HEXIM1 IS AN INHIBITOR OF TWO TRANSCRIPTION FACTORS

CRITICAL IN CANCER: THE ANDROGEN RECEPTOR AND HYPOXIA

INDUCIBLE FACTOR-1 ALPHA

by

I-JU YEH

Submitted in partial fulfillment of the requirements

For the degree of Doctor of Philosophy

Dissertation Advisor: Dr. Monica M. Montano

Department of Pharmacology

CASE WESTERN RESERVE UNIVERSITY

January, 2015

TABLE OF CONTENTS

Title page i

Signature sheet ii

Table of contents 1

List of Table 5

List of Figure 7

Acknowledgements 8

List of abbreviations 9

Abstract 14

CHAPTER I: Introduction, review of literature and statement of purpose

Introduction 16

Review of literature 17

Statement of purpose 38

1 Figures 39

CHAPTER II: HEXIM1 Plays a Critical Role in the Inhibition of the Androgen

Receptor by Antiandrogens

Abstract 44

Introduction 45

Materials and methods 47

Results 51

Discussion 59

Acknowledgments 64

Figures 65

CHAPTER III: Summary, Discussion and Future Direction

Summary and Discussion 82

Future Directions 85

Figures 88

2 CHAPTER IV: Introduction, review of literature and statement of purpose

Introduction 89

Review of literature 89

Statement of purpose 110

Figures 112

CHAPTER V: HEXIM1 down-regulates hypoxia-inducible factor-1 protein stability

Abstract 117

Introduction 118

Material and methods 121

Results 127

Discussion 135

Acknowledgments 140

3 Figures 141

CHAPTER VI: Summary, Discussion and Future Direction

Summary and Discussion 155

Future Direction 162

Concluding Remarks 165

Figures 166

BIBLIOGRAPHY 170

4 LIST OF TABLES

Table 1. Nuclear receptor family 168

Table 2. The regulation of HDAC family on HIF1α activity 169

5 LIST OF FIGURES

Figure I-1 Androgen-deprivation therapy is the first-line choice in

treating prostate cancer 39

Figure I-2 Functional domains of Androgen Receptor (AR) 40

Figure I-3 Regulation of androgen synthesis and the drugs that

have been developed to target androgen production and

the activation of the AR 41

Figure I-4 The progression to CRPC is correlated with the changes in

several genes and pathways 42

Figure I-5 The functional domains of HEXIM1 43

Figure II-1 Expression levels of HEXIM1 in normal prostate and

prostate cancer tissue 65

Figure II-2 Physical interaction between HEXIM1 and AR 67

Figure II-3 Recruitment patterns of HEXIM1 and transcriptional

elongation factors in LNCaP cells 69

Figure II-4 HEXIM1 inhibited FOXA1 recruitment and H3K4me2

Enrichment 72

Figure II-5 HEXIM1 regulated KDM5B expression 75

Figure II-6 HEXIM1 regulated KDM5B recruitment 77

Figure II-7 Altered HEXIM1 expression resulted in altered cellular

response to an anti-androgen 79

Figure II-8 Expression levels of HEXIM1 in normal prostate and

prostate cancer tissue 81

Figure II-9 HEXIM1 was not recruited to control PSA region 81

Figure III-1 Model of HEXIM1-KDM5B regulation in prostate cancer 88

6 Figure IV-1 Cellular processes regulated by hypoxia 112

Figure IV-2 Functional domains of HIF family proteins 113

Figure IV-3 Post-translational modifications of HIF1α 114

Figure IV-4 Glycolysis is catalyzed by PKM I in normal cell

andsPKMII in tumor cell,srespectively 115

Figure IV-5 Hydroxylation catalyzed by prolyl hydroxylases 2

(PHD2) in normoxia 116

Figure V-1 HEXIM1 destabilizes HIF1α protein 141

Figure V-2 HEXIM1 interacted with HIF1α and down-regulation of

22222222222222222222222222222222222222222222222222222222222222222222222222222HEXIM1 resulted in decreased levels of hydroxylated

and ubiquitinated HIF1α and attenuated the

22222222222222222222222222222222222222222222222222222222222222222222222222222HIF1α–pVHL interaction 144

Figure V-3 HEXIM1 up-regulated expression and interaction with

PHD3 147

Figure V-4 Down-regulation of HEXIM1 resulted in an enhanced HIF1α

2222222222222222222222222222222222222222222222222222222222222222222222222222HDAC1 interaction and deacetylation of HIF1α 150

Figure V-5 HEXIM1-regulated expressions of HIF1α-regulated genes 152

Figure V-6 HEXIM1 inhibited hypoxia-induced invasion of

MDA-MB-231 cells 154

Figure VI-1 Hypothesis models 166

Figure VI-2 Overall model for HEXIM1 in prostate cancer and breast

cancer 167

7 ACKNOWLEDGEMENTS

The completion of this dissertation is contributed by many people’s efforts, especially my advisor, Dr. Monica. M. Montano. Dr. Montano guided me the research direction when I was trapped in the scientific puzzles. She is like the bright lamp guiding me in the dark and always enlightens me how to cope with the problems, not only in the scientific works, but also in my future career plan. I really appreciate the tolerance of her for the mistakes I made on my work every now and then. Besides, I also want to thank my dissertation committees for their support: Dr. Noa Noy as my committee chair, Dr. David Danielpour, Dr. Scott Welford and Dr. Bing-Cheng Wang.

I am thoroughly grateful for the assistance and the suggestions from you, without these, this work would not have been completed.

I also want to thank Dr. Shigemi Matsuyama who patiently taught me cell-based techniques; Dr. John Mieyal, not only for the help in my second year of study, but also the guidance for me choosing the lab. I would like to thank my previous lab members, Ndiya, thank you for teaching me the experimental skills and gave me a fantastic project to carry on; Nui, I really appreciate that every time you go to the temple and never forget to pray for me; Nirmala, Yan and Heather, thank you for all the caring about my life.

Finally, I want to thank my family for their support. Thanks for the support from my parents. I am so sorry that I could not be at your side and I appreciate for your understanding. Moreover, I am highly grateful for my fiancé, Yen-Shan Chen. Thanks for all the comfort and support and always show me the bright side. Your support is always the motivation keeps me going on. Thank you!

8 LIST OF ABBREVIATIONS

2-OG 2-Oxoglutarate

17-AAG 17-N-allylamino-17-demethoxygeldanamycin

ACTH Adrenocorticotropic Hormone

ADPC Androgen-Dependent Prostate Cancer

AF-1/AF-2 Activation Function-1/2

AIDS Acquired ImmunoDeficiency Syndrome

AML Acute Myeloid Leukemia

APC/C Anaphase Promoting Complex/Cyclosome

AR Androgen Receptor

ARD1 Arrest-Defective protein 1

ARE Androgen-Response Element

ARNT Aryl-hydrocarbon Nuclear Translocator

bFGF Basic Fibroblast Growth Factor

BH Domain Bcl-2 homolog

BMDC Bone marrow- Derived Cells

BPH Benign Prostatic Hyperplasia

CDK1 …..Cyclin-Dependent Kinase 1

CDC20 Cell-Devision Cycle Protein 20

COPA Cancer Outlier Profile Analyses

CRPC Castration-Resistance Prostate Cancer

CSC Cancer Stem Cell

CTD C-Terminal Domain

CXCL12 C-X-C motif chemokine 12

CXCR4 Chemokine Receptor type 4

9 CZ Central Zone

DBD DNA-Binding Domain

DHEA Dehydroepiandrosterone

DHT Dihydrotestosterone

DRE Digital Rectal Exam

ECM Extra-Cellular Matrix

EDG1 Estrogen Down-regulated Gene 1

EGF/EGFR Epidermal Growth Factor/ Epiderminal Growth Factor Receptor

EPO Erythropoietin

ER Estrogen Receptor

EZH2 Enhancer of Zeste Homolog 2

FRAP FKBP-rapamycin-associated protein

FIH-1 Factor Inhibiting HIF1

GLUT-1 Glucose Transporter-1

GnRH Gonadotropin-Releasing Hormone

GPD1L Glycerol-3-phosphate Dehydrogenase 1-lik

HAF HIF-associated factor

HDAC Histone-Deacetylase

HEXIM1 Hexamethylene-Bis-Acetamide (HMBA)-Inducible protein 1

HIF1α Hypoxia-Inducible Factor 1 alpha

HMBA Hexamethylene-Bis-Acetamide

HRE Hypoxia-Response Element

HSP Heat-Shock Protein

IGF-1 Insulin-like Growth Factor 1

IL-6 Interleukin 6

10 IRS-1 Insulin Receptor Substrate 1

α-KG α- Ketoglutarate

KDM5B Histone Lysine Demethylase

LBD Ligand Binding Domain

LDHA Lactate Dehydrogenase A

LH Luteinizing Hormone

LHRH Luteinizing-Hormone-Releasing Hormone

LOXL2 Lysyl oxidase-like 2

LSD1/LSD2 Lysine-Specific Demethylase 1/2

MAPK Mitogen-Activated Protein Kinase

MDM2/HDM2 Mouse Double Minute 2 Human homolog

MMP9 Matrix Metallopeptidase 9

MTA1 Metastasis-Associated 1

NADPH Nicotinamide Adenine Dinucleotide Phosphate-oxidase

NLS Nuclear Localization Signal

NOS Nitrous Oxide Synthase

NTD N-Terminal Domain

ODD Oxygen-Dependent Degradation

PAS PER-ARNT-SIM

PC Prostate Cancer

PCAF p300/CBP-associated factor

PDK1 Phosphoinositide-dependent kinase-1

PEP Phosphoenolpyruvate

PHD Prolyl Hydroxylase Domain

PI3K/AKT Phosphoinositide222223-OH22222kinase/protein22222kinase22222B

11 PKC Protein Kinase C

PLCγ Phosphoinositide phospholipase C γ

PLGA Poly Lactic-co-Glycolicacid

PPAR Peroxisome Proliferator-Activated Receptor

PR Progesterone Receptor

PSA Prostate-Specific Antigen

P-TEFb Positive Transcription Elongation Factor b

PZ Peripheral Zone

RACK1 Receptor for activated C-kinase 1

RAR Retinoid Acid Receptors

RBC Red Blood Cell

RNA pol II RNA polymerase II

ROS Reactive oxygen species

SARM Selective Androgen Receptor Modulators

SDF1 Stromal cell-Derived Factor 1

SENP Sentrine-specific Protease

SH2/SH3 Src Homology 2/3

snRNP Small Nuclear Ribonucleoproteins

SUMO Small Ubiquitin-like Modifier

TAD Transactivation Domain

TIMP1 TIMP Metallopeptidase Inhibitor 1

TGFα/β Transforming growth factor α/β

TR Thyroid Receptor

TZ Transition Zone

UB Ubiquitin

12 UBE2C Ubiquitin-conjugating enzyme 2C

UbH2B H2B monoubiquitination

VDR Vitamin D receptor

VEGF Vascular Endothelial Growth Factor

VHL von Hippel-Lindau

13 HEXIM1 Is An Inhibitor of Two Transcription Factors Critical in Cancer: The

Androgen Receptor and Hypoxia Inducible Factor-1 Alpha

Abstract

by

I-JU YEH

Prostate cancer (PC) is the second leading cause of death in the United States. It is diagnosed in 80% of men by the time they reach age eighty. Clinically, androgen-ablation therapies are used in the treatment of androgen-dependent prostate cancer (ADPC). Initially these therapies effectively reduce tumor progression; however, ADPC ultimately regress into castration resistant prostate cancer (CRPC).

The mechanisms for how ADPC gradually becomes CRPC are not well-defined.

Studies in our lab revealed that down-regulation of HEXIM1 expression in human prostate cancer cell line (LNCaP) enhance the proliferation rate of the cells. Moreover,

HEXIM1 expression is attenuated in human prostate tumors in comparison to normal prostate tissues. Our data also indicates that HEXIM1 not only interacts with AR but also regulates the AR response to anti-androgen in LNCaP cells. HEXIM1 enhanced recruitment of KDM5B to AR target genes in LNCaP cells. Our studies also show that the mechanism of HEXIM1 regulation of AR transcription involves modulating histone modifications through KDM5B recruitment to AR target genes. Based on these observations, we hypothesize that the loss of HEXIM1 results in

14 androgen-independent action of AR.

The other part of my dissertation is focused on the inhibition of Hypoxia Inducible

Factor 1-alpha (HIF1α.) by HEXIM1. We previously reported that mice expressing

truncated HEXIM11-312 have a higher incidence of mammary tumors with increased vascularization and levels of VEGF and HIF1α. VEGF is an important mediator of angiogenesis and its expression is regulated by HIF1α. Stabilization of the HIF1α protein is regulated by various post-modifications, such as hydroxylation and acetylation. Under normoxia, the half-life of HIF1α is less than 5 minutes. Two key proline residues of HIF1α, Pro402 and Pro564, can be hydroxylated by prolyl hydroxylase (PHDs) and recognized by von Hippel-Lindau (pVHL). The binding of pVHL to HIF1α results in the transport of HIF1α to the proteasome to be degraded. Acetylation of HIF1α by ARD1 (arrest-defective protein 1) was also reported to promote HIF1α degradation. The critical role of HIF1α in tumor metastasis arises not only from the fact that it is a potent activator of angiogenesis but also of invasion and metabolic reprogramming.

The major goal of my study is to investigate the mechanism of HEXIM1 regulation of HIF1α protein expression and the functional consequences of HEXIM1 inhibition of HIF1α. We hypothesize that under hypoxic condition, HEXIM1 regulates HIF1α protein stability by regulating post translational modifications of HIF1α.

15 CHAPTER I

INTRODUCTION

Statistic from the American Cancer Society indicate that approximately 218,000 men were diagnosed with and 32,000 men died of prostate cancer in 2010 [1]. This report indicates that prostate cancer is the second leading cause of cancer death in

American men [2]. Diagnoses of prostate cancer include biopsy following a digital rectal exam (DRE) or prostate-specific antigen (PSA) screening. The classification of prostate cancer into different stages is based on Gleason score. In the metastatic stage, the five years survival rate is only 30.6% [3].

The growth of prostate cancer is highly dependent on androgen [4-6]. The initial state of prostate cancer is androgen-dependent (ADPC); thereby androgen deprivation is a standard treatment for prostate cancer. However, 80-90% of prostate cancer eventually develops into castration resistant prostate cancer (CRPC) within 2-3 years after androgen ablation therapy (Figure I-1). Previous studies indicated that the androgen receptor (AR) plays an essential role in both ADPC and CRPC [5, 7].

Characterization of the role of AR in ADPC and CRPC is important for the development of therapy for prostate cancer.

16 REVIEW OF LITERATURE

Androgen Receptor (AR)-Class I steroid receptor family

The AR (Androgen Receptor) belongs to class I nuclear receptor family [8, 9].

Nuclear hormone receptors belong to a family of intracellular transcription factors that regulate multiple cellular behaviors such as development, reproduction, cell cycle progression and apoptosis [9-11]. Forty-eight genes that encode nuclear receptor family have been identified [10, 12]. Based on the sequence homology, nuclear receptor family can be classified into three categories (Table 1). Class I is the steroid receptor family and includes the progesterone receptor (PR), estrogen receptor (ER), glucocorticoid receptor (GR), AR and mineralocorticoid receptor (MR). Class II is the thyroid/retinoid receptor family and includes the thyroid receptor (TR), vitamin D receptor (VDR), retinoic acid receptor (RAR) and peroxisome proliferators-activated receptors (PPARs). Class III includes orphan receptors with no known identified cognate ligands. Ligand-independent actions of orphan receptor have also been reported [13, 14], . Some orphan receptors have been reported to be involved in variety of metabolic processes, such as lipid homeostasis [11, 15].

AR shares structural similarity of several functional domains with class I nuclear receptor family (Figure I-2) [8] . AF-1 (transactional activation), included in N terminus domain (NTD), modulates the ligand-independent action. The AF-1

17 sequence is less conserved (<15%) across all members of the nuclear receptor family

[12]. While AF-1 functions in ligand-independent manner, it synergistically cooperates with ligand binding domain (LBD, AF-2) to regulate transcription. Post translational modifications of AR also modulate the activities of AF-1 and AF-2 [16].

DNA binding domains (DBD) are highly conserved and comprised of two zinc-finger motifs [12]. The zinc atoms in this region are essential in maintaining a stable structure. Loss of zinc atoms results in the loss of DNA binding activity.

The adjoining flexible hinge region (H) connects the DBD to the LBD at the carboxyl terminus. In the AR, this region contains a nuclear localization signal (NLS), which is required for the translocation of the AR into the nucleus [17]. In the absence of ligand, AR is mainly located in cytoplasm and associated with Heat Shock Protein

(HSP) 90, 70, 56 and 23, which stabilize the AR and prevent the AR from being degraded [18]. Upon ligand binding, AR dissociates from the HSP complex, translocates into nucleus, and binds to the AR response element in the promoter and enhancer regions of its target genes [12, 18]. Cytoskeleton protein Filamin (FlinA) also interacts with DBD, LBD and hinge domains that, in turn, facilitate AR translocation [18].

The AR LBD is mainly composed of twelve helices and the ligand binding pocket is formed by helices 3, 5 and 10 [19]. Upon ligand binding, helix 12 (H12) is

18 repositioned, creating an interface for the recruitment of coactivators [20]. On the other hand, binding of antagonists to AR, repositions H12 into a different conformation that results in recruitment of corepressors [20]. Unlike other nuclear receptors, the primary site of interaction between AR and coactivators is NTD, whereas in other nuclear receptors this interaction domain is located in the C-terminal

LBD (CTD) [19]. While LxxLL-motif containing coregulators are the preferred interacting partners of other nuclear receptors and are required for transactivation, AR preferentially binds to FxxLF motifs which has been found in the NTD and several coregulators [19, 21]. The intramolecular (NTD-CTD) and intermolecular interactions of AR result in dimerization of AR, which occurs independently of ligand-binding.

Binding to ARE stabilizes AR dimerization. Whether other factors are involved in the facilitation of AR dimerization is still under investigation [18].

