 Discussion ...... 153 GH increases mitochondrial respiration ...... 153 GH increases aerobic glycolysis ...... 155 Lactate production in melanoma ...... 157 Relevance with respect to published studies on metabolic flux in melanoma cells ...... 157 Conclusions ...... 159 Declaration of interest ...... 159 Funding ...... 159 Author contributions ...... 160 Acknowledgements ...... 160 References ...... 160 Figures ...... 163 Chapter 5: Key findings and roadmap for future studies ...... 169 Quantification of gene expression in the NCI60 panel and human metastatic melanoma tumors ...... 170 Key findings of gene expression studies ...... 171 Future work on gene expression ...... 172 Effect of GH on proliferation and signal transduction pathways ...... 173 Key findings of the effect of GH on proliferation ...... 173 Future work on GH treatment of melanoma cells ...... 174 Effect of GH on metabolic programming of the melanoma line MDA-MB-435 ...... 175 Key findings of the effect of GH on metabolism ...... 176 Future work on GH and metabolism ...... 176 Hypothetical model of GH action in melanoma cells...... 177 Final thoughts ...... 181 Appendix 1: Table summary of research published on GH and cell lines included in the NCI60 panel...... 183 Appendix 2: Guidelines for primer design...... 190 Appendix 3. Examples of primer validation and temperature optimization...... 192 Appendix 4: Expression of mRNA for the IGF/INSULIN pathway in the NCI60 panel...... 196 Appendix 5: Correlations within our gene expression data...... 200 Appendix 6: Production and validation of recombinant human GH and GHA...... 201

LIST OF TABLES

Page

Table 2.1. Cell lines and cancer types included in the NCI60 ...... 96 Table 2.2. Primer sequences and optimal annealing temperatures for real-time RT qPCR ...... 96 Table 2.3. Geometric mean of relative expression and 95% CI by cancer type and for the entire panel ...... 96 Table 3.1. Summary of GH-induced changes in phosphorylation status of signaling proteins ... 131

LIST OF FIGURES

Page

Figure 1.1. The human GH gene locus...... 20 Figure 1.2. Mechanism of GHR signaling...... 22 Figure 1.3. Signaling pathways activated by GH...... 23 Figure 1.4. A new somatomedin hypothesis...... 24 Figure 1.5. Previous publications that investigated GH in NCI60 cell lines ...... 27 Figure 1.6. GHR mRNA expression in breast cancer...... 29 Figure 1.7. GHR protein expression in breast cancer...... 30 Figure 1.8. GH treatment of breast cancer cells...... 32 Figure 1.9. Expression of GH, PRL and GHR in prostate cancer cells...... 33 Figure 1.10. GH signaling in LNCaP prostate cancer cells...... 34 Figure 1.11. GH and GHR protein expression in a CNS cancer...... 35 Figure 1.12. Mammary cancer studies in the spontaneous dwarf rat...... 38 Figure 1.13. Breast cancer study in GHR-/- mice...... 40 Figure 1.14. Prostate cancer study in lit/lit dwarf mice...... 42 Figure 1.15. Expression of GHR in prostate cancer rat model...... 43 Figure 1.16. Summary of prospective studies of associations between height and cancer...... 44 Figure 2.1. Overall mRNA expression in the NCI60 panel by gene...... 97 Figure 2.2. Relative mRNA expression in the nine tumor types of the NCI60 cell panel...... 98 Figure 2.3. Relative mRNA expression by cell line in the NCI160 panel...... 99 Figure 2.4. GHR mRNA expression in human melanoma tumors...... 100 Figure 2.5. GHR, IGF-1 and IGF-1R mRNA levels in metastatic melanoma samples analyzed by sex (A) and tumor grade (B)...... 101 Supplementary figure 2.1. Geometric mean GHR, PRLR, GH and PRL mRNA expression with 95% CI...... 103 Supplementary figure 2.2. IGF1 (A) and IGF1R (B) mRNA expression in human melanoma tumors...... 104 Supplementary figure 2.3. Expression of PRLR (A), GH (B) and PRL (C) from CCLE...... 106 Figure 3.1. GH dose-response of melanoma cell lines ...... 132 Figure 3.2. Effect of low (0.05 nM) and high-dose (100 nM) GH treatment on proliferation .... 133 Figure 3.3. Basal signal stransduction activation in three melanoma cell lines...... 134 Figure 3.4. Effect of GH treatment on Erk, Akt and mTOR activation in three melanoma lines...... 135 Figure 3.5. Effect of GH treatment STAT1, STAT3 and STAT5 activation in three melanoma lines...... 136 Figure 3.6. Correlation of basal Erk activity with GH-induced proliferation at 100 nM ...... 137 Supplementary figure 3.1. A representative cell dilution curve that validates the PrestoBlue proliferation assay...... 138 13

Supplementary figure 3.2. The impact of low and high GH treatment on the expression of GHR, IGF1 and IGF2 in MDA-MB-435...... 139 Figure 4.1. An overview of oxidative phosphorylation and aerobic glycolysis...... 163 Figure 4.2. Mitochondrial Stress Test...... 164 Figure 4.3 Bioenergetic parameters measured during the mitochondrial stress test...... 165 Figure 4.4. Glycolysis stress test...... 166 Figure 4.5. Bioenergetic parameters measured during the glycolysis stress test...... 167 Figure 4.6. GH-induced lactate production...... 168 Figure 5.1. Hypothetical model of GH action in melanoma...... 178

CHAPTER 1: INTRODUCTION

General overview

An estimated 580,350 Americans will die of cancer in 2013, making it the second leading cause of death in the United States (American Cancer Society 2013). There also will be over 1.66 million new diagnoses of cancer in 2013. The lifetime chance of getting cancer is nearly 1 in 2 if you are a male, and 1 in 3 if you are female (American Cancer Society 2013). Despite many advances in cancer detection and treatment, alarming rates of cancer incidence and mortality persist. The most common cancer treatments have numerous harmful and debilitating side effects and are often ineffective. Thus, there remains a great need to better understand the factors that promote carcinogenesis in order to develop novel pharmacologic cancer treatments.

In 2000, a seminal paper described the six major hallmarks of cancer that allow unimpeded cell growth (Hanahan & Weinberg 2000). These characteristics include the ability to sustain proliferative signaling, evade growth suppressors, activate invasion and metastasis, enable replicative immortality, induce angiogenesis and resist cell death. In 2011, an update to this model was published by the same authors in which they add metabolism as an emerging hallmark of cancer (Hanahan & Weinberg 2011). The last 10 years has seen new vigor in the investigation of an old idea: that cancer cells possess unique metabolic profiles compared to non-proliferative cells (Koppenol et al. 2011). The so-called Warburg effect describes the preference of many cancer cells to metabolize glucose into lactate to produce energy, bypassing the efficient ATP- producing pathways of the mitochondria. The factors that alter the metabolic state of cancer cells are not elucidated in great detail, although many hormones are known to regulate pathways that can influence metabolism. 15

While cancer is fundamentally a disease associated with inherited and somatic gene mutations, the precise factors that permit such abnormal cells to thrive and contribute to the hallmarks of cancer described above is largely unknown. Although much research attention on the causes of cancer has focused on inherited gene mutations, there are data to suggest other etiologies are perhaps more important in the process of carcinogenesis. While we have identified at least 473 “cancer genes”, for example, the American Cancer Society states that only 5% of cancers have a strong hereditary factor (American Cancer Society 2013). Additionally, carcinogenesis is strongly influenced by environmental factors even for those cases associated with mutations in prototypical oncogenes, such as BRCA1/2 or p53 (Holly & Perks 2008). For example, healthy adults are reported to have p53 mutations in 4% of skin cells, without tumorigenesis occurring (Vineis 2003). It is becoming increasingly recognized that while genetic mutations contribute to cancer risk, perhaps these mutations are more permissive then causative.

Exposure to carcinogenic environmental factors, which could include diet and exposure to pollutants, may perhaps act as the master controller of carcinogenesis. Genetic and environmental factors affect many aspects of physiology, including modulation of hormone signaling pathways that control cell growth, proliferation, and apoptosis. Disregulation of these normal hormonal biological functions is increasingly recognized to contribute to carcinogenesis, as explained in a review on the topic, “It is becoming apparent that many human cancer cells respond to a wide range of hormones, and indeed themselves express many hormones which could play an important role in facilitating the evolution and progression of the cancer,” (Holly & Perks 2008).

The idea that environmental factors, including the influence of hormones, are critical in the biology of cancer has been receiving substantial coverage in scientific journals, including a recent article in Science that describes the connection between obesity and cancer (Taubes

2012a). In a related article, the director of a Harvard-associated cancer center discusses results 16

which he states, “support the idea that anything that lowers insulin and IGF levels will inhibit tumor growth,” (Taubes 2012b). Our laboratory developed one of the safest and most effective ways to lower circulating insulin-like growth factor 1 (IGF-1) levels, so we are particularly interested in this approach.

Growth hormone (GH) is a classic endocrine hormone that regulates several biological functions, including cell survival, cell growth, body growth, and metabolism. Endocrine GH is produced in the pituitary gland and is released into the blood where it travels to target tissues. At the target tissue, GH binds to the extracellular domain of the GH receptor (GHR) and/or the prolactin receptor (PRLR) to initiate intracellular signaling events and exert its biological effect on the cell. GH was originally thought to act primarily via IGF-1, whose expression is induced by

GH in the liver and other tissues. This relationship is so close, that GH signaling is often referred to as the GH/IGF-1 axis. In recent years, it has become increasingly clear that GH also has IGF-1- independent functions, although the exact contribution of GH or IGF-1 to a particular biological effect of GH is sometimes difficult to dissect. Research also suggests that GH may act in an autocrine/paracrine fashion in some tissues, in addition to its well established endocrine roles.

Importantly, a GHR antagonist (GHA) has been developed in our laboratory at Ohio

University that effectively inhibits GH action. This antagonist has been developed into a successful pharmaceutical, Somavert (Pegvisomant for injection), currently approved for the treatment of acromegaly (Higham & Trainer 2008). Acromegaly, a condition of chronic GH excess, is usually caused by a somatotroph-derived pituitary adenoma that secretes high amounts of GH. The disease is characterized by elevated circulating GH levels that results in high IGF-1 levels in the blood. Pegvisomant is a competitive inhibitor of GH that will bind to GHRs without causing a signaling event. When carefully titrated, treatment with Pegvisomant can reduce the elevated IGF-1 levels to normal in up to 97% of patients (van der Lely et al. 2001). It is well 17

tolerated in a wide range of doses and a highly effective intervention for GH excess. Presumably,

Pegvisomant could also be used to reduce GH and IGF-1 signaling in other disease conditions where inhibition of the somatotrophic axis may be clinically beneficial.

As early as 1972, GH was detected in human cancer tissue (Beck & Burger 1972).

Around the same time, hypophysectomy was shown to enhance treatment for advanced breast cancer (VanGilder & Goldenberg 1975). There is accumulating evidence that GH may play a role in the development and/or progression of cancer. In humans, there are multiple lines of converging research that implicate GH signaling in carcinogenesis. Growth parameters such as birth weight and final adult height, indirect indicators of GH/IGF-1 action, correlate positively with risk for certain cancers and for overall cancer mortality (Lawlor et al. 2003, Gunnell et al.

2001). Analyses of low-frequency alleles for GH and related genes have shown an association of the pathway with cancer risk (Mulhall et al. 2005, Rudd et al. 2006, Menashe et al. 2010). In fact, a recent study examined more than 528,000 SNPs in 421 pathways to determine the pathways associated with breast cancer (Menashe et al. 2010). The growth hormone signaling pathway had the third highest association with breast cancer. Patients with acromegaly, a condition of elevated

GH and IGF-1 levels, are at increased risk for certain types of cancer (Renehan & Brennan 2008).

Studies of pituitary GH-treated patients, largely industry-sponsored, show a potential increase in cancer incidence and mortality (Swerdlow et al. 2002). At the other physiological extreme, patients who are deficient in GHR are protected from cancer, with no cancer deaths reported in a

22 year study of Laron patients in an Ecuadorian cohort (Guevara-Aguirre et al. 2011). There have also been studies that indicate elevated circulating GH or IGF-1 levels in cancer patients implicating a potential role for endocrine GH (Emerman et al. 1985, Renehan et al. 2004,

Rowlands et al. 2009, Okamoto et al. 2010). 18

Biomolecular characterization of tissue biopsies and explants from cancer patients has shown expression of GH and/or GHR in multiple cancer types, including breast cancer

(Decouvelaere et al. 1995, Lincoln et al. 1998, Mertani et al. 1998, Gebre-Medhin et al. 2001,

Gregoraszczuk et al. 2001), prostate cancer (Untergasser et al. 1999, Weiss-Messer et al. 2004), meningioma (Friend et al. 1999) and colorectal cancer (Yang et al. 2004). Expression of both GH and GHR in some of these tumor tissues indicates the capacity for autocrine/paracrine GH action.

In vitro tissue culture experiments and in vivo studies using laboratory animals have provided further evidence for a role of GH signaling in cancer. The most striking evidence comes from animal studies that have been performed in rodents with reduced or absent GH signaling. In these studies, the cancer progression and mortality are severely reduced in GH/IGF-1 signaling deficient animals that have genetic or chemically induced breast or prostate cancer (Yang et al.

1996, Pollak et al. 2001, Swanson & Unterman 2002, Shen et al. 2007, Thordarson et al. 2004,

Zhang et al. 2007, Majeed et al. 2005, Anzo et al. 2008, Wang et al. 2005, Wang et al. 2008).

Even with this strong epidemiologic and experimental evidence suggesting an important role for GH in carcinogenesis, the research on this topic has been very limited. The work that has been done has primarily focused on breast and prostate cancer, with very little attention to other types. Even the basic research of GH/IGF-1 pathway component expression in cancer cells and tissues has only been performed on a limited number of cancer types and cell lines. Of the studies that have found an effect of GH, the mechanism of GH signaling is rarely explored. Only a handful of researchers have studied the effect of GH treatment or GHR inhibition on cancer cells.

GH background

The growth hormone gene family

Growth hormone (GH) is a cytokine hormone with pleiotropic biological effects, playing a major role in growth, development, metabolism and aging. In humans, the GH gene is found on chromosome 17 and is arranged in a cluster with a total of five highly homologous genes thought to arise from a common ancestral gene, as shown in figure 1.1 (Yoo et al. 2006). In addition to

GH-normal (GH-N), the four other genes in this locus are GH-variant (GHV), chorionic somatomammotropin A and B (CS-A, CS-B), and chorionic somatomammotropin-like (CSL).

These genes are sometimes referred to by other names. CS-A is also known as placental lactogen

(PL), GH-N is also called GH1, and GH-V is referred to as GH2. The five genes each contain five exons and share approximately 95% sequence identity (Chen et al. 1989). Each gene can express three to five alternative splice variants (NCBI RefSeq HuRef database accessed April 3, 2010).

GH-N was originally shown to be expressed “exclusively” in the pituitary gland, the primary site of endocrine GH production. Transcripts from the remaining four genes have been isolated from placental tissues; however, they have received little research attention since discovery and the existence of meaningful expression of these genes is unclear. Another member of this family is the highly conserved prolactin (PRL) gene, located on chromosome 6 in humans. PRL shares

56% amino acid similarity with GH (Cooke et al. 1981) and even closer structural similarity, as reviewed by Goffin et al (Goffin et al. 1996).

Figure 1.1. The human GH gene locus. CS = chorionic somatomammotropin, L = like, N = normal, V = variant. GH-V is also referred to as placental GH and GH2. CS-A is also known as placental lactogen (PL) (figure based on Yoo et al. 2006).

Regulation of pituitary GH

Endocrine GH regulation is under dual control by GH releasing hormone (GHRH) and somatostatin (SST), which are delivered from neurons in the hypothalamus to the anterior pituitary. Three levels of GH feedback on somatotrophs have been described. A long-loop feedback negatively regulates GH production via systemic IGF-1 produced by the liver and other

GH-responsive organs in response to GH. A second level of regulation is the short-loop feedback system, whereby GH travels to the hypothalamus and increases expression of SST while decreasing expression of GHRH. The third level of regulation is an ultra-short feedback system whereby autocrine/paracrine GH action within the pituitary negatively regulates GH production.

Peripheral hormones including ghrelin and leptin also regulate pituitary GH production. A good review of the multi-factorial regulation of GH synthesis and release by the pituitary can be found in (Gahete et al. 2009).

GH activation of signaling via cell surface receptors

GHR and PRLR are members of the class I cytokine receptor superfamily and share a similar tertiary structure and signaling mechanism (Goffin & Kelly 1997). Proteolytic cleavage of 21

human GHR gives rise to a circulating GH binding protein (GHBP) that is thought to prolong the serum half-life of GH (Goffin & Kelly 1997). GHR is now known to exist at the cell surface as a pre-formed dimer that contains two binding sites for GH, referred to as site 1 and 2 (Strous &

Gent 2002). Functional binding of GH to these sites causes a rotation of the receptor subunits, as shown in figure 1.2, which triggers intracellular phosphorylation events that mediate downstream effects (Brown et al. 2005). In the early 1990s, GHAs were developed that inhibit GH action by competitive binding to GHR, illustrated in figure 1.2 and reviewed by Kopchick et al (Kopchick et al. 2002). Various GHAs have been designed, including human (h) GH-G120R, all of which share a mutation of a glycine residue that is critical for GH binding to site 2 of the GHR. Human

GHA (G120R) was successfully developed into a pharmaceutical that is currently used to treat patients with acromegaly (Trainer et al. 2000).

GH from various species has different receptor affinities. Human GH binds to GHR and

PRLR with equal affinity (Goffin et al. 1999). Mouse GH does not bind to either mouse or human

PRLR. To our knowledge, no one has made and tested a mouse GHA and its binding ability to various GHRs. It is expected that it would act similar to mGH, in not binding to PRLR. These differences are important in studies that wish to determine the mechanism of GH action in human cells. 22

Figure 1.2. Mechanism of GHR signaling. The GHR is predimerized at the membrane, though inactive. Its activation is achieved by conformational changes induced upon ligand binding. In such a model, receptor antagonists prevent the appropriate conformational from occurring. D1 and D2 denote domain 1 and domain 2 of the GHR subunit. Site 1 and site 2 are ligand binding locations, while site 3 is a potential receptor subunit interaction site. Modified from (Tallet et al. 2008).

Signaling pathways affected by GH

GH is known to modulate the activity of a variety of signal transduction pathways that are relevant to carcinogenesis (see figure 1.3; Chhabra et al. 2011). These include the phosphoinositol-3 kinase (PI3K)/Akt/mTOR pathway, through which GH has been shown to promote protein synthesis (Hayashi & Proud 2007). The Janus kinase 2 (JAK2)/signal transducer and activator of transcription 3 and 5 (STAT3, STAT5) pathways have been shown to regulate 23

cell survival and other biological pathways that could facilitate metastases, as recently reviewed

(Devarajan & Huang 2009). Lastly, the mitogen-activated protein kinase (MAPK) pathway, a mediator of cell proliferation, differentiation and survival, is capable of activation by GH signaling (Love et al. 1998, Roberts & Der 2007). These pathways are also utilized by other hormones, including PRL, IGFs, insulin and various other growth factors and cytokines. The existence of multiple JAK and STAT proteins further complicates signaling downstream of GHR.

Figure 1.3. Signaling pathways activated by GH. Upon GH binding to GHR, JAKs are phosphoryled, enabling activation of several signaling pathways and following cellular processes (Lanning & Carter-Su 2007, Ratkaj et al. 2010, Kaabi 2012).

Endocrine GH Action: the somatomedin hypothesis

Based on research from the 1950s, the somatomedin hypothesis postulated that GH acts exclusively through a secondary intermediate factor in the serum, simply called somatomedin, to stimulate somatic growth (Daughaday et al. 1972). Later research identified multiple potential

“somatomedin” factors, including that of somatomedin C, now known as IGF-1 (Klapper et al. 24

1983). In the somatomedin hypothesis, the anterior pituitary produces GH and releases it into the serum, where it acts on the liver to stimulate production of IGF-1, which acts as an endocrine hormone on target tissues to affect the known functions of GH, including anabolism of bone and muscle and lipolysis of adipose tissue (Le Roith et al. 2001). More recent research of the

GH/IGF-1 axis has necessitated several modifications to the original hypothesis, including incorporation of the negative feedback loop that regulates pituitary GH production, evidence for direct GH action independent of IGF-1 and local IGF-1 production that acts in an autocrine/paracrine fashion (Kaplan & Cohen 2007), as depicted in figure 1.4.

Figure 1.4. A new somatomedin hypothesis. The effects of GH and IGF-1 on growth and metabolism. GH stimulates IGF-1 production from the liver. This endocrine IGF-1 feeds back to the pituitary to inhibit GH production as well as exerts growth-promoting effects in the periphery. Additionally, circulating GH acts locally on several tissues, in part via autocrine/paracrine IGF-1 (Kaplan & Cohen 2007).

Based largely on data which show normal somatic growth in liver-specific IGF-1 knockout mice, Kaplan et al. argue for a modified somatomedin hypothesis that emphasizes the role of local GH-induced IGF-1 production and the simultaneous augmentative and counteractive actions of GH and IGF-1 at target tissues.

Autocrine/paracrine GH

While pituitary endocrine GH production has been the primary topic of GH research for many years, there is evidence for local production of GH that may be capable of signaling in an autocrine/paracrine fashion (Harvey 2010). Tissues in which GH mRNA and/or protein have been identified include neural, immune, reproductive, skeletal, gastrointestinal, and integumentary, among others (Harvey 2010). One example of exptression in reproductive tissue is present by

Mol and colleagues, who identified expression of GH in mammary tissue of dogs and cats, as well as normal, benign and malignant human mammary tissue (Mol et al. 1996). In multiple studies described below, cancer cells and tissues from a variety of studies have been shown to express GH mRNA and/or protein. In the laboratory of Dr. Peter Lobie, an innovative approach has compared MCF7 cells transfected with a transgene to express GH with exogenous treatment with GH (Liu et al. 1997). Cells with the GH transgene display enhanced STAT activation, proliferation and mitogenic capacity and unique properties compared to cells treated with exogenous GH protein. Further, these cells display an invasive phenotype when implanted into immune-compromised mice (Kaulsay et al. 1999). It is unclear exactly how endogenous GH action may differ from exogenous application of GH. It has been hypothesized that endogenous

GH may activate GHR in the cytoplasm, but escape the silencing mechanisms normally responsible for turning the GH signaling cascade off, or that endogenous GH may activate GHR and JAK2 within the nucleus to regulate gene expression (Waters & Conway-Campbell 2004). It 26

is also possible that intracellular GH acts through an unknown mechanism. Further research is needed to establish the role of autocrine/paracrine GH in carcinogenesis, but if it is shown to play a role in promoting cancer, it will be critical to understand the mechanism of action of autocrine/paracrine GH in order to develop therapeutics that can effectively inhibit locally produced GH in addition to endocrine GH.

GH and cancer literature review

The central role of GH in postnatal somatic growth has been well established. In recent years, data have been steadily accumulating from a wide range of studies that implicate GH signaling in the process of oncogenesis. These studies include in vitro tissue culture, animal studies in dogs, mice and rats as well as epidemiological data from various human populations.

GH stimulation has been implicated in various types of cancers, including neoplasia in the breast, prostate, colon, lung and brain (see literature review below for references).

Evidence from cell culture and human biopsies

Overview of data from cell lines included in the NCI60 panel

A comprehensive literature search was conducted to determine the previous body of work reported on NCI60 cell lines and GH. The number of NCI60 cell lines for each tumor type is as follows: six breast cancer, two prostate cancer, six leukemia, nine non-small cell lung cancer, seven colon cancer, six CNS cancer, nine melanoma, seven ovarian, and eight renal cancer. The name and tumor type of all 60 cell lines can be found in Chapter 2. Only articles that contained the name of at least one of the NCI60 cell lines and the term “growth hormone” in the title or abstract were included in the preliminary search. A list of more than 200 potentially relevant articles were examined and narrowed down to 41 articles that reported relevant research, 27

defined as primary literature on either, A) expression of GH and/or GHR in a NCI60 cell line, or

B) treatment of a NCI60 cell line with GH and/or GHA (figure 1.5). Only 15 of these articles were published within the past 10 years. Some of the key characteristics of this literature are discussed in the following paragraph, while the most relevant data are included in the later sections of this literature review.

Figure 1.5. Previous publications that investigated GH in NCI60 cell lines, presented by cancer type. Data is based on Pubmed searches conducted in January 2012.

A summary of the findings of each of the 41 articles is presented in a table in appendix 1.

Only 11 of the 60 cell lines are discussed in these 41 reports. Limited data have also been published for GH and GHR mRNA expression in these cell lines. GHR mRNA has been shown to be present in 3 breast cancer cell lines (T47D, MCF7, MDA-MB231), one leukemia cell line

(HL60), two prostate cell lines (PC3, DU145) and one CNS cancer cell line (SNB-19) 28

(Decouvelaere et al. 1995, Costoya et al. 2000, Untergasser et al. 1999, Friend et al. 2001).

According to these publications, seven of the NCI60 cell lines have been treated with GH (MCF7,

T47D, MDA-MB231, Molt7, HL60, PC3, and DU145). Direct treatment with GHA has only been performed for one of the 60 cell lines, MCF-7, which showed a decrease in proliferation (Zatelli et al. 2009). Xenografts of five of the 60 cell lines have been implanted into animals, which were then treated with GHA. MCF-7 xenografts showed a 2-fold decrease in proliferation and a 2-fold increase in apoptosis, while no effect was observed for MDA-MB231 or MDA-MB435 xenografts (Divisova et al. 2006). The two xenografts of colon cancer cell lines provide mixed results, with COLO-205 cells showing a decrease in tumor volume and increased apoptosis, while

HT29 cells show no effect (Dagnaes-Hansen et al. 2004). Only a handful of these publications examined downstream signaling molecules to determine which signaling pathway(s) participate in GH signaling in cancer cells.

Breast cancer

While previous research had pointed to a role for hGH and/or PRL in the growth of breast cancer cells, gene expression data had not been shown until a 1995 report (see figure 1.6;

Decouvelaere et al. 1995). Decouvelaere and colleagues examined four breast cancer cell lines and 29 biopsy samples by reverse transcriptase PCR (RT-PCR). They showed, for the first time, that GHR is expressed in breast cancer cell lines (MCF-7, T-47D and MDA-MB-231), and perhaps more importantly, in all but one of 29 human tumor biopsies examined, implicating the possibility of a direct role of GH in carcinogenesis.

Figure 1.6. GHR mRNA expression in breast cancer. A Southern blot of RT-PCR products confirms expression of GHR in ductal carcinomas (lanes 1, 3-8 and 10), a lobular carcinoma (9) and a colloid carcinoma (2). Modified from (Decouvelaere et al. 1995).

Evidence for a direct growth stimulatory effect of PRL was shown in the 1980s.

Inhibition of PRLR and/or GHR was shown to reduce the growth rate of breast cancer cell lines in a Genentech study using GH analogs and GHR antagonists (Fuh & Wells 1995). In this study, four human cell lines (MCF-7, T-47D, FDC-P1 and BT474) were treated with WT hGH (which can activate both GHR and PRLR), a GHR-specific hGH analog, a PRLR-specific hGH analog or either the antagonist G120R-hGH or G120R-hPL. Using SK-BR3 cells, it was shown that stimulation of the PRLR (and not GHR) by a GH analog could stimulate cell growth in the presence of zinc. This effect was abrogated by treatment with G120R-hGH. The cell lines examined were shown to have unique growth responses depending on the antagonist applied. Not only did this study implicate GH-induced PRLR signaling in breast cancer proliferation, it established that inhibitors of PRLR may be effective at reducing the growth rate of breast tumors.

As antibodies to GHR became available, researchers were able to show expression of the receptor in both normal and neoplastic breast tissue (Lincoln et al. 1998, Mertani et al. 1998).

Gebre-Medhin et al were the first to quantify GHR protein expression in breast carcinomas and normal tissue by Western blot less than a decade ago (Gebre-Medhin et al. 2001). All but two out of 47 tumors showed GHR expression. Normal adjacent mammary tissue was also biopsied for 17 30

patients and analyzed. The tumor samples contained more than double the GHR protein as adjacent normal mammary tissues, indicating increased GH sensitivity in carcinogenic breast tissue. A striking example of their findings is shown in figure 1.7, where strong GHR-staining is seen in epithelial cells of breast tumors (a, b and c), while stromal cells and normal tissues display weak staining.

Both non-malignant and malignant breast cancer tissue produces local GH, PRL and IGF-

1 in response to progesterone treatment (Gregoraszczuk et al. 2001). While PRL and IGF-1 protein expression induction by progesterone varied in malignant tissue depending on estrogen receptor (ER) and progesterone receptor status (PR), GH was increased regardless of ER or PR status. In malignant tissue that contains PR and not ER, GH, IGF-1 and PRL were significantly increased in response to progesterone treatment. This report implicates local autocrine and/or paracrine GH, IGF-1 and PRL action in the growth of breast tumors, particularly in PR+ tumors.

Figure 1.7. GHR protein expression in breast cancer. GHR (mAb 263) immunostaining of invasive ductal carcinomas of the breast (a and b) and an atypical medullary carcinoma (c). Strong cytoplasmic immunostaining of the epithelial component and weak staining of some stromal cells are seen (b and c). Adjacent normal breast tissue is negative (d) or weakly positive. Taken from (Gebre-Medhin et al. 2001). 31

Given the potential role of GH in promoting breast cancer, recent studies have examined whether GH can interfere with chemotherapeutic drugs. Autocrine expression of GH in breast cancer cells by stable gene transfer has been shown to confer aggressive tumorigenic properties, in excess of that observed by treatment with exogenous GH. MCF-7 cells stably transfected with hGH were analyzed for sensitivity to tamoxifen (Mojarrad et al. 2009). WT MCF-7 cells displayed growth inhibition upon tamoxifen treatment, while MCF-hGH cells did not. Thus, autocrine hGH expression was able to counteract the effect of the anti-cancer drug tamoxifen. In another study (Zatelli et al. 2009), exogenous application of GH to MCF-7 cells increased cell viability and partially reduced the effectiveness of Doxorubicin treatment, under starved conditions. This effect could be replicated by IGF-1 treatment alone. Importantly, while an IGF-1 blocking antibody could inhibit the effect of IGF-1 treatment, it had no effect on GH’s ability to counteract Doxo treatment, suggesting a mode of action that is not mediated by IGF-1.

Pegvisomant treatment was able to abrogate the effect of GH on Doxo effectiveness under starvation conditions. Astonishingly, treatment with pegvisomant alone was able to reduce DNA synthesis and luciferase expression from a c-fos reporter plasmid (a proto-oncogene induced by

GH treatment), as shown in figure 1.8. The effect of pegvisomant indicates that GHR is constitutively active in these cells, that autocrine GH is produced by these cells, and/or that pegvisomant has direct effects on the cell independent of GHR. Why a starved condition was necessary to see the effects of GH is not clear. Additionally, in the body of the paper the authors state that the GH dose was 1 ug/ml, while some of the figure legends state 1 mg/ml. If any of the treatments were at 1 mg/ml, the dose would far exceed physiological concentrations. 32

Figure 1.8. GH treatment of breast cancer cells. The effect of GH, Doxo and pegvisomant on MCF-7 cell synthesis under starved conditions. Taken from (Zatelli et al. 2009).

Prostate Cancer

Tissue samples of human prostate cancer (PCa) and benign hyperplasia of the prostate

(BPH), BPH primary cultures, and prostatic cancer cell lines were investigated for the role of

GH/PRL signaling in prostate carcinogenesis (Untergasser et al. 1999). By RT-PCR, none of the tissue samples were shown to express PRL or GH/PL (see figure 1.9). Immuno-assays did not detect GH, PL or PRL in BPH, PCa or the prostate cancer cell lines examined, including androgen independent (PC3, DU145) and androgen sensitive (LnCap) cell lines. In contrast, all tissues examined expressed PRLR, the full-length GHR, and some also expressed the exon 3 deletion variant of GHR (see figure 1.9), which encodes a short region of the extracellular domain of GHR (GHRd3). To determine if the receptors were active, BPH primary cell cultures were established and treated with hGH, hPL and hPRL. All three hormones were shown to increase cell proliferation at low concentrations (0.01-0.1 nM). 33

Figure 1.9. Expression of GH, PRL and GHR in prostate cancer cells. A: expression of mRNA for hGHR (including GHRd3 in some samples), PRL-R in placenta (lane 1), BPH (lanes 2-6) and PCa (lanes 7-8) specimens. Similar results for PCa cell lines were described but not shown in the manuscript. Modified from (Untergasser et al. 1999). B: expression of mRNA for GHR full- length and GHRd3 isoforms (lower) and mRNA encoding 22kDa GH-N and 20kDa GH-N isoforms (upper). A = ALVA41, D=DU-145, L=LNCaP, P=PC-3, N=normal prostate and C=negative control. Modified from (Chopin et al. 2002).

In contradiction to the work of Untergasser and colleagues, Chopin et al demonstrated the potential for autocrine/paracrine GH signaling in prostate cancer cell lines (Chopin et al. 2002).

As shown in figure 1.9, by both RT-PCR and Northern analysis, GH and GHR expression was found in four cell lines tested; PC-3, LNCaP, DU-145 and ALVA41 (androgen dependent).

Intriguingly, normal prostate expressed mRNA for the 22kDa isoform of GH, while cancer cell expressed both the 22kDa and 20kDa isoforms (figure 1.9). GHR mRNA was also expressed in two variants, the expected 5.3 kb full-length transcript and an additional 2.1 kb alternatively spliced variant. Expression of GH and GHR was verified by immunohistochemistry. This work revealed the potential for autocrine/paracrine GH signaling in normal and cancerous prostate tissue, and also opened up the possibility that the isoforms of GH and GHR may be relevant to prostate carcinogenesis. The authors mention the portion of Untergasser’s 1999 publication that 34

agree with their results, but failed to address the discrepancy between the two groups in regards to

GH expression.

A later examination studied GHR expression in BPH and prostate adenocarcinoma patient tissues, as well as prostate cell lines LNCaP, PC3 and DU145 (Weiss-Messer et al. 2004). mRNA expression for full-length GHR and a truncated GHR isoform was detected in BPA adenocarcinoma samples as well as in the PCa cell lines. Interestingly, GHR expression was upregulated in the adenocarcinomas, compared to BPHs. The androgen sensitive PCa line LNCaP was assayed for GH sensitivity. While no effect of GH on proliferation was observed (in the presence or absence of androgen), GH was shown to induce intracellular signaling by activation of JAK2, STAT5A, p42/p44 MAPK and Akt/PKB , as shown in figure 1.10. These results indicate that multiple GHR signaling pathways are active in PCa cells and responsive to GH.

Given the limits of quantitation of RT-PCR experiments (without using real-time qPCR methods), the increased expression of GHR in adenocarcinomas requires further validation.

Figure 1.10. GH signaling in LNCaP prostate cancer cells. hGH- induced activation of GHR, and dose-dependent phosphorylation of JAK2, STAT5a, MAPK and AKT in LNCaP cells. Modified from (Weiss-Messer et al. 2004).