Non-genomic responses of AR to androgens have been reported. Upon ligand binding, androgen-bound AR interacts with caveoline-1 in the cytoplasm, that, in turn, triggers activation of PI3K/Akt signaling in prostate cancer cells [22]. Caveoline-1 is an integral membrane protein primarily found in plasma membrane caveolae and is critical for endocytosis and signaling [22]. The activation of kinase pathway provides a potential survival mechanism for prostate cancer cells [1, 5, 23]. However, these non-genomic activities of AR have not been validated in animal model systems [24].

19 Overview of prostate and prostate-related disease a. Prostate structure

The prostate is a male reproductive glands located below the bladder and in front of the rectum. The prostate can be divided into three zones: Central Zone (CZ),

Transitional Zone (TZ) and Peripheral Zone (PZ) [25]. Seventy percent of prostate cancers develop from the largest zone, the PZ whereas the CZ is responsible for only

1-5% of prostate cancers. However, prostate cancer that develops from the CZ tends to be more aggressive. Growth of TZ results in benign prostatic hyperplasia (BPH)

[25]. b. Prostate Function

The prostate gland is composed of muscle and fibrous tissue and secretes seminal fluid. This fluid is ejected into the prostatic urethra by peristaltic contractions of the muscular wall [25]. In addition to seminal fluid, other proteins such as prostatic acid phosphatase, prostate-specific antigen (PSA) and proteolytic enzymes are also secreted by the prostate [26]. PSA helps to maintain the semen in a liquid form and is a biomarker of BPH and prostate cancer [26]. c. Hormonal regulation of the prostate

The hypothalamus controls multiple functions in body by functionally linking the nervous and endocrine systems through the pituitary gland [27]. The hypothalamus

20 secretes luteinizing hormone releasing hormone (LHRH), which stimulates the pituitary gland to synthesize luteinizing hormone (LH) and adrenocorticotropic hormone (ACTH). LH stimulates the testis to produce testosterone, whereas ACTH stimulates the adrenal gland to synthesize adrenal androgens such as dihydroepiandrosterone (DHEA) [28]. Both testicular and adrenal androgens have effects on the prostate, although the effects of testicular androgens are predominant.

In the prostate, testosterone can be converted to DHT by 5α-reductase. DHT is the most potent substrate for AR. After stimulation by DHT, the activated AR promotes prostate maturation (Figure I-3) [28, 29]. d. Prostate Diseases

Prostatitis. Prostatitis is the inflammation of prostate gland which can be caused by bacterial infection or unknown reasons [30]. Symptoms of prostatitis include pain, urination problems, and sexual dysfunction. Prostatitis can be classified into four types: inflammatory/non-inflammatory chronic prostatitis (type I), acute prostatitis, chronic bacterial prostatitis (type II) or chronic pelvic pain syndrome (CP/CPPS, type

III) and asymptomatic inflammatory prostatitis (type IV) [30]. Both type I and type II are bacterial infections and can be treated efficiently with antibiotics. Both Type III and type IV are non-bacterial infection. Type III accounts for 90-95% of diagnoses and causes pain. Although type IV is asymptomatic, leukocytes can be detected in

21 patient’s urine. The causes of nonbacterial prostatitis are not known. Possible causes include a previous bacterial infection, irritation from chemicals, nerve problem involving the lower urinary tract, and parasites [31].

BPH (Benign Prostatic Hyperplasia). BPH, also called benign enlargement of the prostate (BEP), denotes the enlargement of prostate stromal and epithelial cells and is initiated in the TZ [32]. BPH usually occurs in men over age of 50. Causes of BPH initiation are unknown. Some studies indicate that DHT is one of the main causative factors of BPH [33]. Due to the TZ that surrounds the prostate gland, the enlarged prostate will grow inwardly and squeeze and obstruct the urethra. In order to urinate, the bladder needs to contract more to eject urine. Thus, the bladder wall will become thicker and irritable and causes the bladder to contract more frequently in order to empty its content. Gradually, the bladder becomes weaker and loses its urinary function [34]. However, having BPH is not a strong reliable indicator for the risk of developing prostate cancer because BPH and prostate cancer are two different diseases [35].

Prostate Cancer. Prostate cancer refers to cancer that initiates from the prostate. In contrast to BPH, prostate cancer develops from the outer PZ and grows outwardly to invade the surrounding tissue such as the seminal vesicles and bladder [3]. Several factors contribute to prostate cancer, including age, race, family history, diet

22 inflammation and infection [3]. The progression of prostate cancer is correlated with changes in several genes and signaling pathways (Figure I-4) [36].

Androgen Receptor and Prostate Cancer

AR signaling pathways are critical for the development of male reproductive tissues [28]. In particular, AR transcriptional activity is involved in the regulation of growth and differentiation of male reproductive tissues, and maintenance of normal male reproductive function [6, 37]. Androgens and the AR are also required for the survival of malignant prostate cancer cells [5, 6]. Since AR is a ligand-dependent receptor, androgen-ablation therapy is typically the first choice for the treatment of

ADPC. However, the recurrence of Prostate Cancer (PC) is inevitable because prostate cancer gradually becomes resistant to androgen-ablation therapy and develops into castration resistant prostate cancer (CRPC). The term “castration resistant prostate cancer” is preferred over the term “androgen-independent prostate cancer” because the growth of advanced PC is still dependent on AR function [5, 6,

37]. AR is proposed to mediate distinct transcriptional programs in ADPC and CRPC through different combinations of co-regulators [7]. A winged-helix transcription factor, FoxA1, is known to play an essential role in the development of different cancer types, such as prostate, breast, lung and bladder cancers [38]. High expression of FoxA1 is associated with poor prognosis of prostate cancer. FoxA1 is involved in

23 AR-mediated transcription [38-40], and FoxA1 interacts with the AR through AR’s

DNA-binding domain. Epigenetic markers, such as H3K4me1, H3K4me2 and

H3K4me3 levels, are markers for FoxA1 recruitment to the enhancer region of AR target genes in prostate cancer cells [7, 41]. Clinical relevance is further supported by studies using patient specimens that showed H3K4 methylation was significantly increased in CRPC [42].

PSA is a 34 kDa glycoprotein which belongs to human kallikrein-related peptidase family [43]. PSA is synthesized exclusively by prostatic epithelia, and the amount of

PSA in serum can be used as an indicator of BPH and prostate cancer. Screening for

PSA levels has been routinely used since 1990 [26]. Serum PSA of at least 10 ng/ml is an indicator of BPH. For metastatic prostate cancer, the level of serum PSA always exceeds 200 ng/ml [26, 43, 44]. By PSA screening, 40% of PC mortality is reduced

[45]. However, a concern regarding PSA screening is that it cannot distinguish indolent prostate cancer from aggressive prostate cancer resulting in over-treatment of prostate cancer. A study published in New England Journal of Medicine indicated that

2/3 of men diagnosed with prostate cancer are actually at low risk of mortality [46].

The combination of PSA screening, Gleason score are more reliable determinants of prostate cancer treatment. Based on histological examination of prostate cancer tissues, a Gleason score from 2 to 10 is assigned [47].

24 AR is required in CRPC

Because androgens and AR are required for the survival of malignant prostate cancer cells, androgen-ablation therapy or surgical castration can effectively prevent cancer cell growth at the onset. Unfortunately, despite continued treatment, tumors recur to a more aggressive and metastatic phenotype, a major contributor being reinstatement of proliferation capability. This type of recurrent PC is referred to as

CRPC. Several reports indicate that the AR plays a critical role in CRPC cells [6, 24,

48, 49]. Knockdown of AR using AR shRNA in xenografted LNCaP cells resulted in delayed tumor progression and attenuated tumor growth in immunocomprised mice

[50]. However, the mechanistic basis for AR regulation of the progression from

ADPC to CRPC is not well-defined. Understanding the molecular basis for the development of CRPC is crucial for developing successful therapies. Several possible mechanisms are discussed below.

Mechanisms involved in the development of CRPC that require AR a. AR mutation

The rate of somatic AR mutation in CRPC is approximately 8-25% [51]. AR mutations in the ligand-binding domain, amino terminus, DNA-binding domain (DBD) and C-terminal ligand domain (CLD) have been identified in CRPC patients [52]. The

N-terminal domain contains activation function-1 (AF-1, amino acid 101-370), which

25 is important in ligand independent transcriptional activation. E231G mutation in this region causes malignant transformation and metastatic progression [53]. E255K interferes with the interaction between AR and E3 ubiquitin ligase that in turn, increases the stability and prolongs the nuclear localization of AR even in the absence of ligand [54]. A novel R615S mutation has been found in the DBD (amino acid

559-624). This mutation results in the loss of the sensitivity of the AR to androgens

[55]. This unresponsiveness results in the impaired development of male secondary sexual characteristics and mild spermatogenic defect despite the presence of

Y-chromosome. Mutations located in LBD, such as W741C/L and F876, results in activation by AR antagonists [56]. In particular, non-steroid ligands, such as bicalutamide and flutamide that behave as AR antagonists in androgen-dependent PC, activate AR and attenuate hormone ablation therapy efficacy in CRPC. Q798E,

M715V and H874Y exhibit altered ligand binding specificity [57]. Spliced variant forms of AR (ARVs) that lack the DBD have been found in prostate cancer patients

[52]. Level of ARVs is increased during the progression to CRPC. Increased expression of ARVs in xenografted cancer cells resulted in enhanced tumor proliferation in castrated mice [58]. Interestingly, full-length AR is necessary for tumor growth–promoting effect of ARVs and resistance to hormone ablation therapy

[59].

26 b. AR over expression/amplification

AR gene amplifications have been identified in approximately 30% of patients with

CRPC [60]. Higher AR protein level facilitates tumor cell survival under low androgen concentrations resulting from hormone ablation therapy [60]. Paradoxically, patients with amplified AR levels have twice the median survival time than patients without amplified AR [52, 60]. c. Altered expression of AR co-activators or co-repressors

Aberrant expression levels of co-regulators contribute to CRPC [61]. Well characterized regulators include chaperones, histone and chromatin modifying proteins and factors that bridge the transcriptional machinery to AR [62].

Chaperones, such as heat shock protein 90 (HSP90), interact with AR to allow proper folding and increase AR stability, thereby maintaining AR in a conformation required for high affinity ligand binding [63]. Inhibition of HSP90 impairs the ligand binding and abrogates the translocation of AR to the nucleus. Thus, HSP90 has been a target for cancer therapy. For example, 17-Allyamino-17-demethoxygeldanamycin, a small molecule of HSP90 inhibitor, is now in Phase I clinical trial for treatment of patients with advanced solid tumors [63, 64].

Epigenetic changes resulting from AR-mediated transcription have been extensively investigated. AR coactivators, SRC-1 and TIF2, belong to the p160

27 protein family members and possess histone acetyl-transferase activities [18].

Increased expressions of these two co-activators in CRPC cells not only increased the recruitment of CBP/p300, another acetyl-transferase [65], but also induced recruitment of the methyltransfase CARM-1 to the ARE of AR target genes [24].

Consequently, enhanced acetylation and methylation of histone 3 at promoter regions of AR target genes facilitate the activity of the basal transcriptional machinery [66].

The expression of CARM-1 regulated AR target genes result in primarily non-hormone dependent activity [66].

Co-regulators that modulate AR transcription activity also contribute to the progression of CRPC. FoxA1 belongs to wing helix-turn-helix transcription factor family [38]. FoxA1 is a pioneer protein that engages the ER-mediated transcriptional program by creating a less compact chromatin structure that enhances recruitment of co-regulators [67]. Some reports indicate that methylated H3K4 define

FoxA1 binding regions [40]. Expression of FoxA1 in prostate cancer correlates with epigenetic modifications in CRPC as well [39].

Binding of androgens alters the position of the AF-2 helix in the LBD of the receptor, resulting in the recruitment of co-activators and dissociation of co-repressors

[20, 61]. In the presence of antagonist, AF-2 helix orients differently and results in a binding surface for co-repressors, such as NCOR1, NCOR2/SMRT and CoREST [68].

28 Down-regulation of co-repressors in aggressive tumors is associated with the resistance to antagonists [69]. d. Induction of downstream signaling pathways

Several anti-apoptotic or survival pathways are found to be activated in CRPC [70].

These signaling pathways contribute to resistance to androgen ablation therapy.

Insulin-like growth factor-1 (IGF-1), epidermal growth factor (EGF), interleukin-6

(IL-6) and Wnt signaling have been reported to regulate expression of androgen responsive genes by cross talk with AR in CRPC in an androgen-independent manner

[18]. The PI3K/Akt (a ser/thr protein kinase) pathway has been reported to inactivate the pro-apoptotic factor, Bad [71]. Thus, activation of PI3K pathway is one of the pivotal survival signals in prostate cancer [72]. e. Post-translational modifications of AR

Post-translational modifications of AR include phosphorylation, ubiquitination, methylation, acetylation, and SUMOylation (small ubiquitin-like modifier) [73].

Expressions of several enzymes that are responsible for these modifications are altered in prostate cancer.

Tyrosine phosphorylation of AR is triggered by growth factors and is independent of androgens [74, 75]. This modification correlates with high Gleason grade. Ack1, a kinase that phosphorylates the AR at Y267 and Y363 in the transactivation domain, is

29 inappropriately activated in CRPC [48, 76].

RNF-6, an E3 ubiquitin 3 ligase, modifies K845 and K847 of AR and forms polyubiquitin chains [77]. This modification induces a structural change in the AR that results in recruitment of cofactors to the ARE (androgen-response element), selectively modulating a subset of AR target genes. RNF-6 is highly expressed in

CRPC and is associated with the progression to CRPC [77, 78].

Recent findings suggest that SUMOylation abrogates AR activity [73]. AR activity is enhanced when SENP (sentrine-specific protease), a SUMO-protease enzyme, catalyzes the removal of SUMO from AR. SENP1 expression was found to be elevated in patients with CRPC [79]. g. Epigenetic regulations of AR in CRPC

Several epigenetic alterations, such as methylation and acetylation, are frequently found in prostate cancer [73]. These modifications are thought to alter chromatin structure, and allow recruitment of co-regulators and regulation of transcription [80,

81]. Acetylation of histone 3 or 4 (H3Ace or H4 Ace) promotes transcriptional activity by “relaxing” the chromatin structure. Whether histone methylation is involved in transcriptional activation or repression depends on the modified lysine site.

H3K4, H3K36 and H3K79 are usually associated with transcriptional activation whereas H3K9, H3K27 and H4K20 are usually associated with transcriptional

30 repression [82, 83]. Histone methylation can exist in mono-, di- and tri- form. The levels of methylation are also thought to regulate transcriptional activation or repression [42, 84]. Several enzymes have been identified to be involved in adding or removing the methyl residues, implying that histone methylation is a dynamic process.

Overexpression of certain histone modifiers has been linked to prostate cancer progression. For example, LSD1 (Lysine-specific demethylase 1, a H3K4/K9 demethylase), and EZH2 (Enhancer of zeste homolog 2, a H3K27 methylase), are both over-expressed in CRPC and are potential prognostic markers of recurrence [81,

83]. Increased H3K4 methylation levels have been correlated with prostate cancer progression, however the H3K4-modifying enzymes responsible for the increased levels of H3K4me levels in CRPC are not well-defined [81].

Two classes of histone demethylases have been discovered [85]. Class I, LSD1 and

LSD2, reverse histone H3K4 methylation by an oxidative demethylation reaction that uses flavin as a cofactor. This class removes only mono- and di-methyl-lysine residues. Unlike class I, class II can remove all three methylation states on histone lysines. Class II catalyzes demethylations by an oxidative reaction and requires Fe2+ and -KG as cofactors. Class II is characterized by a conserved JmjC (Jumonji C) domain and can reverse methyl-lysine states. JmjC domain was first identified by the conserved amino-acid residues in the Jarid2 (Jumonji), Jarid2C (Smcx) and Jarid1A

31 (RBP2) proteins. The JARID family has two subgroups, JARID1 and JARID2.

JARID2 has no detectable demethylase activity whereas JARID1 specifically catalyzes di- and tri-demethylation on H3K4 [85]. Four members in JARID1 subgroup have been identified in mammals [85]. JARID1B, also known as KDM5B or PLU-1, is up-regulated in 90% of primary and metastatic breast cancers [86, 87].

KDM5B promotes breast cancer progression by repression of BRCA1, a tumor suppressor gene [88]. However, it has been reported that KDM5B expression is gradually lost in advanced melanomas [89]. Furthermore, KDM5B represses the expression of CCL14, an epithelial chemokine, resulting in the suppression of angiogenesis and metastasis in breast cancer cells [90].

As mentioned in previous section, FoxA1 has been associated with epigenetic modifications in CRPC. Besides pioneering the AR pathway, FoxA1 reprograms the

AR binding sites in CRPC [39]. H3K4 methylations have been reported as markers for FoxA1 recruitment to the enhancer regions of AR target genes in CRPC, and sequentially induce AR binding. A consequence of the regulation of AR by FoxA1 is the induction of a distinct transcriptional program in CRPC [7]. AR selectively binds to the enhancer region of and transcriptionally upregulates M-phase cell cycle genes,

CDK1, CDC20 and UBE2C in CRPC. In the presence of CDK1, CDC20 is phosphorylated and activated resulting in the ubiquitination and degradation of

32 anaphase-promoting complex/cyclosome (APC/C) and cyclin B-mediated metaphase-anaphase transition [91]. UBE2C is an APC/C specific-E2 ubiquitin-conjugating enzyme that inactivates the M-phase checkpoint and promotes

M-phase cell cycle progression [7]. Down-regulation of UBE2C gene expression attenuates progression of CRPC [92]. f. Prostate Cancer Stem cells

Prostate cancer stem cells (CSCs) are undifferentiated cells that have tumor-initiating capability [93]. Cell markers such as CD133 and 21 integrin can be used to distinguish between prostate cancer stem cells and non prostate cancer stem cells. It has been reported that increasing cellular PSA expression is reciprocally correlated with tumor grade [93, 94]. Moreover, the expression of cellular PSA in prostate cancer has been correlated with the differentiation state of prostate cancer [94,

95]. Prostate cancer cells with low PSA expression were also characterized as prostate cancer stem cells by the expression of stem cell markers [94]. Stem cells also exhibited resistance to chemotherapy, suggesting that CSC populations may contribute to the progression of CRPC [94]. Targeting prostate cancer stem cells is a promising therapeutic strategy for treating CRPC.

33 Therapeutic strategies

Strategies for treating the prostate cancer include gonadotropin-releasing hormone

(GnRH) agonists, inhibition of testosterone synthesis, inhibitors of DHT, and AR antagonists (Figure I-3) [96]. GnRH, also known as LHRH

(Luteinizing-hormone-releasing hormone), stimulates the synthesis of LH and ACTH.

GnRH agonists such as Leuprolide, are synthetic peptides which bind to GnRH receptors and results in GnRH receptor desensitization [97].

Abiraterone, Ketoconazole and indomethacin, inhibit CYP17A1 [98]. CYP17A1 is expressed in adrenal, testicular and prostatic tumor tissues and promotes the synthesis of testosterone [99]. Inhibitors of DHT synthesis, such as Finasteride and Dutasteride, inhibit 5 reductase, which catalyzes the conversion of testosterone to DHT [98].

AR antagonists include Nilutamide, Flutamide, Bicalutamide and MDV3100

(Figure I-3) [96]. The structure of the antagonist liganded-full length AR has not been resolved. Thus, a precise mechanism of how antagonists inhibit AR activity remains elusive. Some antagonists such as Bicalutamide switch from antagonist to agonist in

CRPC [100]. They are so-called Selective-AR-modulators (SARMs), and the mechanistic bases for their selective antagonist/agonist activities are not well-defined.

Some studies indicate that the antagonistic property of SARMs is due to either the increased recruitment of co-repressors or the ineffective recruitment of co-activators

34 [101]. For example, bicalutamide has been reported to stimulate the assembly of transcriptionally inactive AR, which is unable to recruit coactivators such as SRC-1 or

-2) on DNA [102]. One report favored enhanced recruitment of co-activators rather than loss of co-repressors as the major contributor to the agonistic property of

Bicalutamide in CRPC [103]. Molecular dynamic-based simulation of

AR-Bicalutamide complex has been conducted. One of the simulated complexes revealed that H12 of AR induces an additional binding pocket next to the hormone binding site and results in the distortion of the co-activator binding site [104]. Our studies indicate that the recruitment of HEXIM1 to AR target genes is higher in the

Bicalutamide-treated ADPC cell line. This indicates that Bicalutamide may not only induce the conformational change of AR but also induce the recruitment of co-repressors, such as HEXIM1, to modulate AR activity.

Hexamethylene-inducible gene 1 (HEXIM1), also identified as Estrogen down-regulated gene 1 (EDG1)

Hexamethylene-bis-acetamide (HMBA)-inducible protein 1 (HEXIM1) is a 359 amino-acid protein. HEXIM1 expression was initially discovered because its expression was induced by HMBA in vascular smooth muscle cell [105]. Studies indicated that HEXIM1 inhibits transcriptional elongation process by preventing the interaction between P-TEFb (Positive Transcription Elongation Factor b, composed of

35 cyclin T1 and CDK9) complex and RNA pol II [106]. The interaction between

HEXIM1-7SK snRNA complex and cyclin T1 is essential for the transcriptional repression activity of HEXIM1 [107, 108]. HEXIM1 has three functional domains: (1) a N-terminal proline-rich domain that acts as a self-inhibitory domain, (this domain prevents HEXIM1 from interacting with P-TEFb in the absence of 7SK snRNA); (2) a basic-rich nucleus localization signal (NLS) domain that interacts with 7SK snRNA; and (3) a C-terminal domain that interacts with ERα, P-TEFb and Tat. Tat is a HIV viral protein that binds to P-TEFb and controls host transcriptional elongation [109].

HEXIM1 has been shown to inhibit Tat activity and is a promising pharmaceutical target for Acquired Immune Deficiency Syndrome (AIDS) treatment [110, 111]. The interacting domains of HEXIM1 are shown in Figure I-5.

Hexamethylene-inducible gene 1 (HEXIM1) and Prostate Cancer

In 2011, Mascareno et. al. reported that HEXIM1 was a modulator of AR phosphorylation [112]. However whether the modulation of AR phosphorylation by

HEXIM1 results in a change in AR activity was not reported. Moreover, in contrast with our data and the Oncomine datasets, the authors indicated that the expression of

HEXIM1 is up-regulated in BPH and adenocarcinomas compared to normal prostate tissue. The expression data of Mascareno et. al. is also not consistent their own findings in mice. Loss of one HEXIM1 allele in the TRAMP mice

36 (TRAMP/HEXIM1+/-), accelerated tumor progression. They also observed redistribution of HEXIM1 from the nuclei to the cytoplasm in the

TRAMP/HEXIM1+/- mice, suggesting that the influence of HEXIM1 on the AR is attenuated. However, they did not provide a mechanistic basis for this observation.

HEXIM1 was first discovered as an inhibitor of transcription elongation factor

(P-TEFb), which is composed of CDK9 and cyclin T1. Phosphorylation of the AR by

CDK9 has been reported to promote AR transcriptional activity and prostate cancer cell growth [113]. Thus, the modulation of AR phosphorylation observed by

Mascareno et al. can be attributed to the inhibition of CDK9 by HEXIM1 and not by direct regulation of HEXIM1 by AR as we have observed.

Studies in the Montano laboratory also revealed that downregulation of HEXIM1 expression in an androgen-dependent human prostate cancer cell line (LNCaP), enhanced proliferation rate. Most importantly, HEXIM1 plays a critical role in the inhibition of AR activity by anti-androgens. In support, Oncomine analyses indicated decreased expression of HEXIM1 in CRPC compared to ADPC. Systemic analysis of the cancer outlier profile supported our expression data and indicated that HEXIM1 is a potential transcriptional suppressor in both prostate cancer and breast cancer [114].

We found that decreased HEXIM1 expression resulted in reduced KDM5B expression and enhanced H3K4me2 expression in LNCaP cells. Our studies indicate that

37 HEXIM1 is required for the ability of KDM5B to inhibit methylation of H3K4 on and

FoxA1 recruitment to AR target genes.

STATEMENT OF PURPOSE

This project focuses on the mechanism of AR activity regulated by HEXIM1 and the role of HEXIM1 in the transition from ADPC to CRPC. We found that HEXIM1 and its operational partner regulate methylation level on H3K4, that is essential for the adpc to crpc transition.

Coregulators are essential for regulating AR activity. Enzymes that control the access to DNA regulatory regions in target genes by modifying histone tails have been observed. Cooperation between HEXIM1, AR and KDM5B that results in the alteration of histone methylation has been discovered in our work. In the absence of

HEXIM1, elevating the methylation level on the enhancer region of target genes results in the activation of a CRPC transcriptional program exerted by AR.

Deciphering the coregulators that trigger the transition of ADPC to CRPC would provide a very real possibility for the treatment of CRPC.

38

Figure I-1. Androgen-deprivation therapy is the first-line choice in treating

Prostate Cancer. However, PC recurs in approximately two years and is castration-resistant.

Figure I-1 adapted from: http://www.webedcafe.com/extern/program_media/goldjournal.net/2013/prostate_can cer

39

Figure I-2. Functional domains of Androgen Receptor (AR). AF, activation function; DBD, DNA-binding domain; H, hinge; LBD, ligand-binding domain

Figure I-2 adapted from Annu. Rev. Physiol. 2007. 69:201–20

40

Figure I-3. Regulation of androgen synthesis and the drugs that haven been developed to target androgen production and the activation of the AR.

41

Figure I-4. The progression to CRPC is correlated with the changes in several genes and pathways. The survival of ADPC is dependent on AR whereas the survival of CRPC includes pathways that both involve and bypass AR.

Figure I-4 adapted from Debes and Tindall, 2004

42

Figure I-5. The functional domains of HEXIM1. Shown are regions in HEXIM1 that have been reported to interact with indicated proteins

43 CHAPTER II

HEXIM1 Plays a Critical Role in the Inhibition of the Androgen Receptor by

Antiandrogens

This work has been published in the Biochemical Journal

(Yeh, I-Ju et al., Biochemical Journal 2014)

Abstract

We show that Hexamethylene-bis-acetamide-inducible protein 1 (HEXIM1) functions as an androgen receptor (AR) co-repressor as it physically interacts with the

AR and is required for the ability of antiandrogens to inhibit androgen-induced target gene expression and cell proliferation. OncomineTM database and IHC analyses of human prostate tissues revealed that expression of HEXIM1 mRNA and protein are down-regulated during the development and progression of prostate cancer. Enforced downregulation of HEXIM1 in parental hormone-dependent LNCaP cells results in resistance to the inhibitory action of antiandrogens. Conversely, ectopic expression of

HEXIM1 in the castration resistant prostate cancer (CRPC) cell line, C4-2, enhances their sensitivity to the repressive effects of the anti-androgen, bicalutamide. Novel insight into the mechanistic basis for HEXIM1 inhibition of AR activity is provided by our studies showing that HEXIM1 induces expression of the histone demethylase,

KDM5B, and inhibits histone methylation, resulting in the inhibition of FOXA1

44 licensing activity. This is a new mechanism of action attributed to HEXIM1, and distinct from what have been reported so far to be involved in HEXIM1 regulation of other nuclear hormone receptors, including the estrogen receptor.

Introduction

The androgen receptor (AR) is a member of the nuclear steroid receptor family critical for the development of male reproductive tissues and skeletal muscle. AR mediates these effects by regulating gene transcription following androgen binding and receptor activation [115]. Deregulated expression and activation of AR is known to be critical for prostate cancer progression [115]. AR transcriptional activity results in the regulation of genes responsible for promoting growth, inhibiting apoptosis, and possibly enhancing metastasis of the prostate tumors [115]. Androgens have been shown to regulate numerous genes including prostate specific antigen (PSA), epidermal growth factor receptor (EGFR), cyclin dependent kinase 2 and 4, p21WAF1

(presumed to play an anti-apoptotic role in prostate cancer), cyclin Ds, survivin, and the androgen receptor (AR) coactivator ARA70 [115].

AR plays a role in all the stages of prostate cancer, from initiation, development and resistance to hormone therapy [116]. Owing to the dependence of prostate cancer on AR and androgens, the most common therapy for prostate cancer, in particular in advanced prostate cancer, is androgen ablation via orchodectomy or AR antagonists

45 [116]. Although prostate tumors are typically first responsive to such treatment, they invariably become resistant to anti-androgen therapy. However, AR silencing studies show that castration-resistant prostate cancer (CRPC) cells retain dependence on AR

[116]. This supports the model that the progression to CRPC is functionally linked to the constitutive activation of AR, which may result from some combination of the enhanced expression of AR and AR coactivator(s), reduced expression of AR co-repressor(s), and production of androgens in tumors [116].

We have previously reported that Hexamethylene-bis-acetamide inducible protein 1

(HEXIM1) inhibits the activity of Estrogen Receptor alpha (ERα) in vitro by intercepting an interaction between ERα and positive transcription elongation factor b

(P-TEFb) [117], a protein complex comprised of cyclin T1 and cyclin-dependent kinase 9 (CDK9). ERα directly binds to the cyclin T1 component of this complex

(reviewed in ref. [118]), recruiting P-TEFb to ER target genes whereby P-TEFb phosphorylates the carboxy terminal domain of RNA polymerase II (RNAP II) at serine 2 to promote productive phases of transcriptional elongation [118]. By associating with both ERα and the 7SK snRNP complex, HEXIM1 inhibits the co-recruitment of both ERα and cyclin T1 to the promoter region of ERα target genes.

As a result, HEXIM1 inhibits the phosphorylation of the carboxy terminal domain of

RNAP II and thereby prevents ER-mediated transcriptional elongation in breast cells.

46 HEXIM1 inhibits mammary tumorigenesis, partly by inhibiting ER-dependent transcription of cyclin D1 and vascular endothelial growth factor (VEGF) [119, 120].

We now report on the role of HEXIM1 as a putative co-repressor of AR that is required for the inhibition of AR by antiandrogens. The mechanism for HEXIM1 inhibition of AR involves decreases in the levels of active histone marks, notably histone 3 dimethylated at lysine 4 (H3K4me2) that guide FOXA1 recruitment to

AR-regulated cell cycle genes. These M-phase cell cycle genes were previously shown to be critical of AR induction of CRPC [121]. While HEXIM1 also inhibited recruitment of cyclin T1 to the AR target gene PSA, regulation of active histone marks also plays an important role in HEXIM1 regulation of AR activity. Thus the mechanism of HEXIM1 regulation of AR transcription is distinct from what we have reported so far for ERα.

Materials and Methods

Cell culture and transfections.

LNCaP cells were obtained from the American Tissue Culture Collection, C4-2 and

C4-2B cells [122] were obtained from Dr. Leland Chung, and LAPC4 cell were obtained from Dr. Robert Reiter. LNCaP and C4-2 cells were maintained as previously described [123], and LAPC4 were grown in Iscove's medium (Invitrogen) supplemented with 10% fetal bovine serum (FBS) + 1 nM R1881. To examine

47 androgenic responses, cells were cultured in their respective basal media containing

10% dextran-coated charcoal-extracted FBS (Invitrogen). Construction of expression vectors for control miRNA or HEXIM1 miRNAs (clones 35 and 609) were described previously [119]. LNCaP cells were transfected with expression vectors containing either the HEXIM1 miRNA insert or a control miRNA insert as previously described

[119]. Following blasticidin selection, cells expressing the highest level of GFP were flow-sorted and expanded. C4-2 cells were transfected with control or expression vector for FLAG-HEXIM1 [117] as previously described [119].

Coimmunoprecipitation.

Endogenous proteins were co-immunoprecipitated and analyzed as previously described [117].

In vitro Translation and Protein-protein interaction assays.

In vitro transcription and translation of AR were performed using the Promega

TNT kit (Promega, Madison, Wisconsin) according to the manufacturer’s recommendation. GST-pull down assays were previously described [124].

Western Blot.

Cell lysates were analyzed by western blot as previously described [119].

Anti-HEXIM1 was generated in the Montano laboratory [125]. Primary antibody against UBE2C (H-90; cat# sc-99146), AR (441; cat# sc-7305) and KDM5B (H-180;

48 cat# sc-67035) were obtained from Santa Cruz Biotechnology. Anti-GAPDH was obtained from Millipore.

Immunohistochemistry.

Human prostate tissue samples were obtained from the Cooperative Human Tissue

Network (CHTN) and a tissue microarray from US Biomax, Inc., # PR8011. All samples were confirmed to be AR positive by IHC. We carried out immunohistochemical staining to detect HEXIM1 levels as previously described [69,

125].

Chromatin-Immunoprecipitation.

ChIP assays were carried out as previously described [117]. The primers sequences used were:

PSA-ARE (forward)5’-TCCTGAGTGCTGGTGTCTTAG-3’ proximal

(reverse)5’-AGCCCTATAAAACCTTCATTCC-3’

PSA-ARE (forward)5’-CATGTTCACATTAGTACACCTTG-3’ enhancer

(reverse)5’-TCTCAGATCCAGGCTTGCTTAC-3’

PSA-coding (forward)5’-CACACCCGCTCTACGATATGAG-3’

(reverse)5’-GAGCTCGGAAGGCTCTGA-3’

UBE2C (forward)5’-TGCCTCTGAGTAGGAACAGGTAAGT-3’ enhancer1

(reverse)5’-TGCTTTTTCCATCATGGCAG-3’

49 UBE2C (forward)5’-CCACAAACTCTTCTCAGCTGGG-3’ enhancer2

(reverse)5’-TTCTTTCCTTCCCTGTTACCCC-3’

CDK1 enhancer (forward)5’-GGGAAAGAGAAGCCCTACACTTG-3’

(reverse)5’-GGGCTGTGCTACTTCTCTGGG-3’ CDC20 enhancer (forward)5’-GGAGTTGTGAGAACACCCGG-3’

(reverse)5’-AACACCCAGGTACACCCTCG-3’

KDM5B (forward)5’-GGCAACCCATGTCTATCACAAGAGG-3’ enhancer

(reverse)5’-CTGGATACTTTGATACTCATCTG-3’

Reverse Transcription (RT) PCR Analyses.

Cells were subjected to reverse transcription-PCR (RT37 PCR) analyses as previously described [119]. The primers sequences used were:

PSA (forward)5’-TGTGTGCTGGACGCTGGA-3’

(reverse)5’-CACTGCCCCATGACGTGAT-3’

UBE2C (forward)5’-TGGTCTGCCCTGTATGATGT-3’

(reverse)5’-AAAAGCTGTGGGGTTTTTCC-3’

CDK1 (forward)5’-CCTAGTACTGCAATTCGGGAAATT-3’

(reverse)5’-CCTGGAATCCTGCATAAGCAC-3’

CDC20 (forward)5’-CCTCTGGTCTCCCCATTAC-3’

(reverse)5’-ATGTGTGACCTTTGAGTTCAG-3’

50 KDM5B (forward)5’-CATCACTGGCATGTTGTTCAAATTC-3’

(reverse)5’-GAATGTAGTAAGCCACAAGAAGC-3’

Proliferation assay.

LNCaP cells transfected with expression vector for control miRNA or HEXIM1 miRNA were plated onto 96 well plates. Cells were treated with 10 nM R1881, 10 uM bicalutamide, or both for 7 days. Cell proliferation was assessed using the MTT based

Cell Growth Determination Kit from Sigma-Aldrich according to the manufacturer’s protocol.

Data analyses

Statistical significance was determined using Student’s t test comparison for unpaired data.

Results

Normal prostate epithelial cells expressed nuclear HEXIM1, while HEXIM1 expression is decreased in prostate cancer cells.

Our initial test for a role of HEXIM1 in prostate tumor progression was the examination of HEXIM1 expression in human prostate tissues. We have thoroughly tested for use in immunohistochemistry the HEXIM1 antibody used in these experiments [69, 125-127]. Strong nuclear HEXIM1 expression was evident in normal prostate tissues, and statistically significant decreased expressions were

51 observed in benign prostatic hyperplasia (BPH) and tumors. Moreover, HEXIM1 expression was higher in well or moderately differentiated tumors (Gleason grade <7) when compared to poorly or undifferentiated tumors (Gleason grade >7, Figure II-1A,

Figure II-8, p=0.0096). Human prostate tissue samples with Gleason score ≥7 is associated with shorter time to CRPC among patients with metastatic disease [128,

129]. Our analyses of HEXIM1 expression in human tissues were also consistent with other studies as analyzed by Oncomine [130-132]. Microarray gene expression data from another laboratory (also analyzed by Oncomine) showed a statistically significant decrease in the expression of HEXIM1 in CRPC when compared to benign and prostate carcinomas (Figure II-1B) ref. [131]. We also observed lower levels of

HEXIM1 expression in CRPC cell lines (C4-2, C4-2B) relative to the androgen-dependent cell lines (LNCaP, LAPC4) (Figure II-1C). These cancer cell lines were selected for such comparison because they also express the AR, a characteristic of the vast majority of prostate cancers.