Cancers of the CNS

Meningioma specimens have been analyzed for a potential role of GH in this CNS cancer

(Friend et al. 1999). By PCR and ribonuclease protection assays, all 14 human meningioma samples studied were found to express GHR mRNA, including some tumors that expressed the

GHRd3 isoform of the receptor. Primary cell cultures were established and treated with B2036 (a

GHR-specific GHA that does not bind PRLR), which inhibited cell growth in the presence of

10% serum. Cells were also shown to proliferate when treated with IGF-1. GH and GHR are also expressed in the neuroblastoma cell line N1E-115, as shown in figure 1.11 (Grimbly et al. 2009).

Immunohistochemical analysis reveals strong, co-localized cytoplasmic GH and GHR staining, as well as localization of GHR to the nucleus. Treatment of these cells with exogenous mouse GH

(mGH) resulted in increased neurite growth, indicating biological activity of the expressed GH and GHR. These studies establish the potential for a role of GHR in CNS tumorigenesis, including a possible autocrine/paracrine function, and suggest that GHA therapeutic treatment may be beneficial.

Figure 1.11. GH and GHR protein expression in a CNS cancer. N1E-115 neuroblastoma cells were stained for GH (red), GHR (green) and nuclei (DAPI, blue). Taken from (Grimbly et al. 2009).

Other Cancer Cells

There is a smattering of isolated reports using cell lines from other types of cancers on the potential role of GH signaling in carcinogenesis. For example, the bile duct cancer cell line

GBC939 has been shown to proliferate and to moderately increase IGF-1 production when exposed to GH (Cai et al. 2008). Human colorectal cancer (CRC) specimens have also been examined (Yang et al. 2004). RT-PCR performed on 42 tumor samples showed that a majority

(33 of 42) of the tumors expressed GHR, while 22 of the 42 non-cancerous colorectal mucosa from the same patients showed expression. Immunohistochemistry showed GHR expression in

83% of the CRC samples and nearly 70% of non-cancerous control tissues collected from the same patients. Tumor stage and differentiation status were both associated with GHR expression, suggesting that higher levels of GHR signaling may lead to more aggressive tumors.

Evidence from animal studies

Over the past 20 years, the proliferation and characterization of transgenic and gene mutant laboratory animal models has proven extremely valuable in understanding the role of the

GH/IGF-1 axis in tumor growth and progression (reviewed by Yakar et al. 2005). With genetic manipulation or the identification of spontaneous mutations, the regulation of specific pathways is possible. A variety of animal studies have been conducted to elucidate the role of the GH/IGF-

1 axis in carcinogenesis.

Breast Cancer

In the first study of its kind, Yang et al crossed the immune-deficient scid mouse with the

GH and IGF-1 deficient lit/lit mouse, which has an inactivating mutation in the GH releasing hormone receptor gene (Yang et al. 1996). MCF7 xenografts were injected into scid/scid lit/lit 37

mice and littermate controls, all of which were given estradiol supplementation. Tumors in the normal mice had grown nearly twice as large as tumors in the dwarf mice. Perhaps due to estradiol supplementation, serum IGF-1 levels were high in control mice (904ng/ml) in this study, while only 49.3 ng/ml in the lit/lit animals. It is possible that the high IGF-1 levels impacted the results of this study.

To study the hypothesis that IGF-1 levels may impact breast carcinogenesis, GHA transgenic mice were subject to chemical carcinogenesis. GHA mice express a GHR antagonist protein, and thus have decreased GH signaling (Chen et al. 1991). Consequently, GHA animals are dwarf with low IGF-1 levels. GHA mice were given a gavage of dimethylbenz (a)anthracene

(DMBA) and monitored for tumor incidence (Pollak et al. 2001). At the end of the 39 week study period, only 32% of control animals remained without tumors, while 68% of GHA animals were tumor free. This study dramatically showed that inhibition of the GHR signaling could provide striking protection from breast cancer in vivo. In a similar fashion, the Spontaneous Dwarf Rat

(SDR), which has an inactivating mutation in the GH gene Ghdr/dr, were subject to chemical carcinogenesis with N-methyl-N-nitrosourea (MNU) or DMBA (Swanson & Unterman 2002).

Throughout the duration of the study, the Ghdr/dr rats showed little tumor incidence, while WT or heterozygous mice accumulated mammary tumors, as shown in figure 1.12. The Swanson laboratory followed up this work with a SDR study that examined the effect of GH administration following MNU treatment (Shen et al. 2007). In this study, Ghdr/dr rats were injected with GH for

2 or 3 months following MNU exposure. Tumor occurrence and burden were similar between

WT animals and the GH-treated Ghdr/dr rat group at two months following MNU exposure. Upon withdrawal of GH from Ghdr/dr rats, all tumors rapidly regressed. During that same time period, tumors in WT rats continued to grow and new tumors formed. After a one month period without

GH, re-administration for three weeks caused tumors in Ghdr/dr rats to reoccur at the original 38

tumor sites and resurgence of tumor growth. Examinations revealed a significant decrease in proliferation and increase in apoptosis in mammary tumors of Ghdr/dr mice after withdrawal of GH treatment.

Figure 1.12. Mammary cancer studies in the spontaneous dwarf rat. Upper panels: latency and multiplicity of mammary tumors following MNU (left) or DMBA (right) treatment. Taken from (Swanson & Unterman 2002). Lower panels: Tumor incidence in SDR rats following treatment with NMU (left). Body weight of SDR animals throughout the study (right). Taken from (Thordarson et al. 2004).

In a complementary study, Ghdr/dr rats were injected with MNU and treated with GH with or without estradiol (E2) and progesterone (P4), with E2 and P4 alone, or with IGF-1 (Thordarson et al. 2004). GH-treated Ghdr/dr rats showed a 100% tumor incidence by 10 weeks of treatment, 39

while IGF-1 treated Ghdr/dr rats reached 50% tumor incidence by week 14. Ghdr/dr animals treated with E2 + P4 had no tumors, while, paradoxically, GH + E2 + P4 treated animals also showed almost no tumor incidence, as shown in figure 1.12, lower left. The steroid hormones seemed to block GH action almost completely. This could be explained by decreased serum IGF-1 levels in the GH + E2 + P4 treated animals (515.8 ng/ml), compared to GH treated animals (822.9 ng/ml).

SDR control rats had an IGF-1 level of 534 ng/ml (my analysis, not that of the authors).

Interestingly, both groups receiving GH achieved considerable growth throughout the study (see figure 1.12, lower right), while the steroid treated and IGF-1 treated groups had minimal body growth. PRL levels in this study provided more interesting data. Dwarf rats treated with IGF-1 or

E2 + P4 had 143.1 ng/ml and 192.4 ng/ml, respectively. These levels were significantly higher than the other treatment groups or controls. Even though the E2 + P4 treatment animals displayed elevated PRL levels at the time of carcinogenesis administration, they did not experience any tumor burden, suggesting a strong role for GH but not PRL in this model of mammary carcinogenesis.

Pegvisomant was studied for its impact on cancer cell xenografts in athymic nude mice

(Divisova et al. 2006). Treatment of wild-type FVB/N mice with pegvisomant severely reduced mammary development, while reducing mammary gland GH and IGF-1 signaling. P-Jak2, p-

STAT5, p-IGF-1R and p-IRS-1 were all decreased following 6 weeks of treatment with pegvisomant. Mice were given 250ug/g pegvisomant daily following xenograft growth to a 100-

200mm volume. MCF-7 xenografts stopped growing upon treatment, while MCF-7 control animals showed a steady increase in tumor size.

Mice with a germline GHR gene disruption (GHR-/-) have also been used to examine the role of the GH/IGF-1 axis in mammary carcinogenesis. Mice that develop mammary tumors due to a transgene for SV40 large T antigen (TAg), targeted to the epithelium of mammary and 40

prostate glands, were crossed with GHR-/- animals (Zhang et al. 2007). While TAg/GHR+/+ animals developed an average of nearly 10 tumors per animal, TAg/GHR-/- animals only showed about 3 tumors per animal. Even more striking then the number of tumors was the size.

TAg/GHR+/+ tumors grew to more than 10 times the volume of those found in GHR disrupted animals (see figure 1.13). It remains to be determined whether the lack of GHR signaling is directly responsible for the decrease in tumor burden in the GHR-/- animals, or if other factors such as lower circulating IGF-1 levels are responsible.

Figure 1.13. Breast cancer study in GHR-/- mice. There were significantly fewer tumors (A) and of smaller size (B) in Tag/GHR-/- animals. Taken from (Zhang et al. 2007).

Prostate Cancer

Transgenic adenocarcinoma of the mouse prostate (TRAMP) mice develop prostate carcinoma due to targeted expression of SV40 early genes in prostatic epithelium. The Pollak lab crossed these mice with lit/+ mice to generate GHRHR-deficient (and thus serum GH and IGF-1)

TRAMP mice (Majeed et al. 2005). TRAMP control mice developed more severe carcinomas then TRAMP/lit/lit animals. By the end of the 35-week study, 74% of the TRAMP controls died or had to be euthanized due to tumor burden, in contrast to only 18% of TRAMP/lit/lit animals.

Intriguingly, GHRHR gene expression was not detected in prostate tissue from TRAMP control animals, suggesting that GHRHR’s influence on carcinogenesis is indirect in this model, 41

presumably acting through stimulation of endocrine GH and/or IGF-1. The TRAMP mouse was crossed with another mouse model with reduced serum IGF-1 levels, the liver-specific IGF-1 gene-deleted mouse (LID) (Anzo et al. 2008). LID-TRAMP mice had a 90% reduction in serum

IGF-1 levels compared to control TRAMP animals, from 177ng/ml to 17ng/ml. Circulating GH levels were 3.5 fold higher in LID-TRAMP mice, consistent with the reduced circulating IGF-1.

Unexpectedly, LID-TRAMP mice had the same rate of pathologic tumor occurrence and metastasis as TRAMP controls and had equivalent survival rates during the study, as shown in figure 1.14. This is in contrast to previous models of prostate cancer in mice with disturbed GH and/or IGF-1 signaling (Wang et al. 2005, Majeed et al. 2005). Immunoblots of tumor proteins revealed increased GHR expression and Akt phosphorylation in LID-TRAMP mice, with no changes in tumor IGF-1, and IGF-2 levels were below the detection limit. The findings of this study suggest the possibility that insulin (elevated in LID mice) or IGF-1-independent GH action is permissive for tumorigenesis in conditions of low systemic IGF-1 levels. If this is in fact the case, treatment with GHA would be an effective intervention, as it acts directly at the GHR to inhibit GH action.

Figure 1.14. Prostate cancer study in lit/lit dwarf mice. Top panel: survival of TRAMP control mice and TRAMP/lit/lit mice with GHRHR disruption. Bottom panels: survival (left) and metastasis (right) curves for LID-TRAMP and L/L-TRAMP controls. Modified from (Majeed et al. 2005) and (Anzo et al. 2008).

The T antigen oncogene has also been used to study prostate cancer. TAg-transgenic mice were crossed with GHR-/- animals (Wang et al. 2005) to create dwarf GH-insensitive

TAg/GHR-/- offspring. While seven out of eight of the Tag/GHR+/+ mice developed prostatic neoplasia, only one out of eight of the Tag/GHR-/-mice exhibit abnormal cell type. This resistance to prostate cancer is accompanied by decreased proliferation and increased apoptosis in the prostate epithelial cells. In a follow up study, TAg transgenic rats were crossed with the GH- deficient SDR rat (Wang et al. 2005, Wang et al. 2008). In this study, prostate tumor burden was reduced in Tag/Ghdr/dr animals compared to controls. By 52 weeks of age, 100% of the Tag/Gh+/+ rats had developed invasive prostate adenocarcinomas, while most of the Tag/Ghdr/dr did not.

GHR expression (mRNA and protein) was increased in the prostates of Tag/Gh+/+ rats as carcinogenesis progressed (see figure 1.15), but IGF-1 and IGF-1R dropped. This report 43

highlights the role GHR plays in prostate cancer in this model suggests that IGF-1 is not a mediator of GH action in this regard.

Figure 1.15. Expression of GHR in prostate cancer rat model. Prostatic GHR mRNA expression in TAg rats during the development of carcinogenesis, relative to expression in cancer-free control animals (A). IHC analysis of prostate GHR protein in TAg/Gh+/+ rats during progression of carcinogenesis (B-D) and in Non-TAg/Gh+/+ control rats that do not develop tumors (E-G). Modified from Wang et al. (2008).

Evidence from human studies (epidemiological and clinical)

There are multiple lines of evidence from human studies that implicate the GH/IGF-1 axis in cancer incidence or severity. One of the earliest studies was published in 1972 (Beck &

Burger 1972), in which human bronchogenic and gastric tumors were found to contain immunoreactive GH, indicating the tumors take up endocrine GH or produce GH in an autocrine/paracrine fashion. Much of the research to date is correlative data that associates growth parameters or certain low-frequency polymorphisms for GH signaling and related genes with cancer. 44

Birth weight can be an indirect indicator of in utero hormone exposure. An approximate

2-fold increase in breast cancer risk has been reported between high birth weight females (Lawlor et al. 2003, Okasha et al. 2002). Studies of the association of birth weight with prostate and other cancers have been less clear. There have been literally hundreds of studies that investigate final height and the risk of various cancers. There are mixed and sometimes conflicting results, as shown in figure 1.16. A systematic review of these studies (Gunnell et al. 2001) found a 22% increase in breast cancer with increased stature (> 5’9” vs. < 5’3”), a 20% increase in prostate cancer (> 6’1” vs. < 5’7”) and a colorectal cancer excess of 20–60% for taller individuals. Leg length was the growth parameter that was most strongly associated with cancer risk.

Figure 1.16. Summary of prospective studies of associations between height and cancer. A positive association is defined as at least a 20% higher risk of cancer in the tallest compared to the shortest category of height; an inverse association is defined as at least a 20% lower risk of cancer in the tallest compared to the shortest category. Reproduced from Okasha et al. (2002).

Studies of patients with acromegaly, a condition of chronic GH excess, have indicated possible increased risk for several cancers, including colon, rectal and thyroid, while there are conflicting data on breast, leukemia and other cancers (reviewed by Renehan & Brennan 2008). A large cohort of patients treated with human pituitary GH have been examined for cancer risk compared to the general population (Swerdlow et al. 2002). Patients receiving GH therapy had higher incidence of colorectal cancer and Hodgkin’s Disease. Overall cancer mortality was increased three-fold, while there was an approximate 11-fold increase in mortality from both colorectal cancer and Hodgkin’s disease. These results are alarming in some respects, but the limited size of the study makes interpretation difficult.

Research has also been undertaken to assess circulating hormone levels in cancer patients. Early evidence suggested elevated circulating GH levels in 40% of breast cancer patients

(Emerman et al. 1985). In a meta-regression analysis, an association of high circulating IGF-1 levels with prostate cancer and premenopausal breast cancer was identified (Renehan et al. 2004).

A more recent meta-analysis was conducted that included both retrospective and prospective studies on the association of circulating IGF-1 levels with prostate cancer risk (Rowlands et al.

2009). On average, each standard deviation increase in IGF-1 levels increased the risk of prostate cancer by 21%. A prospective study of 58 patients with various solid tumors found a disturbance of the GH/IGF-1 axis associated with poor performance status, an indicator of general patient well being (Okamoto et al. 2010). In this study, a state of relative GH resistance was observed in patients with the poorest performance status, who had lower IGF-1 levels and higher GH levels than patients with better performance scores.

A number of studies have examined a role of single-nucleotide polymorphisms (SNPs) and low-penetrance alleles of GH and related genes in relation to cancer risk or burden.

Mammograms from 348 women were analyzed, along with serum GH and two GH SNPs 46

(Mulhall et al. 2005). Both GH alleles were associated with increased breast tissue density, a positive indicator of potential breast cancer risk. One of these SNPs, GH1-75A, was also associated with elevated serum GH levels, presumably by affecting transcription rates from the

GH1 gene. These data implicate GH in susceptibility to breast cancer. Similar work has shown that variants in the GH/IGF-1 axis are associated with lung cancer predisposition (Rudd et al.

2006). Recently, pathway analysis was used to exam a genome-wide association study to investigate potential pathways involved in breast cancer (Menashe et al. 2010). In this way, it is the cumulative significance of individual SNPs from a single pathway that determines significance. The analysis included more than 528,000 SNPs of nearly 4,000 genes involved in

421 pathways. There were 21 pathways with significant association with breast cancer. Of these pathways, the GH signaling pathway was ranked third in strength of association with breast cancer. To my knowledge, this type of analysis has not been conducted for other types of cancer.

One important caveat with SNP analysis is that the biological effect of the SNP is often unknown; does an individual SNP increase or decrease the activity of the affected gene or pathway?

Contradictory data

Like most fields of research, there is certainly contradictory data regarding GH and cancer. In a rat model of metastasizing prostate adenocarcinoma, GH treatment was shown to inhibit pulmonary metastasis while supporting host growth (Torosian & Donoway 1991). There has even been an effort over the years to use GH as an intervention in cancer patients preoperatively to prevent catabolism. Wolf and colleagues observed an increase in protein synthesis in a group of 16 cancer patients treated with rhGH for 3 days followed by euglycemic insulin clamp and administration of amino acids and glucose preoperatively (Wolf et al. 1992).

Patients receiving GH also had higher levels of insulin, IGF-1 and glucose than controls, making 47

it difficult to determine if the effect on anabolism was directly from GH action or due to other factors. This study was too small to adequately assess the safety of this treatment. In another study, athymic mice were injected with human MIA PaCa-2 pancreatic cancer cells and treated with growth hormone (Harrison et al. 1996). Tumor growth rate and final size did not change with GH treatment, while protein synthesis in the liver was increased. An in vitro study used 20 solid human tumors of various etiologies that were cultured and treated with rhGH (Fiebig et al.

2000). No cell lines from these tumors exhibited stimulation by GH under the treatment conditions. A major flaw of this study is the experiment was conducted with media that contained a high concentration of fetal calf serum, 20%, which contains a high level of growth stimulatory factors that would undoubtedly mask any potential growth-stimulatory effect of GH. This important limitation was not pointed out in any section of the manuscript for this study, whose last author’s affiliation was listed as “Pharmaceutical Consulting” and has no other GH-related publications listed in Pubmed. While some epidemiologic studies suggest an association between

GH and breast cancer risk, not all data is congruent. A prospective case-control study was performed on GH/IGF-1 and breast cancer risk in the Nurse’s Health Study II (Schernhammer et al. 2006). In this study, no association between circulating levels of GH, IGF-1, IGF-BP1, or

IGF-BP3 was detected. The lack of association with IGF-1 was surprising in the context of previous studies suggesting such an effect. There is also considerable debate over the potential for

GH therapy under approved indications to contribute to or cause carcinogenesis. Holly and Perks, in an article entitled “Growth hormone and cancer: are we asking the right questions?”, point out that the safety studies cited by industry have an average follow up of “just a few years”, and that most of the subjects are children or young adults (Holly & Perks 2006). They further point out that the major cancers in which GH has so far been implicated are breast, prostate and colorectal. 48

All have an incidence of near zero in teenagers and young adults, which means any impact of GH treatment may not be known for decades.

Summary

For many years, evidence has been accumulating that implicates endocrine hormones in carcinogenesis. GH, with a central role in metabolism and cell growth, is one of several hormones that have been examined. Nearly 40 years ago, GH was found in human tumors (Beck and Burger

1972). Indirect indicators of GH action during development, such as birth weight and adult height, have provide evidence that those with indicators of elevated GH action may be at increased risk of certain types of cancers (Okasha et al. 2002; Gunnell 2001). Recent clinical epidemiological research and prospective studies have supported the idea that elevated GH action is associated with increased risk for cancer (Renehan et al. 2004; Rowlands et al. 2009; Okamoto et al. 2010).

Animal studies have provided striking data on the GH pathway and carcinogenesis, with a consistent finding of reduced carcinogenesis in dwarf animals that have reduced GH signaling.

This is true for both oncogene induced and chemically induced breast and prostate tumorigenesis and includes studies using Spontaneous Dwarf Rats, Ghdr/dr rats, GHA-transgenic mice, lit/lit dwarf mice, and GHR-/- mice and GHA treatment of WT mice.

Studies have been conducted using cancer cell lines, human tumor biopsies and laboratory animals that express strong oncogenes or have been treated with chemical carcinogens.

Some of the first studies to examine GHR mRNA expression in cancer cells found expression in multiple breast cancer cell lines, 28 out of 29 human tumor biopsies (Decouvelaere et al. 1995), several prostate cancer cell lines and tumor biopsies (Untergasser et al. 1999, Chopin et al. 2002), as well as in meningioma specimens (Friend et al. 2001, Friend et al. 1999) and colorectal cancer 49

biopsies (Yang et al. 2004). Similarly, GHR protein expression has been shown in both breast cancer tissue samples and adjacent normal tissues (albeit at twice the concentration in cancerous tissue) (Gebre-Medhin et al. 2001), prostate cancer cell lines (Chopin et al. 2002), a neuroblastoma cell line (Grimbly et al. 2009) and colorectal cancer biopsies (Yang et al. 2004).

The capacity to express GH mRNA or protein has been shown in malignant breast cancer tissue

(Gregoraszczuk et al. 2001), prostate cancer cell lines (Chopin et al. 2002) and a neuroblastoma cell line (Grimbly et al. 2009). Expression of both GHR and GH in cancer cell lines and tumor tissues suggests the possibility that local autocrine/paracrine GH action could play a role in the process of neoplasia, tumor survival and maintenance, and/or metastasis (Gregoraszczuk et al.

2001).

A small number of studies have examined stimulation or inhibition of GHR in cancer cell lines. In the first study of its kind, breast cancer cell lines were treated with various hGH/hPRL analogs and hPRLR antagonists (Fuh & Wells 1995). It was concluded that stimulation of PRLR in breast cancer cell lines by PRL or GH analogs could stimulate proliferation, an effect that was reversed by co-treatment with G120R-hGH. In a much more recent study, GH was shown to increase the viability of MCF-7 cells, while treatment with the GHA pegvisomant could reverse this effect (Zatelli et al. 2009). Surprisingly, pegvisomant was able to reduce new cell growth when given alone, an indication that basal GHR activation is occurring or that pegvisomant has actions independent of GHR. Similarly, proliferation of primary cell lines from BPH samples has been shown to increase with GH treatment (Untergasser et al. 1999). GH treatment of bile duct cancer cell lines has also been shown to increase with GH treatment (Cai et al. 2008).

Limited studies have examined GH signaling in cell lines that are included in the NCI60 cancer cell line panel. There are 41 relevant publications that include studies of GH with one or more of these cell lines. Out of these 41 studies, 26 were conducted on 4 breast cancer cell lines. 50

Only 11 of the 60 cell lines are included in these publications (see appendix1). mRNA or protein expression of either GH or GHR was shown in 11 different publications covering 9 of the 60

NCI60 cell lines. A total of 7 of the NCI60 cell lines have been treated with GH (MCF7, T47D,

MDA-MB231, Molt7, HL60, PC3, and DU145). Some of the cell lines show increased proliferation with GH, while others do not. A single cell line has been directly treated with GHA

(MCF7), while 5 cell lines were implanted as xenografts and the animals were then subject to

GHA treatment (MCF7, MDA-MB231, MDA-MB435, COLO205 and HT29). As perhaps the most intensely studied groups of cell lines in the world, it is surprising that there is not more literature on GH signaling and cell lines of the NCI60 panel. On the up-side, the dearth of research in this area makes it an ideal platform from which to begin an investigation and generate novel and interesting data to advance the field of endocrine hormones and cancer.

Comment on PRLR signaling and cancer

In addition to GHR, PRLR is also able to be activated by binding GH. While an in-depth analysis of PRL signaling in cancer is beyond the scope of this dissertation, a brief discussion of data regarding PRLR in cancer is warranted. Several excellent reviews have been published on this topic (Bernichtein et al. 2010, Fernandez et al. 2010, Swaminathan et al. 2008). The involvement of PRLR activation is specifically implicated in breast, and to a lesser extent, prostate cancer. Expression of PRLR has been reported in human breast cancer cell line T47D, as well as the colon cancer line HT-29 (Nagano et al. 1995). Human tumor biopsies from breast and prostate cancer patients have been shown to express PRLR (Mertani et al. 1998, Touraine et al.

1998, Leav et al. 1999, Gill et al. 2001). High levels of circulating PRL have been associated with a 40% increased risk for breast cancer for women in the higher quartile relative to the lowest quartile (Tworoger et al. 2007), but no association between PRL levels and prostate cancer risk 51

has been detected (Stattin et al. 2001). The in vivo animal studies described above in which tumor incidence is severely reduced in GHR-/- animals, argues against a strong role of PRLR in cancer, at least in the absence of GH signaling. In GHR-/- mice, PRL levels are elevated (Chandrashekar et al. 1999) while PRLR remains intact. Thus, higher than normal PRLR activation is unable to overcome the cancer suppressive properties of GHR inactivation. While the focus of this dissertation will be on GH and GHR, some data in Chapter 2 will report expression of PRLR and

PRL in cancer cells.

Hypotheses and research objectives

What role does GH have in cancer biology? This fundamental question remains largely unanswered and there is an urgent need to research this topic. The National Cancer Institute’s

(NCI) Developmental Therapeutics Program (DTP) has developed a human cell panel of 60 different tumor-derived cell lines representative of nine types of cancer. I propose to utilize this cell panel to characterize GH signaling and determine the sensitivity of these cell lines to GH and to inhibition of GHR signaling.

The following hypotheses will guide this research:

Hypothesis 1: Cancer cells possess GH signaling capability

Hypothesis 2: Cancer cells possess autocrine/paracrine GH signaling

Hypothesis 3: GH will increase proliferation and affect cell signaling of cancer cells

Hypothesis 4: Treatment with GH will alter cancer cell metabolism

From these hypotheses, the following research objectives will be undertaken:

Research Objective 1: Produce recombinant GH and GHA proteins 52

Research Objective 2: Characterize mRNA expression of GHR, PRLR, GH and PRL in

human cell lines from the NCI60 cell line panel and perform COMPARE analysis

Research Objective 3: Analyze a subset of NCI60 cancer cell lines (that express GHR

and/or PRLR) for sensitivity to GH

Research Objective 4: Analyze cell signaling in cell lines that respond to GH treatment

Research Objective 5: Assess the effect of GH on metabolic parameters, including the

Warburg effect

Research approach

In the late 1980s, the US National Cancer Institute’s Developmental Therapeutics

Program (DTP) developed a human cell line panel known as the NCI60 to be used as an anticancer drug screen. The panel includes 60 human cell lines from nine tumor types: breast, prostate, CNS, leukemia, colon, lung, renal, ovarian and melanoma. The NCI60 cell panel is an ideal platform to investigate the role of GH in a broad range of human cancers. It is the most extensively studied collection of cell lines in existence (Shoemaker 2006). This panel has been tested for sensitivity to more than 100,000 compounds (Weinstein 2006). During the first decade of existence, the panel was used as a primary drug screen for the discovery of anti-cancer compounds. Following a comprehensive program review in 1997, the function of the NCI60 has transitioned to a research tool to support the cancer research community. Recently, microarray experiments have been conducted to provide gene expression data for the cell panel, including

Affymetrix HG-U95A and HG-U133A arrays (Shankavaram et al. 2007). All data are freely available in public databases. Importantly, NCI-DTP will provide researchers with RNA to allow characterization of potential molecular targets and will assist in data analysis to determine potential compounds that may modulate molecular target pathways. The COMPARE algorithm 53

will be applied to gene expression datasets. In COMPARE analysis, the expression pattern across the NCI60 panel is correlated with existing expression or other data in order to provide clues to the molecular mechanism of action of the compound (Zaharevitz et al. 2002). Sensitivity patterns to the more than 100,000 compounds that have been tested against the NCI60 cell panel can be compared to data from new expression experiments or new compound sensitivity (i.e. GHA) tests.

While the existing gene and protein expression data is valuable in and of itself, there are significant limitations in regards to GH signaling. There are 41 reports of GH/GHR expression and/or activation in cell lines that are included in the NCI60 panel, as summarized and referenced in appendix 1. Only 12 of the NCI-60 cell lines are reported to have been examined for GH and/or GHR by RT-PCR and/or immuno-assays (Decouvelaere et al. 1995, Costoya et al. 1996,

Giesbert et al. 1997, Costoya et al. 2000, Reiter et al. 1995, Untergasser et al. 1999, Chopin et al.

2002, Weiss-Messer et al. 2004, Bidosee et al. 2009, Nagano et al. 1995, Friend et al. 2001).

Only 6 of the cell lines are reported to have been treated with GH and/or GHA either directly or as zenografts (Fuh & Wells 1995, Divisova et al. 2006, Giesbert et al. 1997, Untergasser et al.

1999, Dagnaes-Hansen et al. 2004, Estrov et al. 1991). Of the investigations reported, 26 were conducted on breast cancer cell lines, six on leukemia lines, 5 on prostate, 3 on colon cells and 1 on a CNS cell line appendix 1. One could suggest use of the existing microarray data to select which of the NCI60 cell lines that might be responsive to GH. In reality, this does not appear to be a valid approach. The data shows nearly identical GHR gene expression across all 60 lines as derived from publicly available databases (Shankavaram et al. 2007), but clear differences in sensitivity to GH have been reported among cell lines. For example, the leukemia cell lines Molt4 and HL60 have both been shown to proliferate when treated with GH (Estrov et al. 1991,

Giesbert et al. 1997) while prostate cancer cell lines PC3 and DU145 have not been shown to 54

respond to GH treatment (Untergasser et al. 1999). All four cell lines show similar GHR expression based on the publicly available data (Shankavaram et al. 2007).

This research project has been conducted using the National Cancer Institute’s 60 human cancer cell line panel (NCI60), covering nine tumor types, including breast, prostate, CNS, leukemia, colon, lung, renal, ovarian and melanoma. All 60 cell lines have been assayed for mRNA expression of GHR, PRLR, GH and PRL. Recombinant human GH and GHA have been produced in E. coli cells and used to treat cancer cell lines. Gene expression data was used to select which cell lines were subject to hormone treatment. This subset of cell lines was treateded with multiple doses of GH and monitored for changes in proliferation. Several cell lines that responded to GH treatment were assayed to determine the activation status of signaling pathways.

Finally, expression data was subject to COMPARE analysis to determine if any of the 100,000+ compounds that have already been screened against the NCI60 cell panel may interact in some fashion with the GH/IGF-1 axis.

Limitations of the study

Cancer is a heterogeneous disease that occurs in a complex biological environment. This research did not utilize primary cell cultures or in vivo experiments. Cell lines continuously cultured in the laboratory for many years may not sufficiently represent the behavior of cells in an actual tumor. Thus, it is unknown if the results from these studies conducted in vitro will have relevance to in vivo cancer development and progression. Further limitations stem from the fact that each cell line that responds to GH/GHA may have a different sensitivity that could be difficult to assay accurately, particularly with regard to the single dose protocol for the high- throughput NCI DTP Cell Screen service. While the NCI60 cell panel is perhaps the most intensely studied samples in the world, by limiting this study to cell lines from this panel, other 55

cancer cell lines that may be sensitive to GH will be missed. Given these limitations, if a cell line does respond to a GH treatment by exhibiting growth, it sets the foundation for in vivo research and possibly clinical trials in the future for the use of GHA to treat the corresponding human cancer.

Value of the research

This study brings further understanding to the role of GH in carcinogenesis. More specifically, this research provides a comprehensive picture of the potential for GH signaling in the NCI60 cell lines. This research undoubtedly confirms previous work, while providing novel data from cancer cell lines that have never been examined for GH signaling or sensitivity. The results of this study could precipitate in vivo animal studies and eventually clinical investigations into the use of GHR inhibitors as a pharmacological treatment for cancer. And finally, data from these studies will be applied to algorithms tested against public NCI60 databases that could identify compounds that interact with GHR signaling, potentially providing a foundation for the development of novel biomarkers and/or pharmaceutical therapeutics that may one day be used to treat human disease.

Publication of data and dissertation outline

Three manuscripts have been prepared for submission to relevant scientific journals. The first manuscript (Chapter 2) describes the mRNA expression of GHR, PRLR, GH and PRL in the

NCI60 panel. The second manuscript (Chapter 3) presents the response of several cell lines to treatment with GH. Finally, the third manuscript (Chapter 4) reports the effect of GH on metabolic parameters. These manuscripts, in addition to this first chapter and a final section on key findings and future research, make up the dissertation. 56

References

American Cancer Society 2013 Cancer Facts & Figures 2013.

Anzo M, Cobb LJ, Hwang DL, Mehta H, Said JW, Yakar S, LeRoith D & Cohen P 2008 Targeted deletion of hepatic Igf1 in TRAMP mice leads to dramatic alterations in the circulating insulin-like growth factor axis but does not reduce tumor progression. Cancer research 68 3342-3349.

Beck C & Burger HG 1972 Evidence for the presence of immunoreactive growth hormone in cancers of the lung and stomach. Cancer 30 75-79.

Bernichtein S, Touraine P & Goffin V 2010 New concepts in prolactin biology. The Journal of endocrinology 206 1-11. (doi:10.1677/JOE-10-0069; 10.1677/JOE-10-0069).

Bidosee M, Karry R, Weiss-Messer E & Barkey RJ 2009 Regulation of growth hormone receptors in human prostate cancer cell lines. Molecular and cellular endocrinology 309 82- 92.

Brown RJ, Adams JJ, Pelekanos RA, Wan Y, McKinstry WJ, Palethorpe K, Seeber RM, Monks TA, Eidne KA, Parker MW & Waters MJ 2005 Model for growth hormone receptor activation based on subunit rotation within a receptor dimer. Nature structural & molecular biology 12 814-821. (doi:10.1038/nsmb977).

Cai HH, Sun YM, Bai JF, Shi Y, Zhao HL & Miao Y 2008 Relationship between the GH-IGFs axis and the proliferation of bile duct cancer cell line QBC939 in vitro. Hepatobiliary \& pancreatic diseases international : HBPD INT 7 76-81.

Chandrashekar V, Bartke A, Coschigano KT & Kopchick JJ 1999 Pituitary and testicular function in growth hormone receptor gene knockout mice. Endocrinology 140 1082-1088.

Chen EY, Liao YC, Smith DH, Barrera-Saldana HA, Gelinas RE & Seeburg PH 1989 The human growth hormone locus: nucleotide sequence, biology, and evolution. Genomics 4 479-497.

Chen WY, Wight DC, Mehta BV, Wagner TE & Kopchick JJ 1991 Glycine 119 of bovine growth hormone is critical for growth-promoting activity. Molecular endocrinology (Baltimore, Md.) 5 1845-1852.

Chhabra Y, Waters MJ & Brooks AJ 2011 Role of the growth hormone–IGF-1 axis in cancer. Expert Review of Endocrinology & Metabolism 6 71 -84. (doi:10.1586/eem.10.73).

Chopin LK, Veveris-Lowe TL, Philipps AF & Herington AC 2002 Co-expression of GH and GHR isoforms in prostate cancer cell lines. Growth hormone \& IGF research 12 126-136. 58

Cooke NE, Coit D, Shine J, Baxter JD & Martial JA 1981 Human prolactin. cDNA structural analysis and evolutionary comparisons. The Journal of biological chemistry 256 4007-4016.

Costoya JA, Rios R, Garcia-Barros M, Gallego R, Garcia-Caballero T, Senaris R, Arce VM & Devesa J 2000 Role of growth hormone receptor in HL-60 cell survival. Molecular cell biology research communications : MCBRC 4 26-31. (doi:10.1006/mcbr.2000.0252).

Costoya JA, Vidal A, Garcia-Barros M, Arce V & Devesa J 1996 Expression of the human growth hormone normal gene (hGH-N) in proliferating and differentiated HL-60 cells. Experimental cell research 228 164-167.