Because of the decreased expression of HEXIM1 in CRPC, we examined if

HEXIM1 can regulate the response to antiandrogens. We used two different miRNA clones to downregulate HEXIM1 expression in LNCaP cells (Figure II-1D).

Decreased HEXIM1 expression resulted in increased cell growth and a complete loss of the ability of bicalutamide to repress basal or R1881-stimulated growth, although

52 these cells retained some sensitivity to growth stimulation by R1881 (Figures II-1E and F). Downregulation of HEXIM1 also resulted in attenuation of repressive effects of the antiandrogen MDV3100 on cell proliferation (Figure II-1G).

HEXIM1 interacted with the AR.

The AR has an established role in prostate cancer and CRPC, and further support for a role of HEXIM1 in this disease were our co-immunoprecipitation assays showing that endogenous AR interacted with endogenous HEXIM1 (Figure II-2A).

The interaction was attenuated in cells transfected with HEXIM1miR (Figure II-2B).

The interaction between HEXIM1 and AR was ligand independent, and was observed in the absence of AR ligands and in the presence of AR agonists and antagonists

(Figures II-2A and B). The interaction was validated using in vitro

Glutathione-S-Transferase (GST) pulldown assays to determine if HEXIM1 and AR directly interacted. GST-HEXIM1 bound to Sepharose beads associated with in vitro translated AR. In vitro translated AR did not interact with GST-alone. We have reported that amino acids 150-177 of HEXIM1 was required for the interaction with

 ERα [117], and have now determined that amino acids 150-177 was also involved in the interaction of the AR with HEXIM1 (Figure II-2C).

HEXIM1 regulated recruitment of transcriptional elongation factors to AR target genes.

53 To test the potential for HEXIM1 regulation of AR mediated transcription, we conducted chromatin immunoprecipitation analyses (ChIP) to examine the recruitment of HEXIM1 to the promoter of AR target genes in response to R1881 or bicalutamide and concomitantly assessed whether downregulating the expression of

HEXIM1 would alter the recruitment of transcriptional elongation factors to those promoters. Our data supported that HEXIM1 was recruited to the androgen response element (ARE) containing regions of the PSA promoter in LNCaP cells, and that such recruitment was increased following treatment with the anti-androgens bicalutamide, nilutamide, and MDV3100 (Figures II-3B and II-3C). Downregulation of HEXIM1 in

LNCaP cells by synthetic HEXIM1miRs (Figure II-3A), resulted in decreased recruitment of HEXIM1 to this promoter region as expected (Figure II-3B), and enhanced recruitment of cyclin T1 (Figure II-3D) and serine 2 phosphorylated RNAP

II (S2P RNAPII, Figure II-3E) to the coding region of the PSA gene following treatment with bicalutamide. These results supported that recruitment of HEXIM1 by treatment with bicalutamide inhibited the recruitment of both cyclin T1 and RNAPII to the coding region of PSA. Bicalutamide did not induce HEXIM1 recruitment to the non-ARE containing region of the PSA gene (Figure II-9).

In contrast to LNCaP cells, bicalutamide did not induce recruitment of endogenous

HEXIM1 to the PSA promoter in C4-2 cells, an androgen-independent derivative of

54 the LNCaP cell line (Figure II-3G). Transfection vector for Flag-tagged HEXIM1 resulted in enhanced HEXIM1 recruitment (Figure II-3F), the consequences of which are described below.

HEXIM1 modified histone marks on the enhancer regions of AR target genes.

It has been proposed that the role of AR in androgen-independent cancer cells is not to direct the androgen-dependent gene expression program without androgens, but rather to execute a distinct program resulting in androgen-independent growth [7]. In that study, AR selectively upregulated M-phase cell-cycle genes including CDC20,

CDK1, and UBE2C in an androgen-independent variant of LNCaP cells, abl.

FOXA1 was shown to direct AR-enhancer binding and activation of CDC20, CDK1, and UBE2C [7]. FOXA1 recruitment occurred primarily on H3K9me2-poor but

H3K4me1/2-rich regions, with H3K4me1/2 guiding FOXA1 cell type-specific recruitment through direct physical interactions.

ChIP assays for AR, FOXA1 and H3K4me2 were performed in control and

HEXIM1miR LNCaP cells after treatment with vehicle, R1881, or bicalutamide. We observed that the down-regulation of HEXIM1 resulted in increased recruitment of

AR and FOXA1, and increased levels of H3K4me2 in the enhancer regions of CDC20 and CDK1 in bicalutamide-treated groups (Figures II-4A, B, and C). We observed enhancement of FOXA1 and levels of H3K4me2 on the PSA promoter in

55 bicalutamide treated cells upon downregulation of HEXIM1, however the level of increase was less than that observed with the enhancer regions of CDC20 and CDK1.

While AR occupancy in bicalutamide-treated LNCaP cells was lower relative to

R1881-treated cells, downregulation of HEXIM1 resulted in increased AR occupancy in bicalutamide-treated cells relative to R1881-treated cells (Figure II-4A). However recruitment of R1881-liganded AR was decreased upon downregulation of HEXIM1, suggesting that while HEXIM1 attenuated recruitment of bicalutamide-liganded AR, it enhanced recruitment of R1881-liganded AR.

We observed similar recruitment patterns of FOXA1 in androgen-independent derivative C4-2 cells as those observed in LNCaP-HEXIM1miR cells (Figure II-4D).

Increased expression of HEXIM1 in C4-2 cells (Figure II-3F) attenuated FOXA1 recruitment (Figure II-4D). These results suggested that decreased levels of HEXIM1 in CRPC may contribute to dysregulated AR activation or activation of a distinct AR transcriptional program.

HEXIM1 regulated KDM5B expression and recruitment to AR target genes.

While we observed significantly increased recruitment of HEXIM1 to AR binding sites in the CDK1 or CDC20 genes in the presence of bicalutamide (Figure II-5A), the increase was not particularly impressive. We thus searched for other factors involved in HEXIM1 downregulation of H3K4me2 on AR target genes. We screened known

56 H3K4me2 demethylases for regulation by HEXIM1. KDM5B has been reported to be upregulated in prostate cancers and to enhance AR transcriptional activity [133].

Conversely, KDM5B was shown to be part of a repressive complex on PR target genes [134]. Progestin-induced displacement of KDM5B and MLL2/MLL3-mediated

H3K4 trimethylation during the initial chromatin remodeling events was required for progesterone gene activation [134]. Moreover, an Oncomine dataset indicated downregulation of KDM5B in CRPC (fold change: -2.7, p-value: 6.6x10-5, ref [130]).

It is possible that KDM5B interaction with HEXIM1 resulted in a different function depending on gene context.

Our ChIP-seq analyses and validation by standard ChIP indicated recruitment of

HEXIM1 to the enhancer region of the KDM5B gene (Figure II-5B). Downregulation of HEXIM1 resulted in decreased expression of KDM5B (Figures II-5C and D).

These results suggested that HEXIM1 was involved in the direct transcriptional regulation of KDM5B.

We then determined if KDM5B can modulate H3K4me2 on AR target genes, and the relative role of HEXIM1 in regulating KDM5B. Enforced expression of KDM5B

(Figure II-6A) resulted in decreased levels H3K4me2 on the promoter of AR target genes, and downregulation of HEXIM1 resulted in attenuation of the ability of

KDM5B to downregulate levels of H3K4me2 on AR target genes but not on the

57 control GAPDH gene (Figure II-6B). The basis for the regulation of KDM5B by

HEXIM1 was revealed by our observation that downregulation of HEXIM1 resulted in decreased recruitment of KDM5B (Figure II-6C). Thus, our studies implicated

HEXIM1 not only in the upregulation of expression of a H3K4me2 demethylase but also the recruitment of KDM5B to AR target genes as a mechanism for HEXIM1 downregulation of H3K4 methylation on AR target genes. The regulation of KDM5B recruitment may not just be due to HEXIM1 regulation of KDM5B expression because co-immunoprecipitation assays indicated that endogenous KDM5B interacted with endogenous HEXIM1 (Figure II-6D)

Modulation of HEXIM1 expression resulted in altered gene expression in response to R1881 and bicalutamide.

We next examined the functional consequences of HEXIM1 expression on

AR-dependent gene expression. Downregulation of HEXIM1 repressed CDK1,

CDC20, and UBE2C expression in R1881-treated cells, but enhanced expression of

CDC20, CDK1, and UBE2C mRNAs in vehicle- or bicalutamide-treated cells (Figure

II-7A). The attenuation of R1881-induced gene expression when HEXIM1 expression was downregulated reflected the downregulation of AR recruitment to AR target genes upon decreased HEXIM1 expression. Because of the emerging role of UBE2C in CRPC [135], we also examined regulation of UBE2C protein levels by HEXIM1.

58 The effects of HEXIM1 downregulation on the expression of UBE2C mRNA in vehicle- and bicalutamide-treated cells were reflected at the protein level (Figure

II-7B). Downregulation of HEXIM1 enhanced expression of PSA mRNA and protein in vehicle-, R1881-, or bicalutamide-treated cells

Discussion

While androgen ablation therapy is initially effective for the majority of men with metastatic prostate cancer, resistance to such treatment invariably develops through mechanisms that remain incompletely understood [136, 137]. Most CRPCs express

AR and are dependent on AR for growth and survival, through mechanisms reported to involve constitutive activation of AR [49, 138]. Comparative gene expression profiling of androgen-regulated gene changes in LNCaP cells with AR-regulated genes in the LNCaP androgen-independent variant, abl, strongly suggest that AR controls the expression of a set of genes in CRPC that is unique from those regulated by androgens in androgen-dependent prostate cancer (ADPC) [7]. The latter study provides compelling support that CPRC represents an alteration in the function of AR rather than just its constitutive activation.

While studies on AR coregulatory factors in prostate cancer are not novel, a role for

HEXIM1 in the inhibitory actions of antiandrogens has not been previously reported.

We also identified a new mechanism of action of HEXIM1 involving epigenetic

59 regulation through the H3K4 demethylase, KDM5B that subsequently inhibited

FOXA1 licensing activity. On a related note, H3K4me1, H3K4me2, and H3K4me3 levels were reported to be significantly increased in CRPC [81].

Previous studies demonstrated that P-TEFb binds to liganded AR, which recruits

P-TEFb to the promoters of androgen target genes to facilitate their transcriptional elongation [139, 140]. Our co-immunoprecipitation and GST pulldown experiments indicated that HEXIM1 directly interacted with and prevented AR from recruiting

P-TEFb to AR target genes, similar to the action of HEXIM1 on ER target genes [117,

119]. HEXIM1 may repress androgenic responses by inhibiting the release of P-TEFb from the 7SK small nuclear RNA protein complex (snRNP) [107, 108], sequestering

P-TEFb in a transcriptionally inactive complex that is unable to promote transcriptional elongation of AR target genes through phosphorylation of RNAPII

[106, 141]. We observed regulation of cyclin T1 recruitment by HEXIM1 on the PSA coding region, but did not observe an ensuing significant change in PSA mRNA levels.

On the other hand HEXIM1 regulated H3K4me2 levels and FOXA1 recruitment on

AR target genes, CDK1 and CDC20, with ensuing changes in mRNA levels. Thus our data suggest that HEXIM1 regulation of H3K4 methylation and FOXA1 recruitment may be more critical in HEXIM1 regulation of AR target gene expression.

There is an established link between transcriptional elongation and histone

60 methylation [142-146]. H2B monoubiquitination (UbH2B) plays a critical role in

H3K4 methylation [147, 148]. Conversely, H2B monoubiquitination depends upon the early steps of transcriptional elongation, and CDK9 activity is essential for maintaining global and gene-associated levels of histone H2B monoubiquitination

[144]. However HEXIM1 may act independently of P-TEFb. Results showing

P-TEFb independent actions of HEXIM1 would not be unexpected as our studies indicate that HEXIM1 induced phenotypic effects that are independent of its ability to inhibit P-TEFb [120, 127]. In particular, inhibition of ER-mediated activation of

VEGF gene transcription by HEXIM1 was independent of the ability of HEXIM1 to inhibit P-TEFb.

It is important to note that HEXIM1 regulation of transcriptional elongation and histone modifications do not necessarily translate to global regulation of gene expression. HEXIM1 does not bind directly to DNA but is recruited through its interaction with other transcription factors, allowing for its selective regulation.

Moreover, the genomic targeting of KDM5B is mediated by sequence‐specific DNA binding and by binding to posttranslationally modified histones [88].

In addition to its overexpression in breast cancers, KDM5B dysregulation has been reported in several types of solid tumors. It has been reported that, functionally,

KDM5B plays an important role in the proliferative capacity of breast cancer cells

61 through repression of tumor suppressor genes, including BRCA1 [86]. However,

KDM5B was shown to be part of a repressive complex on PR target genes [134].

Moreover, KDM5B has putative tumor suppressive activity, partly, due to its ability to bind and stabilize hypophosphorylated pRb, leading to maintenance of pRb-mediated cell-cycle control [149]. In agreement with this, it has been observed that the expression of KDM5B is lost in the majority of advanced and metastatic melanomas

[89, 150]. Importantly, KDM5B suppressed mammary angiogenesis and metastasis

[90]. It is likely that KDM5B interacting partners, like HEXIM1, influence function.

While we observed overall repression of the recruitment of bicalutamide-liganded

AR and associated factors, HEXIM1 enhanced recruitment of R1881-liganded AR and R1881-induced mRNA expression. However these ligand specific effects of

HEXIM1 did not solely dictate HEXIM1’s effects on cell proliferation. HEXIM1 inhibited R1881-induced proliferation, implicating that the overall effect of HEXIM1 on R1881-induced proliferation can be attributed to regulation of other

R1881-regulated genes. Decreased recruitment of R1881-liganded AR upon downregulation of HEXIM1 (Figure II-4A) was reminiscent of the impact of PTEN loss on AR activity. The upregulation of AR activity by PTEN was attributed to downregulation of EZH2 expression that resulted in increased dependence on androgens [151]. It has also been reported that AR induced transcription of genes that

62 promoted differentiation (eg. CDK1, ref. [152]) and inhibited those that promote metastasis [153].

The role of HEXIM1 as a tumor suppressor in prostate cancer was supported by high nuclear HEXIM1 expression in normal prostate tissue and decreased expression of HEXIM1 in BPH and tumors. Further decreased expression was observed during the progression from well-differentiated tumors to poorly differentiated tumors.

Further validation was provided by datasets in the Oncomine database that indicated decreased expression of HEXIM1 in CRPC relative to hormone-dependent prostate carcinomas. Cancer Outlier Profile analyses (COPA), the algorithm that led to the discovery of TMPRSS2 and ETS family gene fusion events in prostate cancer, was recently modified to allow for the identification of down-regulated outliers [114].

As a result HEXIM1 was identified as a potential tumor suppressor in prostate cancer

[114]. These sets of data are in sharp contrast to a recent report that HEXIM1 expression was absent in normal prostate but highly expressed in adenocarcinoma of the prostate [112]. Moreover, there is incongruence between that group’s expression data and animal data showing heterozygosity for HEXIM1 accelerated tumor progression in the TRAMP model of prostate cancer. While the group reported that heterozygosity for HEXIM1 resulted in increased phosphorylation of AR, an interaction between AR and HEXIM1 and the functional consequences of increased

63 AR phosphorylation at serine 81 was not reported. The phosphorylation of AR at serine 81 by CDK9 has already been previously reported [113] and the upregulation of AR phosphorylation in heterozygote HEXIM1 mice was likely through its inhibition of CDK9, rather than direct effects of HEXIM1 on AR transcriptional activity as we reported herein. Moreover there are conflicting reports on the functional relevance of the phosphorylation at serine 81 in AR transcriptional activity, and that perhaps its role in AR transcription is gene context dependent [74]. Finally this paper did not report on HEXIM1 regulation of anti-androgenic responses.

Acknowledgments

We are grateful to Dr. Ralf Janknecht (University of Oklahoma) for providing the KDM5B expression vector.

64 A. B.

C. D.

E. F.

G.

65 Figure II-1. Expressions levels of HEXIM1 in normal prostate and prostate cancer tissue. (A) Sections obtained from normal prostate tissues, benign prostatic hyperplastic tissues, and prostate tumors were stained for endogenous HEXIM1. The staining score was the product of the intensity of HEXIM1 nuclear staining and percentage of HEXIM1 positive cells. (B) Oncomine analyses of microarray gene expression of HEXIM1 in human benign carcinomas (n=6), prostate carcinomas (n=

7), and hormone refractory metastatic carcinomas (n=6) ref. (17). For (A) and (B) p values were generated using student's t-test (C) Western blot analyses of HEXIM1 expression in LNCaP, LAPC4, C4-2, and C4-2B cells. LNCaP cells were stably transfected with control or two different HEXIM1miR clones (D) and (E) separately or (F) and (G) together and plated onto 94-well plates, treated as indicated for 6 days, then processed for MTT assays to assess proliferation (lower panel). Figures are representations of at least 3 independent experiments. “a” represents p <0.05 relative to R1881 alone, “b” represents p < 0.05 relative to control transfected cells with the same treatment, “*” represents p <0.05 relative to vehicle

66 A.

B.

C.

67 Figure II-2. Physical interaction between HEXIM1 and AR. (A) LNCaP cells were treated with either vehicle, 10 nM R1881, or 10 mM bicalutamide for 90 minutes (B) LNCaP cells stably transfected with control or HEXIM1miR were treated with vehicle, 10 nm R1881, or 10 mM MDV3100 for 90 min. In (A) and (B) lysates were immunoprecipitated using antibodies against HEXIM1 or AR and analyzed for co-immunoprecipitation of HEXIM1 or AR by Western blotting. Normal rabbit IgG was used as a specificity control. Input lanes represent 25% of the total protein. (C) left panel: In vitro translated and [35S]methionine-labeled androgen receptor (AR) was incubated with GST alone, GST-HEXIM1, or GST-HEXIM1150-177 bound to

Sepharose. Bound protein was eluted and analyzed by 12.5% SDS-polyacrylamide gel electrophoresis. “Input” is input lane and represents 10% in vitro translated product added to the samples. Right panel: Western blot analyses of GST,

GST-HEXIM1, or GST-HEXIM1150-177 expression. Figures are representations of at least 3 independent experiments.

68 A. B.

C.

D. E.

69 F. G.

Figure II-3. Recruitment patterns of HEXIM1 and transcriptional elongation factors in LNCaP cells. LNCaP cells stably transfected with control or HEXIM1miR were treated with vehicle, 10 nm R1881, or 10 mM bicalutamide, nulatimide, flutamide or MDV3100 for 90 min. (A) Cells were processed for Western blot analyses of HEXIM1 relative to GAPDH loading control. Also shown are ChIP analyses of lysates immunoprecipitated with antibodies against (B) and (C) HEXIM1,

(D) cyclin T1, (E) serine 2 phosphorylated RNAPII, or control non-specific rabbit immunoglobin, followed by PCR amplification of the proximal ARE containing region or the coding region of the PSA promoter. C4-2 cells transfected with control vector or expression vector for Flag-tagged HEXIM1 (fl-HEXIM1) were treated with vehicle, 10 nm R1881, or 10 mM bicalutamide for 90 min. (F) Cells were processed for Western blot analyses of HEXIM1 relative to GAPDH loading control. (G) Also shown are ChIP analyses of lysates immunoprecipitated with antibodies against

70 HEXIM1. Error bars indicate standard error of the mean of 3 independent experiments. “*” represents p <0.05 relative to vehicle treated cells, “a” represents p <0.05 relative to R1881 alone, “b” represents p < 0.05 relative to control transfected cells with the same treatment.

71 A.

B.

72

C.

73 D.

Figure II-4. HEXIM1 inhibited FOXA1 recruitment and H3K4me2 enrichment.

LNCaP cells stably transfected with control or HEXIM1miR were treated with vehicle, 10 nm R1881, or 10 mM bicalutamide for 90 min. Results show ChIP analyses of lysates immunoprecipitated with antibodies against (A) AR (B) FOXA1 or (C) H3K4me2 or control non-specific rabbit immunoglobin and PCR amplification of the enhancer regions of CDC20, CDK1, or PSA. (D) C4-2 cells transfected with control vector or expression vector for Flag-tagged HEXIM1 (fl-HEXIM1) were treated with vehicle, 10 nm R1881, or 10 mM bicalutamide for 90 min. Shown are

ChIP analyses of lysates immunoprecipitated with antibodies against FOXA1 or control non-specific rabbit immunoglobin and PCR amplification of the enhancer region of CDK1. Error bars indicate standard error of the mean of 3 independent experiments. “a” represents p <0.05 relative to R1881 alone, “b” represents p < 0.05 relative to control transfected cells with the same treatment.

74 A.

75 Figure II-5. HEXIM regulated KDM5B expression. (A) LNCaP cells stably transfected with control or HEXIM1miR were treated with vehicle, 10 nm R1881, or

10 mM bicalutamide for 90 min. Results show ChIP analyses of lysates immunoprecipitated with antibodies against HEXIM1 or control non-specific rabbit immunoglobin and PCR amplification of the enhancer regions of CDK1 or CDC20.

“*” represents p <0.05 relative to vehicle treated cells. (B) ChIP analyses of lysates from LNCaP cells immunoprecipitated with antibodies against HEXIM1 or control non-specific rabbit immunoglobin and PCR amplification of the

-73339/-72922 region of KDM5B. (C) RNA was harvested from LNCaP cells stably transfected with control or HEXIM1miR and subjected to RT-PCR to assess KDM5B mRNA levels using GAPDH as control. (D) Western blot analyses of endogenous

KDM5B relative to GAPDH loading control in LNCaP cells stably transfected with control or HEXIM1miR. Figures are representations of at least 3 independent experiments.

76 A.

B.

C.

77 D.

Figure II-6. HEXIM1 regulated KDM5B recruitment. LNCaP cells stably transfected with control or HEXIM1miR were treated with 10 mM bicalutamide for 90 min. (A) Western blot analyses of endogenous KDM5B or FLAG-KDM5B relative to GAPDH loading control. ChIP analyses of lysates immunoprecipitated with antibodies against (B) H3K4me2 or (C) KDM5B or control non-specific rabbit immunoglobin and PCR amplification of the enhancer regions of CDK1 or GAPDH.

Error bars indicate standard error of the mean of 3 independent experiments. “a” represents p <0.05 relative to non-FLAG-KDM5B transfected cells, “ b ” represents p < 0.05 relative to control miRNA transfected cells. (D) Lysates from

LNCaP cells were immunoprecipitated using antibodies against HEXIM1 or KDM5B and analyzed for co-immunoprecipitation of HEXIM1 or KDM5B by Western blotting. Normal rabbit IgG was used as a specificity control. Input lanes represent

25% of the total protein. Figures are representations of at least 3 independent experiments.

78 A.

79 B.

Figure II-7. Altered HEXIM1 expression resulted in altered cellular response to an anti-androgen (A) LNCaP cells stably transfected with control or HEXIM1miR were treated with vehicle, 1 nM R1881, and/or 10 M bicalutamide for 4 hours. RNA was harvested and subjected to RT-PCR to assess CDK1, CDC20 and UBE2C mRNA levels using GAPDH as control. (B) LNCaP cells were stably transfected with control or HEXIM1miR and processed for Western blot analyses of UBE2C and PSA levels.

Figures are representations of at least 3 independent experiments. “a” represents p

<0.05 relative to R1881 alone, “b” represents p < 0.05 relative to control transfected cells with the same treatment.

80

Figure II-8. Expressions levels of HEXIM1 in normal prostate and prostate cancer tissue. Sections obtained from normal prostate tissues, benign prostatic hyperplastic tissues, and prostate tumors were stained for endogenous HEXIM1.

Figure II-9. HEXIM1 was not recruited to control PSA region. LNCaP stably transfected with control or HEXIM1 miR were treated with vehicle, 10 nm R1881, or

10 M bicalutamide. Shown are ChIP analyses of lysates immunoprecipitated with antibodies against HEXIM1 or control non-specific rabbit immunoglobin, followed by

PCR amplification of the non-ARE containing region of PSA. The image is representative of 3 independent experiments.

81 CHAPTER III

Summary, Discussion and Future Direction

Summary and Discussion

HEXIM1 expression is decreased in BPH and in more advanced tumor stages when compared to normal prostate tissue (Figure II-1A). The result is also confirmed in both Oncomine analysis and in ADPC and CRPC cell lines (Figure II-1B and C).

Downregulation of HEXIM1 (Figure II-1D) in LNCaP cells results in attenuation of the ability of antagonists to inhibit cell proliferation (Figure II-1E, F and G). The role of HEXIM1 in regulating AR activity in ADPC and CRPC was investigated. The interaction between AR and HEXIM1 is located in the NTD 150-177 of HEXIM1

(Figure II-2C) and the interaction between HEXIM1 and AR is ligand-independent.

The pattern of recruitment of HEXIM1 to the PSA promoter region and cyclin T1,

RNA pol II S2P recruitment to the PSA coding region suggested that HEXIM1 inhibits the recruitment of transcriptional elongation factors to AR target genes

(Figure II-3).

Global analysis of AR binding sites in LNCaP and the CRPC variant LNCaP-abl cell lines revealed that AR executes a distinct transcriptional program in LNCaP-abl cell that results in androgen-independent growth [7]. The expressions of CDC20,

CDK1 and UBE2C have been shown to be up-regulated in LNCaP-abl cell line [154].

82 H3K4me1/2-rich enhancer regions on these AR target genes guide FoxA1 binding, resulting in AR recruitment and activation. We hypothesized that HEXIM1 inhibits

AR activity through the inhibition of histone H3K4 methylation and FOXA1 licensing activity.

We generated stably transfected LNCaP cells wherein HEXIM1 expression is downregulated using HEXIM1 miRNA. Our ChIP data showed that recruitment of

AR, FoxA1 and H3K4me2 to CDK1 and CDC20 enhancer regions are significantly increased in LNCaP-HEXIM1 miR cells in both control and bicalutamide-treated cells

(Figure II-4). Similar recruitment patterns were observed in C4-2 and

LNCaP-HEXIM1 miR (Figure II-4) suggested that downregulation of HEXIM1 contributes to the antagonist-resistant phenotype in LNCaP cells. Our results imply that downregulation of HEXIM1 results in decreased sensitivity of AR to agonists.

Our HEXIM1-CHIP-seq data revealed that HEXIM1 was recruited to the regulatory region of KDM5B and downreulgation of HEXIM1 resulted in reduced mRNA and protein expression of KDM5B (Figure II-5B, C and D). Moreover, we observed an interaction between KDM5B and HEXIM1, and KDM5B recruitment to

CDK1 enhancer is reduced in LNCaP-HEXIM1 miR cells (Figure II-6C). While overexpression of KDM5B in LNCaP cells reduced H3K4me2 levels on AR target genes, overexpression of KDM5B in LNCaP-HEXIM1 miR cells did not result in a

83 reduction in H3K4me2 level (Figure II-6B). Together these results suggest that

HEXIM1 assists in the recruitment of KDM5B to certain AR target genes and its ability to modulate H3K4 methylation in the enhancer regions of these genes.

Downregulation of HEXIM1 in LNCaP results in enhanced expression of AR target genes and enhanced proliferation of bicalutamide treated cells (Figure II-7), suggesting that HEXIM1 is an essential factor in the ability of antagonist to inhibit

AR.

Based on the data discussed above, we generated a working model describing the role of HEXIM1 in promoting the progression ADPC to CRPC (Figure III-1). In

ADPC, HEXIM1 modulates AR transcriptional activity by inducing KDM5B expression and recruitment of KDM5B to the regulatory region of AR target genes, thereby demethylating H3K4. In CRPC, HEXIM1 expression is attenuated. H3K4 methylation levels are increased, inducing recruitment of FoxA1 to the regulatory region of AR target genes and executing a transcriptional program distinct from that in existence in ADPC.

In contrast, different from the observation in bicalutamide-treated cells, the recruitment of AR is reduced in LNCaP-HEXIM1 miR cells after treatment with the agonist R1881 (Figure II-4A). These results were reflected in the reduced mRNA expression of AR target genes in R1881-treated LNCaP-HEXIM1 miR cells (Figure

84 II-7). The data revealed that although HEXIM1 attenuated the antagonist-ligaded AR recruitment to the enhancer regions of cell cycle progression factors (CDK1 and

CDC20), somehow HEXIM1 enhances the agonist-ligaded AR recruitment to the same sites. The reason for this phenomenon is still elusive. Resolving distinct coregulators under agonist or antagonist-ligaded AR by Proteomics and Mass spectrometry should be applied for further investigation.

Future Directions

(1) Ascertain whether the HEXIM1 inhibition of H3K4 methylation is P-TEFb

related

HEXIM1 has been known to inhibit the P-TEFb activity by interacting with cyclin

T1 and preventing the CDK9 from phosphorylating RNA polII [108, 110]. This regulation is also supported by the enhanced recruitment of cyclin T1 to PSA coding region in LNCaP-HEXIM1 miR cells. To determine if HEXIM1 inhibition of P-TEFb activity and H3K4 methylation is interrelated, we will downregulate cyclin T1 in

LNCaP-control miR and LNCaP-HEXIM1 miR cells. If HEXIM1 inhibition of

P-TEFb activity and H3K4 methylation is interrelated, the increase in H3K4me2 levels on the enhancer region of CDK1, CDC20 and UBE2C observed in

LNCaP-HEXIM1 miR cells should be attenuated with concurrent downregulation of cyclin T1. The changes in H3K4me2 levels should be reflected in the changes in

85 expression of CDC20, CDK1 and UBE2C.

To further investigate whether P-TEFb complex is related to HEXIM1 inhibition of

H3K4 methylation, we will check cyclin T1 recruitment to the enhancer region of

CDK1 and CDC20. Cyclin T1-ChIP assay will be performed in LNCaP-control miR and LNCaP-HEXIM1 miR cells with and without exogenous expression of KDM5B.

If P-TEFb is related to the HEXIM1 regulation of H3K4 methylation, the results of cyclin T1-ChIP should be correlated with H3K4me2 (Figure II-6B).

(2) Other potential mechanism for HEXIM1 regulation of AR in PC

HEXIM1 suppresses the transcription elongation process by forming an abortive complex with 7SK snRNA and P-TEFb (composed of CDK9 and cyclin T1) [108].

The direct interaction of HEXIM1 and GR (glucocorticoid receptor) [155], ER [119,

120] and AR [156], suppresses these receptor-mediated transcriptional activity in either P-TEFb-dependent or –independent manner. A recent study demonstrated that

GR could substitute for AR in the activation of a transcriptional program that is necessary for maintaining antagonist-resistant prostate cancer [157]. Thus the role of

GR in the regulation of AR transcription by HEXIM1 should be investigated.

We will first examine HEXIM1 recruitment to AR genes in LNCaP cell wherein

GR is exogenously overexpressed. The enrichment and the co-occupancy of HEXIM1,

GR and AR on the enhancer region of AR target genes will be evaluated by re-CHIP

86 and Co-immunoprecipitation experiments. Co-localization of HEXIM1 and AR would be observed in the enhancer regions of AR target genes whereas overexpression of

GR would attenuate the enrichment of HEXIM1 and AR co-localization.

Further investigation of antagonist response in cells will be the next step. After exogenous expression of GR in LNCaP cells will be followed by treatment with vehicle of AR antagonist. Insensitivity to antagonist in this cell model is expected, and the sensitivity should be restored by exogenous expression of HEXIM1. To further confirm the result, resistance to the antagonist should be evident in HEXIM1 attenuated LNCaP cells and the sensitivity to the antagonist should be restored after downregulation of GR expression.

87

Figure III-1. Model of HEXIM1-KDM5B regulation in PC. In ADPC, HEXIM1 modulates AR transcriptional activity by inducing KDM5B expression and recruitment of KDM5B to the regulatory region of AR target genes. As a result, there is a decrease in H3K4 methylation levels. This action prevents AR from executing a

CRPC transcriptional program. In the absence of HEXIM1, H3K4 methylation levels are increased, resulting in FoxA1 recruitment and induction of a CRPC transcriptional program by AR.

88 CHAPTER IV

INTRODUCTION

Average oxygen tension in normal tissue is approximately 7% whereas in

tumors it is approximately 1.5%. Some areas in breast cancer tissue are even

lower than 1%. The brain is the most hypoxia-sensitive organ whereas skeletal

muscle is considered the most hypoxia-tolerant organ [158]. Hypoxia usually

occurs during embryonic development as well as patho-physiological

conditions, such as ischemic diseases (stroke), atherosclerosis and cancers.

Overall, maintaining oxygen homeostasis is critical for cell survival [159].

Major responses to hypoxia can be categorized into acute and chronic

responses (Figure IV-1). Ion channels, mitochondria and hemeprotein

responses are regarded as acute responses whereas chronic response involves

the modulation of transcription under hypoxia [159]. The major modulator of

cellular responses to hypoxia is Hypoxia-inducible factor (HIF).

REVIEW OF LITERATURE

A closer look of hypoxic environment---Acute hypoxic response a) Ion channels

It has been reported that ion channels are sensitive to the hypoxia [160].

+ 2+ + O2-sensitive ion channels include K -selective, Ca and Na channels [161].

89 O2-sensitive ion channels are mostly found in excitable neurosecretory cells where

they mediate the fast cardio respiratory adjustment in response to the hypoxia. O2 sensors (e.g. NADPH, O2-dependent hydroxylases, HIF, etc…) have been reported to be closely associated with channel oligomers. The switch between the oxy- and deoxy- conformation of these sensors results in the allosteric regulation of the

channels [161]. However, detailed mechanisms of how O2 sensor interacts with ion channels are not well-defined. b) Mitochondria

Mitochondria are hypoxia-sensing sites because they are the primary sites of O2 consumption [162, 163]. It is postulated that in hypoxia, metabolic activity, ATP hydrolysis and mitochondria respiration will be reduced in the mitochondria due to the Warburg effect [164]. HIF1α is stabilized in the mitochondria through the inhibition of prolyl hydroxylases (PHD) [164]. c) Hemeproteins

Hemeprotein is a protein containing a heme group with a reduced iron atom that

can bind O2. Hypoxia causes an allosteric conformational change in the hemeprotein

[159]. Hypoxia shifts the inactive (oxy-) form of hemeprotein to an active form

(deoxy-). In addition to O2, CO and NO can also bind to the iron atom in the heme protein. Once bound to the heme group, these molecules can modulate several

90 biological functions and have anti-inflammatory and anti-apoptotic effects [159, 165].

Chronic hypoxic responses—Transcription factors (HIFs)

Adaptation to hypoxic environment is mediated by a key transcription factor called

Hypoxia-inducible factor, HIF. Physiological responses regulated by HIF include hematopoiesis, angiogenesis, glucose utilization, iron transport, cell proliferation, resistance to oxidative stress, survival, apoptosis, extracellular matrix homeostasis and tumor progression [166]. HIF family comprises two groups, α and β. Each group has three members: 1(αβ), 2(αβ) and 3(αβ) (Figure IV-2). Three alpha subunits (1α,

2αand 3α) are all hypoxia inducible [167]. The function of HIF3α is not well-defined.

Some studies indicate that HIF3α acts to inhibit HIF1 function [168]. In contrast to

HIF1α which is ubiquitously expressed, HIF2α and HIF3α have distinct expression patterns. Distinct expression patterns suggest that their functions are not redundant [168]. Studies indicated that N-TAD (N-terminal Transactivation Domain) confers target gene specificities to HIF1α and HIF2α, whereas the C-TAD

(C-terminal Transactivation Domain) promotes the expression of the common target genes of HIF1α and HIF2α (Figure IV-2) [159, 169, 170].