Dagnaes-Hansen F, Duan H, Rasmussen LM, Friend KE & Flyvbjerg A 2004 Growth hormone receptor antagonist administration inhibits growth of human colorectal carcinoma in nude mice. Anticancer Research 24 3735-3742.

Daughaday WH, Hall K, Raben MS, Salmon WD,Jr, van den Brande JL & van Wyk JJ 1972 Somatomedin: proposed designation for sulphation factor. Nature 235 107.

Decouvelaere C, Peyrat JP, Bonneterre J, Djiane J & Jammes H 1995 Presence of the two growth hormone receptor messenger RNA isoforms in human breast cancer. Cell growth \& differentiation : the molecular biology journal of the American Association for Cancer Research 6 477-483.

Devarajan E & Huang S 2009 STAT3 as a central regulator of tumor metastases. Current Molecular Medicine 9 626-633.

Emerman JT, Leahy M, Gout PW & Bruchovsky N 1985 Elevated growth hormone levels in sera from breast cancer patients. Hormone and metabolic research = Hormon- und Stoffwechselforschung = Hormones et metabolisme 17 421-424. (doi:10.1055/s-2007- 1013563).

Estrov Z, Meir R, Barak Y, Zaizov R & Zadik Z 1991 Human growth hormone and insulin-like growth factor-1 enhance the proliferation of human leukemic blasts. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 9 394-399.

Fernandez I, Touraine P & Goffin V 2010 Prolactin and human tumourogenesis. Journal of neuroendocrinology 22 771-777. (doi:10.1111/j.1365-2826.2010.02011.x; 10.1111/j.1365- 2826.2010.02011.x). 59

Fiebig HH, Dengler W & Hendriks HR 2000 No evidence of tumor growth stimulation in human tumors in vitro following treatment with recombinant human growth hormone. Anti-Cancer Drugs 11 659-664.

Friend KE, Khandwala HM, Flyvbjerg A, Hill H, Li J & McCutcheon IE 2001 Growth hormone and insulin-like growth factor-I: effects on the growth of glioma cell lines. Growth hormone & IGF research : official journal of the Growth Hormone Research Society and the International IGF Research Society 11 84-91. (doi:10.1054/ghir.2000.0183).

Friend KE, Radinsky R & McCutcheon IE 1999 Growth hormone receptor expression and function in meningiomas: effect of a specific receptor antagonist. Journal of neurosurgery 91 93-99.

Fuh G & Wells JA 1995 Prolactin receptor antagonists that inhibit the growth of breast cancer cell lines. The Journal of biological chemistry 270 13133-13137.

Gahete MD, Duran-Prado M, Luque RM, Martinez-Fuentes AJ, Quintero A, Gutierrez-Pascual E, Cordoba-Chacon J, Malagon MM, Gracia-Navarro F & Castano JP 2009 Understanding the multifactorial control of growth hormone release by somatotropes: lessons from comparative endocrinology. Annals of the New York Academy of Sciences 1163 137-153. (doi:10.1111/j.1749-6632.2008.03660.x).

Gebre-Medhin M, Kindblom LG, Wennbo H, T\ornell J & Meis-Kindblom JM 2001 Growth hormone receptor is expressed in human breast cancer. The American journal of pathology 158 1217-1222.

Giesbert S, Panzer S, Kovar H, Fischer S, Printz D, Gadner H & Panzer-Grumayer ER 1997 Acute leukemias express a functional receptor for the human growth hormone. Annals of Hematology 74 253-257.

Gill S, Peston D, Vonderhaar BK & Shousha S 2001 Expression of prolactin receptors in normal, benign, and malignant breast tissue: an immunohistological study. Journal of clinical pathology 54 956-960.

Goffin V, Bernichtein S, Carri\`ere O, Bennett WF, Kopchick JJ & Kelly PA 1999 The human growth hormone antagonist B2036 does not interact with the prolactin receptor. Endocrinology 140 3853-3856.

Goffin V & Kelly PA 1997 The prolactin/growth hormone receptor family: structure/function relationships. Journal of mammary gland biology and neoplasia 2 7-17.

Goffin V, Shiverick KT, Kelly PA & Martial JA 1996 Sequence-function relationships within the expanding family of prolactin, growth hormone, placental lactogen, and related proteins in mammals. Endocrine reviews 17 385-410. 60

Gregoraszczuk EL, Milewicz T, Kolodziejczyk J, Krzysiek J, Basta A, Sztefko K, Kurek S & Stachura J 2001 Progesterone-induced secretion of growth hormone, insulin-like growth factor I and prolactin by human breast cancer explants. Gynecological endocrinology : the official journal of the International Society of Gynecological Endocrinology 15 251-258.

Grimbly C, Martin B, Karpinski E & Harvey S 2009 Growth hormone production and action in N1E-115 neuroblastoma cells. Journal of molecular neuroscience : MN 39 117-124.

Guevara-Aguirre J, Balasubramanian P, Guevara-Aguirre M, Wei M, Madia F, Cheng CW, Hwang D, Martin-Montalvo A, Saavedra J, Ingles S, de Cabo R, Cohen P & Longo VD 2011 Growth hormone receptor deficiency is associated with a major reduction in pro-aging signaling, cancer, and diabetes in humans. Science translational medicine 3 70ra13. (doi:10.1126/scitranslmed.3001845; 10.1126/scitranslmed.3001845).

Gunnell D, Okasha M, Smith GD, Oliver SE, Sandhu J & Holly JM 2001 Height, leg length, and cancer risk: a systematic review. Epidemiologic reviews 23 313-342.

Hanahan D & Weinberg RA 2000 The hallmarks of cancer. Cell 100 57-70. (doi: 10.1016/S0092- 8674(00)81683-9).

Hanahan D & Weinberg RA 2011 Hallmarks of cancer: the next generation. Cell 144 646-674. (doi:10.1016/j.cell.2011.02.013; 10.1016/j.cell.2011.02.013).

Harrison LE, Blumberg D, Berman R, Ng B, Hochwald S, Brennan MF & Burt M 1996 Effect of human growth hormone on human pancreatic carcinoma growth, protein, and cell cycle kinetics. The Journal of surgical research 61 317-322.

Harvey S 2010 Extrapituitary growth hormone. Endocrine 38 335-359. (doi:10.1007/s12020-010- 9403-8; 10.1007/s12020-010-9403-8).

Hayashi AA & Proud CG 2007 The rapid activation of protein synthesis by growth hormone requires signaling through mTOR. American journal of physiology.Endocrinology and metabolism 292 E1647-55. (doi:10.1152/ajpendo.00674.2006).

Higham CE & Trainer PJ 2008 Growth hormone excess and the development of growth hormone receptor antagonists. Experimental physiology 93 1157-1169 (doi:10.1113/expphysiol.2008.042515).

Holly JM & Perks CM 2008 Cancer as an endocrine problem. Best practice & research.Clinical endocrinology & metabolism 22 539-550. (doi:10.1016/j.beem.2008.07.007).

Holly J & Perks C 2006 Growth hormone and cancer: are we asking the right questions?*. Clinical endocrinology 64 122-124. 61

Kaabi YA 2012 Growth hormone and its receptor: a molecular insight. Saudi Journal for Health Sciences 1 61-68

Kaplan SA & Cohen P 2007 The somatomedin hypothesis 2007: 50 years later. The Journal of clinical endocrinology and metabolism 92 4529-4535.

Kaulsay KK, Mertani HC, Tornell J, Morel G, Lee KO & Lobie PE 1999 Autocrine stimulation of human mammary carcinoma cell proliferation by human growth hormone. Experimental cell research 250 35-50. (doi:10.1006/excr.1999.4492).

Klapper DG, Svoboda ME & Van Wyk JJ 1983 Sequence analysis of somatomedin-C: confirmation of identity with insulin-like growth factor I. Endocrinology 112 2215-2217.

Kopchick JJ, Parkinson C, Stevens EC & Trainer PJ 2002 Growth hormone receptor antagonists: discovery, development, and use in patients with acromegaly. Endocrine reviews 23 623- 646.

Koppenol WH, Bounds PL & Dang CV 2011 Otto Warburg's contributions to current concepts of cancer metabolism. Nature reviews.Cancer 11 325-337. (doi:10.1038/nrc3038; 10.1038/nrc3038).

Lanning NJ & Carter-Su C 2006 Recent advances in growth hormone signaling. Reviews in endocrine & metabolic disorders 7 225-235.

Lawlor DA, Okasha M, Gunnell D, Smith GD & Ebrahim S 2003 Associations of adult measures of childhood growth with breast cancer: findings from the British Women's Heart and Health Study. British journal of cancer 89 81-87. (doi:10.1038/sj.bjc.6600972).

Le Roith D, Bondy C, Yakar S, Liu JL & Butler A 2001 The somatomedin hypothesis: 2001. Endocrine reviews 22 53-74.

Leav I, Merk FB, Lee KF, Loda M, Mandoki M, McNeal JE & Ho SM 1999 Prolactin receptor expression in the developing human prostate and in hyperplastic, dysplastic, and neoplastic lesions. The American journal of pathology 154 863-870. (doi:10.1016/S0002- 9440(10)65333-3).

Lincoln DT, Sinowatz F, Temmim-Baker L, Baker HI, Kolle S & Waters MJ 1998 Growth hormone receptor expression in the nucleus and cytoplasm of normal and neoplastic cells. Histochemistry and cell biology 109 141-159.

Liu N, Mertani HC, Norstedt G, Tornell J & Lobie PE 1997 Mode of the autocrine/paracrine mechanism of growth hormone action. Experimental cell research 237 196-206. (doi:10.1006/excr.1997.3789). 62

Love DW, Whatmore AJ, Clayton PE & Silva CM 1998 Growth hormone stimulation of the mitogen-activated protein kinase pathway is cell type specific. Endocrinology 139 1965- 1971.

Majeed N, Blouin MJ, Kaplan-Lefko PJ, Barry-Shaw J, Greenberg NM, Gaudreau P, Bismar TA & Pollak M 2005 A germ line mutation that delays prostate cancer progression and prolongs survival in a murine prostate cancer model. Oncogene 24 4736-4740.

Menashe I, Maeder D, Garcia-Closas M, Figueroa JD, Bhattacharjee S, Rotunno M, Kraft P, Hunter DJ, Chanock SJ, Rosenberg PS & Chatterjee N 2010 Pathway Analysis of Breast Cancer Genome-Wide Association Study Highlights Three Pathways and One Canonical Signaling Cascade. Cancer research (doi:10.1158/0008-5472.CAN-09-4502).

Mertani HC, Garcia-Caballero T, Lambert A, Gerard F, Palayer C, Boutin JM, Vonderhaar BK, Waters MJ, Lobie PE & Morel G 1998 Cellular expression of growth hormone and prolactin receptors in human breast disorders. International journal of cancer.Journal international du cancer 79 202-211.

Mojarrad M, Momeny M, Mansuri F, Abdolazimi Y, Tabrizi MHH, Ghaffari SHH, Tavangar SMM & Modarressi MHH 2009 Autocrine human growth hormone expression leads to resistance of MCF-7 cells to tamoxifen. Medical oncology (Northwood, London, England) .

Mol JA, van Garderen E, Rutteman GR & Rijnberk A 1996 New insights in the molecular mechanism of progestin-induced proliferation of mammary epithelium: induction of the local biosynthesis of growth hormone (GH) in the mammary glands of dogs, cats and humans. The Journal of steroid biochemistry and molecular biology 57 67-71.

Mulhall C, Hegele RA, Cao H, Tritchler D, Yaffe M & Boyd NF 2005 Pituitary growth hormone and growth hormone-releasing hormone receptor genes and associations with mammographic measures and serum growth hormone. Cancer epidemiology, biomarkers \& prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology 14 2648-2654.

Nagano M, Chastre E, Choquet A, Bara J, Gespach C & Kelly PA 1995 Expression of prolactin and growth hormone receptor genes and their isoforms in the gastrointestinal tract. The American Journal of Physiology 268 G431-42.

Okamoto I, Munakata M, Miyazaki M, Satoh T, Takahata T, Takamatsu Y, Muto O, Koike K, Ishitani K, Mukaiyama T, Sakata Y, Nakagawa K & Tamura K 2010 Disturbance of the growth hormone-insulin-like growth factor-1 axis associated with poor performance status in patients with solid tumors. Japanese journal of clinical oncology 40 222-226.

Okasha M, Gunnell D, Holly J & Davey Smith G 2002 Childhood growth and adult cancer. Best practice & research.Clinical endocrinology & metabolism 16 225-241. (doi:10.1053/beem.2002.0204). 63

Pollak M, Blouin MJ, Zhang JC & Kopchick JJ 2001 Reduced mammary gland carcinogenesis in transgenic mice expressing a growth hormone antagonist. British journal of cancer 85 428- 430.

Ratkaj I, Stajduhar E, Vucinic S, Spaventi S, Bosnjak H, Pavelic K & Pavelic SK 2010 Integrated gene networks in breast cancer development. Functional \& integrative genomics 10 11-19.

Reiter E, Kecha O, Hennuy B, Lardinois S, Klug M, Bruyninx M, Closset J & Hennen G 1995 Growth hormone directly affects the function of the different lobes of the rat prostate. Endocrinology 136 3338-3345.

Renehan AG & Brennan BM 2008 Acromegaly, growth hormone and cancer risk. Best practice & research.Clinical endocrinology & metabolism 22 639-657. (doi:10.1016/j.beem.2008.08.011).

Renehan AG, Zwahlen M, Minder C, O'Dwyer ST, Shalet SM & Egger M 2004 Insulin-like growth factor (IGF)-I, IGF binding protein-3, and cancer risk: systematic review and meta- regression analysis. Lancet 363 1346-1353. (doi:10.1016/S0140-6736(04)16044-3).

Roberts PJ & Der CJ 2007 Targeting the Raf-MEK-ERK mitogen-activated protein kinase cascade for the treatment of cancer. Oncogene 26 3291-3310. (doi:10.1038/sj.onc.1210422).

Rowlands MA, Gunnell D, Harris R, Vatten LJ, Holly JM & Martin RM 2009 Circulating insulin-like growth factor peptides and prostate cancer risk: a systematic review and meta- analysis. International journal of cancer.Journal international du cancer 124 2416-2429. (doi:10.1002/ijc.24202).

Rudd MF, Webb EL, Matakidou A, Sellick GS, Williams RD, Bridle H, Eisen T, Houlston RS & GELCAPS Consortium 2006 Variants in the GH-IGF axis confer susceptibility to lung cancer. Genome research 16 693-701.

Schernhammer ES, Holly JM, Hunter DJ, Pollak MN & Hankinson SE 2006 Insulin-like growth factor-I, its binding proteins (IGFBP-1 and IGFBP-3), and growth hormone and breast cancer risk in The Nurses Health Study II. Endocrine-related cancer 13 583-592.

Shankavaram UT, Reinhold WC, Nishizuka S, Major S, Morita D, Chary KK, Reimers MA, Scherf U, Kahn A, Dolginow D, Cossman J, Kaldjian EP, Scudiero DA, Petricoin E, Liotta L, Lee JK & Weinstein JN 2007 Transcript and protein expression profiles of the NCI-60 cancer cell panel: an integromic microarray study. Molecular cancer therapeutics 6 820- 832. (doi:10.1158/1535-7163.MCT-06-0650).

Shen Q, Lantvit DD, Lin Q, Li Y, Christov K, Wang Z, Unterman TG, Mehta RG & Swanson SM 2007 Advanced rat mammary cancers are growth hormone dependent. Endocrinology 148 4536-4544. (doi:10.1210/en.2007-0513). 64

Shoemaker RH 2006 The NCI60 human tumour cell line anticancer drug screen. Nature reviews.Cancer 6 813-823. (doi:10.1038/nrc1951).

Stattin P, Rinaldi S, Stenman UH, Riboli E, Hallmans G, Bergh A & Kaaks R 2001 Plasma prolactin and prostate cancer risk: A prospective study. International journal of cancer.Journal international du cancer 92 463-465.

Strous GJ & Gent J 2002 Dimerization, ubiquitylation and endocytosis go together in growth hormone receptor function. FEBS letters 529 102-109.

Swaminathan G, Varghese B & Fuchs SY 2008 Regulation of prolactin receptor levels and activity in breast cancer. Journal of mammary gland biology and neoplasia 13 81-91. (doi:10.1007/s10911-008-9068-6; 10.1007/s10911-008-9068-6).

Swanson SM & Unterman TG 2002 The growth hormone-deficient Spontaneous Dwarf rat is resistant to chemically induced mammary carcinogenesis. Carcinogenesis 23 977-982.

Swerdlow AJ, Higgins CD, Adlard P & Preece MA 2002 Risk of cancer in patients treated with human pituitary growth hormone in the UK, 1959-85: a cohort study. Lancet 360 273-277.

Taubes G 2012a Cancer research. Unraveling the obesity-cancer connection. Science (New York, N.Y.) 335 28, 30-2. (doi:10.1126/science.335.6064.28; 10.1126/science.335.6064.28).

Taubes G 2012b Cancer research. Cancer prevention with a diabetes pill? Science (New York, N.Y.) 335 29. (doi:10.1126/science.335.6064.29; 10.1126/science.335.6064.29).

Tallet E, Rouet V, Jomain JB, Kelly PA, Bernichtein S & Goffin V 2008 Rational design of competitive prolactin/growth hormone receptor antagonists. Journal of mammary gland biology and neoplasia 13 105-117.

Thordarson G, Semaan S, Low C, Ochoa D, Leong H, Rajkumar L, Guzman RC, Nandi S & Talamantes F 2004 Mammary tumorigenesis in growth hormone deficient spontaneous dwarf rats; effects of hormonal treatments. Breast cancer research and treatment 87 277- 290.

Torosian MH & Donoway RB 1991 Growth hormone inhibits tumor metastasis. Cancer 67 2280- 2283.

Touraine P, Martini JF, Zafrani B, Durand JC, Labaille F, Malet C, Nicolas A, Trivin C, Postel- Vinay MC, Kuttenn F & Kelly PA 1998 Increased expression of prolactin receptor gene assessed by quantitative polymerase chain reaction in human breast tumors versus normal breast tissues. The Journal of clinical endocrinology and metabolism 83 667-674.

Trainer PJ, Drake WM, Katznelson L, Freda PU, Herman-Bonert V, van der Lely AJ, Dimaraki EV, Stewart PM, Friend KE, Vance ML, Besser GM, Scarlett JA, Thorner MO, Parkinson 65

C, Klibanski A, Powell JS, Barkan AL, Sheppard MC, Malsonado M, Rose DR, Clemmons DR, Johannsson G, Bengtsson BA, Stavrou S, Kleinberg DL, Cook DM, Phillips LS, Bidlingmaier M, Strasburger CJ, Hackett S, Zib K, Bennett WF & Davis RJ 2000 Treatment of acromegaly with the growth hormone-receptor antagonist pegvisomant. The New England journal of medicine 342 1171-1177.

Tworoger SS, Eliassen AH, Sluss P & Hankinson SE 2007 A prospective study of plasma prolactin concentrations and risk of premenopausal and postmenopausal breast cancer. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 25 1482-1488. (doi:10.1200/JCO.2006.07.6356).

Untergasser G, Rumpold H, Hermann M, Dirnhofer S, Jilg G & Berger P 1999 Proliferative disorders of the aging human prostate: involvement of protein hormones and their receptors. Experimental gerontology 34 275-287. van der Lely AJ, Hutson RK, Trainer PJ, Besser GM, Barkan AL, Katznelson L, Klibanski A, Herman-Bonert V, Melmed S, Vance ML, Freda PU, Stewart PM, Friend KE, Clemmons DR, Johannsson G, Stavrou S, Cook DM, Phillips LS, Strasburger CJ, Hackett S, Zib KA, Davis RJ, Scarlett JA & Thorner MO 2001 Long-term treatment of acromegaly with pegvisomant, a growth hormone receptor antagonist. Lancet 358 1754-1759.

VanGilder JC & Goldenberg IS 1975 Hypophysectomy in metastatic breast cancer. Archives of surgery (Chicago, Ill.: 1960) 110 293-295.

Vineis P 2003 Cancer as an evolutionary process at the cell level: an epidemiological perspective. Carcinogenesis 24 1-6.

Wang Z, Luque RM, Kineman RD, Ray VH, Christov KT, Lantvit DD, Shirai T, Hedayat S, Unterman TG, Bosland MC, Prins GS & Swanson SM 2008 Disruption of growth hormone signaling retards prostate carcinogenesis in the Probasin/TAg rat. Endocrinology 149 1366- 1376.

Wang Z, Prins GS, Coschigano KT, Kopchick JJ, Green JE, Ray VH, Hedayat S, Christov KT, Unterman TG & Swanson SM 2005 Disruption of Growth Hormone Signaling Retards Early Stages of Prostate Carcinogenesis in the C3(1)/T Antigen Mouse. Endocrinology 146 5188- 5196.

Waters MJ & Conway-Campbell BL 2004 The oncogenic potential of autocrine human growth hormone in breast cancer. Proceedings of the National Academy of Sciences of the United States of America 101 14992-14993.

Weinstein JN 2006 Spotlight on molecular profiling: "Integromic" analysis of the NCI-60 cancer cell lines. Molecular cancer therapeutics 5 2601-2605. (doi:10.1158/1535-7163.MCT-06- 0640). 66

Wolf RF, Pearlstone DB, Newman E, Heslin MJ, Gonenne A, Burt ME & Brennan MF 1992 Growth hormone and insulin reverse net whole body and skeletal muscle protein catabolism in cancer patients. Annals of Surgery 216 280-8; discussion 288-90.

Yakar S, Leroith D & Brodt P 2005 The role of the growth hormone/insulin-like growth factor axis in tumor growth and progression: Lessons from animal models. Cytokine \& growth factor reviews 16 407-420.

Yang XF, Beamer WG, Huynh H & Pollak M 1996 Reduced growth of human breast cancer xenografts in hosts homozygous for the lit mutation. Cancer research 56 1509-1511.

Yang X, Liu F, Xu Z, Chen C, Li G, Wu X & Li J 2004 Growth hormone receptor expression in human colorectal cancer. Digestive diseases and sciences 49 1493-1498.

Yoo EJ, Cajiao I, Kim JS, Kimura AP, Zhang A, Cooke NE & Liebhaber SA 2006 Tissue- specific chromatin modifications at a multigene locus generate asymmetric transcriptional interactions. Molecular and cellular biology 26 5569-5579. (doi:10.1128/MCB.00405-06).

Zaharevitz DW, Holbeck SL, Bowerman C & Svetlik PA 2002 COMPARE: a web accessible tool for investigating mechanisms of cell growth inhibition. Journal of molecular graphics & modelling 20 297-303.

Zatelli MCC, Minoia M, Mol\`e Daniela, Cason V, Tagliati F, Margutti A, Bondanelli M, Ambrosio MRR & degli Uberti E 2009 Growth hormone excess promotes breast cancer chemoresistance. The Journal of clinical endocrinology and metabolism 94 3931-3938.

Zhang X, Mehta RG, Lantvit DD, Coschigano KT, Kopchick JJ, Green JE, Hedayat S, Christov KT, Ray VH, Unterman TG & Swanson SM 2007 Inhibition of estrogen-independent mammary carcinogenesis by disruption of growth hormone signaling. Carcinogenesis 28 143-150.

CHAPTER 2: EXPRESSION OF GROWTH HORMONE, PROLACTIN, AND

COGNATE RECEPTORS IN THE NCI60 CANCER CELL LINE PANEL

Elahu S Gosney1,2, Riia K Junnila1, John J Kopchick1,2,3

1Edison Biotechnology Institute, Ohio University, 1 Water Tower Drive, Athens, Ohio

2Molecular and Cellular Biology Program, Dept of Biological Sciences, Ohio University, Athens

Ohio

3Dept of Biomedical Sciences, Heritage College of Osteopathic Medicine, Ohio University,

Athens, Ohio

Corresponding author: John J Kopchick, 1 Water Tower Drive, Athens, Ohio 45701; [email protected]

Short title: GHR, PRLR, GH and PRL mRNA in NCI60 Cell Lines

Keywords: Growth hormone receptor, melanoma, growth hormone, prolactin, breast cancer

Abstract

Accumulating evidence implicates growth hormone receptor (GHR) action in carcinogenesis. While multiple studies show evidence for expression of GH and GHR mRNA and protein in human cancer tissue, only a few cancer types have been investigated and there is a lack of GH and GHR quantification in these studies. The National Cancer Institute Developmental

Therapeutics Program has a panel of 60 cancer cell lines that has been extensively studied. This panel includes nine types of human cancer: breast, CNS, colon, leukemia, melanoma, non-small 68

cell (NSC) lung, ovarian, prostate and renal. We utilized this panel to assess expression of GHR,

GH, prolactin receptor (PRLR) and prolactin (PRL) mRNA by real-time qPCR. Both GHR and

PRLR show a broad range of expression both within and among most cancer types. Of the nine cancer types examined, melanoma stood out with universally high expression of GHR.

Expression of GH and PRL was barely detectable in most cell lines and when it was detected, the levels were low. Utilizing the existing NCI60 databases, a COMPARE analysis revealed a negative correlation between GHR mRNA expression and sensitivity to tubulin-inhibiting compounds and a positive correlation with sensitivity to the mTOR inhibitor rapamycin. A panel of cDNA from human metastatic melanoma biopsies confirmed GHR gene expression in melanoma tissue in vivo. In these human biopsies, the level of both GHR and insulin-like growth factor 1 receptor (IGF1R) mRNA was elevated in advanced stage IV tumor samples compared to stage III. In conclusion, we have identified high levels of GHR mRNA expression in several melanoma cell lines and a broad range of expression in human melanoma tumor biopsies. Based on this data, the GH pathway could be an important therapeutic target in melanoma.

Introduction

Growth hormone (GH) is secreted by the somatotrophs of the anterior pituitary. In addition, GH can be expressed in other cell types, such as lymphocytes and neurons, where it may act in an autocrine and/or paracrine manner (Harvey 2010, Wu et al. 2011). The GH gene family, composed of five highly homologous genes, includes a GH variant (GH2) that is best known for its expression in placental tissue (Yoo et al. 2006). Once GH binds and activates its receptor

(GHR), downstream signaling events occur that regulate a variety of biological processes and functions. While GH was once thought to act primarily through a key downstream effector, insulin-like growth factor 1 (IGF-1), there is clear evidence that GH can have direct effects 69

independent of IGF-1, for example the promotion of insulin resistance (Moller & Jorgensen

2009). The pleiotropic effects of GH-induced intracellular signaling are initiated by modulation of multiple pathways that include JAK/STAT, PI3K/AKT and MAPK (Chhabra et al. 2011). In addition to its effects through GHR, human (h) GH can also initiate signaling via the prolactin receptor (PRLR), which has also been implicated in carcinogenesis (Bernichtein et al. 2010).

Collectively, these pathways are known to play a role in mitogenic, anti-apoptotic and metabolic actions of GH; thus, the direct actions of GH could be involved in key processes of carcinogenesis (Chhabra et al. 2011, Cohen et al. 2000, Perry et al. 2006, Clayton et al. 2011).

It is increasingly recognized that cancer cells not only respond to their hormonal environment but also have the ability to directly alter it by producing hormones themselves

(Holly & Perks 2008). While the role of GH in somatic growth is well established, its function in the initiation and promotion of tumor growth remains to be elucidated. Several lines of evidence from human and animal studies have converged over the past 40 years to support a role of GH in carcinogenesis. Early evidence was reported in 1972, when a report described the presence of GH in lung and stomach cancer biopsies (Beck & Burger 1972). Shortly thereafter, pituitary ablation was shown to increase the effectiveness of treatment for advanced breast cancer (VanGilder &

Goldenberg 1975, Bundred et al. 1986). Indirect indicators of GH action, such as higher birth weight and adult height, positively associate with cancer risk and mortality (Gunnell et al. 2001,

Lawlor et al. 2003). Patients who have a condition of GH excess, acromegaly, have a higher incidence of colon polyps (Martino et al. 2004) and appear to have an increased risk of developing colon cancer. The risk of other malignancies may also be increased; however, this result is controversial (Renehan & Brennan 2008, Jenkins 2006). At the other physiological extreme of GH action, Laron Syndrome patients have an inherited mutation that inactivates GHR, resulting in GH insensitivity and low IGF-1 levels (Laron & Kopchick 2011). These patients very 70

rarely develop tumors and do not experience cancer mortality (Guevara-Aguirre et al. 2011,

Shevah & Laron 2007). Further evidence implicating GH in cancer has come from a large genome-wide association study, which found that the GH induced signaling pathway has the third highest association with breast cancer risk out of 421 pathways studied (Menashe et al. 2010).

These data from human studies are supported by numerous animal studies, which show a decrease in cancer incidence and tumor burden in several rodent lines with reduced GH action

(Ikeno et al. 2009, Yang et al. 1996, Pollak et al. 2001, Majeed et al. 2005, Wang et al. 2008,

Swanson & Unterman 2002). Collectively, these data strongly suggest a role for GH in cancer.

Multiple studies have shown both GH and GHR mRNA and protein expression in human cancer biopsies and explants, such as breast, prostate and colorectal cancer tissues (Wu et al.

2011, Decouvelaere et al. 1995, Lincoln et al. 1998, Mertani et al. 1998, Gebre-Medhin et al.

2001, Weiss-Messer et al. 2004). Generally, these studies have detected gene expression by means of non-quantitative RT-PCR and/or immunohistochemistry (IHC). A lack of quantification and the limited number of cancer types investigated to date has left many questions unanswered.

The present study was designed to quantitatively assess expression of GHR, PRLR, GH and PRL mRNA in multiple human cell lines that represent several types of cancer. To accomplish this, we have selected to utilize the NCI60 human tumor cell line panel, a group of cell lines assembled by the National Cancer Institute Developmental Therapeutics Program. This intensely studied group of cancer cell lines represents nine types of human cancer: breast, CNS, colon, leukemia, melanoma, non-small cell (NSC) lung, ovarian, prostate and renal (Weinstein 2006, Shoemaker

2006). We have quantified the expression of GHR, PRLR, GH and PRL mRNA across the NCI60 panel. The results have been examined for correlations to existing NCI60 data to reveal potential novel relationships to expression of other genes and sensitivity to anti-cancer compounds. The 71

data presented here lay the groundwork for future studies that may add to the understanding of the role of GH and PRL signaling in carcinogenesis.

Materials and methods

NCI60 cancer cell line RNA samples

Total RNA from the 60 human cancer cell lines included in the NCI60 panel was provided by the National Cancer Institute’s Development Therapeutics Program (table 2.1). RNA quality and quantity was assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies) by the Ohio University Genomics Facility (Athens, Ohio).

Real-time RT qPCR

Five µg of total RNA was reverse transcribed to cDNA using the Maxima First Strand cDNA Synthesis Kit for RT-qPCR as per the manufacturer’s protocol (Thermo Scientific K1642).

All PCR reactions were performed with the same batch of cDNA, which was diluted 1:10 prior to use. Each PCR reaction contained 10 µl of diluted cDNA. A Maxima SYBR Green/Fluorescein qPCR Master Mix (2X) (Thermo Scientific K0242) was used to perform real-time RT qPCR with an iCycler iQ Real-Time PCR Detection System (Bio-Rad). All reactions were performed in duplicate in a final volume of 25 µl. Cycling parameters were as per the manufacturer’s protocol, with the exception of the annealing temperature (see table 2.2), which was optimized for each primer pair. Additionally, the extension steps were reduced to 15 seconds for all primer pairs except GHR, which had a 30 second extension. In order to include all 60 samples in duplicate, two 96-well plates were used to assay each gene. To normalize for plate-to-plate variability, four internal reference control (IRC) reactions were performed on each plate in which a 1:100 cDNA 72

dilution from MCF7 was assayed for GHR expression. All real-time RT qPCR data was scaled by setting the IRC data to 100. Two reference genes, SDHA and HPRT, were found to be stably expressed in all cell lines and were used to normalize expression data.

Primer design and validation

Primers were designed using Primer3 software (Rozen & Skaletsky 2000) hosted online at http://primer3.wi.mit.edu/ by the Whitehead Institute (Massachusetts Institute of Technology).

In order to minimize the possibility of amplification from potential genomic DNA contamination, all primer pairs were designed to generate amplicons that contain exon-exon junctions.

Amplicons between 100 and 350 base pairs in length were preferred to favor high PCR efficiency and short amplification time. Ten or more primer pairs for each gene were evaluated for amplicon secondary structure using mfold software(Zuker 2003) hosted online by The RNA Institute,

University at Albany, State University of New York (http://mfold.rna.albany.edu). Primers with lower secondary structure were selected for PCR analysis. Primer pairs were analyzed for specificity using Primer-BLAST software hosted by the National Center for Biotechnology

Informatics (Ye et al. 2012). Primer sequences and annealing temperatures are shown in table 2.2.

Primers were synthesized by Sigma Life Sciences (Sigma-Aldrich). Annealing temperature optimization was performed using an annealing temperature gradient from 57.5 °C to 65 °C. Each primer pair was analyzed for amplification efficiency by using 6 serial 5-fold cDNA dilutions. All primer sets have an amplification efficiency of greater than 90%. PCR products were subject to melt curve analysis with iCycler software and were also analyzed by agarose gel electrophoresis to verify specificity. 73

Human melanoma biopsy cDNA arrays

Based on data collected from the NCI60 panel, we chose to investigate expression of

GHR in human melanoma biopsies. Additionally, IGF-1 and IGF-1R mRNA was quantified in these tumor samples. Arrays containing cDNA from human tumor samples were purchased from

Origene Technologies. Detailed patient information is provided from Origene to allow analysis of data by tumor grade and sex. A Melanoma cDNA Array I (Origene MERT101) containing 40 melanoma tumor samples were analyzed for GHR expression. A melanoma cDNA array was also analyzed for HPRT expression in order to normalize GHR expression data. These arrays consist of 96-well PCR plates which contain quantities of human tumor cDNA that have been normalized to beta actin expression levels. In order to compare expression levels between cell lines and human tumor samples, cDNA samples from the NCI60 melanoma cell lines were added to the human melanoma array plate and included in the real-time qPCR analysis of the arrays. HPRT expression was used to normalize and scale expression values.

Broad-Novartis Cancer Cell Line Encyclopedia

The Broad Institute and the Novartis Institutes for Biomedical Research have collaborated to generate gene expression data on 957 cancer cell lines in an ongoing project termed the Cancer Cell Line Encyclopedia (CCLE; Barretina et al. 2012). The CCLE provides public access to data generated through this project, including gene expression data from microarray studies. This public database is accessible at http://www.broadinstitute.org/ccle.

Through the CCLE Terms of Access, we declare that, “those who carried out the original analysis and collection of the data bear no responsibility for the further analysis or interpretation of it.”

Gene Expression data was extracted from CCLE_Expression_Entrez_2012-10-18.res and is 74

presented as gene-centric RMA-normalized (Robust Multi-array Average) mRNA expression data.

COMPARE correlations with existing NCI60 data

The NCI60 panel has been treated with thousands of chemical compounds including many with known anti-cancer activity and subject to numerous gene expression analyses including microarrays (Weinstein 2006). COMPARE analysis is an algorithm developed by NCI to search for correlations between new and existing data, such as gene expression, in order to discover similar patterns across the NCI60 panel (Zaharevitz et al. 2002). This can provide clues as to the mechanism of action of compounds that have already been shown to be effective at inhibiting growth of NCI60 cell lines and can also reveal novel and interesting links in gene expression patterns. The gene expression data obtained in this experiment were subject to

COMPARE analysis. Expression data were analyzed against the GI50 (the dose at which cell growth is inhibited by 50%) information for compounds and also against existing microarray data from 5 separate Affymetrix arrays. In COMPARE, the expression pattern of a given gene is compared to a database of GI50 values to identify compounds with a similar or inverse pattern.

In this analysis, a positive correlation between a gene and a compound indicates that cells with higher expression of that gene are more easily killed by treatment with that compound, while a negative correlation is indicative of resistance to that compound.

Data analysis

Raw data from the iCycler software was imported into qBasePLUS v2.3 (Biogazelle). Cq values were converted to quantities by the delta Cq method with correction for amplification efficiency and scaled to the IRC to correct for plate to plate variation. Expression data was then 75

normalized to the reference genes (HPRT and SDHA). Finally, data was scaled by setting the mean IRC value for each plate to 100 in order to allow comparison across all cell lines and between genes. Statistical analysis was conducted with GraphPad Prism 5 (GraphPad Software) and qBasePLUS. Data was subjected to the Kruskal-Wallis rank sum test for non-parametric data with Dunn’s multiple comparison post-hoc test. Correlations were performed by Spearman’s rank-order analysis. A P-value <0.05 was considered significant. The geometric mean was used to describe average expression levels between cancer types due to data spanning several orders of magnitude. Expression values less than one were set to one for display purposes. By correcting for amplification efficiency, plate to plate variation, and scaling all data in an identical manner, direct comparisons regarding the magnitude of gene expression were made between different genes.

Results

GHR, PRLR, GH and PRL expression across the NCI60 panel

GHR, PRLR, GH and PRL mRNA levels were examined in the NCI60 cell line panel using real-time RT qPCR. Data were analyzed by the geometric mean and scaled to allow comparison across all cell lines and among genes. GHR shows the highest mean mRNA level across the entire panel at 28.5, followed by PRLR with a level of 11.2 (see table 2.3 and figure

2.1). Using several primer pairs, we were unable to detect transcripts from the GH1 gene, and thus it will not be discussed further in the Results section. However, expression of GH2 mRNA was detected in several NCI60 cell lines. GH2 and PRL show the lowest levels at 1.1 and 1.4, respectively. mRNA levels differ significantly between GHR and GH2, GHR and PRL, PRLR and

GH2, PRLR and PRL (Dunn’s Multiple Comparison Test P<0.0001) but not between GHR and 76

PRLR or GH2 and PRL. The expression pattern of each of the four genes across the NCI60 panel was tested for correlations. Two associations were found, namely, a positive correlation between expression of GHR and PRL (P<0.01, r=0.394) and between PRLR and GH2 (P<0.05, r=0.299).

GHR, PRLR, GH and PRL expression between 9 human cancer types

GHR mRNA is widely expressed in the NCI60 panel cell lines, with detection in 55 out of 60 cell lines (figure 2.2A). Expression of GHR varies greatly between cancer types. Those with the highest levels of GHR mRNA include melanoma, lung and prostate, which have mean levels of 1348.0, 81.1 and 57.6, respectively (table 2.3 and supplementary figure 2.1). In melanoma, all cell lines express high levels of GHR mRNA. Renal, leukemia and colon cancer have the lowest levels of GHR expression, with mean values of 8.1, 2.8 and 1.0, respectively. The seven colon cancer cell lines all fall within the lowest quartile for GHR expression. With the exception of colon cancer and melanoma, all cancer types contain both high and low GHR expressing cell lines

(figure 2.2A). GHR mRNA expression shows significant differences across cancer types

(P<0.0001). Post-hoc analysis shows that colon cancer cell lines are significantly different from lung, and melanoma is statistically different from leukemia, colon and renal (P<0.05).

PRLR mRNA expression is highest in breast cancer cell lines (figure 2.2B and table 2.3).

CNS, which expresses 44 times less PRLR than breast, has the second highest levels of PRLR mRNA. NSC lung and melanoma cancer types have nearly equal mean PRLR mRNA levels, each

~88-fold lower than breast cancer. As with GHR, leukemia and renal lines are among the three lowest in terms of PRLR expression. The two prostate cell lines express the lowest amount of

PRLR mRNA of the 60 cell lines studied. While four cancer types have cell lines with GHR levels below the minimum threshold of one, seven types had cell lines below this threshold for PRLR

(see figure 2.3). The differences in PRLR expression are statistically significant across the nine 77

cell line types (P<0.001). Post-hoc analysis shows significant differences between breast and renal, leukemia and CNS, as well as leukemia and breast (P<0.05).

In order to evaluate the potential for paracrine and autocrine action in cancer cells, we assessed mRNA expression of GH2 and PRL in the NCI60 panel (figure 2.2C and D and table

2.3). Expression of both GH2 and PRL mRNAs were very low in all cancer types examined. GH2 mRNA expression is the lowest of all four genes tested. Expression of PRL varies significantly by cancer type (P<0.05), while GH2 does not. Post-hoc analysis shows that mean PRL expression is statistically different between the melanoma and colon groups (P<0.05). While GH2 is not significantly different between any cancer types, colon stands out with the most number of cell lines (four out of seven) that express GH2 mRNA above the minimum threshold of one (figure

2.2C). Mean PRL mRNA levels are below the minimum threshold in five of nine cancer types

(table 2.3). The four cancer types with PRL levels above the minimum threshold are melanoma, with a mean relative expression of 2.7, followed by NSC lung, leukemia and breast.

GHR, PRLR, GH and PRL expression within cancer types: differences between the cell lines

Individual mRNA expression levels for each cell line are shown in figure 2.3 organized by cancer type. Within most cancer types, GHR mRNA levels vary greatly among cell lines

(figure 2.2A and figure 2.3). Unique among the nine cancer types, all melanoma cell lines have high levels of GHR mRNA. Out of the 12 cell lines with the highest GHR mRNA levels, eight are melanoma lines, with highest levels observed in UACC62 and MALME3M. Within breast cancer cell lines, T47D and MCF7 show the highest levels of GHR mRNA, with T47D being 11-fold higher than MCF7. GHR mRNA levels were relatively low in the remaining breast cancer cell lines, including two with little or no GHR expression (MDA-MB-468 and MDA-MB-235).

Prostate cancer had one cell line (PC3) with high GHR levels, and one with relatively low levels 78

(DU-145). In ovarian cancer, four cell lines show relatively low levels of GHR mRNA, while three are within the top third. All NSC lung cancer lines show GHR mRNA levels above the minimum threshold, with NCIH-226 being among the highest of all cell lines. A498 is the highest of the renal cell lines, while RXF-393 and 786-O both have very low GHR mRNA levels. The

CNS cell lines all display low to medium GHR expression. GHR mRNA is low in nearly all colon and leukemia cell lines with the exception of the leukemia line SR.

PRLR mRNA levels show great variation between cell lines, both within and among cancer types. This gene has the highest single expression value of the four genes tested, exceeding a level of 100,000 in T47D cells, and at the same time, is expressed at a level below 1 in more than a quarter of the cell lines. PRLR mRNA is expressed at the highest levels in breast cancer cell lines. Expression is greatest in T47D, MCF7, BT-549 and MDA-MB-468 cells, which together make up four of the top five PRLR expressing cell lines in the panel with the melanoma line, SK-MEL-5, being the other cell line in the top five. Only one leukemia cell line, SR, expressed PRLR above the minimum threshold, and even in this line, SR expression of PRLR was more than 30,000-fold lower than that found in T47D cells.

GH2 and PRL mRNA is expressed at a much lower level than either GHR or PRLR.

There are 47 and 44 cell lines that show GH2 and PRL, respectively, below the minimum threshold level of one. This includes 11 in which PRL is undetectable. The cell lines with the most PRL mRNA are SK-MEL-2 and UACC-257, both melanoma lines. Five of the nine cancer types have a single cell line with GH2 expression above the minimum (see figure 2.2C).

Although four out of seven colon cancer cell lines express GH2 above a level of one, expression levels were still quite low. Cancer lines that express PRL above the minimum include four melanoma, four NSC lung, three leukemia, two breast, one renal and one CNS cell line. 79

Expression of GHR in metastatic melanoma samples

Due to the high levels of GHR detected in all NCI60 melanoma cell lines, the potential in vivo role of GHR in human melanoma biopsies was explored using commercially available cDNA microarrays. These arrays contain three non-tumor skin sample controls and 40 samples from both male and female stage III and stage IV melanomas. GHR is expressed in most tumor samples (figure 2.4), with more than a fourth expressing GHR mRNA at levels greater than 100 relative expression units. GHR expression is undetectable in six samples. The three controls all express high levels of GHR mRNA.

GHR, IGF1 and IGF1R mRNA levels in stage III and IV metastatic melanoma samples by sex

The detection of GHR in human melanoma biopsies led us to further explore the

GH/IGF-1 axis in these tumor samples. Levels of IGF1 and IGF1R mRNA were quantified using the same commercially available cDNA microarrays (supplementary figure 2.2). These arrays contain cDNA from 21 stage III (regional) and 19 stage IV (distant) metastatic melanoma tumor samples. Sixteen of these tumors are female, while 24 are male. When categorized by sex, both

GHR and IGF1 are nearly double in tumors from male patients, while IGF1R does not differ

(figure 2.5A). In stage IV tumor samples, mRNA expression of both GHR and IGF1R are markedly increased compared to stage III samples while IGF1 did not appear to show a substantial difference (figure 2.5B). The difference in GHR expression in advanced stage IV tumors was significant (one-tailed Mann-Whitney test P<0.05).

NCI60 COMPARE: correlations with microarray and GI50 data

COMPARE was undertaken in order to examine potential links between the genes examined here and existing NCI60 data including sensitivity to anti-cancer compounds as well as metabolomic and microarray expression data. 80

GHR expression was correlated with the GI50 for 11 chemical compounds that are effective growth inhibitors of NCI60 cell lines (Pearson correlation P<0.001). Of these, seven of the compounds are known or predicted to act by inhibition of tubulin. All seven of these correlations are negative, indicating that lower GHR expression indicates a higher susceptibility to tubulin-inhibiting compounds, or conversely, that high GHR mRNA levels indicate a resistance to tubulin-inhibitors. Also, GHR expression is positively associated with rapamycin sensitivity

(Pearson correlation P<0.001). Similar to GHR, PRLR expression shows negative correlations with a high number of compounds thought to act via tubulin inhibition. PRL expression is positively associated with sensitivity to several aromatase inhibitors, while GH does not have any significant correlations with GI50 data.

There are 530 genes in the existing NCI60 microarray database with expression patterns that show a significant correlation with GHR mRNA levels (P-value <0.001). Of these, 515 are positively correlated, while 15 show a negative correlation. Our GHR qPCR data shows the highest correlation with GHR expression from the microarray data, providing independent confirmation of the real-time qPCR performed in this study. Included in the significant correlations are GHR, PRLR, prolactin-induced protein, transforming growth factor beta (TGF-B)

3, TGF-B activated kinase 1 (also known as MAP3K7 binding protein 3), MAPK 8, MAPK15,

MAP3K13, collagen triple helix repeat containing 1, tumor necrosis factor receptor superfamily member 10c, tubulin tyrosine ligase-like family member 4, collagen type IX alpha 3 and melanoma inhibitory activity. PRLR has 626 correlations to microarray data (P<0.001), all but seven of which are positive. Among these are PRLR, GHR, prolactin-induced protein, TGF-B3,

MAP3K13, MAPK15, tumor necrosis factor receptor superfamily member 10c, melanoma associated antigen (mutated) 1-like 1, and melanoma antigen family F 1. PRL has 294 microarray correlations with a P-value < 0.001, among these are melanoma antigen family C 1, melanoma 81

antigen family C 2, melanoma antigen family D 2, melanoma antigen family B 1, melanoma antigen family A 8, and melanoma antigen family A 11.

Comparison to CCLE Data

The CCLE is a collection of gene expression data on 957 cancer cell lines. GHR gene expression data from the CCLE database is shown in figure 2.6. In agreement with our studies, melanoma is one of the highest GHR expressing cancer types while those with the lowest levels in the CCLE include leukemia (T-cell and B-cell acute lymphoblastic leukemia) and colon

(colorectal in the CCLE). Our finding that PRLR mRNA expression is highest in breast cancer cell lines was confirmed in the CCLE data. Figures generated from the CCLE database for expression of PRLR, PRL and GH are shown in supplementary figure 2.3.

Discussion

GHR and PRLR activation can modulate numerous intracellular signaling pathways that are central to the processes of tumorigenesis (Chhabra et al. 2011, Cohen et al. 2000, Perry et al.

2006, Clayton et al. 2011, Goffin et al. 1996, Goffin & Kelly 1997). Multiple lines of evidence from in vitro, animal and human studies converge to support the hypothesis that GH could promote the growth of certain types or sub-types of cancer. To date, a majority of the investigations into the role of GH in cancer have been limited to breast and prostate cancer, and most of these reports do not quantify expression of GH or GHR. In the currents study, we utilized the NCI60 cancer cell line panel to characterize cell lines from nine types of human cancer for the potential ability to respond to GH through GHR and PRLR signaling. Surprisingly, the results presented here are the first to report high GHR mRNA levels in melanoma cell lines and in a subset of human melanoma tumors. The increased GHR and IGF1R mRNA in tumors from male 82

patients and the elevation of GHR and IGF1 mRNA in advanced stage IV tumors are novel findings with potential relevance to the development of personalized therapeutic interventions.

GH and PRL mRNA levels were low in nearly all cell lines, suggesting that most of the NCI60 cancer cell lines possess little to no autocrine GH or PRL activity. Additionally, we have obtained intriguing gene expression profiles and correlations with existing data that will prove useful to direct research in the future in terms of therapeutic intervention. Below, we will discuss the most important results of our study and their implications.

GHR is highly expressed in melanoma cell lines

The most striking result from this study is the universally high GHR expression in melanoma cell lines (figure 2.2). In the NCI60 panel, mean levels of GHR mRNA were more than

16 times higher in melanoma than in the next highest cell type. Our finding has been independently replicated by the recently published Broad-Novartis CCLE, which performed microarray analysis on 947 cancer cell lines. Public data from the CCLE found melanoma to be the second highest for GHR expression out of 36 cancer cell line classifications (figure 2.6). Both

GHR protein and mRNA are expressed in normal human skin (Lobie et al. 1990, Tavakkol et al.

1992, Mertani et al. 1995). Several IHC studies of human biopsies have shown GHR expression in different types of skin cancer (Lincoln et al. 1998, Ginarte et al. 2000, Stanimirovic et al.

2005, Lincoln et al. 1999). Intriguingly, there are two recent case studies that lend support to the hypothesis that GHR could play a causative role in melanoma. In the first report, a 26 year old female was diagnosed with malignant melanoma following multiple hormone replacement therapy that included both estrogen and GH (Caldarola et al. 2010). While it is impossible to determine if or which hormone treatment caused or was implicated in the melanoma, the authors state that a role of GH “could not be ruled out”. In another striking report, a husband and wife 83

were both diagnosed with melanoma within 2 weeks of each other (Handler et al. 2012). Both patients had received daily injections of human GH for 3 months as part of an “anti-aging regimen”. Although a cause and effect issue cannot be distilled here, the short length of time between the start of GH exposure and diagnosis of melanoma in these two patients is striking.

PRLR is highly expressed in breast cancer cell lines

While there was not a single cancer type that had high PRLR mRNA expression in all cell lines, several types showed high levels in a subset of cell lines, including NSC lung, CNS, melanoma and ovarian. PRLR has also been observed in several ovarian cancer cell lines (Tan et al. 2011), although we could not find reports of PRLR mRNA expression in NSC lung, CNS or melanoma cancer cell lines. We found that PRLR mRNA was by far the highest in breast cancer cell lines, which is in agreement with a large body of literature showing high PRLR expression and activity in mammary carcinoma tissue (reviewed by Swaminathan et al. 2008). While PRLR has been extensively studied with regard to breast cancer, our novel findings of high PRLR mRNA levels in several other cancer types suggests that research should expand to include cancers of these other tissues.

GHR and IGF1 mRNA is higher in tumors from males while GHR and IGF1R mRNA is elevated in advanced stage IV metastatic tumors

We observe expression of GHR mRNA in a majority of human melanoma biopsy samples examined. This is in agreement with previous data from IHC studies (Lincoln et al. 1998,

Ginarte et al. 2000, Stanimirovic et al. 2005, Lincoln et al. 1999). The average expression level of GHR mRNA was substantially higher in the three control skin samples than in the metastatic tumor samples. While GHR expression has been found in skin by others (Lobie et al. 1990), the 84

exact cell types within skin that are responsible for the high GHR mRNA expression is unknown.

The control skin samples are composed of many different cell types, but the tumors presumably derive exclusively from melanocytes. It is possible that melanocytes are not the major contributor to the high GHR expression found in the control skin samples. It is also possible that melanocytes increase GHR expression upon oncogenic transformation. Alternatively, it is possible that GHR expression drops when melanocytes become cancer cells, but a sub-group of melanoma tumors express a level of GHR that confers a benefit to the cancer cell upon metastasis to a distant site.

When we grouped samples by gender, we observed a substantial increase in GHR and IGF-1 mRNA in male patients compared to female. The potential for the GH/IGF1 axis to play a role in metastasis and invasion led us to analyze expression of GHR, IGF1 and IGF1R by cancer stage.

We found higher expression of both GHR and IGF1R in more advanced stage IV samples compared to stage III. This finding may have clinical relevance, given the results of a phase I clinical trial which found a partial response to combination treatment with a humanized IGF-1R antibody and docetaxel in a patient with metastatic melanoma (Macaulay et al. 2013). It has also been shown that malignant melanoma tumors express IGF-1R and treatment of cell lines established from these tumors undergo apoptosis when treated with antibodies that inhibit IGF-1R signaling. The elevated GHR mRNA in advanced metastatic melanoma tumors suggests that inhibiting the GHR axis with GH-suppressive or GHR-inhibiting therapeutics may be worth pursuing in patients whose tumors show high GHR mRNA levels.

Low GHR expression levels in colon cancer cell lines is in contrast with previous data from biopsies

Data from human patients with GH excess (acromegaly) implicate GHR in the development of colon cancer (Renehan & Brennan 2008). Yet, in our analysis of NCI60 colon 85

cancer cell lines, GHR mRNA levels are barely detectable. These data are not in agreement with several IHC studies on human colon cancer biopsies, which report the presence of GHR in colorectal tumor tissues, although not all studies are in agreement (reviewed by Kabeer et al.

2011). Possible explanations for this discrepancy could be that the cell lines are not representative of colon tumors, the level of mRNA does not represent the level of protein or the immunohistochemistry studies do not adequately portray GHR expression. It would be useful to quantify GHR gene expression in colon cancer biopsies and possibly perform IHC using different

GHR antibodies than have been used previously. A subset of the NCI60 colon cancer cell lines expresses PRLR, which makes a direct effect of GH possible through the PRLR. If GH does have a role in colon cancer, our data showing a lack of GHR in colon cancer cell lines suggests that GH may act through PRLR or a downstream factor such as IGF-1 to promote tumorigenesis.

Limited evidence for autocrine and/or paracrine GH and PRL action

Forced expression of GH has been shown to confer unique properties to cancer cells different from the effects of treatment with exogenous GH (Kaulsay et al. 1999, Zhu et al. 2005,

Brunet-Dunand et al. 2009). We were unable to detect GH1 in the NCI60 cell lines. GH2 and

PRL mRNA are only detectable in a few of the NCI60 cell lines, and where they are expressed, it is at relatively low levels. This data suggests that it is unlikely that GH or PRL act in an autocrine or paracrine manner in most of the NCI60 cell lines. It is interesting that PRL expression was observed in lung cancer, as elevated circulating PRL levels have been observed in male lung cancer patients (Bhatavdekar et al. 1994). This introduces the intriguing possibility that tumor- derived PRL may increase endocrine PRL levels. Although there are a few examples of cell lines from various cancer types that express both ligand and receptor mRNA (figure 2.3), most NCI60 cell lines do not possess the capacity for autocrine or paracrine action by these hormones. 86

Cell lines with high GHR and/or PRLR expression tend to be resistant to tubulin inhibiting drugs

We used the COMPARE algorithm to examine our gene expression patterns for correlations to existing NCI60 database information including microarray and GI50 data. Our

GHR and PRLR mRNA expression patterns both show a significant negative correlation with a number of tubulin inhibiting drugs—which implies that cell lines with high GH signaling capacity are resistant to these drugs. One possible explanation could be that GH-induced signaling might increase tubulin expression, as it does in 3T3-F442A preadipocytes and rat embryonic cerebral cortical cells (Guller et al. 1989, Ajo et al. 2003), to levels that are sufficient to resist the effect of tubulin-inhibitors. Intriguingly, GHR was positively correlated with the expression of tubulin tyrosine ligase-like family member 4 (TTLL4), an enzyme that adds glutamate to tubulin and has recently been implicated in the growth of pancreatic cancer cells

(van Dijk et al. 2007, Kashiwaya et al. 2010). Furthermore, in dividing cells, the mitotic spindle has been shown to have high levels of tubulin polyglutamylation (Bobinnec et al. 1998). In addition, downstream signaling molecules in the GHR pathway, specifically JAK2 and STAT1, have been shown to bind to tubulin (Ma & Sayeski 2007). The association of high GHR and/or

PRLR expression and low sensitivity to tubulin inhibiting drugs may warrant the exploration of these receptors as a potential biomarker to screen cancer patients prior to treatment with this class of chemotherapeutics. It also may be of interest to investigate whether treatment with drugs that reduce GH activity may sensitize melanoma and/or breast cancer cells to tubulin inhibitors.

GHR is positively correlated with rapamycin sensitivity and expression of MAPK-related genes

GH is known to activate MAPK signaling pathways (Chhabra et al. 2011). The positive correlation of GHR mRNA levels with expression of several members of the MAPK family in the 87

NCI60 panel (see results) suggests that this pathway is more active in certain subtypes of cancer that may be stimulated by GH. GH has been shown to alter cellular metabolism in hepatoma cells by activation of mTOR complex 1, likely through Akt activation (Hayashi & Proud 2007). We found that cell lines with higher GHR levels are accompanied by higher sensitivity to the mTOR inhibitor rapamycin. This correlation suggests that GHR signaling may be involved in cancer cell growth via mTOR activation.

PRL expression is positively associated with effectiveness of estrogen suppression

We observe a positive correlation between PRL expression in cancer cell lines and their sensitivity to aromatase inhibitors. Estrogen is known to stimulate PRL expression in pituitary lactotrophs (Chaturvedi & Sarkar 2008), and thus it is plausible that estrogen could do the same in cancer cells. One would expect that cells with higher sensitivity to aromatase inhibitors would have elevated estrogen, a known stimulus of PRL expression. The observation that subsets of cancer cell lines, particularly lung and melanoma, show PRL mRNA expression is indeed interesting and warrants further investigation. Interestingly, the use of aromatase inhibitors to block estrogen synthesis is being explored as a potential therapeutic intervention in patients with

NSC lung cancer (reviewed by Miki et al. 2011). The link between PRL expression, estrogen and aromatase inhibitors in cancer cells awaits further exploration.

In summary, we have examined the GH and PRL pathways by quantifying mRNA expression in 60 cell lines from nine human cancer types. Most of the cancer types showed a wide range of GHR mRNA expression, suggesting that different types of cancer may have specific sub- types that could potentially be sensitive to inhibition of GHR signaling. As has been previously shown by others, we found high PRLR mRNA levels in breast cancer cell lines. Several other cancer types displayed high PRLR mRNA levels and should be explored for a potential role of 88

GH and/or PRL. The novel finding of high GHR mRNA expression in melanoma may have significant implications for the treatment of melanoma with drugs that lower GH levels

(somatostatin analogs) or with drugs that inhibit GH action (pegvisomant). Any extrapolations of this data to potential biological significance will require further studies to determine whether the gene expression patterns reported here accurately portray active and relevant signaling pathways in vivo. To this end, we currently have studies underway to further characterize melanoma cell lines for sensitivity to GH and GHR antagonists.

Declaration of interest

JJK is an inventor of US patent 5350836 entitled “Growth hormone antagonists”. The other authors have nothing to declare.

Funding

This research is supported by the State of Ohio’s Eminent Scholar Program, which includes a gift from Milton and Lawrence Goll, by AMVETS, the Diabetes Institute at Ohio

University and by NIH (P01AG031736).

Author contributions

ESG designed and performed a majority of the experiments and was the primary author. RKJ assisted with experiments, data analysis, preparation of figures and manuscript edits. JJK guided the study and participated in discussions of manuscript organization, content, writing and editing. 89

Acknowledgements

We would like to thank Dr. Steven Swanson (Chicago) for the inspiration to perform this study; Susan Holbeck and the National Cancer Institute for providing RNA samples from the

NCI60 panel and for performing COMPARE analysis and Vincent Goffin (Paris) for his guidance on experiments involving PRLR and PRL.

References

Ajo R, Cacicedo L, Navarro C & Sanchez-Franco F 2003 Growth hormone action on proliferation and differentiation of cerebral cortical cells from fetal rat. Endocrinology 144 1086-1097.

Barretina J, Caponigro G, Stransky N, Venkatesan K, Margolin AA, Kim S, Wilson CJ, Lehár J, Kryukov GV, Sonkin D, Reddy A, Liu M, Murray L, Berger MF, Monahan JE, Morais P, Meltzer J, Korejwa A, Jané-Valbuena J, Mapa FA, Thibault J, Bric-Furlong E, Raman P, Shipway A, Engels IH, Cheng J, Yu GK, Yu J, Aspesi P, de Silva M, Jagtap K, Jones MD, Wang L, Hatton C, Palescandolo E, Gupta S, Mahan S, Sougnez C, Onofrio RC, Liefeld T, MacConaill L, Winckler W, Reich M, Li N, Mesirov JP, Gabriel SB, Getz G, Ardlie K, Chan V, Myer VE, Weber BL, Porter J, Warmuth M, Finan P, Harris JL, Meyerson M, Golub TR, Morrissey MP, Sellers WR, Schlegel R & Garraway LA 2012 The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity. Nature 483 7391-307. (doi:10.1038/nature11003).

Beck C & Burger HG 1972 Evidence for the presence of immunoreactive growth hormone in cancers of the lung and stomach. Cancer 30 75-79.

Bernichtein S, Touraine P & Goffin V 2010 New concepts in prolactin biology. The Journal of endocrinology 206 1-11. (doi:10.1677/JOE-10-0069; 10.1677/JOE-10-0069).

Bhatavdekar JM, Patel DD, Chikhlikar PR, Mehta RH, Vora HH, Karelia NH, Ghosh N, Shah NG, Suthar TP & Neema JP 1994 Levels of circulating peptide and steroid hormones in men with lung cancer. Neoplasma 41 101-103.

Bobinnec Y, Moudjou M, Fouquet JP, Desbruyeres E, Edde B & Bornens M 1998 Glutamylation of centriole and cytoplasmic tubulin in proliferating non-neuronal cells. Cell motility and the cytoskeleton 39 223-232. (doi:2-5). 90

Brunet-Dunand SE, Vouyovitch C, Araneda S, Pandey V, Vidal LJ, Print C, Mertani HC, Lobie PE & Perry JK 2009 Autocrine human growth hormone promotes tumor angiogenesis in mammary carcinoma. Endocrinology 150 1341-1352. (doi:10.1210/en.2008-0608).

Bundred NJ, Eremin O, Stewart HJ, Dale BA & Forrest APM 1986 Beneficial response to pituitary ablation following aminoglutethimide. British Journal of Surgery 73 388-389. (doi:10.1002/bjs.1800730525).

Caldarola G, Battista C & Pellicano R 2010 Melanoma onset after estrogen, thyroid, and growth hormone replacement therapy. Clinical therapeutics 32 57-59. (doi:10.1016/j.clinthera.2010.01.011; 10.1016/j.clinthera.2010.01.011).

Chaturvedi K & Sarkar DK 2008 Alteration in G proteins and prolactin levels in pituitary after ethanol and estrogen treatment. Alcoholism, Clinical and Experimental Research 32 806- 813. (doi:10.1111/j.1530-0277.2008.00638.x; 10.1111/j.1530-0277.2008.00638.x).

Chhabra Y, Waters MJ & Brooks AJ 2011 Role of the growth hormone–IGF-1 axis in cancer. Expert Review of Endocrinology & Metabolism 6 71 -84. (doi:10.1586/eem.10.73).

Clayton PE, Banerjee I, Murray PG & Renehan AG 2011 Growth hormone, the insulin-like growth factor axis, insulin and cancer risk. Nature reviews.Endocrinology 7 11-24. (doi:10.1038/nrendo.2010.171; 10.1038/nrendo.2010.171).

Cohen P, Clemmons DR & Rosenfeld RG 2000 Does the GH-IGF axis play a role in cancer pathogenesis? Growth hormone & IGF research : official journal of the Growth Hormone Research Society and the International IGF Research Society 10 297-305. (doi:10.1054/ghir.2000.0171).

Gebre-Medhin M, Kindblom LG, Wennbo H, T\ornell J & Meis-Kindblom JM 2001 Growth hormone receptor is expressed in human breast cancer. The American journal of pathology 158 1217-1222.

Ginarte M, Garcia-Caballero T, Fernandez-Redondo V, Beiras A & Toribio J 2000 Expression of growth hormone receptor in benign and malignant cutaneous proliferative entities. Journal of cutaneous pathology 27 276-282.

Goffin V & Kelly PA 1997 The prolactin/growth hormone receptor family: structure/function relationships. Journal of mammary gland biology and neoplasia 2 7-17. 91

Grunenwald S, Tack I, Chauveau D, Bennet A & Caron P 2011 Impact of growth hormone hypersecretion on the adult human kidney. Annales d'Endocrinologie 72 485 495. (doi:10.1016/j.ando.2011.08.001).

Guller S, Corin RE, Wu KY & Sonenberg M 1989 Growth hormone-induced alteration of morphology and tubulin expression in 3T3 preadipose cells. Biochemical and biophysical research communications 163 895-901.

Gunnell D, Okasha M, Smith GD, Oliver SE, Sandhu J & Holly JM 2001 Height, leg length, and cancer risk: a systematic review. Epidemiologic reviews 23 313-342.

Handler MZ, Ross AL, Shiman MI, Elgart GW & Grichnik JM 2012 Potential role of human growth hormone in melanoma growth promotion. Archives of Dermatology 148 1179-1182. (doi:10.1001/archdermatol.2012.2149; 10.1001/archdermatol.2012.2149).

Harvey S 2010 Extrapituitary growth hormone. Endocrine 38 335-359. (doi:10.1007/s12020-010- 9403-8; 10.1007/s12020-010-9403-8).

Holly JM & Perks CM 2008 Cancer as an endocrine problem. Best practice & research.Clinical endocrinology & metabolism 22 539-550. (doi:10.1016/j.beem.2008.07.007).

Ikeno Y, Hubbard GB, Lee S, Cortez LA, Lew CM, Webb CR, Berryman DE, List EO, Kopchick JJ & Bartke A 2009 Reduced incidence and delayed occurrence of fatal neoplastic diseases in growth hormone receptor/binding protein knockout mice. The journals of gerontology.Series A, Biological sciences and medical sciences 64 522-529. (doi:10.1093/gerona/glp017; 10.1093/gerona/glp017).

Jenkins PJ 2006 Cancers associated with acromegaly. Neuroendocrinology 83 218-223.

Kabeer MA, Widdison AL, Demaine A & Mathew J 2011 Expression of growth hormone receptors in colorectal carcinoma. Diagnostic Histopathology 17 355 361. (doi:10.1016/j.mpdhp.2011.06.004). 92

Kashiwaya K, Nakagawa H, Hosokawa M, Mochizuki Y, Ueda K, Piao L, Chung S, Hamamoto R, Eguchi H, Ohigashi H, Ishikawa O, Janke C, Shinomura Y & Nakamura Y 2010 Involvement of the tubulin tyrosine ligase-like family member 4 polyglutamylase in PELP1 polyglutamylation and chromatin remodeling in pancreatic cancer cells. Cancer research 70 4024-4033. (doi:10.1158/0008-5472.CAN-09-4444; 10.1158/0008-5472.CAN-09-4444).

Laron Z & Kopchick JJ 2011 Laron Syndrome - From Man to Mouse - Lessons from Clinical and Experimental Experience.Springer.

Lincoln DT, Sinowatz F, Kolle S, Takahashi H, Parsons P & Waters M 1999 Up-regulation of growth hormone receptor immunoreactivity in human melanoma. Anticancer Research 19 1919-1931.

Lobie PE, Breipohl W, Lincoln DT, Garcia-Aragon J & Waters MJ 1990 Localization of the growth hormone receptor/binding protein in skin. The Journal of endocrinology 126 467- 471.

Ma X & Sayeski PP 2007 Identification of tubulin as a substrate of Jak2 tyrosine kinase and its role in Jak2-dependent signaling. Biochemistry 46 7153-7162. (doi:10.1021/bi700101n).

Macaulay VM, Middleton MR, Protheroe AS, Tolcher A, Dieras V, Sessa C, Bahleda R, Blay JY, LoRusso P, Mery-Mignard D & Soria JC 2013 Phase I study of humanized monoclonal antibody AVE1642 directed against the type 1 insulin-like growth factor receptor (IGF-1R), administered in combination with anticancer therapies to patients with advanced solid tumors. Annals of Oncology : Official Journal of the European Society for Medical Oncology / ESMO 24 784-791. (doi:10.1093/annonc/mds511; 10.1093/annonc/mds511).

Martino A, Cammarota G, Cianci R, Bianchi A, Sacco E, Tilaro L, Marzetti E, Certo M, Pirozzi G, Fedeli P, Pandolfi F, Pontecorvi A, Gasbarrini G & De Marinis L 2004 High prevalence 93

of hyperplastic colonic polyps in acromegalic subjects. Digestive diseases and sciences 49 662-666.

Mertani HC, Delehaye-Zervas MC, Martini JF, Postel-Vinay MC & Morel G 1995 Localization of growth hormone receptor messenger RNA in human tissues. Endocrine 3 135-142. (doi:10.1007/BF02990065; 10.1007/BF02990065).

Miki Y, Abe K, Suzuki S, Suzuki T & Sasano H 2011 Suppression of estrogen actions in human lung cancer. Molecular and cellular endocrinology 340 168-174. (doi:10.1016/j.mce.2011.02.018; 10.1016/j.mce.2011.02.018).

Moller N & Jorgensen JO 2009 Effects of growth hormone on glucose, lipid, and protein metabolism in human subjects. Endocrine reviews 30 152-177. (doi:10.1210/er.2008-0027; 10.1210/er.2008-0027).

Perry JK, Emerald BS, Mertani HC & Lobie PE 2006 The oncogenic potential of growth hormone. Growth hormone \& IGF research 16 277-289.

Pollak M, Blouin MJ, Zhang JC & Kopchick JJ 2001 Reduced mammary gland carcinogenesis in transgenic mice expressing a growth hormone antagonist. British journal of cancer 85 428- 430.

Renehan AG & Brennan BM 2008 Acromegaly, growth hormone and cancer risk. Best practice & research. Clinical endocrinology & metabolism 22 639-657. (doi:10.1016/j.beem.2008.08.011).

Rozen S & Skaletsky H 2000 Primer3 on the WWW for general users and for biologist programmers. Methods in molecular biology (Clifton, N.J.) 132 365-386.