In hypoxic conditions, HIF1α heterodimerizes with HIF1β (=aryl-hydrocarbon nuclear translocator=ARNT) and activates downstream target genes by binding to the hypoxia response element (HRE) [159, 169]. Even in normoxic conditions, basal

91 levels of HIF1α have been found to maintain oxygen homeostasis in mammalian tissues [171]. Several processes, such as angiogenesis, glucose metabolism, cell proliferation and even apoptosis, are at least in part, regulated by HIF family [172]. To date, HIF1α has been investigated more than the other two alpha subunits. The regulation of HIF1α stability will be addressed in the following sections.

Classical Pathway of regulation of HIF1α protein

Under normoxic conditions (21% oxygen), the half-life of HIF1α is less than 5 minutes. Two key proline residues, Pro402 and Pro564, can be hydroxylated by prolyl hydroxylase (PHDs) and recognized by von Hippel-Lindau (pVHL). pVHL is part of an E3 ubiquitin ligase (Elongins/BC/Cul2/VHL) and the binding of pVHL to HIF1α results in the transport of HIF1α to the proteasome to be degraded [159, 169-172]. pVHL includes a full-length protein (amino acid 1-213) and N-terminally truncated protein (amino acids 54-213). Both full-length and truncated proteins have similar function and are collectively called pVHL [173-175].

There are three isoforms of PHDs, PHD1-3. Hydroxylation reactions catalyzed by

PHDs require ascorbate as a cofactor and -ketoglutarate as a substrate. The catalytic site of PHD contains a Fe2+ ion, which can be chelated by Co2+ [176]. Expression of

PHD2 and PHD3 are hypoxia inducible suggesting that they are HIF target genes

[177]. In gliomas, PHD2 and PHD3 have been reported to be negative regulators of

92 HIFs that prevent HIF-induced apoptosis in hypoxic conditions [178]. PHD2 is the primary enzyme regulating HIF1α stability in normoxic conditions, whereas PHD1 and PHD3 are critical enzymes contributing to the hydroxylation of HIF1α in hypoxic conditions [179, 180]. The contribution of PHDs to HIF1α stability is also related to the relative abundance of PHDs [181]. In normoxic conditions, P564 is hydroxylated prior to P402 and is required for the hydroxylation of P402. P402 hydroxylation has been reported to be more sensitive to the changes of physiologic oxygen level than

P564 [181]. These results suggest that each proline hydroxylation has a distinct role in mediating HIF1α stability in response to oxygen level. Several post-translational sites in HIF1α are shown in Figure IV-3.

PHDs are not responsible for all the hydroxylation in HIF1α. Another hydroxylation site is at C-terminal domain, Asn803. N803 is hydroxylated by FIH-1

(factor inhibiting HIF1), which is also α-KG (α-Ketoglutarate), Fe2+-dependent enzyme [182]. This hydroxylation site prevents the recruitment of coactivators, such as p300/CBP, to the C-terminal transactivation site and reduces the transcriptional activity of HIF1α [183].

pVHL-dependent pathway is not the only pathway responsible for the degradation of HIF1α. In hypoxic conditions, HIF1α interacts with p53, which recruits MDM2,

(an E3 ubiquitin ligase) and results in the degradation of HIF1α [184].

93 Serendipitously, a protein named Jab1 has been found to compete with p53 for binding to HIF1α to stabilize HIF1α in hypoxic conditions [185]. Furthermore, HAF

(HIF-associated factor), a HIF1α-specific E3 ubiquitin ligase, can destabilize HIF1α in a pVHL-independent and proteasome-dependent manner [186].

Non-classical Pathway of regulation of HIF1α

HIF1α is regulated not only by sensing oxygen but also by other stimuli such as transition metals, nitric oxide (NO), reactive oxygen species, growth factors and physiological stress [176]. a. Transition Metals

In normoxic conditions, HIF1α can be stabilized by transition metals such as cobalt and nickel. Transition metals replace the putative oxygen sensor protein (such as heme protein), locking HIF1α in its deoxygenated state. Cobalt or nickel bound- protoporphyrin IX, an enzyme that catalyzes the incorporation of iron into heme, has lower affinity than iron-bound protoporphyrin IX to oxygen [187]. Thus, cobalt or nickel bound-protoporphyrin IX mimics low oxygen environment, which results in

HIF1α activation [176]. Other reports have demonstrated that substituting the ferrous ion with cobalt or nickel inactivates PHD activity thereby stabilizing HIF1α in normoxic conditions [188]. b. Nitric Oxide (NO)

94 NOS (NO synthase), an enzyme responsible for synthesis of NO, is a HIF1 target gene. Activation of HIF1α increases NO production. NO has been reported to attenuate the enzymatic function of PHDs in normoxic conditions. However, NO restricts mitochondrial respiration by competing with oxygen, thereby inhibiting the mitochondrial enzyme cytochrome c oxidase (complex IV). This restriction increases the availability of non-respiratory oxygen that should reactivate PHDs and result in the destabilization of HIF1α [176, 189, 190]. Thus, it is still not clear how NO regulates HIF1α stability. c. Reactive Oxygen Species (ROS)

ROS contributes to metastasis, angiogenesis and radiation resistance of tumor cells.

Previous reports indicate that PHD activity can be inhibited by ROS, thereby enhancing HIF1α stability [191]. However, it has been demonstrated that treating the

cells with strong oxidizing reagents, such as H2O2 and diamide in hypoxic conditions results in attenuated expression of HIF1α and HIF1 DNA binding activity [176]. d. Growth Factors (GFs) and Mechanical stress

Growth factors activate signal pathways that contribute to HIF1α expression in normoxic conditions in many cancer cells. Insulin, IGF, Epidermal Growth Factor

(EGF), and IL-1 are all reported to activate the PI3K/AKT/FRAP (phosphoinositide

3-OH kinase/protein kinase B /FKBP-rapamycin-associated protein) pathway, which

95 increases HIF1α protein expression by enhancing HIF1α mRNA synthesis [169, 172,

176, 192]. For example, IGF-1 induces the phosphorylation of transcriptional regulators, such as 4E-BP1, p70 S6 kinase and eIF-4E, which results in enhanced

HIF1α expression [193]. HER2, a tyrosine kinase receptor, is overexpressed in breast cancer cells and stimulates the PI3K/AKT/FRAP pathway, leading to non-hypoxic expression of HIF1α [194]. Interestingly, mechanical stress also induces HIF1α expression in normoxic conditions. Increased blood pressure causes the wall of aorta to stretch which activates PI3K/AKT pathway and results in increased HIF1α accumulation [195]. However, the basis for how cells sense stress and activate the

PI3K/AKT pathway is not well-defined. e. microRNAs (miRs)

miR-210, a direct HIF1 target gene, has been reported to stabilize HIF1α and increase HIF1 transcriptional activity. This positive feedback regulation is mediated by GPD1L (glycerol-3-phosphate dehydrogenase 1-lik), which is down-regulated by miR-210 [196]. GPD1L increases PHD activity through an unknown mechanism, thereby resulting in degradation of HIF1α in hypoxic conditions [196]. f. Rack1 (Receptor for activated C-kinase 1) and HSP90 (pVHL-independent)

Heat Shock Protein 90 (HSP90) is a chaperone protein that stabilizes HIF1α by binding to the PAS domain in HIF1α [197]. Inactivation of HSP90 (using 17 AAG,

96 17-N-allylamino-17-demethoxygeldanamycin) leads to improper folding of HIF1α

[198]. However, HSP70 has been reported to interact with HIF1α and promote HIF1α degradation through the proteasome pathway [199, 200].

RACK1 (receptor of activated protein C kinase) was identified as a PKC (Protein

Kinase C) anchoring protein that anchors PKC to its targets such as -integrin [201].

Rack1 has been reported to destabilize HIF1α in an O2/PHD/pVHL-independent manner by competing for the binding of HSP90 to HIF1α [202]. Another

O2/ubiquitin/pVHL-independent regulation of HIF1α is through SHARP1, a down-stream target of Tap63. Tap63 is a tumor suppressor protein [203]. SHARP1 enhances HIF1α degradation by promoting the interaction between HIF1α and proteasome [204]. Depletion of SHARP1 is frequently found in triple-negative breast cancer [205].

Post-translational modifications of HIF1α a. Acetylation (proteasome-dependent manner)

Mouse ARD1 (arrest-defective protein 1) functions as a protein acetyl transferase that catalyzes HIF1α acetylation at K532 and destabilizes HIF1α in normoxic conditions. In hypoxic conditions, K532 is de-acetylated by HDAC1

(Histone-deacetylase 1), which is recruited by MTA1 (Metastasis-associated protein 1) and thereby stabilizing HIF1α. Whether the regulation of ARD1 is

97 hydroxylation-related or pVHL-dependent is unknown [206]. However, other researchers reported that the interaction between human ARD1 and HIF1α neither results in acetylation of HIF1α nor induces HIF1α degradation [207]; therefore the regulation of HIF1α stability by acetylation is still controversial [208]. Another acetyl transferase, PCAF (p300/CBP-associated factor), acetylates HIF1α at K674. This acetylation activates HIF1α transcription activity by promoting the recruitment of p300 [209]. However, K674 can be de-acetylated by SIRT1 deacetylase, which prevents the recruitment of p300 and represses HIF1 target genes [209, 210]. b. Sumoylation (proteasome-dependent manner)

Sumoylation of HIF1α is another mode of regulation of HIF1α stability in hypoxic conditions. Reports suggest that HIF1α can be sumoylated in hypoxic conditions

[211]. Sumoylation of HIF1α results in HIF1α degradation in a pVHL/proteasome-dependent and hydroxylation-independent manner [212].

De-sumoylation by SENP1, which is up-regulated in tumor cells, results in stabilization of HIF1α [213]. c. Ubiquitination (Ub)

Besides the classical model of pVHL-mediated Ub, HIF1α can also be ubiquitinated by HDM2 in the presence of p53. HDM2 is an E3 Ub-ligase that mediates Ub-dependent degradation of p53 [214]. Reports indicate that HIF1α can be

98 recruited to HDM2 by interacting with p53 and this interaction resulted in HIF1α ubiquitination by HDM2 and degradation by the proteasome [184]. In the absence of p53, however, HDM2 positively regulates HIF1α expression by activating the

PI3K/Akt pathway [192]. d. Phosphorylation and S-nitrosation

The p42/44 Mitogen-Activated protein kinase (MAPK) family has been shown to phosphorylate both HIF1α and HIF2α [215] . Phosphorylation of HIF1α increases the transcriptional activity of HIF1 [170, 215]. However, specific phosphorylation sites have not been identified.

S-nitrosation is the process of transferring a NO moiety to a protein. S-nitrosation of cystine 800 in HIF1α has been shown to increase the interaction between HIF1α and p300 thereby increasing HIF1α transcriptional activity [216].

Responses to chronic hypoxia mediated by HIF

Several downstream hypoxic responses are modulated by HIFs, and include erythropoiesis, differentiation, proliferation, energy metabolism and angiogenesis. a) Erythropoiesis

The expression of genes involved in erythropoiesis can be induced under chronic hypoxic condition. For example, the erythropoietin (EPO) gene is induced in hypoxia and is required for the formation of red blood cells (RBCs). Increasing the abundance

99 of RBCs enhances the transport of O2 [217]. b) Differentiation

A hypoxic environment has been reported to maintain stem cells in an undifferentiated state [218]. Cooperation between Notch signaling and HIF1α maintains the undifferentiated cell state [201]. Notch signaling pathway is triggered by cell-cell contact. Binding of the ligands from the adjacent cells to Notch receptor releases the intracellular domain (ICD). Cofactors, such as p300/CBP, are recruited to form a complex with ICD that, in turn, activates Notch target genes, including those that promote cell proliferation. HIF1α has been reported to interact with and stabilize

ICD [219, 220].

Oct-4 and MYC have been shown to be directly regulated by HIF2α [221, 222].

Activation of Oct-4 is essential for HIF2α induced differentiation of embryonic stem cells [221]. Enhanced MYC transcription, through HIF2α promotes hypoxic cell proliferation in hypoxic condition [222]. c) Proliferation

Growth factors (GF), such as insulin-like growth factor 2 (IGF-2) and transforming growth factor  (TGF-α), are HIF1 target genes [223]. Binding of these GFs to their cognate receptors IGFR and EGFR, respectively, activate downstream pathways such as PI3K/Akt and MAPK (mitogen-activated protein kinase) pathways, which increase

100 HIF1α expression and promote cell proliferation [223]. d) Energy metabolism

A hypoxic environment induces a switch to aerobic glycolysis [224]. This is the so-called Warburg Effect (Figure IV-4). Increased utilization of glucose during hypoxia is regulated by HIF-1. Several genes that are regulated by HIF-1 are necessary for glucose metabolism and includes GLUT-1 (Glucose Transporter-1) which increases glucose uptake, LDHA (Lactate Dehydrogenase A ) which increases lactate production and PDK1 (Phosphoinositide-dependent kinase-1) which decreases

O2 consumption [225]. e) Angiogenesis

Angiogenesis is a multiple-step process by which new blood vessels develop from existing vessels. An essential factor for angiogenesis is VEGF, which promotes proliferation and migration of vascular endothelial cells, resulting in enhanced formation of new blood vessels [226]. VEGF expression is induced by HIF family members; however, their relative contributions to VEGF expression is cell-type specific [227]. Other angiogenesis promoting factors that are also HIF1 target genes, include endothelin 1, TGF-β3, and angiopoietins [228, 229]. f) Cell death

Hypoxia not only promotes cell proliferation but also apoptosis [230-232]. Bcl-2

101 family members are known to regulate apoptosis. Bcl-2 family members can be categorized by the number of Bcl-2 homolog (BH) domains. Pro-apoptotic members include Bax, Bak and BH3-only whereas anti-apoptotic proteins include Bcl-2 and

Bcl-XL [233]. HIF1 has been shown to induce the expression of pro-apoptotic members [230, 234]. However, chronic hypoxic conditions have been shown to be insufficient to contribute to cell death [159] and the presence of other factors, such as generation of ROS and a reduction in ATP level, are necessary [159, 235, 236].

Investigators are still debating the role of HIF1 in cell death. A proposed role is that in prolonged anaerobic condition, if cell survival cannot be restored by HIF1-induced protective mechanisms, then HIF will activate the cell-death pathway [230, 237].

HIF1 and tumor metastasis

Tumor metastasis is the main cause of death in cancer patients. Efficient inhibition of tumor metastasis allows for increased survival of cancer patients. Angiogenesis, the formation of blood vessels from pre-existing vessels, allows tumor cells to rapidly proliferate. Due to the rapid proliferation, the environment of the tumor is hypoxic. In this regard, previous reports indicate that HIF1 is not only a potent activator of angiogenesis, but also a critical factor in invasion and metabolic reprogramming [238,

239].

Metastasis involves the loss of cellular adhesion, increased cellular motility and

102 invasion, entrance of cells to the bloodstream by the circulation, exit from the bloodstream and incorporation into a distant target tissue. These processes require the cancer cells to cooperate with multiple factors to overcome barriers that are natural defenses of our body [240, 241]. As suggested by Stephen Paget in 1889, the interaction between tumor cells and the host environment determines the final outcome [242, 243]. For example, TGFβ is released from the bone matrix and promotes the establishment of a suitable environment for cancer cells to survive. This is referred to as the “vicious cycle” in bone metastasis [244].

The most common organs that breast cancer cells colonize are bone, lung, lymph node and brain [245]. However, how primary breast tumor cells take the initial step toward metastasis is not well-defined. It has been proposed that within a population of tumor cells, a subpopulation exists with high metastatic potential that can be selected by specific markers, such as CD44+CD24- [246]. This also suggests that metastasis might be an intrinsic feature of primary tumor cells.

Tumor Angiogenesis

Angiogenesis, the process of new blood vessel formation from pre-existing vessels, increases the availability of nutrients and oxygen to tumor cells. Angiogenic factors stimulate blood vessels’ growth. However, due to the rapid growth rate, the newly developed blood vessels are relatively fragile and irregularly shaped, resulting in

103 insufficiency of nutrient and oxygen [247, 248]. Thus, the microenvironment around tumor cells is hypoxic.

Multiple pathways are crucial in tumor angiogenesis with the HIF family being a major regulator. Many signaling pathways, such as PLCγ (Phosphoinositide phospholipase C) and PI3K signaling, are activated by HIF1 downstream target genes under hypoxic condition and result in increased vascular permeability, endothelial cell proliferation, and tumor cell migration [248-250].

HIF1 target genes involved in tumor survival

HIF1 target genes induce multiple signal pathways and promote cancer cell survival. So far, more than hundreds of HIF1 target genes have been identified. The relevance of certain HIF1 targets in cancer is discussed below. a. SDF1 and CXCR4.

SDF1 (stromal cell-derived factor 1), also named CXCL12, is a highly conserved chemokine and is the sole ligand for CXCR4, (albeit there is one report that indicated that SDF1 binds to CXCR7 as well [251]).The SDF1/CXCR4 complex promotes angiogenesis, metastasis and stem cell mobilization [252]. SDF1/CXCR4 pathway is a critical regulator of breast cancer cell metastasis [253, 254]. A clinically relevant antagonist of CXCR4 is AMD3100, which has been used to treat AML (acute myeloid leukemia) patients. The anti-tumor efficacy of AMD3100 is under investigation [255].

104 b. VEGF.

Over-expression of VEGF contributes to tumor angiogenesis, metastasis, and has been correlated with poor survival rates. All members of the VEGF family bind to tyrosine kinase receptors [228]. Clinically-relevant anti-VEGF drugs include monoclonal antibodies or small peptides that inhibit the tyrosine kinase receptor [227].

However, combinations of other treatments are necessary due to the high recurrence rate in tumors [256]. The high recurrence rate of anti-VEGF drugs in treating tumors can be partly attributed to compensatory angiogenic pathways [257]. Up-regulation of the compensatory pathways impairs the efficacy of anti-VEGF drugs, thus, HIF1α is a better drug target than VEGF [223, 258]. c. HSulf1.