Shevah O & Laron Z 2007 Patients with congenital deficiency of IGF-I seem protected from the development of malignancies: a preliminary report. Growth hormone \& IGF research 17 54-57.

Shoemaker RH 2006 The NCI60 human tumour cell line anticancer drug screen. Nature reviews.Cancer 6 813-823. (doi:10.1038/nrc1951). 94

Stanimirovic A, Cupic H, Bosnjak B, Tomas D, Kruslin B & Belicza M 2005 TP53, Bcl-2 and growth hormone receptor expression in cutaneous squamous cell carcinoma. Acta dermatovenerologica Croatica : ADC 13 201-205.

Swanson SM & Unterman TG 2002 The growth hormone-deficient Spontaneous Dwarf rat is resistant to chemically induced mammary carcinogenesis. Carcinogenesis 23 977-982.

Tan D, Chen KE, Khoo T & Walker AM 2011 Prolactin increases survival and migration of ovarian cancer cells: importance of prolactin receptor type and therapeutic potential of S179D and G129R receptor antagonists. Cancer letters 310 101-108. (doi:10.1016/j.canlet.2011.06.014; 10.1016/j.canlet.2011.06.014).

Tavakkol A, Elder JT, Griffiths CE, Cooper KD, Talwar H, Fisher GJ, Keane KM, Foltin SK & Voorhees JJ 1992 Expression of growth hormone receptor, insulin-like growth factor 1 (IGF-1) and IGF-1 receptor mRNA and proteins in human skin. The Journal of investigative dermatology 99 343-349. van Dijk J, Rogowski K, Miro J, Lacroix B, Edde B & Janke C 2007 A targeted multienzyme mechanism for selective microtubule polyglutamylation. Molecular cell 26 437-448. (doi:10.1016/j.molcel.2007.04.012).

VanGilder JC & Goldenberg IS 1975 Hypophysectomy in metastatic breast cancer. Archives of surgery (Chicago, Ill.: 1960) 110 293-295.

Weinstein JN 2006 Spotlight on molecular profiling: "Integromic" analysis of the NCI-60 cancer cell lines. Molecular cancer therapeutics 5 2601-2605. (doi:10.1158/1535-7163.MCT-06- 0640).

Wu ZS, Yang K, Wan Y, Qian PX, Perry JK, Chiesa J, Mertani HC, Zhu T & Lobie PE 2011 Tumor expression of human growth hormone and human prolactin predict a worse survival outcome in patients with mammary or endometrial carcinoma. The Journal of clinical endocrinology and metabolism 96 E1619-29. (doi:10.1210/jc.2011-1245; 10.1210/jc.2011- 1245).

Yang XF, Beamer WG, Huynh H & Pollak M 1996 Reduced growth of human breast cancer xenografts in hosts homozygous for the lit mutation. Cancer research 56 1509-1511. 95

Ye J, Coulouris G, Zaretskaya I, Cutcutache I, Rozen S & Madden TL 2012 Primer-BLAST: a tool to design target-specific primers for polymerase chain reaction. BMC bioinformatics 13 134-2105-13-134. (doi:10.1186/1471-2105-13-134; 10.1186/1471-2105-13-134).

Zaharevitz DW, Holbeck SL, Bowerman C & Svetlik PA 2002 COMPARE: a web accessible tool for investigating mechanisms of cell growth inhibition. Journal of molecular graphics & modelling 20 297-303.

Zhu T, Starling-Emerald B, Zhang X, Lee K, Gluckman PD, Mertani HC & Lobie PE 2005 Oncogenic Transformation of Human Mammary Epithelial Cells by Autocrine Human Growth Hormone. Cancer Res 65 317-324.

Zuker M 2003 Mfold web server for nucleic acid folding and hybridization prediction. Nucleic acids research 31 3406-3415.

Tables

Table 2.1. Cell lines and cancer types included in the NCI60. Breast CNS Colon Leukemia Melanoma NSC Lung Ovarian Prostate Renal MCF7 SF-268 COLO 205 CCRF-CEM LOX IMVI A549/ATCC IGROV1 PC-3 786-0 MDA-MB-231 SF-295 HCC-2998 HL-60(TB) MALME-3M EKVX OVCAR-3 DU-145 A498 MDA-MB-468 SF-539 HCT-116 K-562 M14 HOP-62 OVCAR-4 ACHN HS 578T SNB-19 HCT-15 MOLT-4 MDA-MB-435 HOP-92 OVCAR-5 CAKI-1 BT-549 SNB-75 HT29 RPMI-8226 SK-MEL-2 NCI-H226 OVCAR-8 RXF 393 T-47D U251 KM12 SR SK-MEL-28 NCI-H23 NCI/ADR-RES SN12C SW-620 SK-MEL-5 NCI-H322M SK-OV-3 TK-10 UACC-257 NCI-H460 UO-31 UACC-62 NCI-H522

Table 2.2. Primer sequences and optimal annealing temperatures for real-time RT qPCR. Gene Sense primer Antisense primer Annealing Efficiency Amplicon GHR GTGATGCTTTTTCTGGAAGTGA TCAGGGCATTCTTTCCATTC 59.1 °C 90.2% 196/262 bp PRLR GCGAACACTGAGGATGCTTT GCAGATGCCACATTTTCCTT 60.0 °C 97.7% 103 bp GH2 GTCCTGTGGACAGCTCACCT CTCGACCTTGTCCATGTCCT 57.5 °C 91.3% 302 bp PRL CATCAACAGCTGCCACACTT TTCCAGGATCGCAATATGCT 60.0 °C 92.8% 114 bp SDHA GCAACAGAAGAAGCCCTTTG GTTTTGTCGATCACGGGTCT 58.1 °C 98.5% 108 bp HPRT CTTTGCTGACCTGCTGGATT TCCCCTGTTGACTGGTCATT 59.1 °C 98.0% 117 bp

Table 2.3. Geometric mean (gmean) of relative expression and 95% CI by cancer type and for the entire panel. Tumor type GHR PRLR GH2 PRL gmean 95% CI gmean 95% CI gmean 95% CI gmean 95% CI Breast 34.2 0.6-1904.0 1401.0 42.9-45703.0 1.3 0.8-2.0 1.2 0.9-1.8 CNS 18.3 4.3-78.2 31.6 9.0-110.7 1.0 0.9-1.2 1.0 1.0-1.1 Colon 1.0 1.0-1.1 5.7 1.6-19.9 1.7 0.6-4.3 1.0 1.0-1.0 Leukemia 2.8 0.3-27.7 1.2 0.7-2.1 1.1 0.8-1.5 1.8 0.9-3.9 Melanoma 1348.0 518.7-3503.0 15.8 2.2-115.8 1.0 1.0-1.0 2.7 0.9-8.2 NSC Lung 81.1 16.4-400.0 15.9 2.8-91.7 1.1 0.9-1.3 1.8 1.0-3.2 Ovarian 46.3 5.9-362.3 4.2 0.5-38.5 1.0 1.0-1.0 1.0 1.0-1.0 Prostate 57.6 0.0-1.5x1011 1.0 1.0-1.0 1.0 1.0-1.0 1.0 1.0-1.0 Renal 8.1 1.7-38.3 2.6 1.2-5.7 1.2 0.8-1.9 1.0 0.9-1.2 Overall 28.5 13.6-59.8 11.2 5.6-22.4 1.1 1.0-1.3 1.4 1.1-1.7

Figures

Figure 2.1. Overall mRNA expression in the NCI60 panel by gene. Boxes are interquartile range, whiskers show 10-90 percentiles and dots indicate individual cell lines outside of this range. The line indicates median values. Values of zero are not shown.

Figure 2.2. Relative mRNA expression in the nine tumor types of the NCI60 cell panel. Expression of GHR (A), PRLR (B), GH (C) and PRL (D) was analyzed by real-time RT qPCR using the delta Cq method. Results are normalized to two reference genes and corrected for plate- to-plate variability. Each symbol represents the expression of a single cell line, organized by cancer type. 99

Figure 2.3. Relative mRNA expression by cell line in the NCI160 panel. GHR (dark blue), PRLR (orange), GH2 (light blue) and PRL (yellow) are displayed left to right. Cell lines are organized alphabetically within each cancer type.

100

Figure 2.4. GHR mRNA expression in human melanoma tumors. A human melanoma tumor cDNA array was analyzed for GHR expression by real-time RT qPCR using the delta Cq method. Results are normalized to one reference gene and scaled to be comparable to the quantification of GHR in NCI60 cell lines as presented in figures 1 and 2. Three normal skin (white bars), 21 stage III melanoma and 19 stage IV melanoma (black bars) biopsies are shown.

101

Figure 2.5. GHR, IGF-1 and IGF-1R mRNA levels in metastatic melanoma samples analyzed by sex (A) and tumor grade (B). Geometric mean gene expression with 95% CI is shown relative to male (a) or tumor stage III (b) groups.

102

Figure 2.6. CCLE GHR mRNA expression levels from 947 cell lines in 36 tumor classifications. Expression levels are shown by box-and-whisker plots. The median expression level is indicated by a line, the inter-quartile range is shown by a box, and bars show 1.5× the inter-quartile range. The number of samples for each cancer type (n) is shown in parentheses. Gene expression data was extracted from CCLE_Expression_Entrez_2012-10-18.res and is presented as gene-centric Robust Multi-array Average-normalized mRNA expression levels. This data was extracted from the publicly accessible CCLE database (see Materials and methods).

103

Supplementary figures

GHR PRLR 10000 10000

1000 1000

100 100

10 10

1 1

GeoMean Relative Expression Relative GeoMean GeoMean Relative Expression Relative GeoMean

CNS CNS Breast Colon Renal Breast Colon Renal OvarianProstate OvarianProstate LeukemiaMelanomaNSC Lung LeukemiaMelanomaNSC Lung

GH PRL 10000 10000

1000 1000

100 100 10 10

1 GeoMean Relative Expression Relative GeoMean 1 CNS

Breast Colon Renal Expression Relative GeoMean OvarianProstate Leukemia MelanomaNSC Lung CNS Breast Colon Renal OvarianProstate LeukemiaMelanomaNSC Lung

Supplementary figure 2.1. Geometric mean GHR, PRLR, GH2 and PRL mRNA expression with 95% CI.

104

IGF1 in Human Melanoma Tumor Samples Skin Stage III Stage IV 100000

10000

1000

100

10 Relative expression Relative

1 2 3 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9

10 11 12 13 14 15 16 17 18 19 20 21 10 11 12 13 14 15 16 17 18 19

IGF1R in Human Melanoma Tumor Samples Skin Stage III Stage IV 100000

10000

1000

100

10 Relative expression Relative

1 2 3 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9

10 11 12 13 14 15 16 17 18 19 20 21 10 11 12 13 14 15 16 17 18 19

Supplementary figure 2.2. IGF1 (A) and IGF1R (B) mRNA expression in human melanoma tumors. A human melanoma tumor cDNA array was analyzed for GHR expression by real-time RT qPCR using the delta Cq method. Results are normalized to one reference gene and scaled to be comparable to the quantification of GHR in NCI60 cell lines as presented in figures 1 and 2. Three normal skin (white bars), 21 stage III melanoma and 19 stage IV melanoma (black bars) biopsies are shown.

105

106

Supplementary figure 2.3. Expression of PRLR (A), GH (B) and PRL (C) from CCLE.

107

CHAPTER 3: GROWTH HORMONE MODULATES PROLIFERATION AND

INTRACELLULAR SIGNALING IN MELANOMA CELL LINES

Elahu S Gosney1,2, Riia K Junnila1, John J Kopchick1,2,3

1Edison Biotechnology Institute, Ohio University, Athens, Ohio

2Molecular and Cellular Biology, Program Dept of Biological Sciences, Ohio University, Athens

Ohio

3Dept of Biomedical Sciences, Heritage College of Osteopathic Medicine, Ohio University,

Athens, Ohio

Corresponding author: John J Kopchick, 1 Water Tower Drive, Athens, Ohio 45701, [email protected]

Short title: GH action and signaling in melanoma cell lines

Keywords: Growth hormone, growth hormone receptor, melanoma, signaling pathways, cell proliferation, STAT1, STAT3, STAT5, MAPK, Erk1/2, PI3K, Akt, mTOR

Abstract

Advanced metastatic melanoma is a disease with high mortality and limited treatment options. In a previous study, we found high expression of growth hormone receptor (GHR) mRNA in human tumors and in all nine of the melanoma cell lines included in the National

Cancer Institute’s NCI60 cancer cell line panel. GH is capable of modulating many intracellular 108

signaling pathways that are implicated in carcinogenesis, including JAK/STAT, MAPK/Erk,

PI3K/Akt and mTOR. The impact of GHR activation on melanoma cells has not been reported previously. In this study, we report the effect of GH treatment on proliferation of the NCI60 melanoma lines MALME-3M, MDA-MB-435, SK-MEL-5, SK-MEL-28 and UACC-62. A high dose of GH increases proliferation in a majority of the lines tested. GH elicits a biphasic response in two of the cell lines, in which a high dose stimulates proliferation while a low dose inhibits it.

In three of the cell lines, we examined the effect of GH on signal transduction pathways. Erk activation was unaffected by GH with up to 60 minutes of treatment. GH altered the phosphorylation state of Akt, mTOR, STAT1, STAT3 and STAT5 in a manner that varied with cell line, GH dose and the duration of treatment. Taken together, we show that melanoma cell lines possess active GHRs that can modulate multiple signaling pathways and alter cell proliferation. Further work is necessary to determine the biological significance of GH action in melanoma.

Introduction

Metastatic melanoma is an aggressive disease that has been increasing in incidence in recent years (American Cancer Society 2013). Advances in understanding the biomolecular profile of melanoma have delineated sub-types that may be vulnerable to treatment with targeted therapeutics (Chandra & Pavlick 2012). Several cell types in skin, including melanocytes, express growth hormone receptor (GHR) (Lobie et al. 1990, Tavakkol et al. 1992, Mertani et al. 1995,

Chhabra et al. 2011), which is known to modulate a variety of signal transduction pathways involved in carcinogenesis (Lobie et al. 1990, Tavakkol et al. 1992, Mertani et al. 1995, Chhabra et al. 2011). Growth hormone (GH) induced signaling has been investigated for its role in prostate (Weiss-Messer et al. 2004, Bidosee et al. 2009, Fuhrman et al. 2005, Wang et al. 2005) 109

and breast cancer (Brunet-Dunand et al. 2009, Ratkaj et al. 2010, Shen et al. 2007, Zhang et al.

2007), but studies in other cancer types are scarce.

We recently investigated the expression of GHR mRNA in 60 cancer cell lines from nine different types of cancer (manuscript in submission). Melanoma shows the highest and most consistent GHR expression in this dataset. Our study also quantified mRNA expression of GHR in stage III and IV human metastatic melanoma tumor samples. A majority of these tumors express GHR, although a subset of samples did not. We observed an increase of GHR expression in males compared to females and a statistically significant increase in GHR in stage IV vs. stage

III samples. Others have reported finding GHR protein expression in human melanoma cell lines and tumor samples by immunohistochemistry (Lincoln et al. 1999, Ginarte et al. 2000). Two recent case studies offer intriguing evidence that GH may play a causative role in the development of melanoma. In one case, a 26 year old patient developed malignant melanoma following multiple hormone replacement therapy, which included GH (Caldarola et al. 2010). In the second report, a middle-aged couple taking GH for a supposed anti-aging benefit were both diagnosed with melanoma within three months of the start of their ill-advised “fountain of youth” regimen (Handler et al. 2012). Although a causative link was impossible to prove, the short period of time between the initiation of GH exposure and the melanoma diagnoses was striking.

GH is known to activate a variety of intracellular signal transduction pathways. The best- characterized of these is the Janus-kinase 2 (JAK2) and signal transducer and activator of transcription 5 (STAT5) signaling cascade. GH activation of JAK2/STAT5 is known to induce expression of IGF1 in the liver and regulates a host of other genes (Waters & Brooks 2012). The role of STAT5 in carcinogenesis is not clearly defined. STAT5 has been shown to induce proliferation of breast cancer cell lines and confer resistance to chemotherapy in vitro. However, nuclear STAT5 appears to be a positive prognostic indicator for breast cancer and STAT5 110

disruption in mice leads to liver cancer (Furth et al. 2011, Mueller et al. 2011). GH has been shown to regulate the activity of other STAT proteins including STAT1 and STAT3 (Smit et al.

1996, Yoo et al. 2011). While STAT1 has not been thoroughly examined in melanoma, it has been shown to have a beneficial role in mediating the anti-cancer activities of interferon-γ, yet an in vivo model shows that STAT1 could promote melanoma metastasis (Yan et al. 2011, Schultz et al. 2010). STAT3 has been suggested as a therapeutic target in melanoma due to its role in promoting tumor cell growth, metastasis and tumor angiogenesis (Kortylewski et al. 2005, Wang et al. 2012). Thus, current evidence for the role of STATs in the induction or progression of a variety of cancers is mixed.

GH has been observed to activate MAPK/Erk signaling in some cell types but not others

(Love et al. 1998, Weiss-Messer et al. 2004). PI3K/Akt signaling is known to mediate anti- apoptotic effects and has also been shown to be activated by GH in hepatoma and prostate cancer cells (Jin et al. 2008, Weiss-Messer et al. 2004). Constitutive activation of MAPK/Erk and

PI3K/Akt pathways is often observed in human melanomas, making them attractive therapeutic targets (Yajima et al. 2012, Russo et al. 2009). Downstream of PI3K/Akt, mTOR is a central regulator of metabolic programming that alters a range of processes from autophagy to the

Warburg effect (Zoncu et al. 2011, Sun et al. 2011). mTOR has been shown to be activated by

GH in hepatoma cells and has recently been described as a target for novel melanoma therapeutics (Hayashi & Proud 2007, Populo et al. 2012).

There is a convergence of evidence from epidemiological, in vitro, animal and human studies that implicates a potential role for GH in carcinogenesis. A limited number of cancer cell lines have been subject to treatment with GH. Most of these studies have been conducted with breast and prostate cell lines. Our previous finding of high GHR mRNA expression in human melanoma cell lines and tumors has led us to hypothesize that GH may promote the growth of 111

melanoma cells by modulation of STAT, MAPK, PI3K/Akt and/or mTOR signaling. The effect of GH on growth and intracellular signaling pathways in melanomas has not been previously reported. Therefore, we set out to examine the effect of GH on cell proliferation and signal transduction pathways in NCI60 melanoma lines. We report the effect of GH treatment on cell proliferation and the activation status of Erk1/2, Akt, STAT1, STAT3 and STAT5 intracellular signaling pathways in melanoma cell lines.

Materials and methods

NCI60 melanoma cell lines, culture

Melanoma cell lines MALME-3M, MDA-MB-435, SK-MEL-5, SK-MEL-28 and

UACC-62 were obtained from the National Cancer Institute’s Development Therapeutics

Program. All cell lines were maintained in complete media composed of RPMI 1640 media containing 2.5 mM L-glutamine (Thermo Scientific SH30027.FS) and supplemented with 5% fetal bovine serum (FBS; Thermo Scientific SH30071.03) and 1x antibiotic-antimycotic containing penicillin, streptomycin and amphotericin B (Gibco 15240-062) at 37°C in presence of

5% CO2.

Proliferation assays

Recombinant human GH (hereafter referred to as GH) was produced using a method modified from Patra et al. (2000) and validated for bioactivity by detection of STAT5 activation in mouse L-cells that express a truncated mouse GHR and also in mouse liver in vivo (see appendix 6). Human melanoma cells in passage 3-8 were seeded on 96-well plates at 2,000 to

6,000 cells/well depending on the cell line and allowed to attach overnight in 100 µl of full 112

media. On the second day, media was replaced with RPMI 1640 media containing 0%, 0.5% or

5% FBS. The following day, GH diluted in serum-free media was added to each well in a volume of 10 µl and cells were allowed to grow for 48 hours. The degree of proliferation was determined by the resazurin-based PrestoBlue Cell Viability Reagent (Invitrogen A-13261). In this assay, 10

µl of reagent was added to each well and plates were then gently mixed on a plate shaker for 10-

15 seconds and incubated at 37°C for 20-60 minutes. Fluorescence was read on a SpectraMax M2 microplate reader (Molecular Devices). In pre-experimental optimization trials, differences in apparent cell proliferation were noted across the plate horizontally and vertically. To overcome this bias, cells were seeded in a random pattern and treatments were distributed equally across the plate. The ability of PrestoBlue to accurately portray cell number was validated for each cell line in cell dilution tests (supplementary figure 3.1). An initial dose-response study was designed to determine the GH concentration(s) most likely to elicit a biological effect in subsequent in-depth studies. Dose-response data over several orders of magnitude of GH treatments was obtained for all five cell lines, after which three (SK-MEL-5, UACC-62 and MDA-MB-435) were selected for a more comprehensive analysis at a low (0.05 nM) and high (100 nM) concentration of GH. The dose-response studies were conducted in a single experiment with four technical replicates, while subsequent studies were performed with six technical replicates and at least three independent experiments.

Analysis of MAPK/Erk, PI3K/Akt and mTOR signaling

MDA-MB-435, SK-MEL-5 and UACC-62 cell lines were treated short-term with GH, after which phosphorylation of Erk1/2, Akt and mTOR was assessed by ELISAs. Cells were seeded on 12-well plates and grown to 75-90% confluency, then starved for 16 hours in RPMI without serum. The starved cells were treated in duplicate wells for 10, 30 or 60 minutes with the 113

same GH doses used in the main proliferation assays, receiving 0.05 nM, 100 nM GH or no treatment (control). Total protein was collected by adding 150 µl of cold RIPA buffer per well and shaking the plates for 15 minutes at 4°C. Samples were centrifuged for 10 minutes at 15,000 rpm on a tabletop microcentrifuge at 4°C. Supernatants were stored for later use at -80°C. Total and phospho-Akt (S473) and total and phospho-Erk1/2 (T202/Y204 of Erk1 and Y185/Y187 of

Erk2) ELISAs were purchased from Ray Biotech (PEL-Akt-S473-002 and PEL-Erk-T202-002).

These ELISA kits include cell lysates that serve as a positive control (A431 cells treated with epidermal growth factor). ELISAs for total and phosphorylated (S2448) mTOR were obtained from Cell Signaling (7974S and 7976S). Manufacturers’ protocols were followed to complete the assays. Absorbance was measured on a Spectramax M1 microplate reader (Molecular Devices).

In-cell STAT signaling assay

Cell-based ELISA assays were utilized to measure total and phosphorylated STAT levels.

MDA-MB-435, SK-MEL-5 and UACC-62 cells were grown on 24-well plates (Seahorse

Biosciences, 100777-004) until they reached 75-90% confluence. These plates were utilized for this assay because they have specialized wells with sloped walls that allow gentle washes and a growth area similar to a well in a 96-well plate. Cells were switched to serum-free media and starved for 16 hours, and then treated with 0, 0.05 nM or 100 nM GH for 10, 30 or 60 minutes.

Untreated controls were assayed in duplicate, while GH-treatment was done in single wells at each time point. In order to allow statistical tests with limited data points, the 10 and 30 minutes time points were combined for statistical analysis. Total and phosphorylated STAT1 (Y701),

STAT3 (Y705) and STAT5 (Y694) were measured by cell-based ELISA (Ray Biotech, CBEL-

Stat-SK) as per the manufacturer’s protocol. 114

Data analysis

All data was analyzed by GraphPad Prism 5 (GraphPad Software). In all cell proliferation assays, untreated samples were set to one to normalize the data. For the low and high GH proliferation experiments, data from multiple independent experiments (n=3-7) were analyzed by randomized block design ANOVA (referred to as “repeated measures ANOVA” in Graphpad

Prism). One-way ANOVA was used for analysis of basal phosphorylation rates. For the in-cell

STAT ELISAs, the 10 and 30 minute time points were combined and analyzed by one-way

ANOVA. A t-test was used to analyze GH modulation of Erk, Akt and mTOR phosphorlation. A

P-value of <0.05 was considered significant. All ANOVA analyses were followed by Tukey’s post- hoc test.

Results

Dose Response of five melanoma cell lines to GH

Five melanoma cell lines were treated with different concentrations of GH to assess dose- response. This assay was not designed to obtain precise dose response data, but to provide a basis to determine which GH dose(s) and serum condition(s) to follow up with in more robust cell proliferation assays (presented in the next section). This preliminary dose response assay was not subject to statistical analysis, so data must be interpreted with caution. In this initial proliferation test, the effect of serum concentration during GH treatment was examined with 0%, 0.5% or 5%

FBS. Serum concentration appears to impact the response to GH treatment, with a trend for a greater proliferative effect in reduced-serum conditions observed in most cell lines. MALME-3M, however, appears to show a slight decrease in proliferation with increasing GH doses, independent of the serum concentration (figure 3.1A-C). With 0.5% and 5% serum, the dose 115

response curve has a slight “U”-shape with a minimum at 5 nM GH, indicative of a potential biphasic response. While MALME-3M appears to be affected by treatment with GH, the magnitude of the effect is small. SK-MEL-28 without serum (figure 3.1D) shows a modest decrease at all doses with a minimum at 0.05 nM GH. This cell lines also shows minor decreases in proliferation at most doses of GH, but there is no clear dose-response pattern (figure 3.1E and

1F).

SK-MEL-5 appears to show an increase in proliferation with high GH in 0% and 0.5% serum but a decrease in 5% serum (figure 3.1G, H and I). Treatment of UACC-62 with GH (0% serum) induces a linear dose-dependent increase in cell growth (figure 3.1J). A similar, though smaller, effect is seen with 0.5% serum (figure 3.1K). MDA-MB-435 cells show the greatest differences upon GH treatment. Under serum-free conditions, MDA-MB-435 exhibits a biphasic dose response with a minimum at 0.05 nM GH and a maximum at 100 nM GH (figure 3.1M). A blunted effect is observed with 0.5% serum in MDA-MB-435 (figure 3.1N). The response with

5% serum was different from 0% and 0.5% and difficult to assess due to a high growth rate (data not reported). SK-MEL-5, UACC-62 and MDA-MB-435 cells were selected for further study.

Effect of low and high GH treatment on proliferation of SK-MEL-5, UACC-62 and MDA-MB-435

Based on the results from the preliminary dose-response study described above, three cell lines (SK-MEL-5, UACC-62 and MDA-MB-435) were treated with a low (0.05 nM) and high

(100 nM) dose of GH with either 0% or 0.5% FBS (figure 3.2). In SK-MEL-5, there is a decrease in proliferation at the low GH dose and an increase with high GH treatment in both 0% and 0.5% serum (figure 3.2A and B). While the decrease in the presence of 0.5% serum is only 5.9%, it is statistically significant (P<0.05). In UACC-62, there is no significant difference in proliferation between the low-dose and untreated groups (figure 3.2C and D). The high dose of GH, however, 116

significantly increases proliferation under both serum conditions, showing a 13.8% increase in serum-free media and 15.2% with 0.5% serum (P<0.05 and P<0.001, respectively). In MDA-

MB-435, which showed the greatest changes in the initial dose-response study, we observe a clear biphasic response. In serum-free media, MDA-MB-435 shows a decrease of 8.3% with low dose

GH (P<0.05; figure 3.2E and F) and an increase of 32% with the high GH dose. No significant differences were observed in MDA-MB-435 in the presence of 0.5% serum (figure 3.2F).

Basal activation of Erk, Akt and mTOR in SK-MEL-5, UACC-62 and MDA-MB-435

Substantial differences exist in the degree of basal phosphorylation of Erk, Akt and mTOR (figure 3.3; statistical analysis by one-way ANOVA). The three cell lines differ significantly in the level of basal Erk phosphorylation (P<0.05). SK-MEL-5 shows the highest ratio of P-Erk/Erk, followed by UACC-62, while MDA-MB-435 displays the lowest. For P-Akt,

MDA-MB-435 is again the lowest, while UACC-62 shows significantly higher levels than either of the other cell lines (P<0.05). In UACC-62, P-mTOR/mTOR is significantly lower than in

MDA-MB-435 (P<0.05) and both were significantly lower than SK-MEL-5 (P<0.01). Basal phosphorylation of STAT3 is similar in SK-MEL-5 and UACC-62, while MDA-MB-435 is significantly lower than both of the other lines (P<0.05). Both STAT1 and STAT5 show similar patterns of basal phosphorylation. UACC-62 has significantly higher rates of basal STAT1 and

STAT5 phosphorylation than the other two lines (P<0.05).

MAPK/Erk

Activation of MAPK/Erk signaling was assessed by measuring total and phosphorylated

Erk1/2 by ELISA. As shown in figure 3.4A-C, there is no discernible effect of GH treatment on

Erk1/2 phosphorylation in any of the cell lines examined (SK-MEL-5, UACC62, MDA-MB-435). 117

PI3K/Akt

In order to examine the impact of GH on PI3K/Akt signaling, we determined the levels of total and phosphorylated Akt. In SK-MEL-5, no substantial changes are observed with 0.05 nM

GH (figure 3.4D), but the 100 nM GH dose increases Akt phosphorylation after 60 minutes, although this difference was not statistically significant. Both the low and high doses of GH show a trend for decreased Akt phosphorylation in UACC-62 after 60 minutes of treatment (figure

3.4E). In MDA-MB-435, treatment with 0.05 nM GH causes a significant decrease in Akt phosphorylation at 30 minutes (P<0.05). By 60 minutes, P-Akt returned to baseline levels. A similar trend is noted for treatment with 100 nM GH in this cell line.

mTOR

To assess whether GH treatment could affect metabolic pathways regulated by mTOR, we determined its phosphorylation status. mTOR activation is significantly induced in SK-MEL-5 by high dose GH as early as 10 minutes after treatment (P<0.05) and appears to be maintained for at least 30 minutes (figure 3.4G). At 30 minutes, low dose GH also increases mTOR phosphorylation. In UACC-62, mTOR phosphorylation is increased from baseline with both doses at all time points measured (figure 3.4H). The increase reached significance with low dose

GH at 10 minutes (P<0.05). In contrast, phosphorylation of mTOR in MDA-MB-435 is inhibited at 10 minutes by low dose GH. Both doses significantly decrease mTOR phosphorylation 60 minutes after treatment (figure 3.4I; t-test P<0.05).

118

STAT1, STAT3 and STAT5

We examined the effect of low and high dose GH treatment on phosphorylation of

STAT1, STAT3 and STAT5 at 10, 30 and 60 minutes (figure 3.5). In all cases, the 10 and 30 minutes time points were similar, therefore they were combined for statistical analysis. In SK-

MEL-5 cells, high dose GH elicits a significant increase in P-STAT1 (P<0.05). The increase appears as early as 10 minutes and lasts at least 30 minutes (figure 3.5A). In contrast, the other two cell lines both show a trend for a decrease in P-STAT1 with GH treatment at both doses after

10 and 30 minutes (figure 3.5B and C), an effect which is more pronounced in UACC-62 than

MDA-MB-435. In UACC-62, P-STAT1 returns to basal levels by 60 minutes, while it declines further in MDA-MB-435.

In all three cell lines, P-STAT3 levels are consistently decreased with both doses of GH at 10 and 30 minutes (figure 3.5D, E and F). At most time points, the reduction in P-STAT3 is greater with the low dose than with the high dose. In SK-MEL-5 and UACC-62, the low dose significantly decreases the level of STAT3 phosphorylation (P<0.01 and P<0.05, respectively).

The high dose elicits a significant decrease in UACC-62 (P<0.05).

STAT5 is the best characterized STAT associated with GH signaling. Phosphorylation of

STAT5 increases in SK-MEL-5 and MDA-MB-435 with high dose GH treatment (figure 3.5G;

P<0.05). The low dose does not significantly alter phosphorylation in either cell line. In UACC-

62, the low dose treatment transiently lowers P-STAT5 (P<0.01). In this cell line, the high dose does not initially show an effect but appears to increase P-STAT5 levels by 60 minutes.

Discussion

Metastatic melanoma is an aggressive disease that causes nearly 10,000 deaths each year in the United States (American Cancer Society 2013) and more than 65,000 deaths worldwide

(Lucas et al. 2006). We have previously uncovered high GHR mRNA expression in nine 119

melanoma cell lines from the NCI60 panel (manuscript in submission). There is substantial evidence that GH can regulate many intracellular signaling pathways that influence key processes in carcinogenesis. In this study, we present the first evidence that GH treatment of melanoma cells can modulate the activity of several intracellular signaling molecules and alter cell proliferation.

GH exerts a biphasic effect on proliferation in melanoma cells

Given the high GHR expression in melanoma cell lines, we anticipated that there would be a robust proliferative response to GH treatment. In our cursory dose response experiments with five melanoma cell lines (MALME-3M, SK-MEL-28, SK-MEL-5, UACC-62 and MDA-MB-

435) we observed moderate changes in proliferation. While the effect of GH treatment on proliferation was not great in absolute terms, the consistency and reproducibility of the findings are striking. In several dose-response experiments, we observed a biphasic effect of GH. This led us to conduct a more robust study to determine the effect of high and low GH dose on cell proliferation with SK-MEL-5, UACC-62 and MDA-MB-435. We chose to use a 100 nM (2,200 ng/ml) GH dose for our high-dose treatment because it caused the greatest increase in proliferation in the initial dose-response study, and we chose 0.05 nM (1.1 ng/ml) as our low-dose because it had an inhibitory effect on proliferation in some cell lines. The low dose is within the range of circulating GH levels in adult humans, while the high dose is much greater than physiological serum concentrations (Van den Berg et al. 1996). The production of autocrine and/or paracrine GH in the proximity of tumors could affect the local GH concentration in tumor tissue, but to our knowledge, the concentration of GH in human tumors has not been reported. In three melanoma cell lines examined in this study (SK-MEL-5, UACC-62 and MDA-MB-435), treatment with high dose GH elicited an increase in proliferation, with the greatest effect seen in 120

MDA-MB-435. GH has been reported to induce proliferation of MCF7 breast cancer and Molt4 leukemia cell lines at doses similar to our high dose (Biswas & Vonderhaar 1987, Giesbert et al.

1997, Costoya et al. 2000, Zatelli et al. 2009). In both SK-MEL-5 and MDA-MB-435, we observed decreases in proliferation with a 0.05 nM dose of GH. While these decreases were not large, they were statistically significant and consistent in repeated, independent experiments. In contrast to our findings, Geisbert et al. reported a proliferative response in leukemia cell lines with a dose similar to our low dose, while others have shown inhibition of apoptosis in colon cancer cell lines at this dose (Giesbert et al. 1997, Bogazzi et al. 2004).

The effect of low and high GH on MAPK/Erk, PI3K/Akt and mTOR signaling

To better understand the action of GH in melanoma, we assessed the modulation of

MAPK/Erk, PI3K/Akt and mTOR signaling pathways with a high and low GH treatment (for a summary of effects, see table 1). We observe significant differences in basal phosphorylation of

Erk1/2 between all three cell lines. Although GH treatment was unable to alter Erk1/2 phosphorylation, lower basal activation was associated with a greater proliferative response to high dose GH treatment (figure 3.6). This suggests that lower MAPK/Erk signaling allows other pathways that are responsive to GH treatment to exert a stronger influence on proliferation rates.

It is possible that GH alters Erk1/2 activation at a different time point or dose other than those measured. All three lines have mutations in the proto-oncogene BRAF that activate the

MAPK/Erk pathway and may explain the lack of an effect of GH on Erk1/2 activation in these cells (Ikediobi et al. 2006).

The PI3K/Akt pathway is known to regulate many cellular processes, including suppression of apoptosis and regulation of metabolic programming. One manner in which Akt influences cellular energy balance and cell growth is by activation of mTOR (Hayashi & Proud 121

2007). mTOR activity is often upregulated in cancer and two drugs that inhibit its action are approved for first-line or combination treatment of advanced metasatic renal cell cancer (Gomez-

Pinillos & Ferrari 2012). As expected, in both SK-MEL-5 and MDA-MB-435, basal P-Akt levels parallel basal P-mTOR levels. UACC-62, however, shows high basal P-Akt and low P-mTOR

(figure 3.3). UACC-62 cells have a homozygous inactivating mutation in the PI3K/Akt suppressor PTEN (Ikediobi et al. 2006), thus the high basal activation of Akt is expected. The disconnect between Akt and mTOR activation in this cell line is an interesting finding that warrants future investigation.

mTOR is a key pathway in melanoma that can promote tumorigenesis (Xie et al. 2013).

With GH treatment, SK-MEL-5 shows an early activation of P-mTOR that is independent of P-

Akt or Erk1/2 activation. In UACC-62, P-mTOR increases progressively at 10 and 30 minutes with GH treatment, while P-Akt levels decrease early and even further at 60 minutes. This suggests a possible negative regulation of mTOR on Akt activity, a mechanism previously shown by Harrington et al. (Harrington et al. 2005). In MDA-MB-435, low dose treatment decreases mTOR activation at 10 minutes independent of Akt phosphorylation, implying a more direct affect of GH signaling on mTOR activity. This observation conflicts with data from obtained for hepatoma cells that show an Akt-dependent activation of mTOR by GH (Hayashi & Proud 2007).

In summary, each cell line has a unique response to GH treatment with regard to Akt and mTOR activation.

The effect of low and high GH on STAT1, STAT3 and STAT5 signaling

GH is well known to modulate the activity of several STAT proteins (Smit et al. 1996).

We examined the effect of GH treatment on STAT1, STAT3 and STAT5 phosphorylation (for a summary of effects, see table 1). STAT5, the most extensively studied STAT associated with GH signaling, was activated in all three cell lines at the high dose of GH, which also stimulated cell 122

proliferation. If STAT5 activation proves to be involved in increasing cell proliferation, it would be consistent with the findings of Hassel et al. who found that STAT5 has a role in anti-apoptotic signaling in melanoma cells (Hassel et al. 2008). STAT5 activation has also been correlated with advanced stages of melanoma in a fish model of the disease (Schartl et al. 2010), which suggests that the GH-induced increase in STAT5 phosphorylation in human melanoma cells could promote invasion and metastasis.

In the three melanoma lines examined, GH showed a consistent negative regulation of

STAT3 phosphorylation at both doses tested. Surprisingly, in two of the cell lines, this effect was greater at low dose. Our observation of GH suppression of STAT3 is in contrast to findings in leukemia cells in which GH induced activation of STAT3 (Costoya 2000). We observed a mixed response of P-STAT1 to GH treatment. In UACC-62 and MDA-MB-435, both high and low dose

GH lowered P-STAT1 levels, while the high GH dose in SK-MEL-5 increased STAT1 phosphorylation. While others have shown that induction of P-STAT1 is associated with anti- proliferative effects (Lesinski 2007), we did not observe any relationship between STAT-1 phosphorylation and cell proliferation.

In the metastatic melanoma cell lines we examined, GH inhibits STAT3 activation at both a high and low dose. GH has a complex cell line specific pattern of STAT1 and STAT5 regulation that includes both suppression and promotion of tyrosine phosphorylation. Given the strong evidence for a role of STAT signaling in metastatic melanoma, further work to unravel the complex effect of GH on the various STAT proteins is warranted.

Other factors that may contribute to a biphasic effect of GH in melanoma

Examination of the effect of low and high dose GH on cell signaling pathways (see below) did not readily explain the observed biphasic response. GH is known to interact with the 123

prolactin receptor (PRLR) and there is evidence for a functional heterodimer of GHR-PRLR that has distinct signaling characteristics (Somers et al. 1994, Langenheim & Chen 2009).

Interestingly, both cell lines that exhibit a biphasic response co-express PRLR along with

GHR, while the cell line that lacks this response also lacks PRLR expression (table 3.1). A preliminary examination of GH-induced changes in gene expression in MDA-MB-435 has revealed differences in the activation of IGF-1 and IGF2 mRNA expression with low and high dose GH treatment (supplementary figure 3.2). The low and high GH doses induced IGF2 expression at different times, while the high GH alone increased IGF1 expression. Although

IGF2 is not generally considered to be GH-regulated, some reports show an effect of GH on

IGF2 expression in vivo in human liver (von Horn et al. 2002). GH also increases IGF2 mRNA expression in prostate cancer cell lines and levels are elevated in patients with prostate and colorectal cancer (Matuschek et al. 2011, Bidosee et al. 2011). Compared to IGF-1, IGF-2 has a much higher binding affinity for the cancer-associated insulin receptor isoform A (Belfiore &

Malaguarnera 2011). Although these findings require further confirmation, the differential effect of low and high GH on these growth factors may underlie the biphasic response we observed in proliferation. Taken together, our work has revealed a biphasic effect of GH in melanoma and adds to the list of cancer cell lines that show increased growth with GH treatment.

Conclusions

This is the first report on the treatment of melanoma cell lines with GH. Our results show that melanoma cells possess active GH receptors that are capable of modulating multiple signaling pathways in a complex manner. Although GH does not induce dramatic changes in the growth of melanoma lines, significant changes are observed in proliferation and signal transduction pathways. Our major finding is a biphasic effect of GH on the proliferation of 124

melanoma cell lines, with a low dose moderately inhibiting growth while a high dose stimulates growth. We observed both activation and inhibition of signal transduction pathways by low and high doses of GH. The fact that GH doses that differ by a factor of 2,000 are both able to elicit changes in the activation state of several signaling pathways in melanoma is somewhat astounding. The opposing effects of low and high GH doses on cell proliferation observed in two of the cell lines indicates that different doses of GH can have dramatically different effects on the biological state of a cancer cell. This biphasic response is not easily explained by the effect of GH on various cell signaling pathways, and its biological significance is unknown. We believe further investigations will bring a better understanding to the role of GH in cancer and may reveal novel relationships that may be exploited in the diagnosis and treatment of metastatic melanoma.

Declaration of interest

JJK is an inventor of US patent 5350836 entitled “Growth hormone antagonists”. The other authors have nothing to declare.

Funding

This research is supported by the State of Ohio’s Eminent Scholar Program, which includes a gift from Milton and Lawrence Goll, by AMVETS, and by NIH (P01AG031736).

Author contributions

ESG designed and performed a majority of the experiments and was the primary author.

RKJ assisted with experiments, data analysis, preparation of figures and manuscript edits. JJK guided the study and participated in discussions of manuscript organization, content and editing. 125

Acknowledgements

We would like to thank Susan Holbeck and the National Cancer Institute for providing cell lines from the NCI60 panel.

References

American Cancer Society 2013 Cancer Facts & Figures 2013.

Belfiore A & Malaguarnera R 2011 Insulin receptor and cancer. Endocrine-related cancer 18 R125-47. (doi:10.1530/ERC-11-0074; 10.1530/ERC-11-0074).

Bidosee M, Karry R, Weiss-Messer E & Barkey RJ 2011 Growth hormone affects gene expression and proliferation in human prostate cancer cells. International journal of andrology 34 124-137. (doi:10.1111/j.1365-2605.2010.01064.x; 10.1111/j.1365- 2605.2010.01064.x).

Bidosee M, Karry R, Weiss-Messer E & Barkey RJ 2009 Regulation of growth hormone receptors in human prostate cancer cell lines. Molecular and cellular endocrinology 309 82- 92.

Biswas R & Vonderhaar BK 1987 Role of serum in the prolactin responsiveness of MCF-7 human breast cancer cells in long-term tissue culture. Cancer research 47 3509-3514.

Bogazzi F, Ultimieri F, Raggi F, Russo D, Vanacore R, Guida C, Brogioni S, Cosci C, Gasperi M, Bartalena L & Martino E 2004 Growth hormone inhibits apoptosis in human colonic cancer cell lines: antagonistic effects of peroxisome proliferator activated receptor-gamma ligands. Endocrinology 145 3353-3362.

Chandra S & Pavlick AC 2012 Targeted therapies for metastatic melanoma. Dermatologic clinics 30 517-524. (doi:10.1016/j.det.2012.05.001; 10.1016/j.det.2012.05.001). 126

Chhabra Y, Waters MJ & Brooks AJ 2011 Role of the growth hormone–IGF-1 axis in cancer. Expert Review of Endocrinology & Metabolism 6 71-84. (doi:10.1586/eem.10.73).

Fuhrman B, Barba M, Sch\unemann HJ, Hurd T, Quattrin T, Cartagena R, Carruba G & Muti P 2005 Basal growth hormone concentrations in blood and the risk for prostate cancer: a case- control study. The Prostate 64 109-115.

Furth PA, Nakles RE, Millman S, Diaz-Cruz ES & Cabrera MC 2011 Signal transducer and activator of transcription 5 as a key signaling pathway in normal mammary gland developmental biology and breast cancer. Breast cancer research : BCR 13 220. (doi:10.1186/bcr2921; 10.1186/bcr2921).

Giesbert S, Panzer S, Kovar H, Fischer S, Printz D, Gadner H & Panzer-Grumayer ER 1997 Acute leukemias express a functional receptor for the human growth hormone. Annals of Hematology 74 253-257.

Gomez-Pinillos A & Ferrari AC 2012 mTOR signaling pathway and mTOR inhibitors in cancer therapy. Hematology/oncology clinics of North America 26 483-505, vii. (doi:10.1016/j.hoc.2012.02.014; 10.1016/j.hoc.2012.02.014).

Harrington LS, Findlay GM & Lamb RF 2005 Restraining PI3K: mTOR signalling goes back to the membrane. Trends in biochemical sciences 30 35-42. (doi:10.1016/j.tibs.2004.11.003).

Hassel JC, Winnemoller D, Schartl M & Wellbrock C 2008 STAT5 contributes to antiapoptosis in melanoma. Melanoma research 18 378-385. (doi:10.1097/CMR.0b013e32830ce7d7; 10.1097/CMR.0b013e32830ce7d7).

Ikediobi ON, Davies H, Bignell G, Edkins S, Stevens C, O'Meara S, Santarius T, Avis T, Barthorpe S, Brackenbury L, Buck G, Butler A, Clements J, Cole J, Dicks E, Forbes S, Gray K, Halliday K, Harrison R, Hills K, Hinton J, Hunter C, Jenkinson A, Jones D, Kosmidou V, 127

Lugg R, Menzies A, Mironenko T, Parker A, Perry J, Raine K, Richardson D, Shepherd R, Small A, Smith R, Solomon H, Stephens P, Teague J, Tofts C, Varian J, Webb T, West S, Widaa S, Yates A, Reinhold W, Weinstein JN, Stratton MR, Futreal PA & Wooster R 2006 Mutation analysis of 24 known cancer genes in the NCI-60 cell line set. Molecular cancer therapeutics 5 2606-2612. (doi:10.1158/1535-7163.MCT-06-0433).

Jin H, Lanning NJ & Carter-Su C 2008 JAK2, but not Src family kinases, is required for STAT, ERK, and Akt signaling in response to growth hormone in preadipocytes and hepatoma cells. Molecular endocrinology (Baltimore, Md.) 22 1825-1841. (doi:10.1210/me.2008- 0015; 10.1210/me.2008-0015).

Kortylewski M, Jove R & Yu H 2005 Targeting STAT3 affects melanoma on multiple fronts. Cancer metastasis reviews 24 315-327. (doi:10.1007/s10555-005-1580-1).

Langenheim JF & Chen WY 2009 Development of a novel ligand that activates JAK2/STAT5 signaling through a heterodimer of prolactin receptor and growth hormone receptor. Journal of receptor and signal transduction research 29 107-112. (doi:10.1080/10799890902845252).

Lincoln DT, Sinowatz F, Kolle S, Takahashi H, Parsons P & Waters M 1999 Up-regulation of growth hormone receptor immunoreactivity in human melanoma. Anticancer Research 19 1919-1931.

Lobie PE, Breipohl W, Lincoln DT, Garcia-Aragon J & Waters MJ 1990 Localization of the growth hormone receptor/binding protein in skin. The Journal of endocrinology 126 467- 471.

Love DW, Whatmore AJ, Clayton PE & Silva CM 1998 Growth hormone stimulation of the mitogen-activated protein kinase pathway is cell type specific. Endocrinology 139 1965- 1971.

Lucas R, McMichael T, Smith W & Armstrong B 2006 Solar Ultraviolet Radiation: Global burden of disease from solar ultraviolet radiation.

Matuschek C, Rudoy M, Peiper M, Gerber PA, Hoff NP, Buhren BA, Flehmig B, Budach W, Knoefel WT, Bojar H, Prisack HB, Steinbach G, Shukla V, Schwarz A, Kammers K, Erhardt A, Scherer A, Bolke E & Schauer M 2011 Do insulin-like growth factor associated proteins qualify as a tumor marker? Results of a prospective study in 163 cancer patients. European journal of medical research 16 451-456.

Mueller KM, Kornfeld JW, Friedbichler K, Blaas L, Egger G, Esterbauer H, Hasselblatt P, Schlederer M, Haindl S, Wagner KU, Engblom D, Haemmerle G, Kratky D, Sexl V, Kenner 128

L, Kozlov AV, Terracciano L, Zechner R, Schuetz G, Casanova E, Pospisilik JA, Heim MH & Moriggl R 2011 Impairment of hepatic growth hormone and glucocorticoid receptor signaling causes steatosis and hepatocellular carcinoma in mice. Hepatology (Baltimore, Md.) 54 1398-1409. (doi:10.1002/hep.24509; 10.1002/hep.24509).

Patra AK, Mukhopadhyay R, Mukhija R, Krishnan A, Garg LC & Panda AK 2000 Optimization of inclusion body solubilization and renaturation of recombinant human growth hormone from Escherichia coli. Protein expression and purification 18 182-192. (doi:10.1006/prep.1999.1179).

Populo H, Soares P & Lopes JM 2012 Insights into melanoma: targeting the mTOR pathway for therapeutics. Expert opinion on therapeutic targets 16 689-705. (doi:10.1517/14728222.2012.691472; 10.1517/14728222.2012.691472).

Ratkaj I, Stajduhar E, Vucinic S, Spaventi S, Bosnjak H, Pavelic K & Pavelic SK 2010 Integrated gene networks in breast cancer development. Functional \& integrative genomics 10 11-19.

Russo AE, Torrisi E, Bevelacqua Y, Perrotta R, Libra M, McCubrey JA, Spandidos DA, Stivala F & Malaponte G 2009 Melanoma: molecular pathogenesis and emerging target therapies (Review). International journal of oncology 34 1481-1489.

Schartl M, Wilde B, Laisney JA, Taniguchi Y, Takeda S & Meierjohann S 2010 A mutated EGFR is sufficient to induce malignant melanoma with genetic background-dependent histopathologies. The Journal of investigative dermatology 130 249-258. (doi:10.1038/jid.2009.213; 10.1038/jid.2009.213).

Schultz J, Koczan D, Schmitz U, Ibrahim SM, Pilch D, Landsberg J & Kunz M 2010 Tumor- promoting role of signal transducer and activator of transcription (Stat)1 in late-stage melanoma growth. Clinical & experimental metastasis 27 133-140. (doi:10.1007/s10585- 010-9310-7; 10.1007/s10585-010-9310-7).

Smit LS, Meyer DJ, Billestrup N, Norstedt G, Schwartz J & Carter-Su C 1996 The role of the growth hormone (GH) receptor and JAK1 and JAK2 kinases in the activation of Stats 1, 3, and 5 by GH. Molecular endocrinology (Baltimore, Md.) 10 519-533.

Somers W, Ultsch M, De Vos AM & Kossiakoff AA 1994 The X-ray structure of a growth hormone-prolactin receptor complex. Nature 372 478-481. (doi:10.1038/372478a0).

Sun Q, Chen X, Ma J, Peng H, Wang F, Zha X, Wang Y, Jing Y, Yang H, Chen R, Chang L, Zhang Y, Goto J, Onda H, Chen T, Wang MR, Lu Y, You H, Kwiatkowski D & Zhang H 2011 Mammalian target of rapamycin up-regulation of pyruvate kinase isoenzyme type M2 is critical for aerobic glycolysis and tumor growth. Proceedings of the National Academy of 129

Sciences of the United States of America 108 4129-4134. (doi:10.1073/pnas.1014769108; 10.1073/pnas.1014769108).

Van den Berg G, Veldhuis J, Frölich M & Roelfsema F 1996 An amplitude-specific divergence in the pulsatile mode of growth hormone (GH) secretion underlies the gender difference in mean GH concentrations in men and premenopausal women. Journal of Clinical Endocrinology & Metabolism 81 2460-2467. von Horn H, Ekstrom C, Ellis E, Olivecrona H, Einarsson C, Tally M & Ekstrom TJ 2002 GH is a regulator of IGF2 promoter-specific transcription in human liver. The Journal of endocrinology 172 457-465.

Wang X, Liu P, Liu H, Yang W, Liu Z, Zhuo Z & Gao Y 2012 Delivery of interferons and siRNA targeting STAT3 using lentiviral vectors suppresses the growth of murine melanoma. Cancer gene therapy 19 822-827. (doi:10.1038/cgt.2012.65; 10.1038/cgt.2012.65).

Waters MJ & Brooks AJ 2012 Growth hormone and cell growth. Endocrine development 23 86- 95. (doi:10.1159/000341761; 10.1159/000341761).

Xie X, White EP & Mehnert JM 2013 Coordinate autophagy and mTOR pathway inhibition enhances cell death in melanoma. PloS one 8 e55096. (doi:10.1371/journal.pone.0055096; 10.1371/journal.pone.0055096).

Yajima I, Kumasaka MY, Thang ND, Goto Y, Takeda K, Yamanoshita O, Iida M, Ohgami N, Tamura H, Kawamoto Y & Kato M 2012 RAS/RAF/MEK/ERK and PI3K/PTEN/AKT Signaling in Malignant Melanoma Progression and Therapy. Dermatology research and practice 2012 354191. (doi:10.1155/2012/354191; 10.1155/2012/354191).

Yan J, Wang ZY, Yang HZ, Liu HZ, Mi S, Lv XX, Fu XM, Yan HM, Zhang XW, Zhan QM & Hu ZW 2011 Timing is critical for an effective anti-metastatic immunotherapy: the decisive 130

role of IFNgamma/STAT1-mediated activation of autophagy. PloS one 6 e24705. (doi:10.1371/journal.pone.0024705; 10.1371/journal.pone.0024705).

Yoo KH, Baik M & Hennighausen L 2011 Context-Specific Growth Hormone Signaling through the Transcription Factor STAT5: Implications for the Etiology of Hepatosteatosis and Hepatocellular Carcinoma. Genes & cancer 2 3-9. (doi:10.1177/1947601911405046; 10.1177/1947601911405046).

Zoncu R, Efeyan A & Sabatini DM 2011 Zoncu R., Efeyan A. & Sabatini D. M. 2011 mTOR: from growth signal integration to cancer, diabetes and ageing. Nature Reviews Molecular Cell Biology 12 21-35.

131

Tables

Table 3.1. Summary of GH-induced changes in the phosphorylation status of signaling proteins. The relative level of GHR and PRLR expression in all three cell lines is indicated (see Chapter 2/manuscript in submission). Cell GHR PRLR [GH] P-Erk P-Akt P-mTOR P-STAT1 P-STAT3 P-STAT5 line mRNA mRNA SK- 0.05 nM ↔ ↔ ↑ ↔ ↓** ↑ higher high MEL-5 100 nM ↔ ↑ ↑* ↑* ↓ ↑* UACC- 0.05 nM ↔ ↓ ↑** ↓ ↓* ↓** un- highest 62 100 nM ↔ ↓ ↑ ↓ ↓* ↑ detected MDA- 0.05 nM ↔ ↓* ↓* ↓ ↓ ↓ low- MB- high middle 435 100 nM ↔ ↓ ↓* ↓ ↓ ↑**

132

Figures

Figure 3.1. GH dose-response of melanoma cell lines MALME-3M (A-C), SK-MEL-28 (D-F), SK-MEL-5 (G-I), UACC-62 (J-L) and MDA-MB-435 (M-O). GH doses varied from 0.002/0.02 nM to 100 nM and experiments were done under three different fetal bovine serum conditions: 0%, 0.5% and 5% serum (error bars show the SD). 133

Figure 3.2. Effect of low (0.05 nM) and high-dose (100 nM) GH treatment on proliferation of SK-MEL-5 (A&B), UACC-62 (C&D) and MDA-MB-435 (E&F). Data is presented as the mean from 3-6 individual experiments, each run with 0% as well as 0.5% serum (error bars represent the SD).

134

Figure 3.3. Basal signal stransduction activation in three melanoma cell lines. Ratio of phosphorylated to total Erk1/2, Akt, mTOR, STAT1, STAT3 and STAT5 were measured in untreated SK-MEL-5, UACC-62 and MDA-MB-435 cell lines.

135

Figure 3.4. Effect of GH treatment on Erk, Akt and mTOR activation in three melanoma lines. Erk1/2 (A-C), Akt (D-F) and mTOR (G-I) activation by 0.05 nM and 100 nM GH was recorded at 0, 10, 30 and 60 minutes after treatment. The OD ratio of phosphorylated to total protein is reported (error bars represent the SD).

136

Figure 3.5. Effect of GH treatment STAT1, STAT3 and STAT5 activation in three melanoma lines. STAT1 (A-C), STAT3 (D-F) and STAT5 (G-I) activation by 0.05 nM and 100 nM GH was recorded at 0, 10, 30 and 60 minutes after treatment. OD ratio of phosphorylated and total STAT protein is plotted. To indicate statistical significance, * (P<0.05) or ** (P<0.01) marks differences between untreated and 0.05 nM GH treated cells, while # marks differences between untreated cells and the 100 nM GH treated cells (P<0.05).

137

30% MDA-MB-435 P=0.1176 r2=0.9663 20% UACC-62

10%

SK-MEL-5 0%

Proliferation increase (100 nM GH) nM (100 increase Proliferation 0.15 0.20 0.25 0.30 0.35 0.40 Basal P-Erk/Erk

Figure 3.6. Correlation of basal Erk activity with GH-induced proliferation at 100 nM (0% serum). Increase in proliferation is displayed in percentages whereas Erk activity is reported as the OD ratio of phosphorylated and total Erk.

138

Supplementary figures

Supplementary figure 3.1. A representative cell dilution curve that validates the PrestoBlue proliferation assay.

139

Supplementary figure 3.2. The impact of low and high GH treatment on the expression of GHR, IGF1 and IGF2 in MDA-MB-435.

140

CHAPTER 4: GROWTH HORMONE REGULATES METABOLIC PROGRAMMING

AND THE WARBURG EFFECT IN METASTATIC MELANOMA CELLS

Elahu S Gosney1,2, Riia K Junnila1, John J Kopchick1,2,3

1Edison Biotechnology Institute, Ohio University, 1 Water Tower Drive, Athens, Ohio

2Molecular and Cellular Biology Program Dept of Biological Sciences, Ohio University, Athens

Ohio

3Dept of Biomedical Sciences, Ohio University Heritage College of Osteopathic Medicine,

Athens, Ohio

Corresponding author: John J Kopchick, 1 Water Tower Drive, Athens, Ohio 45701; [email protected]

Short title: GH potentiates the Warburg effect in melanoma

Keywords: Growth hormone, growth hormone receptor, melanoma, metabolism, Warburg effect, lactate, aerobic glycolysis

Abstract

Almost 90 years ago, cancer cells were shown to have a unique metabolism capable of producing high amounts of energy from lactic acid fermentation. Nearly 80 years, growth hormone (GH) has been known to alter metabolism. Although these two early discoveries were made within years of each other, the effect of GH on cancer metabolism has not been reported. 141

Today, metabolic reprogramming is now considered an emerging hallmark of cancer. At the same time, accumulating evidence strongly suggests a role of GH in carcinogenesis. We have previously shown that melanoma cell lines and tumor biopsies express high levels of GH receptor

(GHR) mRNA and that GH treatment alters signal transduction pathways and affects cell proliferation. In order to determine if GH may alter the metabolic state of cancer cells, we utilized a Seahorse flux bioanalyzer to examine mitochondrial respiration and aerobic glycolysis via measurement of oxygen consumption and extracellular acidification. We observed a biphasic effect of GH on cell metabolism in MDA-MB-435 metastatic melanoma cells. A low dose of GH, within physiological range, caused numerous changes in metabolic programming, including increases in basal respiration and aerobic glycolysis, a higher capacity for mitochondrial and glycolytic energy production, and a general increase in the bioenergetic state. In order to confirm the apparent increase in aerobic glycolysis, lactate production was measured. A low dose of GH induced a dramatic increase in lactate production. Importantly, this effect was able to be reversed by co-treatment with GHR antagonist. The effects of GH were not observed at a high dose. We conclude that a physiological dose of GH is able to energize metastatic melanoma cells and potentiate the Warburg effect. Further studies are urgently needed to determine if the observed effects occur in vivo, and if so, what implications there may be for treatment of this deadly disease.

Introduction

One of the first described actions of growth hormone (GH) was a metabolic one: the suppression of insulin activity, discovered by Nobel-prize winner Bernardo Houssay (Houssay

1936). Since that time, GH has been shown to be involved in the regulation of lipid, glucose and protein metabolism along with resting energy expenditure (Moller & Jorgensen 2009). We 142

recently discovered relatively high expression of growth hormone receptor (GHR) mRNA in human metastatic melanoma tumors and all nine of the melanoma cell lines in the NCI60 panel

(manuscript in submission). Further investigation revealed a biphasic growth response to GH treatment in two of the melanoma cell lines, with the greatest effect observed in MDA-MB-435 cells. In these cells, low dose GH (0.05 nM) modestly inhibits growth whereas a high dose (100 nM) promotes growth (manuscript in preparation/see figure 3.2). Additionally, GH modulates the activation status of several signal transduction proteins in this cell line, including Akt and mTOR.

Inhibitors of the PI3K/Akt/mTOR pathway are under intense study as anti-cancer agents with dozens of clinical trials currently underway (Sheppard et al. 2012). This pathway regulates many biological processes, including metabolic programming (Zaytseva et al. 2012).

In recent years, there has been a renewed interest in an old observation: the fact that cancer cells utilize lactic acid fermentation to generate energy from glucose, even in the presence of oxygen (Koppenol et al. 2011). This observation, known as the Warburg effect (Warburg et al.

1924), was overlooked for decades, but now, metabolic reprogramming is considered an emerging hallmark of cancer biology (Hanahan & Weinberg 2011). A brief refresher in the basics of cellular energy production, depicted in figure 4.1, may be useful in understanding this phenomenon. The utilization of glucose as a fuel source begins with a series of step-wise enzymatic reactions that convert glucose to pyruvate, generating two ATPs for each molecule of glucose. In non-proliferating cells under aerobic conditions, pyruvate is transported into the mitochondria where it is utilized for oxidative phosphorylation through the tricarboxylic acid cycle. This process produces an electrochemical gradient that drives ATP production via ATP synthase. This highly efficient form of energy production consumes oxygen. When oxygen is scarce, cells are unable to undergo oxidative phosphorylation, and so shift to the conversion of pyruvate to lactate in order to generate energy. Highly proliferative cells, including cancer cells, 143

upregulate lactate production even in the presence of oxygen. Lactate is transported out of the cell, carrying with it a proton that acidifies the extracellular space (or media for cells grown in vitro). This process, referred to as “aerobic glycolysis”, yields only 1/18th the amount of ATP from each glucose molecule compared to oxidative phosphorylation, but increases the synthesis of building blocks that are necessary for proliferation, such as amino acids and nucleotides

(Vander Heiden et al. 2009). The term “aerobic glycolysis” is somewhat confusing, because glycolysis is by definition the breakdown of glucose to pyruvate, which can either be converted to lactate and exit the cell or enter the mitochondria and fuel respiration. In keeping with the terminology used in literature, we will refer to energy production through lactic acid fermentation as “aerobic glycolysis”, or referred to as “glycolytic” processes in figures. It was once thought that this change in metabolism was an indirect response to growth factors, which increase energy demand by driving transcription and translation. Strong evidence now exists, though, which shows that growth factors have direct effects on these metabolic pathways in cancer cells (Ward

& Thompson 2012).

A recently established technology, the Seahorse Extracellular Flux Analyzer, has refined and greatly enhanced the biochemical tools available to study metabolism in cancer cells. This instrument allows the simultaneous measurement of the rate of oxygen consumption (OCR) and extracellular acidification (ECAR), surrogate markers of mitochondrial respiration and aerobic glycolysis, respectively. The Seahorse system allows the sequential treatment of live cells with multiple compounds that affect metabolic processes, with continuous monitoring of OCR and

ECAR. For example, treatment with the ATP-synthase inhibitor oligomycin reveals both the amount of oxygen dedicated to mitochondrial ATP production and the ability of a cell to switch to aerobic glycolysis to compensate for a decrease in mitochondrial energy production. Cells can be pre-treated with a drug or protein of interest to determine if changes in metabolic 144

programming occur, a process that typically requires several hours or more of treatment. Effects observed on glycolysis can easily be confirmed by independent methods, such as the direct measurement of lactate production in cell culture media.

There are many clues that GH regulates metabolism in both healthy and diseased tissue.

At the same time, GH may have immediate, direct effects and act through downstream effectors to promote key pathways of carcinogenesis. To date, no one has examined the influence of GH on the Warburg effect, aerobic glycolysis, or the bioenergetic profile of cancer cells. In this study, we build on our previous results that implicate GH in the growth of metastatic melanoma cells.

Here, we examine the effect of GH treatment on metabolic programming in a GH-sensitive metastatic melanoma cell line, MDA-MB-435. We present a novel action of GH: elevation of the bioenergetic state of cancer cells and potentiation of the Warburg effect.

Materials and methods

Cell culture

The metastatic melanoma cell line MDA-MB-435 was provided by the National Cancer

Institute’s Developmental Therapeutics Program. Cells were maintained in complete RPMI 1640 media (Thermo Scientific SH30027.01) containing 2.05 mM L-glutamine, 2 g/l glucose, 5% fetal bovine serum (FBS; Thermo Scientific SH30071.03) and 1x antibiotic-antimycotic containing penicillin, streptomycin and amphotericin B (Gibco 15240-062) at 37°C in presence of 5% CO2.

Cells were split at a 1:10 dilution after reaching 80-90% confluence by manual shaking to collect cells. All assays utilized cells that had been split 3 to 8 times. 145

GH treatment

Recombinant human GH and GH antagonist (GHA) were produced in-house similar to the method of Patra et al. (2000) and their bioactivity was tested by measuring STAT5 activation in vitro utilizing mouse L-cells that express a mouse GHR transgene (see appendix 6).

Recombinant GH was used to treat MDA-MB-435 melanoma cells that were seeded at a density of 48,000 cells per well on 24-well plates (Seahorse Biosciences, 100777-004) in a volume of 500

µl of complete media. Cells were incubated at 37°C in presence of 5% CO2 for 4h. GH was added in an additional 10 µl of media and the cells were incubated for 12 h, after which both a mitochondrial stress test and glycolysis stress test were performed. All treatments were done in triplicate for each assay.

Mitochondrial stress test

A mitochondrial stress test kit (Seahorse Biosciences, 101848-400) was used to analyze metabolic parameters. All reagents were prepared as per the manufacturer’s protocol. Cartridges

(Seahorse Biosciences, 102340-100) were hydrated with XF Calibrant overnight at 37°C without added CO2. Cells were treated with 0, 0.05, 2 or 100 nM GH as described above and then rinsed with XF Assay Media containing 1 mM pyruvate and 25 mM glucose, 675 µl of which was added to each well after rinsing. The plate was then incubated at 37°C without added CO2 for one hour.

The appropriate cartridge injection wells were loaded with 75 µl of each reagent. The plates and cartridges were inserted into the Seahorse XFe24 flux analyzer. Oligomycin, rotenone and antimycin were used at a final concentration of 1 µM, while FCCP was used at 250 mM.

Compounds were injected into the wells sequentially, starting with oligomycin, followed by

FCCP, and a final injection of antimycin and rotenone. Oligomycin is an ATP synthase inhibitor, whereas FCCP is an ionophore that disrupts the proton gradient in the mitochondria and 146

stimulates maximal respiration. The combination of antimycin A and rotenone shuts down the electron transport chain (ETC). OCR and ECAR measurements were collected every nine minutes, starting with two readings at baseline and twice after each treatment. After the Seahorse assay, cells were counted manually using a standard hemocytometer. Although the differences in cell number were small, all data were normalized to the percent change in cell number relative to cells not treated with GH.

Mitochondrial stress test calculations

Key parameters of mitochondrial function and glycolysis were calculated from the mitochondrial stress test as per manufacturer’s recommendations, including basal respiration,

OCR due to ATP synthase, maximal respiration, spare respiratory capacity, non-mitochondrial respiration and proton leak. For all calculations, the mean value of two sequential readings before or after treatment was used. Basal respiration was calculated by subtracting the final OCR readings (non-mitochondrial respiration) from the baseline readings. The amount of OCR due to

ATP production was calculated as the difference between basal respiration and the value obtained after treatment with the ATP synthase inhibitor oligomycin. The mean OCR after FCCP treatment indicates oxygen flux at maximal respiration. The spare respiratory capacity was calculated by subtracting the basal OCR values from the maximal. Non-mitochondrial respiration and proton leak were calculated as suggested by the Seahorse protocols. The change in aerobic glycolysis caused by inhibition of ATP synthase (oligomycin-induced glycolytic flux) was calculated by subtracting basal ECAR from the two readings immediately following oligomycin treatment.

Similarly, FCCP-induced glycolytic flux was calculated by subtracting the ECAR values after oligomycin treatment from those obtained after FCCP treatment. The amount of aerobic glycolysis that is dependent on the electron transport chain (ETC) was calculated as the difference 147

between the final ECAR values and those obtained after FCCP. Spare respiratory reserve capacity was calculated by subtracting basal ECAR from the maximum ECAR values, which were obtained after treatment with oligomycin and FCCP.

Glycolysis stress test

A glycolysis stress test kit (Seahorse Biosciences, 102194-100) was used to assess the effect of altered glycolysis on cellular metabolism. All reagents were prepared as per the manufacturer’s protocol. XF Calibrant was used to hydrate cartridges overnight at 37°C without added CO2. Cells were treated with 0, 0.05 or 100 nM GH as described above. Cells were rinsed with XF Assay Media without pyruvate or glucose. A volume of 675 µl of XF Assay Media lacking pyruvate and glucose was added to each well after rinsing and cells were incubated for one hour at 37°C without added CO2. Cartridge injection wells were filled with 75 µl of each reagent. The plates and cartridges were loaded into the Seahores XFe24 analyzer. The final concentrations of glucose, oligomycin and 2-deoxy glucose were 10 mM, 1 µM and 100 mM, respectively. Compounds were injected in sequence, starting with glucose, followed by oligomycin, and finally, 2-deoxy glucose. OCR and ECAR measurements were measured with the Seahorse XFe24 analyzer and normalized to cell count as for the mitochondrial stress test.