Hsulf1 belongs to the Sulfatase family. Heparan sulftase is an extracellular endoglucosamine-6-sulfatase that removes the sulfate at 6-O position of glucosamine and activates several signaling pathways in the ECM (extracellular matrix), such as the Wnt pathway [259]. Heparin glucosamine is also a docking site for many proteins, such as SDF1 and bFGF2 [260]. Previous report shows that under hypoxic condition,

HSulf1 is down-regulated. Knockdown of HIF1α rescues the down-regulation of

Hsulf1, indicating that HIF1α inhibits HSulf1 expression. The reduction in Hsulf1 expression somehow results in the activation of bFGF2 and enhancement of cell

105 migration and invasion [261]. The role of HSulf1 under hypoxic conditions obviously needs to be further defined. d. Prolyl hydroxylases (PHDs)

Prolyl hydroxylase belongs to 2-oxoglutarate (2-OG)-dependent dioxygenase family and is also called proline dioxygenase. PHD catalyzes the hydroxylation of prolyl residues to hydroxyl prolyl residues during collagen synthesis (Figure IV-5).

There are three members in this family: PHD1, PHD2 and PHD3. They use oxygen and 2-OG as co-substrates and require iron and ascorbate as cofactors. Hypoxia induces the expression of PHD2 and PHD3 suggesting negative feedback regulation of HIF1α [178]. The induction of PHD3 expression is slightly stronger than that of

PHD2 expression in several cancer cell lines [180]. Studies indicate that PHD2 is involved in oxygen regulation of HIF1α in normoxia [179] whereas PHD3 has higher affinity for oxygen than PHD2 in hypoxic conditions [180]. Both PHD2 and PHD3 retain hydroxylase activity at low oxygen concentration [177, 258]. Previous reports suggest that negative feedback regulation of HIF1α by PHD2 and PHD3 establishes a tissue-specific threshold for adjusting HIF1α activation rather than simply accelerating HIF1α degradation after re-oxygenation [177, 178].

It has been reported that PHD1 promotes breast tumorigenesis by controlling cyclin

D1 [262]. Clinical analysis of breast cancer patients indicate that the expression of

106 PHD1 is associated with poorer prognosis and is frequently found in ER-negative tumors. On the other hand, high expression of PHD2 in tumors is correlated with better survival rates [263]. Reduced expression of PHD3 has been observed in larger tumors as well as in poorly differentiated and highly proliferative tumors. Conversely,

PHD3 expression is associated with good prognosis [263]. Other studies suggest that

PHD3 induces apoptosis through a HIF-independent pathway [258, 264]. However,

PHD3 does not always induce apoptosis. It has been reported that PHD3 expression is induced by hypoxia and inflammatory stimuli in human neutrophils. HIF1-induced apoptosis in neutrophils is considered a key mechanism for maintaining appropriate inflammatory response [265, 266]. By increasing the degradation of HIF1α PHD3 promotes the survival of neutrophils in response to hypoxia. e. Loxl2

Lysyl oxidase-like 2 (LOXL2) is a member of the LOX family that is comprised of extracellular matrix (ECM)-modifying enzymes and are essential in promoting cancer metastasis [267]. LOX2 catalyzes the cross-linking between collagens and elastin.

High expression of LOX2 has been associated with poor prognosis in cancer patients

[268]. Moreover, LOXL2 is highly expressed in basal-like breast cancer type. LOXL2 enhances metastatic dissemination by increasing the expression of metallopeptidase inhibitor 1 (TIMP1) and matrix metallopeptidase 9 (MMP9), which collectively

107 increase the degradation and remodeling of ECM in breast cancer cells [268].

Tumor and energy metabolism

Pyruvate kinase catalyzes the last step in glycolysis, converting the substrate PEP

(phosphoenolpyruvate) into pyruvate [269]. There are 4 isozymes of pyruvate kinase in mammals (L, R, M1, and M2). L type is the major isozyme in the liver; R is found in red cells; M1 is the major form in the muscle, heart and brain; and M2 is found in early fetal tissues as well as in most tumor cells [269, 270]. Tumor cells show enhanced glycolytic flux [271]. Enhanced glycolysis is induced by the HIF1 complex through up-regulation of several glycolytic enzymes [272, 273]. Pyruvate kinase M II

(PKM2) is a HIF1 target gene and is frequently over-expressed in tumor cells [274].

PKM2 catalyzes the serial steps in the generation of pyruvate from glucose. Pyruvate then undergoes the aerobic glycolysis generates lactate (Figure IV-4) [274, 275]. The generation of energy by PKM2 is independent of oxygen supply that allows the tumor cells to survive in the hypoxic condition. Positive feedback regulation occurs through prolyl hydroxylase 3, which hydroxylates PKM2 and enhances the interaction between HIF1α and PKM2 [276]. This interaction results in increased HIF1 transcriptional activity. Because of the low oxygen concentration in solid tumors, low oxygen consumption rate of aerobic glycolysis is an alternative way for tumor cells to generate ATP.

108 It has been reported that HIF1 inactivates tricarboxylic acid cycle enzymes by actively transcribing pyruvate dehydrogenase kinase 1 [277]. Activation of pyruvate dehydrogenase kinase 1 also results in the attenuation of the ability of pyruvate dehydrogenase to convert the pyruvate into acetyl-CoA. This hypoxia induced metabolic switch provides the ATP production and prevents the hypoxia-induced apoptosis in tumor cells.

HIF1 has been demonstrated to repress mitochondrial respiration in VHL-deficient renal cell carcinoma by repression of c-myc activity [278]. However, the impairment of mitochondrial respiration may be cell-type specific [279, 280].

Hexamethylene-inducible gene 1 (HEXIM1) and breast cancer (BC)

We initially identified HEXIM1 as a regulator of ER in breast cancer cells [120].

Analysis of clinical samples by microarray revealed that HEXIM1 expression is frequently downregulated in various types of breast cancer samples, such as luminal A, luminal B and basal-type [119]. The Montano laboratory has shown that mammary

tumors that develop in mice expressing C-terminus truncated HEXIM11-312 exhibit increased vascularization, and increased HIF1α and VEGF protein levels [120].

Down-regulation of HEXIM1 in MCF-7 cells results in enhanced HIF1α protein expression while HIF1α mRNA level is not altered [107]. Conversely, overexpression of HEXIM1 in mammary epithelial cells inhibited metastasis in a mouse model of

109 metastatic breast cancer that can be correlated with decreased vascularization, and decreased HIF1α and VEGF protein levels [126].

STATEMENT OF PURPOSE

Clinically, twenty five percent of breast cancer cases are Estrogen Receptor

(ER)-negative. ER-negative breast cancers are more aggressive and the survival rate of ER-negative patients is significantly lower than ER-positive patients [281]. Tumor metastasis is the main cause of death in ER-negative cancer patients. The inhibition of tumor metastasis allows for an increased survival rate of ER-negative patients.

The purpose of our study is to understand the mechanistic basis for the regulation of HIF1α by HEXIM1. In particular, we determined the effect of HEXIM1 on the interactions between HIF1α and factors that are critical for HIF1α post-translational modifications and HIF1α stability. We also determined the functional consequences of HEXIM1 regulation on HIF1α.

HIF is the major regulator in cancer cells, and is a response to a hypoxic environment. A hypoxic environment has been identified as the main cause of therapeutic resistance, which is an inevitable problem in cancer therapy. The role

HIF plays in tumor metastasis has been reported. A better understanding of the mechanistic basis of a hypoxic environment in breast cancer provides a new insight into exploring a novel target which may improve patients’ survival. Our

110 study shows that HEXIM1 has potential as a therapeutic target in treating breast cancer patients by regulating the stability of HIF1α.

111

Figure IV-1. Cellular processes regulated by hypoxia. Maintaining oxygen homeostasis is essential for cell survival. Acute response is related to ion channel regulation. Chronic effect is through transcription regulation by HIFs. If the cell survival cannot be guaranteed, cell death response will be induced.

Journal of Paramedical Sciences (JPS); spring 2010 Vol.1, NO.2 ISSN 2008-496X

112

Figure IV-2. Functional domains of HIF family proteins. bHLH and PAS

(PER-ARNT-SIM) domains are required for dimerization and DNA binding. HIFα family exhibits a high degree of similarity in DNA and ARNT binding domain and

C-TADs. N-TAD confers target gene specificities of HIF1α and HIF2α, whereas the

C-TAD promotes the expression of HIF1α/ HIF2α common target genes.

113

Figure IV-3. Post-translational modifications of HIF1α. In normoxia, several modifications have been found to regulate HIF1α stability, such as hydroxylation and acetylation. Hydroxylation is catalyzed by PHDs that have high Km for oxygen

(100-250uM). pVHL recognizes those modification sites and results in the transport of HIF1α to the proteasome to be degraded. In hypoxia, the hydroxylation of HIF1α is inhibited. HIF1α translocates into nucleus, binds to HIF1 and regulates transcription.

Arthritis Research & Therapy

114

Figure IV-4. Glycolysis is catalyzed by PKM I in normal cell and PKM II in tumor cell respectively. PKM I is a constitutively active enzyme whereas PKM II activity relies on allosteric activation by the upstream metabolite fructose-1.6-bisphosphate. PKMII may lead to reduce the mitochondria density and the expression of proteins involved in oxidative phosphorylation.

115

Figure IV-5. Hydroxylation catalyzed by prolyl hydroxylases 2 (PHD2) in normoxia. PHD2 incorporates one oxygen atom into hydroxylated-proline and the other oxygen into 2-oxoglutarate which converts 2-oxoglutarate into succinate.

http://en.wikipedia.org/wiki/EGLN1

116 CHAPTER V

HEXIM1 down-regulates hypoxia-inducible factor-1α protein stability

This work ha been published in the Biochemical Journal

(Yeh, I-Ju et al., Biochemical Journal 2013 456, 195–204)

Abstract

We have previously reported on the inhibition of HIF1α (hypoxia-inducible factor

α)-regulated pathways by HEXIM1. Disruption of HEXIM1 activity in a knock-in mouse model expressing a mutant HEXIM1 protein resulted in increased susceptibility to the development of mammary tumors, partly by up-regulation of

VEGF (vascular endothelial growth factor) expression, HIF1α expression and aberrant vascularization. We now report on the mechanistic basis for HEXIM1 regulation of HIF1α. We observed direct interaction between HIF1α and HEXIM1, and HEXIM1 up-regulated hydroxylation of HIF1α resulting in the induction of the interaction of HIF1α with pVHL(von Hippel–Lindau protein) and ubiquitination of

HIF1α. The up-regulation of hydroxylation involves HEXIM1-mediated induction of

PHD3 (prolyl hydroxylase 3) expression and interaction of PHD3 with HIF1α.

Acetylation of HIF1α has been proposed to result in increased interaction of HIF1α with pVHL and induced pVHL-mediated ubiquitination, which leads to the proteasomal degradation of HIF1α. HEXIM1 also attenuated the interaction of HIF1α

117 with HDAC1 (histone deacetylase 1), resulting in acetylation of HIF1α. The consequence of HEXIM1 down-regulation of HIF1α protein expression is attenuated expression of HIF1α target genes in addition to VEGF and inhibition of

HIF1α-regulated cell invasion.

Introduction

Breast cancers and other solid tumors are susceptible to hypoxia because they proliferate and outgrow vascular supplies of oxygen and nutrients [273]. The main regulator that orchestrates the cellular response to hypoxia is HIF-1

(hypoxia-inducible factor 1), a heterodimeric transcription factor composed of α- and

-subunits critical for adaptive responses to reduced oxygen [282]. Overexpression of

HIF1α protein in breast cancer correlates with poor prognosis, increased risk of metastasis and decreased survival [283]. The critical role of HIF1α in tumor metastasis arises from the fact that it is a potent activator of angiogenesis, invasion and metabolic reprogramming through its up-regulation of target genes important for these functions {e.g.VEGF (vascular endothelial growth factor), ECM (extracellular matrix)-degrading proteases and GLUT1 (glucose transporter 1), see [284]}. HIF1α is also a mediator of the effects of the tumor microenvironment on the metastatic behavior. HIF1α plays a critical role in the generation of the ‘pre-metastatic niche’ to which the tumor cells metastasize through the recruitment of BMDCs (bone marrow-

118 derived cells). It has also been proposed that hypoxia stimulates expansion of normal and cancer stem cells [285]. The HIF1α pathway is thus an ideal target for cancer therapy, since interfering with a master regulator of the hypoxic response could disrupt multiple processes essential for tumor cell self-renewal, expansion, dissemination and metastatic colonization. The clinical benefits of anti-VEGF therapy are relatively modest and usually measured in weeks or months [286]. In some cases, patients do not respond to anti-VEGF treatments. Hypoxic regions of tumors are believed to be the source of tumor cells that are resistant to radiation, chemotherapy and anti-angiogenic treatment [283]. Anti-angiogenic agents efficiently prune tumor vessels and cause hypoxia [287]. However, metabolic reprogramming to glucose addiction allows tumor cells to generate energy in hypoxic conditions and for tumor stem cells in hypoxic niches to escape anti-angiogenic treatment [287]. Increased intratumor hypoxia also results in the production of redundant angiogenic factors by tumors and acquisition of a more invasive phenotype. HIF1α is a major regulator of these angiogenic actors following hypoxia, and regulates several genes involved in angiogenesis, proliferation and migration of endothelial cells, pericyte recruitment, modification of vascular permeability and recruitment of BMDCs [257]. Thus targeting HIF1α rather than VEGF may offer advantages in late-stage breast cancer.

Regulation of HIF1α stability is mediated by the ODD (oxygen-dependent

119 degradation) domain through various posttranslational modifications [288]. HIF1α is hydroxylated at Pro402 and Pro564 by a family of HIF PHD (prolyl hydroxylase) domain proteins, which require oxygen [289, 290]. Hydroxylated HIF1α subsequently interacts with the tumor suppressor pVHL (von Hippel–Lindau protein), which targets it for proteasomal degradation [223, 288]. ARD1 (arrest-defective protein 1) is another enzyme proposed to modify HIF1α by acetylating the Lys532 residue in the

ODD domain of HIF1α [291]. Acetylation of HIF1α has been reported to result in increased interaction of HIF1α with pVHL and induced pVHL-mediated ubiquitination, which leads to the proteasomal degradation of HIF1α [291]. However, the functional relevance of HIF1α acetylation remains controversial. We have recently reported that re-expression of HEXIM1 [HMBA

(hexamethylene-bis-acetamide)-inducible protein 1] through transgene expression or polymer-mediated delivery of HMBA inhibited metastasis in a mouse model of metastatic mammary cancer that can be correlated with decreased expression of

HIF1α VEGF, compensatory pro-angiogenic factors and vascularization [126]. We now report that HEXIM1 directly regulates HIF1α protein stability by up-regulating hydroxylation, interaction with pVHL and ubiquitination of HIF1α HEXIM1 also regulated HIF1α acetylation by attenuating its interaction with HDAC1 (histone deacetylase 1). As a result, HEXIM1 is able to regulate expression of HIF1α target

120 genes and HIF1α-induced cell invasion.

Material and Methods

Reagents

Antibodies against the following were obtained from Santa Cruz Biotechnology:

CXCR4 (CXC chemokine receptor 4), HDAC1, pan-acetyl, ubiquitin and SDF-1

(stromal-cell-derived factor 1). The anti-HEXIM1 antibody was generated in the

Montano laboratory [125]. Anti-HIF1α and anti-PHD3 antibodies were obtained from

OxyCell Bioresearch. The anti-VHL antibody was from BD Pharmingen. The antibody against HIF2α and HIF1α hydroxylated at Pro564 was obtained from Novus

Biologicals. Anti-HA (haemagglutinin) and anti-tubulin antibodies were from Sigma

Chemicals.

Hypoxia treatment

Cells were placed in an airtight modular incubator chamber (Billup-Rothenburg,

Forma Scientific) that had been equilibrated with a gas mixture containing 1% oxygen,

5% CO2 and 94.5% nitrogen at 37◦C. Hypoxia treatments were for 8 h, except for experiments involving detection of HIF1α target genes that were conducted using 16 h of hypoxia treatment.

Cell culture

MCF-7 and MDA-MB-231 cells were obtained from A.T.C.C. (Manassas, VA,

121 U.S.A.) and maintained as previously described [125]. MDA-MB-231 cells were transfected with control vector or expression vector for FLAG–HEXIM1 as described previously [120]. RCC4 cells (pVHL-deficient or transfected with wild-type pVHL, see [292]) were maintained as described previously [293].

RNAi

MCF7-control miRNA and MCF7-HEXIM1 miRNA cells were generated as described previously [119]. Apolymerase II promoterdriven miRNA expression vector system (Invitrogen) was used. To make pcDNA-HEXIM1 miR, miRNA oligonucleotides were annealed and cloned into the pcDNA 6.2 GW/EmGFP vector

(Invitrogen) according to the manufacturer’s instructions. MCF-7 cells were transfected with pcDNA 6.2-GW/EmGFP-miR expression vectors containing either the HEXIM1 miRNA insert or a control LacZ miRNA insert. Following blasticidin selection, cells expressing the highest level of GFP were flow-sorted and expanded.

The sequence of the miRNA oligonucleotides are: miR clone 35 (forward),

5-TGCTGTACAGTTGCTAGTTTGAGGCTGTTTTGGCCACTG

ACTGACAGCCTCAATAGCAACTGTA-3;

(reverse),

5-CCTGTACAGTTGCTATTGAGGCTGTCAGTCAGTGGCCAA

122 AACAGCCTCAAACTAGCAACTGTAC-3;

miR clone 609 (forward),

5-TGCTGATGAGGAACTGCGTGGTGTTAGTTTTGGCCACTG

ACTGACTAACACCACAGTTCCTCAT-3;

(reverse),

5-CCTGATGAGGAACTGTGGTGTTAGTCAGTCAGTGGCCA

AAACTAACACCACGCAGTTCCTCATC-3.