Glycolysis stress test calculations

Glycolytic parameters were calculated as per the manufacturer’s recommendations. In order to calculate glucose-induced aerobic glycolysis, the difference in ECAR before and after glucose addition was calculated. Glycolytic capacity was analyzed by subtracting the final ECAR readings (following 2-deoxy glucose addition) from the readings after oligomycin treatment.

Glycolytic reserve capacity was calculated as the difference between the oligomycin-dependent 148

ECAR and glucose-induced glycolytic flux. Non-glycolytic acidification was the mean final

ECAR reading after 2-DG treatment. OCR measurements were also analyzed to calculate parameters of respiration during the glycolysis stress test. Glucose-induced oxygen flux was calculated by subtracting the baseline OCR values from the values after glucose treatment. The oligomycin-induced oxygen flux was calculated by subtracting the OCR values after glucose addition from those after oligomycin treatment to shut down ATP synthase.

Lactate assay

To measure cellular lactate production, a fluorescent-based L-lactate assay kit was utilized (Cayman, 700510). All reagents were prepared as per the manufacturer’s protocol. MDA-

MD-435 cells were seeded at 24,000 cells per well in a 96-well tissue-culture treated plate (BD

Falcon) and grown for 24 hours in complete media. Cells were subject to a rinse and then 100 µl of fresh complete media was added. Cells were then treated by addition of GH in a 10 µl volume of complete media for a final concentration of 0, 0.05, 2 or 100 nM GH. After 14 hours of GH treatment, media was collected and subject to the lactate assay. Deproteination was accomplished by ultracentrifugation with a 3kD filter (Amicon Ultra-0.5, Ultracel-3; Millipore, UFC500308) and centrifuged for 20 minutes at 21k x g in a 1.5 ml tube. The reaction mix and standard curve was prepared as per the manufacturer’s recommendations. Media was diluted 1:20 in 1x assay buffer. Reactions were performed in a final volume of 200 µl containing 20 µl of diluted samples.

After a 20 minute incubation at room temperature, fluorescence was read at a 530 excitation and

590 emission on a Spectramax M2 (Molecular Devices). The concentration of lactate was calculated based on the standard curve using SoftMax Pro software (Molecular Devices).

Statistical analysis was performed by t-test, with P<0.05 considered significant. 149

Results

Mitochondrial stress test

The bioenergetic properties of the metastatic melanoma cell line MDA-MB-435 were analyzed by treating cells with GH overnight at different doses and then monitoring OCR and

ECAR after the addition of several modulators of mitochondrial activity (figure 4.2). Cells were treated overnight with GH then the effect of sequential addition of oligomycin, FCCP, and antimycin A + rotenone was collected following basal readings. At each time point, OCR values are higher in the low-dose treatment groups (0.05 and 2 nM GH) compared to untreated or the

100 nM GH dose group (figure 4.2A). OCR is nearly identical at all time points for the 0.05 and 2 nM GH groups. The 100 nM GH group is less responsive to FCCP than the control group, but is otherwise nearly identical to untreated cells. FCCP treatment induces significant differences in

OCR between treatment groups. Specifically, at the 36 minute time point, 0.05 and 2 nM GH treatment caused higher OCR compared to both the untreated and 100 nM GH groups (two-way

ANOVA with Bonferrone post-hoc, P<0.05). At the 45 minute time point, the difference in OCR remains significant for both low-dose GH groups compared to the high-dose GH group (P<0.01).

ECAR, an indicator of non-mitochondrial ATP production through aerobic glycolysis, was also monitored at all time points of the mitochondrial stress test. As with OCR, 0.05 and 2 nM GH treatment groups show higher ECAR values at every time point compared to either the untreated or 100 nM GH treated groups (figure 4.2B). Unlike OCR, the ECAR values are lower for 100 nM GH treated cells than for untreated cells at all time points. ATP synthase inhibition with oligomycin and induction of maximal mitochondrial respiration with FCCP cause additive increases in ECAR (figure 4.2B) and lead to maximum aerobic glycolysis. Full inhibition of the electron transport chain with antimycin and rotenone induces decreased ECAR in all groups, but 150

values remain above basal levels for all treatment groups. At all but the first two (basal readings) time points, the 2 nM GH treatment group is significantly higher than the 100 nM GH group

(P<0.05). Additionally, at the 54 minute time point, the 2 nM GH group is significantly different from the untreated group (P<0.05).

The 0.05 and 2 nM GH treated groups show trends for increased oxygen consumption and capacity in all metabolic parameters measured by the mitochondrial stress test (figure 4.3A-C and 2G-I). On the other hand, the 100 nM GH treated group is similar to or slightly lower than control in these measures. This includes basal respiration, ATP synthesis, maximal respiration, spare reserve capacity, non-mitochondrial respiration and proton leak. The greatest differences are seen in the OCR readings for the linked parameters of maximal respiration and spare-reserve capacity (figure 4.3C and H). The maximal mitochondrial respiration is increased by 49% and

48% in the 0.05 and 2 nM GH groups, respectively, compared to control (figure 4.3C) while the

100 nM GH group is 17% lower than control values. The spare reserve capacity showed a similar pattern, with even greater differences. The 0.05 and 2 nM GH treatment groups are 59% and 50% higher in spare capacity while the 100 nM GH group shows a reduction of 30%.

ECAR was also analyzed during the mitochondrial stress test to determine the interplay between mitochondrial parameters and aerobic glycolysis (figure 4.3D-F and J-K). Basal glycolysis appears to be higher in the 2 nM GH treatment group but similar among all other groups. When ATP synthase is inhibited by oligomycin, aerobic glycolysis increases for all groups. The greatest increase is seen in the 0.05 nM GH treated cells (figure 4.3E), which is statistically greater than control values (P<0.05). Cells treated with the highest GH dose show the smallest increase, a difference which is statistically significant compared to the 0.05 and 2 nM

GH groups. The 0.05 nM GH group also shows the greatest level of ETC-dependent aerobic glycolysis (significantly higher compared to 2 nM and 100 nM GH groups, P<0.05), while the 151

100 nM GH treated group has significantly less ETC-linked glycolysis compared to untreated or the 0.05 nM GH groups (P<0.05; figure 4.3J). Figure 4.3K shows the combined effect of ATP- synthase inhibition and induction of maximal respiration by mitochondrial membrane uncoupling on the rate of aerobic glycolysis. GH treatment at 0.05 nM significantly increases this parameter

(P<0.05) compared to the control and 100 nM GH groups.

Glycolysis stress test

Aerobic glycolysis was analyzed by continuous real-time monitoring of ECAR and OCR in cells that were incubated with GH overnight, starved for one hour and fed with glucose, followed by metabolic manipulation with oligomycin and 2-deoxy glucose. Prior to the glycolysis stress test, MDA-MB-435 cells were treated overnight with 0, 0.05 or 100 nM GH then switched to starvation media, which lacks glucose and pyruvate. After two basal readings, glycolysis is activated by addition of glucose. Subsequently, oligomycin is added to inhibit ATP synthase.

Lastly, 2-deoxy glucose is added to fully inhibit the glycolytic pathway (figure 4.4). At all time points, the 0.05 nM GH cells show higher aerobic glycolysis compared to controls, while 100 nM

GH treated cells are lower (figure 4.4A). Following inhibition of ATP synthase with oligomycin treatment in the 0.05 nM GH group, aerobic glycolysis is significantly higher at the 45 minute time point compared to untreated control cells (P<0.05), and it is higher at the 36 and 45 minute time points compared to the 100 nM GH group (P<0.01; figure 4.4A). The glycolysis stress test primarily provided meaningful data from ECAR readings; however OCR was also collected and analyzed (figure 4.4B). The OCR readings were marked by high variability, and no statistically significant differences were found by two-way ANOVA.

Several parameters of glycolytic flux were examined. Basal aerobic glycolysis, glucose- induced aerobic glycolysis, and glycolytic capacity all show a similar pattern of increased ECAR 152

flux in 0.05 nM GH treated cells and a slight decrease in 100 nM GH cells relative to controls

(figure 4.5A-C). The greatest differences were observed in glycolytic reserve capacity. The 0.05 nM GH treated cells show a 386% increase compared to non-treated controls and a 351% increase in glycolytic reserve compared to cells that received the high GH dose (figure 4.5G). The amount of non-glycolytic acidification appears to be higher in both GH treatment groups (figure

4.5H).

In addition to glycolytic parameters, OCR was also measured during the glycolysis stress test. Basal respiration is similar in all groups, with a minor decrease possible for the 100 nM GH group (figure 4.5D). The addition of glucose seems to increase respiration in the untreated cells, but not in either of the GH treated groups (figure 4.5E). As expected, inhibition of ATP synthase by addition of oligomycin causes OCR to decrease in all groups, with the greatest drop occurring in the 0.05 nM GH cells (figure 4.5F).

GH-induced lactate production

In order to confirm the effects of GH on aerobic glycolysis by an independent experimental technique, lactate production was determined in MDA-MB-435 melanoma cells treated with GH for 14 hours (figure 4.6). We observe a biphasic effect of GH on lactate production. The lowest GH dose, 0.05 nM, induces a significant 65% increase in lactate concentration compared to untreated cells (Student’s t-test, P<0.05). Lactate levels are significantly higher in the 0.05 nM GH group compared to the 100 nM GH treatment (P<0.05).

Treatment with 2 nM GH shows a trend for increased lactate production, with a reading 81% higher than controls. We also tested whether the effect of low-dose GH could be reversed by co- treatment with GH antagonist (GHA). The effect of 0.05 nM GH is severely reduced by the addition of 0.5 nM GHA (P<0.05). 153

Discussion

Cancer cells are known to exhibit severe alterations in cellular metabolism (Vander

Heiden et al. 2009, Koppenol et al. 2011). Whether this bioenergetic phenotype is a cause or a consequence of oncogenic processes has been debated for many years and remains unsettled.

There is an emerging view that these alterations in metabolism are not simply a side effect of cancer, but may be crucial to the ability of cancer cells to thrive, adapt and metastasize (Ward &

Thompson 2012). While the precise factors that regulate glucose flux and energy production in cancer cells are still largely unknown, the integration of signals from various growth factors could play a critical role. In previous work, we identified expression of GHR in human metastatic melanoma tumor samples and cell lines (manuscript in preparation). We have also shown that GH treatment of the metastatic melanoma line MDA-MB-435 affects proliferation and the activation status of several signal transduction pathways in a biphasic manner (manuscript in preparation).

Here, we utilized the Seahorse instrumentation and technology to examine the effect of GH on metabolic parameters in metastatic melanoma cells. We present the first evidence that GH modulates mitochondrial respiration, aerobic glycolysis and potentiates the Warburg effect in human metastatic melanoma cells.

GH increases mitochondrial respiration

In order to understand the effect of GH on mitochondrial respiration in metastatic melanoma, a mitochondrial stress test was performed. In this experiment, we treated MDA-MB-

435 cells overnight with GH and then monitored OCR in real-time with sequential addition of various compounds that alter mitochondrial energy production. A clear biphasic effect of GH is observed. The two low GH doses (0.05 and 2 nM) increased OCR at all time points, while the

100 nM was nearly identical to untreated cells. This suggests that GH at physiological 154

concentrations increases respiration in melanoma cells, thus energizing the cells. In previous work with this cell line, we showed that GH doses in this range elicit small but significant decreases in cell proliferation. The consequence of the increased respiration in these cells, and whether it is related to the decrease in proliferation is unknown. FCCP treatment, which short circuits the proton gradient in the mitochondria, was used to determine the spare respiratory capacity. This parameter showed the greatest effect of GH, with a 59% increase with 0.05 nM GH and a 30% decrease with 100 nM GH. The increased spare capacity at 0.05 nM GH means that

GH exposure at concentrations that tumor cells could experience in vivo (Van den Berg et al.

1996) increases the ability of melanoma cells to meet high energy demands. This increased capacity to make high amounts of ATP under metabolically challenging conditions could provide an advantage to a tumor cell that is undergoing processes with high energy demand, such as cell division, migration or stress response.

In the mitochondrial stress test, we also measured ECAR, a surrogate marker of aerobic glycolysis. As with OCR, ECAR was higher at all time points for the 0.05 and 2 nM GH treated cells, although basal aerobic glycolysis was similar between untreated and 0.05 nM GH groups.

When ATP synthase was inhibited with oligomycin, a decrease in OCR was mirrored by an increase in aerobic glycolysis in all groups, likely a compensatory mechanism to meet the energy demands of the cell. Maximal glycolysis occurred after the sequential addition of oligomycin and

FCCP, which first inhibits ATP synthase, then induces maximal respiration by disruption of the proton gradient in the mitochondria. Aerobic glycolysis was stimulated by inhibition of ATP synthase and further stimulated by inducing maximal oxygen consumption. A state of maximum respiration could lead to a general increase in glucose flux. This would raise pyruvate levels in order to feed oxidative phosphorylation in the mitochondria, and as a by-product, more pyruvate would be available for lactate dehydrogenase to convert to lactate. When the ETC was inhibited, 155

aerobic glycolysis decreased. This effect was greatest for the 0.05 nM GH, which was similar to untreated cells, but the effect was significantly blunted in the 2 and 100 nM groups. In cells treated with 2 or 100 nM GH, the amount of aerobic glycolysis that is dependent on the ETC is much less than in the untreated or 0.05 nM GH cells, suggesting that physiological levels of GH may confer a survival advantage to melanoma cells. This observation offers further evidence in support of the hypothesis that lactate production depends partly on mitochondrial-induced glucose flux, which could be elevated by increased glucose transport into the cell, or by increased activity of glycolytic enzymes.

GH increases aerobic glycolysis

To examine the role of GH on aerobic glycolysis, we performed a glycolysis stress test.

Basal glycolysis in starved cells was similar in all groups. When the cells were fed glucose, aerobic glycolysis increased in all groups, with a greater increase observed in the cells treated with 0.05 nM GH. In order to assess the ability of cells to switch their energy source from mitochondrial respiration to aerobic glycolysis (lactate production), cells were treated with oligomycin to block ATP synthase. The 0.05 nM GH treated cells show a nearly 4-fold increase in aerobic glycolysis to compensate for decreased ATP synthase activity, while control cells or those receiving 100 nM GH show only a minor response. In order to confirm the effect on aerobic glycolysis that we observed by ECAR measurements, we measured lactate production after 14 hours of treatment with GH. Either 0.05 or 2 nM GH substantially increases lactate accumulation in cell culture media while treatment with 100 nM GH did not. The acidification caused by elevated aerobic glycolysis could potentially confer an invasive phenotype and promote metastasis (Estrella et al. 2013). Similar to our results, increased lactate production has previously been observed in two melanoma lines distinct from the one we examined (Scott et al. 2011). 156

Additionally, lactate production has been shown to increase in breast cancer cells in response to

PRL treatment (Varghese et al. 2010). OCR readings were highly variable in the glycolysis stress test. We suspect that cell adhesion may be affected by the starvation conditions of this assay, thereby causing more cells to become detached, which could interfere with OCR readings. Taken together, this data shows that a physiological dose of GH stimulates the Warburg effect in melanoma cells.

How might GH regulate the Warburg effect in metastatic melanoma cells? A key player in the regulation of metabolic programming is pyruvate kinase, specifically the PKM2 splice variant which is preferentially expressed at high levels in cancer cells (Luo & Semenza 2012).

Alternative splicing of pyruvate kinase precursor mRNA to produce the M2 variant can be upregulated by MYC (David et al. 2010). MYC is a transcriptional target of autocrine GH action in mammary epithelial cells that undergo oncogenic transformation (Zhu et al. 2005), and presumably other cell types. As a tetramer, PKM2 phosphorylates PEP to pyruvate, providing a substrate that can be utilized for lactate production (Warburg effect) or mitochondrial respiration.

Upon phospohorylation, PKM2 tetramers dissociate into dimers, which possess different binding affinities and enzymatic functions. These dimers are translocated into the nucleus where they directly bind to and regulate several transcription factors, including STAT3 and HIF-1alpha (Gao et al. 2012, Luo & Semenza 2011). Surprisingly, PKM2 has been shown to directly bind to and become phosphorylated by the human prolactin receptor, which is able to be activated by GH

(Varghese et al. 2010). Whether GHR directly interacts with PKM2 is unexplored. A model can be proposed in which GH binding results in direct or indirect phosphorylation of PKM2 causing it to dissociate from tetramers and form dimers that enter the nucleus where they bind to transcription factors including STAT3 and HIF-1alpha resulting in metabolic reprogramming.

HIF-1alpha activation could positively regulate GHR expression, as has been shown in HEK293 157

cells (Erman et al. 2011). Other transcriptional targets of HIF-1alpha in cancer cells include

PKM2 (Kress et al. 1998) and lactate dehydrogenase A (Semenza et al. 1996), the enzyme responsible for converting pyruvate to lactate. This creates the possibility of a self-renewing feed- forward regulation system that reprograms cell metabolism by upregulation of aerobic glycolysis and elevated glycolytic capacity. Further work should be undertaken to determine if GH affects

PKM2 activity, and what the role of JAK and STATs are if such an affect is observed.

Lactate production in melanoma

Our finding of increased lactate production with GH treatment may be of clinical relevance. Expression of lactate dehydrogenase has been examined in human melanoma lesions

(Zhuang et al. 2010). It was observed that expression of the most efficient isoform of lactate dehydrogenase (LDH), LDH5 was associated with tumor thickness and mitotic score as well as reduced survival. Further, serum LDH, presumably of tumor origin, has been described as a prognostic biomarker for patients with metastatic melanoma (Weide et al. 2012). Further refinement comes from Ho et al. who describe two subgroups of stage IV melanoma, those with normal serum LDH and those with elevated LDH (Ho et al. 2012). This suggests that the GH- induced lactate production that we observed could have important clinical consequences if GH has this effect in vivo, and consequently, clinical trials using GHA therapy in patients with advanced metastatic melanoma may be warranted.

Relevance with respect to published studies on metabolic flux in melanoma cells

Before any reports appeared describing the use of the Seahorse flux analyzer to examine metabolism in melanoma, a study using more traditional techniques was described that indicated the dysregulation of metabolism in melanoma cells. Using isotope labeling, Scott and colleagues 158

were able to determine the ultimate destination of metabolized glucose and glutamine in several melanoma cell lines (Scott et al. 2011). Under basal conditions, they observed an increased glucose consumption and lactate production in melanoma cells compared to melanocytes. They further observed an active tricarboxylic acid cycle in all the melanoma cell lines they examined.

Out of 152 cancer studies published that utilize the Seahorse analyzer, a few have examined metabolism in melanoma cells. The first of these showed that melanoma cells have higher levels of oxidative phosphorylation compared to melanocytes, and that the anti-cancer therapeutic

Elesclomol was an inhibitor of mitochondrial metabolism (Barbi de Moura et al. 2012).

Elesclomol was found to be less effective in melanoma cells that exhibit high rates of aerobic glycolysis, suggesting that metabolic programming could underlie sensitivity to certain therapeutics. This same group subsequently published another study that utilized the Seahorse flux analyser to study metabolic flux in human melanoma tumor slices (Ho et al. 2012). That study indicated two metabolically distinct groups of metastatic stage IV melanomas—one with normal serum LDH that utilizes both aerobic glycolysis and mitochondrial respiration and another associated with high serum LDH levels that largely relies on glycolysis. Other work has identified a role for PGC1alpha in delineating the preference of a melanoma cell line for mitochondrial or glycolytic energy production (Vazquez et al. 2013). Another report examined the potential mechanism behind “glycolysis addiction” observed for some melanoma cell lines (Hall et al.

2013). The authors found that an activating mutation in BRAF, which results in constitutively active Erk1/2, was responsible for the strong reliance on glycolysis in two melanoma cell lines.

While our data are preliminary, the results we obtained using the Seahorse flux analyzer are in line with the recent findings of others who have used the same technology to examine metabolic programming in melanoma cells.

159

Conclusions

Our investigation into the effect of GH on metabolic programming in melanoma has revealed strong activation of aerobic glycolysis and modulation of respiration. There has been no reporting on this effect of GH in the literature. We found that short term GH treatment reprograms melanoma cells by inducing higher basal respiration and glycolysis and stimulating the build-up of respiratory and glycolytic reserve capacity. This may confer a survival benefit by allowing the cells to better handle states of high-energy demand and could promote an invasive phenotype. GHR is widely expressed in healthy and diseased tissue. The alteration of metabolic programming by GH could potentially favor greater biosynthesis of macromolecules needed for tumor growth and metastasis. Our novel finding of a dramatic effect of low-dose GH on aerobic glycolysis, mitochondrial respiration and the Warburg effect could have broad implications for human health if confirmed in other disease states.

Declaration of interest

JJK is an inventor of US patent 5350836 entitled “Growth hormone antagonists”. The other authors have nothing to declare.

Funding

This research is supported by the State of Ohio’s Eminent Scholar Program, which includes a gift from Milton and Lawrence Goll, by AMVETS, and by NIH (P01AG031736).

160

Author contributions

ESG designed and performed a majority of the experiments and was the primary author.

RKJ assisted with experiments, data analysis, preparation of figures and manuscript edits. JJK guided the study and participated in discussions of manuscript organization, content and editing.

Acknowledgements

We would like to thank Susan Holbeck and the National Cancer Institute for providing the MDA-MB-435 cell line from the NCI60 panel.

References

David CJ, Chen M, Assanah M, Canoll P & Manley JL 2010 HnRNP proteins controlled by c- Myc deregulate pyruvate kinase mRNA splicing in cancer. Nature 463 364-368. (doi:10.1038/nature08697; 10.1038/nature08697).

Erman A, Wabitsch M & Goodyer CG 2011 Human growth hormone receptor (GHR) expression in obesity: II. Regulation of the human GHR gene by obesity-related factors. International journal of obesity (2005) 35 1520-1529. (doi:10.1038/ijo.2011.10; 10.1038/ijo.2011.10).

Estrella V, Chen T, Lloyd M, Wojtkowiak J, Cornnell HH, Ibrahim-Hashim A, Bailey K, Balagurunathan Y, Rothberg JM, Sloane BF, Johnson J, Gatenby RA & Gillies RJ 2013 Acidity generated by the tumor microenvironment drives local invasion. Cancer research 73 1524-1535. (doi:10.1158/0008-5472.CAN-12-2796; 10.1158/0008-5472.CAN-12-2796).

Gao X, Wang H, Yang JJ, Liu X & Liu ZR 2012 Pyruvate kinase M2 regulates gene transcription by acting as a protein kinase. Molecular cell 45 598-609. (doi:10.1016/j.molcel.2012.01.001; 10.1016/j.molcel.2012.01.001).

Hanahan D & Weinberg RA 2011 Hallmarks of cancer: the next generation. Cell 144 646-674. (doi:10.1016/j.cell.2011.02.013; 10.1016/j.cell.2011.02.013).

Ho J, de Moura MB, Lin Y, Vincent G, Thorne S, Duncan LM, Hui-Min L, Kirkwood JM, Becker D, Van Houten B & Moschos SJ 2012 Importance of glycolysis and oxidative phosphorylation in advanced melanoma. Molecular cancer 11 76-4598-11-76. (doi:10.1186/1476-4598-11-76; 10.1186/1476-4598-11-76). 161

Houssay BA 1936 The hypophysis and metabolism. New England Journal of Medicine 214 961- 971.

Koppenol WH, Bounds PL & Dang CV 2011 Otto Warburg's contributions to current concepts of cancer metabolism. Nature reviews.Cancer 11 325-337. (doi:10.1038/nrc3038; 10.1038/nrc3038).

Kress S, Stein A, Maurer P, Weber B, Reichert J, Buchmann A, Huppert P & Schwarz M 1998 Expression of hypoxia-inducible genes in tumor cells. Journal of cancer research and clinical oncology 124 315-320.

Luo W & Semenza GL 2012 Emerging roles of PKM2 in cell metabolism and cancer progression. Trends in endocrinology and metabolism: TEM 23 560-566. (doi:10.1016/j.tem.2012.06.010; 10.1016/j.tem.2012.06.010).

Luo W & Semenza GL 2011 Pyruvate kinase M2 regulates glucose metabolism by functioning as a coactivator for hypoxia-inducible factor 1 in cancer cells. Oncotarget 2 551-556.

Moller N & Jorgensen JO 2009 Effects of growth hormone on glucose, lipid, and protein metabolism in human subjects. Endocrine reviews 30 152-177. (doi:10.1210/er.2008-0027; 10.1210/er.2008-0027).

Semenza GL, Jiang BH, Leung SW, Passantino R, Concordet JP, Maire P & Giallongo A 1996 Hypoxia response elements in the aldolase A, enolase 1, and lactate dehydrogenase A gene promoters contain essential binding sites for hypoxia-inducible factor 1. The Journal of biological chemistry 271 32529-32537.

Sheppard K, Kinross KM, Solomon B, Pearson RB & Phillips WA 2012 Targeting PI3 kinase/AKT/mTOR signaling in cancer. Critical reviews in oncogenesis 17 69-95.

Scott DA, Richardson AD, Filipp FV, Knutzen CA, Chiang GG, Ronai ZA, Osterman AL & Smith JW Comparative metabolic flux profiling of melanoma cell lines: beyond the Warburg effect. Journal of Biological Chemistry 286 42626-34. (doi:10.1074/jbc.M111.282046).

Vander Heiden MG, Cantley LC & Thompson CB 2009 Understanding the Warburg effect: the metabolic requirements of cell proliferation. Science Signaling 324 1029.

Varghese B, Swaminathan G, Plotnikov A, Tzimas C, Yang N, Rui H & Fuchs SY 2010 Prolactin inhibits activity of pyruvate kinase M2 to stimulate cell proliferation. Molecular endocrinology (Baltimore, Md.) 24 2356-2365. (doi:10.1210/me.2010-0219; 10.1210/me.2010-0219).

Warburg O, Posener K & Negelein E 1924 Ueber den stoffwechsel der tumoren. Biochemische Zeitschrift 152 319.

Ward PS & Thompson CB 2012 Metabolic reprogramming: a cancer hallmark even warburg did not anticipate. Cancer cell 21 297-308. (doi:10.1016/j.ccr.2012.02.014; 10.1016/j.ccr.2012.02.014).

Weide B, Elsasser M, Buttner P, Pflugfelder A, Leiter U, Eigentler TK, Bauer J, Witte M, Meier F & Garbe C 2012 Serum markers lactate dehydrogenase and S100B predict independently disease outcome in melanoma patients with distant metastasis. British journal of cancer 107 422-428. (doi:10.1038/bjc.2012.306; 10.1038/bjc.2012.306).

Zaytseva YY, Valentino JD, Gulhati P & Evers BM 2012 mTOR inhibitors in cancer therapy. Cancer letters 319 1-7. (doi:10.1016/j.canlet.2012.01.005; 10.1016/j.canlet.2012.01.005).

Zhu T, Starling-Emerald B, Zhang X, Lee K, Gluckman PD, Mertani HC & Lobie PE 2005 Oncogenic Transformation of Human Mammary Epithelial Cells by Autocrine Human Growth Hormone. Cancer Res 65 317-324.

Zhuang L, Scolyer RA, Murali R, McCarthy SW, Zhang XD, Thompson JF & Hersey P 2010 Lactate dehydrogenase 5 expression in melanoma increases with disease progression and is associated with expression of Bcl-XL and Mcl-1, but not Bcl-2 proteins. Modern pathology : an official journal of the United States and Canadian Academy of Pathology, Inc 23 45-53. (doi:10.1038/modpathol.2009.129; 10.1038/modpathol.2009.129).

163

Figures

GLU Growth factors

GLU

2 ATP Oxidative Aerobic PYR phosphorylation glycolysis

LAC LAC O2 + 34 ATP H

Figure 4.1. An overview of oxidative phosphorylation and aerobic glycolysis.

164

Figure 4.2. Mitochondrial Stress Test. A Seahorse XF24 was used measure extracellular flux in MDA-MB-435 melanoma cells. Cells were treated overnight with 0, 0.05, 2 or 100 nM GH followed by measurement of mitochondrial respiration (OCR; panel A) and aerobic glycolysis (ECAR; panel B) at baseline and after sequential treatment with an ATP synthase inhibitor (oligomycin), proton gradient disruptor (FCCP) and ETC inhibitors (antimycin and rotenone). Time points that show a statistically significant difference (P<0.05) between two groups are marked with $ (0 vs. 0.05 nM), € (0 vs. 2 nM) or < (2 vs. 100 nM). 165

Figure 4.3 Bioenergetic parameters measured during the mitochondrial stress test. Metabolic parameters for respiration (A-C, G-I) and aerobic glycolysis (D-F, J, K) were calculated. Basal respiration (A), ATP synthesis (B), maximal respiration (C), basal glycolytic flux (D), oligomycin-induced glycolytic flux (E), FCCP-induced glycolytic flux (F), non-mitochondrial respiration (G), spare respiratory reserve capacity (H), proton leak (I), ETC-dependent glycolytic flux (J) and oligomycin and FCCP-induced glycolysis (K) are shown. Statistically significant differences between any two groups (t-test) are marked by asterisks (* denotes P<0.05, ** P<0.01). 166

Figure 4.4. Glycolysis stress test. Extracellular flux was measured in MDA-MB-435 melanoma cells. After overnight treatment with 0, 0.05 or 100 nM GH, cells were starved for one hour and treated with glucose, ATP synthase inhibitor (oligomycin), and 2-deoxy glucose. Mitochondrial respiration (OCR; panel A) and aerobic glycolysis (ECAR; panel B) were measured at baseline and after treatments. Time points that show a statistically significant difference (P<0.05) between any two groups (two-way ANOVA) are marked as follows: 0 vs. 0.05 nM with $ and 0.05 vs. 100 nM with #.

167

Figure 4.5. Bioenergetic parameters measured during the glycolysis stress test. Metabolic parameters for aerobic glycolysis (A-C, G, H) and respiration (D-F) were calculated. Basal glycolytic flux (A), glucose-induced glycolytic flux (B), glycolytic capacity (C), basal respiration (D), glucose-induced oxygen flux (E), oligomycin-induced oxygen flux (F), glycolytic reserve (G) and non-glycolytic acidification are presented.

168

* * * 10000

5000 [lactate] µM

0 0

2 nM GH 0.05 nM GH 100 nM GH0.5 nM GHA

0.05 nM GH + 0.5 nM GHA

Figure 4.6. GH-induced lactate production. MDA-MB-435 melanoma cells were treated for 14 hours with 0, 0.05, 2, 100 nM GH, 0.5 nM GHA or 0.05 nM GH combined with 0.5 nM GHA. Statistically significant differences (one-way ANOVA with Tukey’s post-hoc) are marked by asterisks (* denotes P<0.05).

169

CHAPTER 5: KEY FINDINGS AND ROADMAP FOR FUTURE STUDIES

A convergence of evidence over the past two decades has implicated GH in carcinogenesis. There are provocative results from a number of epidemiological, in vitro and in vivo studies, although the research to date has several major limitations. First, the epidemiological and genetic studies are simply correlation studies, making definitive or causative conclusions impossible. The animal studies to date have failed to adequately separate the effects of GH from

IGF-1. In vitro studies have been scarce and mostly fail to investigate mechanisms of GH action in cancer cells. Further, there is a myopic focus on prostate and breast cancer that largely ignores other cancer types. Given the known biological effects of GH, it is surprising that more research in this area has not been undertaken. On the other hand, this topic is ripe for novel discoveries and nascent research leads that can direct future work.

Based on existing literature, we developed several hypotheses, as outlined in Chapter 1.

We predicted that cancer cells possess GH signaling capability and autocrine/paracrine GH activity. We further proposed that treatment of certain cancer cells with GH would alter cell signaling and increase proliferation. We also wished to explore a new direction in GH-related cancer research, so we hypothesized that GH could alter cancer cell metabolism. In order to test these hypotheses, we chose to use the NCI60 cell panel, which contains 60 human cell lines from nine cancer types. Our objectives were to produce recombinant GH and GHA, characterize mRNA expression of GHR, PRLR, GH and PRL in the cell lines included in the NCI60 panel, determine the effect of GH treatment on cell signaling and proliferation in a subset of cell lines and to assess the effect of GH on metabolic programming in cancer cells. These objectives have largely been met and our hypotheses have been tested. We found support for all our hypotheses 170

with the exception of the potential for autocrine/paracrine GH and/or PRL activity in cancer cells, as most cells did not express appreciable levels of mRNA for these genes.

Three manuscripts (Chapters 2-4) that detail our experiments and results have been prepared to be submitted for publication. The most important observations from these studies are described below, along with notes on additional supplementary data that were not included in the manuscripts. Some suggestions are given for the direction of future research that may be able to answer new and unresolved questions.

Quantification of gene expression in the NCI60 panel and human metastatic melanoma tumors

We obtained RNA samples from all of the NCI60 panel cell lines. RNA was converted to cDNA and the expression of the GH/PRL and INS/IGF family receptors and ligands was measured by real-time qPCR. Data were normalized and scaled to allow direct comparisons between cells lines and also between each primer set.

As presented in Chapter 2, we analyzed mRNA expression of GHR, PRLR, GH and PRL.

Additional assays were performed to quantify PRLR short isoform 1a, PRLR short isoform 1b,

IGF1R, INSR-A, INSR-B, IGF1, IGF2 and INS (see appendix 4). All primers were designed as outlined in appendix 2 and described in Chapter 2. They were subject to temperature optimization and primer efficiency tests. In order to confirm specificity, PCR products were analyzed by agarose gel electrophoresis and in some cases, DNA sequencing. Several examples of temperature optimization and agarose gel electrophoresis validation that primer sets were subject to are presented in appendix 3. We performed correlation analysis to reveal potential relationships between mRNA transcript expression, as presented in appendix 5. Interestingly, GHR mRNA expression was correlated with expression of IGF1R but not with IGF1. Correlations were also 171

found between the long form of PRLR mRNA levels and the levels of GH, IGF1R, IGF1 and

INSR mRNA. Importantly, the results obtained for high GHR in melanoma were confirmed in the

CCLE, a recently available database of gene expression in a vast number of cancer cell lines. The

CCLE data also confirmed our finding of high PRLR expression in breast cancer cell lines.

In order to determine if the results in vitro may translate to in vivo tumor biology, human tumor biopsies were analyzed for mRNA expression of GHR, IGF1 and IGF1R as described in

Chapter 2. One array containing 40 tumor samples was used for each assay, resulting in weak statistical power. Even so, a significant increase in GHR expression was detected in the more advanced metastatic samples. Further experiments should be undertaken in the future to confirm this finding and determine if it extends to other types of cancer shown to express GHR.

Key findings of gene expression studies

1. GHR and PRLR are expressed at relatively high levels in a subset of tumor cell lines

2. GHR is highly expressed in all melanoma lines of the NCI60 panel

3. GHR is expressed in many melanoma tumor samples

4. GHR and IGF1expression trends to be higher in males, while GHR and IGF1R

mRNA levels were increased in advanced stage IV melanoma samples

5. GH and PRL expression is generally very low in cancer cell lines, offering little

support for the theory of autocrine/paracrine GH or PRL action in cancer cell lines

6. Across the NCI60 cell line panel, global expression of IGF1R is much higher than

INSR, INSR-A is much higher than INSR-B and IGF2 is much greater than IGF1 or

INS 172

7. The pattern of GHR and PRLR expression in the NCI60 panel is correlated with

resistance to tubulin inhibitors, while GHR expression is correlated with sensitivity to

the mTOR inhibitor rapamycin

8. IGF1 and IGF1R are expressed in all melanoma tumor samples

Future work on gene expression

The data obtained for GHR, PRLR, GH and PRL gene expression in the NCI60 panel provide a comprehensive view of mRNA levels of these genes in the 60 included cell lines. The weakness of the NCI60 panel is that some types of cancer are poorly represented and some are not represented at all, such as thyroid. Given the association of thyroid cancer with acromegaly

(as discussed in Chapter 1), a follow-up study should utilize the assays described in Chapter 2 to analyze the potential for GHR and PRLR signaling in thyroid cancer cell lines.

The real question that gene expression data is designed to answer is whether or not the measured gene is relevant in the tested cell or tissue. Ideally, for protein coding genes, this means that the presence of the protein product(s) should be evaluated by immunological or other methods (such as ligand binding assays). The lack of good antibodies for GHR and PRLR make direct measurement of these proteins difficult. We were able to somewhat circumvent this obstacle by indirectly observing biological effects of hormone treatment. Further work with existing or newly developed antibodies should be undertaken to quantify the levels of GHR and

PRLR protein, at least in a subset of cell lines shown to express mRNA for these genes.

Our analysis of human tumor biopsies provided useful data showing that a subset of tumor cells express GHR mRNA with advanced stage tumors exhibiting higher levels.

Additionally, while we found a trend for increased GHR expression in males, statistical significance eluded us due to the limited number of samples in each group. Origene, which we 173

purchased the melanoma cDNA panel from for this assay, offers a second panel with distinct metastatic melanoma samples. Repeating the gene expression studies presented here (Chapter 2) with the other melanoma cDNA plate offered by Origene (or mRNA/cDNA from other melanoma samples) would give more confidence in these results. If resources can be found for additional studies of this type, exploration of GHR mRNA expression in human tumors of other cancer types

(such as NSC lung and thyroid) could expand the possibilities for research into the role of GH in carcinogenesis.

Effect of GH on proliferation and signal transduction pathways

Five of the nine NCI60 melanoma cell lines were treated with a wide range of GH doses.

Proliferation was monitored by PrestoBlue assay, which measures the reducing power of each well as an indirect marker of cell number. Based on the proliferation data, three of the cell lines were selected for further study to determine which cell signaling pathways were affected by GH treatment. We found mixed results showing that GH could modulate the phosphorylation of Akt, mTOR, STAT1, STAT3 and STAT5 in a complex way that depended on dose, length of treatment and cell line. Although no clear pattern from this data could be discerned, of special interest is the modulation of mTOR phosphorylation by GH, which was increased in two cell lines and decreased in the third. It would be of interest to perform additional experiments to determine if the effect of GH is mediated by GHR, PRLR or hybrid receptors. Further studies are necessary to better understand the effect of GH in melanoma.

Key findings of the effect of GH on proliferation

1. Not all cell lines respond the same to treatment

2. The effect of GH on melanoma cell line proliferation is modest 174

3. GH exerts a biphasic effect on the growth of some melanoma lines, with a slight

inhibition of growth at a low dose and an increase at a high dose

4. GH does not activate MAPK signaling in these cell lines

5. GH has a variable effect on AKT and mTOR depending on cell line and GH dose

6. GH can activate STAT5 at a high GH dose and inhibit STAT3 at low and high dose

Future work on GH treatment of melanoma cells

Perhaps the biggest finding from our proliferation studies is that GH has a biphasic effect on the growth of at least one melanoma cell line. The underlying mechanism of this response is unknown and requires further study. One such avenue to explore would be to determine if different doses of GH have different effects on the activation of GHR or PRLR. While GH binds to both receptors, it is possible that at the low dose there is a preference for one receptor over the other, and at the high dose, both are strongly activated (or perhaps even inhibited). The use of selective GHR and PRLR inhibitors could be used to block signaling through one of the receptors while allowing GH to activate the other. An alternative approach would be to repeat the proliferation studies with PRL and compare the results to those presented in this dissertation

(Chapter 3). Due to the fact that our proliferation assay is actually a measure of the reducing potential of the sample, an effect of GH on this factor could skew results. Because of this, it could be of use to repeat the proliferation tests with actual cell counts if an automated cell counter becomes available.

Our preliminary investigations into the effect of GH treatment on cell signaling pathways presented us with perhaps more questions than answers. The lack of a clear pattern across all three cell lines that we tested indicates that the response of melanoma cell lines to GH may be context-dependent. Further, we utilized ELISA-based assays that generate data based on the ratio 175

of total protein to phosphorylated protein. A different type of quantitative data might give a clearer understanding of the effects of GH. A reasonable approach to further explore this would be to perform Western blotting techniques that include well-characterized positive and negative control cell lines in order to quantify the amount of phosphorylated signaling protein. One other possibility would be to repeat the proliferation assays with GH treatment in the presence of specific inhibitors of various signaling pathways to determine which are essential for the proliferative (or anti-proliferative) effects of GH.

All our studies were conducted by treatment of melanoma cells with GH alone (in some cases in the presence of serum). It could be of interest to co-treat melanoma cells with GH or

GHA alongside established chemotherapeutic agents to determine if the GH pathway could be involved in resistance to chemotherapy and to explore the potential for synergistic effects of GH inhibition during therapeutic treatment.

Effect of GH on metabolic programming of the melanoma line MDA-MB-435

The greatest effect of GH on proliferation was observed in MDA-MB-435, in which a low dose of GH decreased growth while a high dose increased it. This cell line was selected for additional studies to examine if the effect of GH on proliferation could be due to alteration of metabolic pathways. There have been no reports in the literature on GH and the Warburg effect.

Our results show dramatic effects of GH treatment on melanoma cell metabolism including potentiating of the Warburg effect. In this work, we have identified a novel action of GH that may be highly relevant to cancer biology. The metabolic effects of GH could promote tumor growth and invasion. 176

Key findings of the effect of GH on metabolism

1. GH exerts a biphasic effect on metabolism in melanoma lines

2. Low-dose GH energizes melanoma cells by inducing a dramatic increase in

respiratory activity and capacity

3. Low-dose GH increases aerobic glycolysis and thus potentiates the Warburg effect

4. Low-dose GH increases the ability of melanoma cells to compensate for energy stress

Future work on GH and metabolism

Although we were able to find statistically significant differences between treated and untreated groups, the major limitations of our metabolic studies are that they were performed in a single experiment with a single cell line. Therefore, it is necessary to repeat the Seahorse and lactate assays with MDA-MB-435 and additional cell lines. Due to the high expression of GHR in all melanoma cell lines of the NCI60 panel and the fact that three of them are preserved in frozen stocks in the Kopchick laboratory, I would suggest including two or more of these cell lines at a minimum. If the metabolic effects are confirmed, the next logical step is to determine if this is a melanoma-specific effect or if it can be generalized to other cell lines that possess high GHR expression.

Assuming that the metabolic effects of GH in cancer cells are confirmed, the next logical step is to examine potential mechanisms of action. One model of the potential mechanism is outlined in detail in the Discussion section of Chapter 4. This model proposes that GH could interact with PKM2 to regulate metabolism. PKM2 is a multi-functional enzyme that possesses pyruvate kinase activity as a tetramer, but is able to translocate to the nucleus upon dissociation into dimers, whereupon it can interact with various transcription factors to regulate metabolic 177

programming. The effect of GH treatment on PKM2 activation as well as interacting partners, such as HIF-1alpha and MYC, should be investigated. Expression should be measured after treatment of melanoma cells with 0.05 nM GH. Commercial antibodies are available and could be used to quantify the abundance of these proteins. As a first step toward this end, we have recently obtained antibodies to total and phosphorylated PKM2, and validated their use in Western blot.

Hypothetical model of GH action in melanoma cells

Based on our data and the body of literature on GH action, a hypothetical model of the effects of GH in melanoma can be proposed, as depicted in figure 5.1. One important caveat to the model is that much of the data regarding GH action in cancer, including that presented in this dissertation, are context-specific and varies with cell line and experimental conditions. Also, not everything in the model is likely to occur within the same cell; it is probable that only a subset of these pathways would be relevant in any given cell line or tumor cell.

We can start from the situation of a melanoma cell exposed to GH. From mRNA expression data, we know that melanoma cells express high levels of GHR in all cell lines

(Chapter 2). We also observed PRLR expression in some of the cell lines examined. In addition to

GHR, GH can also bind and activate PRLR, and there is even evidence that functional heterodimers of GHR-PRLR may exist (Somers et al. 1994, Langenheim & Chen 2009). While these various receptors for GH may have differential effects, it is likely that they initiate similar signaling events. The most established signaling pathway activated by GH is via JAK/STAT5. In our experiments (Chapter 3), STAT5 was clearly activated by GH treatment in several melanoma cell lines. We also show preliminary evidence that GH may increase expression of IGF2 mRNA in a time-specific manner. While IGF-1 is the most well-known growth factor induced by GH, others have shown induction of IGF2 expression in prostate cancer cell lines. Our gene 178

Figure 5.1. Hypothetical model of GH action in melanoma. See text for description. 179

expression data in appendix 4 shows mRNA expression of both IGF1R and INSR-A, as well as

IGF2 and IGF1 in melanoma cell lines. Either if these IGFs can active either receptor (or hybrid receptors) in order to promote cell proliferation and survival (Chao & D’Amore 2008). Another

GH-regulated gene described in the literature is the multi-functional oncogene MYC (Calo et al.

2003). We have not yet measured MYC expression in melanoma cell lines, but this would be of interest. MYC is known to mediate alternative splicing of transcripts from the PK gene to give rise to the cancer-associated PKM2 splice variant (Tamada et al. 2012). The role of PKM2 is described below. Ultimately, both MYC and IGFs have established roles in the promotion of cell growth and proliferation.

Although most studies of GH signaling have focused on the JAK/STAT pathway, there is good evidence to indicate that GH also modulates other intracellular signal transduction pathways, including PI3K/Akt/mTOR and MAPK/Erk (Hayashi & Proud 2007, Love et al. 1998,

Okada & Kopchick 2001). Activation of mTOR by GH, as has been observed by others in hepatoma cells, could be one potential mechanism by which GH could regulate metabolic programming in melanoma cells. The melanoma cells we examined all have activating mutations of the MAPK pathway (Ikediobi et al. 2006), so as expected, we did not observe changes in Erk activation with GH treatment, although as depicted in the figure, GH could activate Erk via SHC in melanoma cells that do not have constitutive activation of this pathway. Activation of mTOR could alter metabolic programming and regulate cell growth, while MAPK signaling in melanoma could promote cell proliferation. While we found an effect of GH on cell proliferation under certain conditions, the absolute magnitude of this effect was modest (Chapter 3). One possible explanation is that the cell lines we used have constitutive activation of the MAPK signaling pathway due to an activation mutation in BRAF (Ikediobi et al. 2006). 180

The most intriguing finding from our data was that GH treatment enhanced both oxidative phosphorylation and aerobic respiration in melanoma cells (Chapter 4). This results in a highly-energized melanoma cell that possesses increased metabolic flexibility and lactic acid production. The high production rate of lactate increases extracellular acidity, which has been associated with an invasive and metastatic phenotype in human melanoma cells (Rofstad et al.

2006). Thus, future studies must be performed to determine if GH enhances metastasis in vivo.

The enhanced metabolism we observed, through both metabolic respiration and aerobic glycolysis, highly suggests that GH treatment is associated with increased glucose uptake, which is at odds with the established role of GH in inhibition of glucose uptake (Moller & Jorgensen

2009). It is possible that cancer cells respond differently in this regard, either through alterations in GH signaling or by the presence of other factors that overcome the inhibition of glucose transport that might be expected. To better examine this question, analysis of glucose transporter expression and glucose consumption under various GH treatment conditions should be undertaken.

We propose a potential mechanism by which GH signaling could potentiate aerobic glycolysis in melanoma cells. First, as mentioned above, MYC, a transcriptional target of GH signaling, promotes expression of the cancer-associated tumor-promoting PKM2 variant transcript (Tamada et al. 2012). PKM2 has emerged as a central regulator of cancer metabolism and a key promoter of the Warburg effect (Luo & Semenza 2012). PKM2 is a multi-functional enzyme (Tamada et al. 2012). As a tetramer, PKM2 is a cytosolic protein that catalyzes the conversion of phosphoenol pyruvate to pyruvate. Upon phosphorylation, however, PKM2 morphs into a dimer with new kinase activity and the ability to enter the nucleus where it interacts with several transcription factors, including HIF-1α. A proteomics-based study identified a direct interaction between PRLR and PKM2 that results in JAK2-dependent phosphorylation of PKM2 181

and an increase in aerobic glycolysis (Varghese et al. 2012). Given the ability of GH to activate the PRLR and the presence of PRLR in many of the melanoma cell lines we examined, it is highly plausible that this modulation of PKM2 form could occur in melanoma. We also further speculate that GHR may possess a similar or identical capacity to modulate PKM2 activity via direct interaction with PKM2. This should be explored with co-immunoprecipitation studies.

The activation of HIF-1α by PKM2 dimer formation and nuclear activity is likely to result in metabolic reprogramming through activation of several genes, including GLUT1, LDH and PKM2 (Tamada et al. 2012). Intriguingly, others have shown that at least in HEK293 cells,

HIF-1α up-reguates expression of GHR (Herman et al. 2011). If this occurs in melanoma cells, it could create a self-potentiating feed-forward loop in which GHR activation increases HIF-1α activity, which promotes further expression of GHR. In fact, this is a possible explanation for the hig levels of GHR mRNA expression we observe in melanoma cell lines. Lastly, vascular endothelial growth factor (VEGF) is induced by HIF-1α, thereby offering a mechanism for promotion of angiogenesis to support tumor growth in vivo.

Final thoughts

The work presented here details many novel findings with regard to a potential role of

GH in cancer. First, we have identified melanoma as a unique cancer type that possesses high- level GHR expression. Second, treatment of melanoma cell lines had a biphasic affect on cell growth, and a multitude of cell-specific and dose-dependent effects on signal transduction pathways. Lastly, and I believe most importantly, we have uncovered a novel action of GH: modulation of metabolic programming and potentiation of the Warburg effect in cancer cells.

This effect was characterized by high basal energy production, increased metabolic flexibility associated with elevated respiratory and glycolytic capacity. The metabolic studies presented here 182

are important first steps that will help us understand the effect of GH in melanoma and elucidate the general role GH may play in metabolic programming, an emerging hallmark of cancer. While this dissertation comes to a close, perhaps the findings contained in it are just the beginning of a new line of research in the GH field. It is our hope that the findings here will bring a better understanding to metastatic melanoma so that we may one day be able to fight this aggressive and deadly disease.

183

APPENDIX 1: TABLE SUMMARY OF RESEARCH PUBLISHED ON GH AND CELL

LINES INCLUDED IN THE NCI60 PANEL.

184

185

186

References for appendix 1 table:

Alam KS, Morimoto M, Yoshizato H, Fujikawa T, Furukawa K, Tanaka M & Nakashima K 1998 Expression and purification of a mutant human growth hormone that is resistant to proteolytic cleavage by thrombin, plasmin and human plasma in vitro. Journal of Biotechnology 65 183-190. Bidosee M, Karry R, Weiss-Messer E & Barkey RJ 2009 Regulation of growth hormone receptors in human prostate cancer cell lines. Molecular and cellular endocrinology 309 82- 92. Biswas R & Vonderhaar BK 1987 Role of serum in the prolactin responsiveness of MCF-7 human breast cancer cells in long-term tissue culture. Cancer research 47 3509-3514. Blatt J, Wenger S, Stitely S & Lee PA 1987 Lack of mitogenic effects of growth hormone on human leukemic lymphoblasts. European journal of pediatrics 146 257-260. Borges S, Moudilou E, Vouyovitch C, Chiesa J, Lobie P, Mertani H & Raccurt M 2008 Involvement of a JAK/STAT pathway inhibitor: cytokine inducible SH2 containing protein in breast cancer. Advances in Experimental Medicine and Biology 617 321-329. (doi:10.1007/978-0-387-69080-3_30). Brunet-Dunand SE, Vouyovitch C, Araneda S, Pandey V, Vidal LJ, Print C, Mertani HC, Lobie PE & Perry JK 2009 Autocrine human growth hormone promotes tumor angiogenesis in mammary carcinoma. Endocrinology 150 1341-1352. (doi:10.1210/en.2008-0608). Chopin LK, Veveris-Lowe TL, Philipps AF & Herington AC 2002 Co-expression of GH and GHR isoforms in prostate cancer cell lines. Growth hormone \& IGF research 12 126-136. Costoya JA, Finidori J, Moutoussamy S, Searis R, Devesa J & Arce VM 1999 Activation of growth hormone receptor delivers an antiapoptotic signal: evidence for a role of Akt in this pathway. Endocrinology 140 5937-5943. Costoya JA, Rios R, Garcia-Barros M, Gallego R, Garcia-Caballero T, Senaris R, Arce VM & Devesa J 2000 Role of growth hormone receptor in HL-60 cell survival. Molecular cell biology research communications : MCBRC 4 26-31. (doi:10.1006/mcbr.2000.0252). Costoya JA, Vidal A, Garcia-Barros M, Arce V & Devesa J 1996 Expression of the human growth hormone normal gene (hGH-N) in proliferating and differentiated HL-60 cells. Experimental cell research 228 164-167. Dagnaes-Hansen F, Duan H, Rasmussen LM, Friend KE & Flyvbjerg A 2004 Growth hormone receptor antagonist administration inhibits growth of human colorectal carcinoma in nude mice. Anticancer Research 24 3735-3742. Decouvelaere C, Peyrat JP, Bonneterre J, Djiane J & Jammes H 1995 Presence of the two growth hormone receptor messenger RNA isoforms in human breast cancer. Cell growth \& differentiation : the molecular biology journal of the American Association for Cancer Research 6 477-483. Dembinski TC, Leung CK & Shiu RP 1985 Evidence for a novel pituitary factor that potentiates the mitogenic effect of estrogen in human breast cancer cells. Cancer research 45 3083- 3089. 187

Divisova J, Kuiatse I, Lazard Z, Weiss H, Vreeland F, Hadsell DL, Schiff R, Osborne CK & Lee AV 2006 The growth hormone receptor antagonist pegvisomant blocks both mammary gland development and MCF-7 breast cancer xenograft growth. Breast cancer research and treatment 98 315-327. Estrov Z, Meir R, Barak Y, Zaizov R & Zadik Z 1991 Human growth hormone and insulin-like growth factor-1 enhance the proliferation of human leukemic blasts. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 9 394-399. Friend KE, Khandwala HM, Flyvbjerg A, Hill H, Li J & McCutcheon IE 2001 Growth hormone and insulin-like growth factor-I: effects on the growth of glioma cell lines. Growth hormone & IGF research : official journal of the Growth Hormone Research Society and the International IGF Research Society 11 84-91. (doi:10.1054/ghir.2000.0183). Fuh G & Wells JA 1995 Prolactin receptor antagonists that inhibit the growth of breast cancer cell lines. The Journal of biological chemistry 270 13133-13137. Giesbert S, Panzer S, Kovar H, Fischer S, Printz D, Gadner H & Panzer-Grumayer ER 1997 Acute leukemias express a functional receptor for the human growth hormone. Annals of Hematology 74 253-257. Kaulsay KK, Mertani HC, Lee KO & Lobie PE 2000 Autocrine human growth hormone enhancement of human mammary carcinoma cell spreading is Jak2 dependent. Endocrinology 141 1571-1584. Kaulsay KK, Mertani HC, Tornell J, Morel G, Lee KO & Lobie PE 1999 Autocrine stimulation of human mammary carcinoma cell proliferation by human growth hormone. Experimental cell research 250 35-50. (doi:10.1006/excr.1999.4492). Kaulsay KK, Zhu T, Bennett W, Lee KO & Lobie PE 2001 The effects of autocrine human growth hormone (hGH) on human mammary carcinoma cell behavior are mediated via the hGH receptor. Endocrinology 142 767-777. Langenheim JF & Chen WY 2009 Development of a novel ligand that activates JAK2/STAT5 signaling through a heterodimer of prolactin receptor and growth hormone receptor. Journal of receptor and signal transduction research 29 107-112. (doi:10.1080/10799890902845252). Lemus-Wilson A, Kelly PA & Blask DE 1995 Melatonin blocks the stimulatory effects of prolactin on human breast cancer cell growth in culture. British journal of cancer 72 1435- 1440. Leroy-Martin B & Peyrat JP 1989 Modulation of prolactin receptors (PRL-R) by lactogenic and steroid hormones in human breast cancer cells in long-term tissue culture (T-47D). Anticancer Research 9 631-636. Leung KC, Brce J, Doyle N, Lee HJ, Leong GM, Sjogren K & Ho KK 2007 Regulation of growth hormone signaling by selective estrogen receptor modulators occurs through suppression of protein tyrosine phosphatases. Endocrinology 148 2417-2423. (doi:10.1210/en.2006-1305). Liu N, Mertani HC, Norstedt G, Tornell J & Lobie PE 1997 Mode of the autocrine/paracrine mechanism of growth hormone action. Experimental cell research 237 196-206. (doi:10.1006/excr.1997.3789). 188

Mertani HC, Zhu T, Goh EL, Lee KO, Morel G & Lobie PE 2001 Autocrine human growth hormone (hGH) regulation of human mammary carcinoma cell gene expression. Identification of CHOP as a mediator of hGH-stimulated human mammary carcinoma cell survival. The Journal of biological chemistry 276 21464-21475. (doi:10.1074/jbc.M100437200). Mojarrad M, Momeny M, Mansuri F, Abdolazimi Y, Tabrizi MHH, Ghaffari SHH, Tavangar SMM & Modarressi MHH 2009 Autocrine human growth hormone expression leads to resistance of MCF-7 cells to tamoxifen. Medical oncology (Northwood, London, England) . Murphy LJ, Sutherland RL & Lazarus L 1985 Regulation of growth hormone and epidermal growth factor receptors by progestins in breast cancer cells. Biochemical and biophysical research communications 131 767-773. Murphy LJ, Vrhovsek E, Sutherland RL & Lazarus L 1984 Growth hormone binding to cultured human breast cancer cells. The Journal of clinical endocrinology and metabolism 58 149- 156. Nagano M, Chastre E, Choquet A, Bara J, Gespach C & Kelly PA 1995 Expression of prolactin and growth hormone receptor genes and their isoforms in the gastrointestinal tract. The American Journal of Physiology 268 G431-42. Reiter E, Kecha O, Hennuy B, Lardinois S, Klug M, Bruyninx M, Closset J & Hennen G 1995 Growth hormone directly affects the function of the different lobes of the rat prostate. Endocrinology 136 3338-3345. Shiu RP, Lima G, Leung CK & Dembinski TC 1986 Intrinsic and extrinsic factors in estrogen action in human breast cancer: role of polyamines and pituitary factors. Journal of steroid biochemistry 24 133-138. Shiu RP & Paterson JA 1984 Alteration of cell shape, adhesion, and lipid accumulation in human breast cancer cells (T-47D) by human prolactin and growth hormone. Cancer research 44 1178-1186. Soletormos G, Fogh JM, Sehested-Hansen B, Spang-Thomsen M, Schioler V, Dombernowsky P & Skovsgaard T 1997 Carcino-embryonic antigen in monitoring the growth of human colon adenocarcinoma tumour cells SK-CO-1 and HT-29 in vitro and in nude mice. European journal of cancer (Oxford, England : 1990) 33 108-114. Tsunekawa B, Wada M, Ikeda M, Uchida H, Naito N & Honjo M 1999 The 20-kilodalton (kDa) human growth hormone (hGH) differs from the 22-kDa hGH in the effect on the human prolactin receptor. Endocrinology 140 3909-3918. Untergasser G, Rumpold H, Hermann M, Dirnhofer S, Jilg G & Berger P 1999 Proliferative disorders of the aging human prostate: involvement of protein hormones and their receptors. Experimental gerontology 34 275-287. Vouyovitch CM, Vidal L, Borges S, Raccurt M, Arnould C, Chiesa J, Lobie PE, Lachuer J & Mertani HC 2008 Proteomic analysis of autocrine/paracrine effects of human growth hormone in human mammary carcinoma cells. Advances in Experimental Medicine and Biology 617 493-500. (doi:10.1007/978-0-387-69080-3_49). 189

Weiss-Messer E, Merom O, Adi A, Karry R, Bidosee M, Ber R, Kaploun A, Stein A & Barkey RJ 2004 Growth hormone (GH) receptors in prostate cancer: gene expression in human tissues and cell lines and characterization, GH signaling and androgen receptor regulation in LNCaP cells. Molecular and cellular endocrinology 220 109-123. Yang XF, Beamer WG, Huynh H & Pollak M 1996 Reduced growth of human breast cancer xenografts in hosts homozygous for the lit mutation. Cancer research 56 1509-1511. Zatelli MCC, Minoia M, Mol\`e Daniela, Cason V, Tagliati F, Margutti A, Bondanelli M, Ambrosio MRR & degli Uberti E 2009 Growth hormone excess promotes breast cancer chemoresistance. The Journal of clinical endocrinology and metabolism 94 3931-3938.

190

APPENDIX 2: GUIDELINES FOR PRIMER DESIGN.

1. Lookup gene in http://www.ncbi.nlm.nih.gov/gene. (get the correct species!)

2. Scroll down to “NCBI Reference Sequences” section a. Click to open mRNA RefSeq (NM_XXXXXXX) b. Most genes have different transcript mRNA variants, make sure you know which one you want, or most likely, you will want primers that amplify all variants, so ensure primers are in regions that are in all variants if possible

3. Copy RefSeq number into http://www.ncbi.nlm.nih.gov/tools/primer-blast/

4. Copy complete mRNA sequence into Primer3 http://frodo.wi.mit.edu/primer3/ a. Following setting changes: “Product Size Range: 80-350” and “Number to Return: 10” i. If mFold structure is bad (see below), restrict size to 80-150 or similar (smaller size = less secondary structure)

5. Look at mRNA sequence page to see if primers are on different exons—only use primer pairs separated by an intron

6. Copy first amplicon (amplified sequence) from Primer3 and past into http://mfold.rna.albany.edu/?q=mfold/DNA-Folding-Form setting the “Na conc. At 10 mM” and “Mg at 2.5 mM” (MUST SELECT “mM”, can change numbers if you know your PCR conditions) a. Click on jpg or pdf links to check secondary structure, most important is the lower dG the better (0 to -5 best, -5 to -10 usually ok), and the less structure near 5’ and 3’ ends the better b. Repeat with primer sets in other areas of gene by opening another primer3 page, copying gene sequence, and copying and pasting the left and right primers into the new page, so we can get the amplicon sequence c. Examine the structure to see if there are certain areas that would be ideal for a primer (low or no secondary structure), and you can try to “force” primer3 to pick a primer in that region by limiting sequence with <> or other techniques.

7. Once a good pair is found based on structure and exon-exon boundaries, copy the primers into PrimerBlast http://www.ncbi.nlm.nih.gov/tools/primer-blast/ to check for specificity 191

(enter organism, mus musculus or homo sapiens). If there are off-target matches of similar or smaller size, and without many mis-matches, you need to find a different primer pair.

8. If for some reason it is not possible to have good primers spanning an exon-exon boundary, you can ask primerblast to design exon-spanning primers, which will be primers that cross exon-exon boundaries and thus should not amplify from contaminating genomic DNA.

192

APPENDIX 3. EXAMPLES OF PRIMER VALIDATION AND TEMPERATURE

OPTIMIZATION.

193

194

195

196

APPENDIX 4: EXPRESSION OF MRNA FOR THE IGF/INSULIN PATHWAY IN THE

NCI60 PANEL.

197

198

199

Average expression across the entire NCI60 panel (arithmetic mean of non-transformed data).

200

APPENDIX 5: CORRELATIONS WITHIN OUR GENE EXPRESSION DATA. target x target y r p PRLR PRLR LONG 0.788 0.000E0 INSR BOTH INSR A 0.886 0.000E0 INSR B INSR BOTH 0.607 5.162E-7 PRLR PRLR BOTH SHORT 0.873 7.224E-7 PRLR SF1A PRLR 0.870 7.565E-7 PRLR LONG PRLR BOTH SHORT 0.867 7.930E-7 PRLR SF1A PRLR LONG 0.824 1.208E-6 PRLR SF1A PRLR BOTH SHORT 0.931 4.918E-6 PRLR SF1A PRLR SF1B 0.719 1.925E-5 PRLR PRLR SF1B 0.522 2.936E-5 IGF1R GHR 0.495 1.494E-4 PRLR SF1B PRLR BOTH SHORT 0.653 1.742E-4 PRLR SF1B PRLR LONG 0.469 3.069E-4 INSR B INSR A 0.433 6.081E-4 GH2 INS 0.426 1.067E-3 GH2 PRLR SF1B 0.402 2.114E-3 PRLR SF1A PRL 0.628 2.126E-3 PRLR SF1B INS 0.377 3.208E-3 GH2 IGF1R 0.376 4.175E-3 IGF2 INS 0.365 4.376E-3 PRL GHR 0.394 5.936E-3 PRLR SF1B IGF1R 0.351 6.236E-3 IGF1R PRL 0.382 6.477E-3 GH2 PRLR LONG 0.359 7.417E-3 IGF1R PRLR LONG 0.347 9.156E-3 GH2 IGF2 0.322 1.499E-2 IGF1 PRLR LONG 0.325 1.681E-2 GH2 PRLR 0.299 2.428E-2 INSR BOTH GHR 0.291 3.153E-2 PRL PRLR BOTH SHORT 0.432 3.568E-2

201

APPENDIX 6: PRODUCTION AND VALIDATION OF RECOMBINANT HUMAN GH

AND GHA.

GH production protocol

A. Bacteria Growth

1. Grow overnight starter culture in LB+Amp (6.6 ml of starter needed for each 330 ml culture) 2. Next Day: Innoculate 3x660 ml LB+Amp at 1:50 starter culture to media 3. Grow to OD600=0.8 (dilute culture to OD600=0.6 with LB+Amp) 4. Take un-induced sample (sample US) 5. Induce with IPTG to final concentration of 1 mM by adding 1000x IPTG stock (1M) 6. Grow an additional 4 hours 7. Take induced sample (sample IS) 8. Pellet bacteria by centrifugation at 8k RPM for 10 minutes 9. Pellets can be frozen at -20 overnight

B. Inclusion Body Isolation

1. Resuspend bacteria pellet in 2.5 ml CLB (pH 8.5 with 1 mM PMSF) for each gram of pellet 2. Add 0.5 mg lysozyme for each ml of resuspended IBs 3. Leave at RT for 15 minutes with mixing 4. Mix well and pull through 22G needle several times and shoot into new tube 5. Pre-weigh centrifuge tubes 6. Sonication a. Sonicate gently 5 x 15 seconds on “5” (until becomes “foamy”) b. Pull through 22G needle and shoot into new tube c. Centrifuge at 20,000xg for 10 min (sample S1 after centrifugation) 7. Tris-DOC Wash I a. Resuspend in 10 mM Tris-HCl 1% DOC, pH 8.5 in same volume as in B1 b. Repeat B5 (only do B6 if still viscous) (sample S2) 8. Tris-DOC Wash II a. Resuspend in 10 mM Tris-HCl 1% DOC, pH 8.5 in same volume as in B1 b. Repeat B5 (only do B6 if still viscous) (sample S3) 9. Tris Wash a. Resuspend in 10 mM Tris-HCl pH 8.5 in 2x the volume used in B1 (sample S4) b. Centrifuge 20,000xg for 10 min 10. H2O Wash I a. Resuspend in H2O in 2x volume as in B1 (sample S5) 202

b. Centrifuge 20,000xg for 10 min 11. H2O Wash II a. Resuspend in H2O in 2x volume as in B1 (sample S6) b. Take a small sample of IBs (40ul) to run on a gel and run BCA on c. Centrifuge 20,000xg for 10 min 12. Weigh pellet and freeze at -80

C. Protein Purity and Concentration Assays

1. Perform BCA assay As Per Protocol on all samples (APP) 2. Run all samples from A and B on a denaturing SDS gel a. Samples should be mixed with 1% SDS (add 1ul of 10% SDS and mix) before adding to Laemmli Sample Buffer and boiled

D. IB Solubilization

1. Pre-weigh centrifuge tube 2. Add 1 ml IBS buffer per 3 mg of protein in the IB pellet 3. Homogenize on ice for 5 min 4. Leave at RT for 15 min 5. Centrifuge 20,000xg for 15 min a. Transfer supernatant to 50 ml conical tube (sample Sol IBs) and weigh pellet (this is the amount if IBs lost)

E. Refolding by Dilution

1. Prepare plastic mixing containers in cold room (appropriate sizes, with stir plate) 2. Perform 5 2-fold dilutions with 100 mM Tris, pH 8.0 a. Just quickly pour in the refolding buffer with stirring b. Wait about one minute between dilution step (after dilutions--sample Ref GH) 3. Filter refolded GH a. Pack refolded GH in water-ice bath b. Pack filter bottle in ice bath c. Filter with 45um filter bottle (Ref Fil GH)

F. Ion Exchange Chromatography with 5ml HiTrap Q-Sepharose Columns

1. Filter Start Buffer (50 mM Tris, pH 8.0) 2. Wash column with 25 ml of Start Buffer 3. Wash with 25 ml Start Buffer w/ 1 M NaCl 4. Equilibrate with 50 ml start buffer a. Check pH/pI of wash through 203

5. Apply refolded GH to column (at about 50% max rate, all others at max rate or 5 ml/min if done by syringe) 6. Wash with 25 ml start buffer a. Monitor OD280 7. Elute with 50 ml elution buffer (linear gradient from 0-500 mM NaCl) a. Collect 4.5 ml fractions (fraction collector set to 165 seconds) b. Monitor OD280 of all fractions 8. Regenerate column a. Wash with 25 ml Start Buffer w/ 1 M NaCl b. Wash with 50 ml Start Buffer

G. Filter Sterilize Protein Prep

1. Filter sterilize protein prep with syringe filter 2. Aliquot depending on cell culture or animal study to be used in

Typical results for induction and ion-exchange chromatography

SDS-PAGE analysis of whole cell lysates from uninduced and induced GH cultures. Protein size ladder (lane 1) with spiked in bGH (indicated by arrow). Bacterial lysates from 3 uninduced

(lanes 2-4) and 3 induced (lanes 5-7) cultures.

204

A B

SDS-PAGE analysis of IEC fractions from hGH isolation. Protein size ladder (A, lane 1),

Solubilized IBs (A, lane 2) and fractions 1-8 (A, lanes 3-10) and 9-17 (B, lanes 2-10).

E6 L-cell STAT5 Activation Assay

GH induced GHR activation is measured by way of STAT-5 tyrosine phosphorylation, a

GH induced intracellular signaling intermediate. This assay uses mouse L-cells that express the mouse GH receptor. The extent of STAT-5 tyrosine (Tyr694) phosphorylation is detected by

Western blot using a STAT-5 antibody specific for phosphorylated Tyrosine at position 694 (Cell

Signaling, #9351). Phosphorylation of STAT-5 Tyr694 is dependent on both the dose of GH and the time of incubation. E6 cells are grown to ~80% confluency in DMEM and starved overnight without serum. GH and/or GHA (see concentrations on graph below) are applied for 10 minutes at 37ºC in serum-free media. Protein is isolated by incubation in 1x RIPA buffer with phosphatase inhibitors (Cell Signaling) followed by brief sonication. Proteins are run on 15% acrylamide SDS-PAGE and blotted using the anti-P-STAT5 primary antibody.

205

! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !

Thesis and Dissertation Services ! !