RT (reverse transcription)–PCR analyses

MCF-7 and MDA-MB-231 cells were subjected to high (21%) or low (1%) oxygen conditions as indicated. All cells were subsequently subjected to RT–PCR analyses as described previously [119]. Total mRNAs were extracted using TRIzol® reagent from Invitrogen as per the manufacturer’s protocol. mRNAs were reverse transcribed using the M-MLV Reverse Transcriptase kit (Invitrogen) following the recommended protocol. cDNAs were PCR-amplified using the primers listed below. The amplified products were run on an agarose gel and visualized by ethidium bromide staining. A

12-bit digital camera captured fluorescence and signal intensities were quantified using the Alphaimager software from Alpha Innotech. The primers used and sequences are:

123 HIF-1α (forward), 5-TGCTAATGCCACCACTACC-3;

(reverse), 5-TGACTCCTTTTCCTGCTCTG-3

VEGF (forward), 5-CTTTCTGCTGTCTTGGGTG-3

(reverse), 5-ACTTCGTGATGATTCTGCC-3

SDF1 (forward),5-CCGCGCTCTGCCTCAGCGACGGGAAG-3

(reverse), 5-CCTGTTTAAAGCTTTCTCCAGGTACT-3

CXCR4 (forward), 5-AGCTGTTGGCTGAAAAGGTGGTCTATG-3

(reverse), 5-GCGCTTCTGGTGGCCCTTGGAGTGTG-3

Hsulf1 (forward), 5-GAGCCATCTTCACCCATTCAAG-3

(reverse), 5-TTCCCAACCTTATGCCTTGGGT-3

GAPDH (forward), 5-TCCACTGGCGTCTTCACC-3

(reverse), 5-GGCAGAGATGATGACCCTTTT-3

Western blot analysis

Cell lysates were analysed by Western blotting as described previously [119]. Total protein was extracted using MPER Mammalian Protein Extraction reagent (Thermo

Fisher Scientific). Proteins were detected using their respective primary antibodies and HRP (horseradish peroxidase)-conjugated secondary antibody. GAPDH or tubulin was used as a loading control. Signals were detected using the ECL Western

Blotting Analysis System (GE Healthcare).

124 Co-immunoprecipitation

Endogenous proteins were co-immunoprecipitated and analysed as previously described [117]. MCF7 and MDA-MB-231 lysates were incubated with Protein G beads that had been preadsorbed with specific antibody or non-specific mouse IgG (as a negative control). The beads were collected by centrifugation and washed with

RIPA buffer (20 mMTris/HCl, pH 7.5, 150 mMNaCl, 0.5% sodium deoxycholate, 1 mM EDTA and 0.1% SDS; filtered) and PBS twice. After the final wash, pellets were resuspended in Western sampling buffer, and immunoprecipitated proteins were separated by SDS/PAGE (10% gel). Western blot analyses were performed as described above.

ChIP assays

Cells were grown in 100-mm-diameter dishes and processed for ChIP analyses as described previously [119]. Briefly, cells were fixed with 1% formaldehyde and lysed in SDS-lysis buffer with protease inhibitors. Lysed cells were sonicated using a

Branson 450 sonicator. Clarified sonicated chromatin was diluted 10-fold in ChIP dilution buffer and used for immunoprecipitation with a given antibody. The antibody–chromatin complexes were pulled down using Protein A beads. The beads were subjected to a series of washes and the antigen–DNA complexes were eluted.

The eluates were reverse cross-linked overnight at 65◦C and the DNA was purified by

125 phenol/chloroform extraction. Ethanol-precipitated pellets were resuspended in water and were used as a template for PCR analysis. PCR-amplified products were run on a

2% agarose gel and visualized by ethidium bromide staining. A 12-bit digital camera captured fluorescence and signal intensities were quantified using the Alphaimager software.

In vitro translation and protein–protein interaction assays

In vitro transcription and translation of HIF1α and HDAC1 were performed using the Promega TNT® kit according to the manufacturer’s recommendations. GST pull-down assays have been described previously [124].

Mouse model

All animal work reported in the present study was approved by the CWRU

Institutional Animal Care and Use Committee. HEXIM1 expression was induced in mammary epithelial cells of PyMT (Polyoma Middle-T antigen) mice by mating

MMTV (murine mammary tumor virus)/HEXIM1 bitransgenic mice with PyMT mice as described previously [126]. HEXIM1 expression was induced by supplementing the drinking water of mice with doxycycline at a final concentration of 2 mg/ml.

MMTV/PyMT/HEXIM1 mice (+−doxycycline) were killed at 17 weeks of age. Total protein from mammary tumors was extracted using M-PER Mammalian Protein

Extraction reagent.

126 Invasion assays

Cell-invasion assays were performed as described previously [126]. Cell-invasion assays were performed using Transwell inserts (8-mm-diameter pore size; Corning

Costar) that were coated with 10% MatrigelTM and placed inside the wells of a 24- well plate. MDA-MB-231-control or MDA-MB-231-fl-HEXIM1 were suspended in

Opti-MEM® (serum-free culture medium, Invitrogen) and placed in the upper chamber of the Transwell insert (50000 cells/well). The lower chambers contained

MEM (minimal essential medium) supplemented with 5% FBS. The cells were allowed to migrate to the lower chamber at 37 ◦C for 48 h. After incubation, cells invading the MatrigelTM were fixed with 3%paraformaldehyde and stained with

0.5% Crystal Violet. Invading cells were counted from five random fields per well under a microscope.

Data analyses

Statistical significance was determined using the Student’s t test comparison for unpaired data. For some comparisons, probability values for the observed differences between groups were based on one-way ANOVA.

Results

HEXIM1 down-regulated HIF1α protein levels

Alterations in HEXIM1 levels were achieved by transfecting breast epithelial

127 MDA-MB-231 or MCF7 cells with control vector (or control miRNA), or expression vector for FLAG–HEXIM1 or HEXIM1 miRNA respectively. Reflecting what we observed in our previous mouse studies [120], modulation of HEXIM1 levels resulted in alterations in HIF1α protein levels in hypoxia-treated MCF7 cells and

MDA-MB-231 cells (Figures V-1A–C). Up-regulation of HIF1α protein levels was observed using two different miRNA clones (Figure V-1A). HIF1α protein levels were not significantly altered by HEXIM1 under normoxia. We also did not observe

HEXIM1 regulation of the mRNA level of HIF1α (Figures V-1A and B), suggesting posttranscriptional regulation. We thus examined whether HEXIM1 regulates HIF1α protein stability. We up-regulated expression of HEXIM1 using an expression vector for FLAG–HEXIM1. Cells were maintained under hypoxia followed by treatment with the protein synthesis inhibitor cycloheximide for different time periods. As shown in Figure V-1C, the stability of HIF1α protein was decreased in cells transfected with FLAG–HEXIM1 compared with control transfected cells. One-way

ANOVA indicates statistical significance with a P value of 0.017. We did not see alterations in HIF2α levels as a result of downregulation of HEXIM1 levels (Figure

V-1D).

HEXIM1 interacted with HIF1α and increased HIF1α ubiquitination by enhancing the interaction of HIF1α with pVHL

128 We then determined whether HEXIM1 can act on HIF1α directly. Using endogenous co-immunoprecipitation experiments, we observed an interaction between HEXIM1 and HIF1α under hypoxic conditions (Figure V-2A). We did not observe a HEXIM1 and HIF1α interaction under normoxic conditions (results not shown). The interaction between HEXIM1 and HIF1α was verified using GST pull-down assays. We then examined whether HEXIM1 can regulate post-translational modifications of HIF1α that targets HIF1α for degradation, in particular hydroxylation and ubiquitination. We observed decreased levels of ubiquitinated and hydroxylated HIF1α upon hypoxia treatment and downregulation of

HEXIM1 using HEXIM1 miRNA (Figures V-2B and C). We then determined whether HEXIM1 promoted the association of HIF1α with pVHL. Down-regulation of HEXIM1 resulted in attenuation of the interaction of HIF1α with pVHL under hypoxic conditions (Figure V-2D). The fact that HEXIM1 attenuation of HIF1α protein stability involved an enhanced interaction of HIF1α with pVHL was supported by our observation that the ability of HEXIM1 to down-regulate HIF1α protein expression was attenuated in pVHL-deficient RCC4 cells when compared with pVHL-transfected RCC4 cells (Figure V-2E).

HEXIM1 regulated expression of and interacted with PHD3

To explore the potential basis for HEXIM1 regulation of the hydroxylation of

129 HIF1α, we examined our microarray dataset [126]. Genes involved in post-translational modifications were shown to be enriched in our pathway analyses of genes differentially expressed in MCF7 control miRNA and MCF7-HEXIM1 miRNA cells (P=10−5). One of the differentially expressed post-translational modifications genes encodes PHD3, which has been reported to down-regulate HIF1α protein stability [177, 180]. Our ChIP-seq analyses and validation by standard ChIP indicated recruitment of HEXIM1 to the −3790/−3677 region of the PHD3 gene

(Figure V-3A). A consequence of the recruitment of HEXIM1 to the PHD3 gene was down-regulation of PHD3 expression under normoxic conditions and up-regulation of

PHD3 expression in hypoxia-treated MDA-MB-231 cells (Figures V-3A and B).

Mechanistically, the opposite effects of HEXIM1 on PHD3 expression during normoxic and hypoxic conditions may be due to the presence of distinct factors in the transcriptional complex with HEXIM1 on the PHD3 gene under normoxic and hypoxic conditions. The HEXIM1-binding site was distinct from the HIF1α-binding site [294] and suggested HIF1α-independent regulation of PHD3 by HEXIM1.

Consistent with the inhibition of PHD3 expression by HEXIM1 under normoxic conditions was the enhancement of PHD3 expression in HEXIM1 miRNA transfected cells (Figure V-3B). Under hypoxia, PHD3 expression was induced through HIF1α in certain cells (including breast cells MCF7 and BTB474) [180]. Up-regulation of

130 HIF1α probably accounted for our finding that down-regulation of HEXIM1 also resulted in upregulation of PHD3 under hypoxic conditions (Figure V-3B). The effect of increased expression of HEXIM1 in the hypoxic environment of advanced breast cancer was also examined. We have recently reported that re-expression of HEXIM1 through transgene expression inhibited metastasis in a mouse model of metastatic mammary cancer, the PyMT transgenic mouse, that can be correlated with decreased expression of HIF1α, VEGF, compensatory pro-angiogenic factors and vascularization [126]. HEXIM1 re-expression in this mouse model resulted in increased PHD3 expression (Figure V-3B). We then determined whether HEXIM1 can influence PHD3 regulation of HIF1α stability. HEXIM1 interacted with PHD3

(Figure V-3C), suggesting that HEXIM1 may play a role in the downregulation of

HIF1α by PHD3 by also promoting an interaction between HIF1α and PHD3. In support, down-regulation of HEXIM1 resulted in up-regulation of HIF1α expression despite the increase in HA-PHD3 transfected cells (Figure V-3D). The above results suggested that there are two mechanisms by which HEXIM1 regulates PHD3, by direct transcriptional up-regulation of the PHD3 gene or by interaction with PHD3.

Both mechanisms are required for HEXIM1 to inhibit HIF1α via PHD3. Two mechanisms would ensure negative regulation of HIF1α activity was maintained.

Up-regulation by HEXIM1 is especially critical in breast cancer cells where

131 expression of PHDs is decreased [295]. In cases where HEXIM1 expression was also down-regulated, as in metastatic breast cancer [126], the up-regulation of PHD3 was likely owing to up-regulation of HIF1α. However, at these low levels, HEXIM1 was unable to effectively mediate an interaction between PHD3 and HIF1α, thus up-regulation of HIF1α was maintained.

HEXIM1 inhibits the HDAC1–HIF1α interaction and increases HIF1α acetylation

We tested other potential modes of regulation of HIF1α by HEXIM1. The acetylation of HIF1α has been reported to induce interaction of HIF1α with pVHL and HIF1α ubiquitination [291]. HDAC1 has been reported to deacetylate HIF1α and, as such, to be a positive regulator of HIF1α stability via direct interaction [206].

However, the role of HDAC1 and ARD1, the acetyltransferase that acetylates HIF1α, in the regulation of HIF1α is controversial, since both overexpression and silencing of

ARD1 were shown to have no impact on HIF1α stability and on the mRNA levels of the downstream target genes of HIF1α [296, 297]. We observed deacetylation of

HIF1α upon down-regulation of HEXIM1 (Figure V-4A). We also observed that

HEXIM1 interacted with HDAC1 (as shown by co-immunoprecipitation of endogenous proteins, Figure V-4B) and GST pull-down assays (Figure V-4C).

Attenuation of this interaction by down-regulating HEXIM1 resulted in enhanced

132 interaction of HIF1α with HDAC1 (Figure V-4B). Although the enhanced

HIF1α–HDAC1 interaction can be attributed to an increased HIF1α level, the deacetylation of HIF1α supports the enhanced HIF1α–HDAC1 interaction. Moreover our GST pull-down assays indicated that the interaction between HEXIM1 and HIF1α was attenuated in the presence of HDAC1, and the interaction between HEXIM1 and

HDAC1 was attenuated in the presence of HIF1α (Figure V-4C). These findings suggested a competition between HEXIM1 and HDAC1 for binding to HIF1α or that

HIF1α provided a platform for an interaction between HEXIM1 and HDAC1.

HEXIM1 then sequestered HDAC1 from the HIF1α protein complex. Thus HEXIM1 can induce a decrease in HIF1α protein levels by inhibiting the ability of HDAC1 to act on HIF1α.

HEXIM1 inhibits the expression of HIF1α target genes

The down-regulation of HIF1α protein levels by HEXIM1 prompted us to examine

HEXIM1 inhibition of other HIF1α-regulated genes besides VEGF. HIF1α promotes metastases of breast cells by up-regulating CXCR4 and SDF1 signaling [298] and both factors are expressed in breast cells [299]. Another transcriptional target, LOXL2

(lysyl oxidase-like 2) [267], is secreted by hypoxic breast tumor cells, accumulates at pre-metastatic sites, cross-links collagen IV in the basement membrane, and is essential for CD11b+ myeloid cell recruitment [300]. Down-regulation of HEXIM1

133 resulted in enhanced hypoxia induced VEGF, SDF-1, CXCR4 and LOXL2 mRNA expression, and CXCR4 and SDF1 protein expression in MCF7 cells (Figure V-5).

We have previously reported on down-regulation of VEGF protein levels by HEXIM1

[120]. HEXIM1 also altered expression of CXCR4, SDF-1 and LOXL2 under normoxic conditions, which suggested regulation by HEXIM1 also involved

HIF1α-independent mechanisms (Figure V-5).

HEXIM1 inhibits hypoxia-induced breast cancer cell invasion

We determined whether HEXIM1 can regulate other known hypoxia-regulated cellular processes through its attenuation of HIF1α-regulated gene transcription. We have previously observed that HEXIM1 inhibited metastasis using a mouse model of metastatic mammary cancer [126]. Conversely, downregulation of HEXIM1 in MCF7 cells using HEXIM1 miRNA resulted in enhanced cell invasion [126]. We tested the possibility that inhibition of HIF1α is a contributing factor in the inhibitory effects of

HEXIM1 on cell invasion. Our invasion assays using Matrigel-TM-coated Boyden chambers indicated that HEXIM1 inhibited invasion of MDA-MB-231 cells (Figure

V-6A). That HEXIM1 inhibition of HIF1α is a critical factor in the inhibitory effects of HEXIM1 on breast cancer cell invasion was supported by experiments where cotransfection of HIF1α or HDAC1 attenuated the inhibitory effects of HEXIM1 on

MDA-MB-231 cell invasion. To verify the inhibition of HIF1α-mediated cell

134 invasion by HEXIM1 we examined inhibition of HSulf-1 expression by HIF1α.

Down-regulation of HSulf-1 expression was shown to mediate hypoxia-mediated enhanced cell migration and invasion [261]. HSulf-1 was also shown to be a direct transcriptional target of HIF1α. Down-regulation of HEXIM1 resulted in further attenuation of HSulf-1 expression observed under hypoxic conditions (Figure V-6B).

Discussion

We have uncovered a novel function of HEXIM1 in directly modulating HIF1α levels in breast cancer cells. The results from the present study indicate that HIF1α is a direct target of HEXIM1 and that HEXIM1 regulates post-translational modifications of the HIF1α protein known to be important for regulating HIF1α protein stability. HEXIM1 is critical for the ability of PHD3 to down-regulate HIF1α protein stability under hypoxic conditions. The functional relevance of this regulation is supported by attenuation of the expression of HIF1α target gene expression and hypoxia-induced cell invasion by HEXIM1. The fact that HIF1α is a direct target of

HEXIM1 and HEXIM1 modulates post-translational modifications of HIF1α supports the role for HEXIM1 as a critical regulator of HIF1α activities in breast cancer cells.

PHD3 levels are relatively low across a wide range of normoxic cells, such that PHD2 was the most abundant enzyme in normoxic culture in all cells [180]. However, PHD3 is induced by hypoxia in certain cells (including breast cell lines MCF7 and BTB474)

135 and under these conditions suppression of PHD3 by siRNA increased the half-life of

HIF1α under normoxia and hypoxia [180]. Moreover, there are differences with respect to HIF1α activation between mice lacking PHD2 and mice lacking pVHL [34], suggesting residual prolyl hydroxylation of HIF1α by PHD1 or PHD3 in cells lacking

PHD2. These results suggest that PHD3 retains significant activity under hypoxic conditions and that the enzyme is important in limiting physiological activation of

HIF (particularly HIF2α) in hypoxia. PHD3 appears to have a lower oxygen Km than

PHD2 [177] and therefore may remain active at intermediate levels of hypoxia that can inactivate PHD2. PHD3 is a HIF1α target up-regulated by HIF1α, and is proposed to be part of the negative-feedback mechanism [301]. The upregulation of

PHD3 by HEXIM1 suggests HIF1α-independent regulation by HEXIM1. In breast tumors, PHD3 expression has been correlated with good prognosis [263] and decreased PHD3 expression has been positively correlated with a basal phenotype

[302]. Moreover, there is no evidence of methylation-induced epigenetic silencing of

PHD1, PHD2 and PHD3 in invasive carcinomas as the basis of increased HIF1α expression [303]. However, PHD1–PHD3 expression did not correlate with HIF1α expression in breast carcinomas [263]. Moreover, in contrast with the expected down-regulation of HIF1α activity by PHD3, PHD3 potentiates HIF1α coactivator function of the pyruvate kinase isoform PKM2 in HeLa cells, resulting in metabolic

136 reprogramming [276]. It is possible that the predominant effects of PHD3 depend on the expression of factors such as HEXIM1 that directs PHD3 to other

HEXIM1-interacting factors such as HIF1α. The opposite effects of HEXIM1 on

PHD3 under normoxia and hypoxia may be opportune given the different roles of

PHD3 targets, in particular Spry2 that has tumor-suppressor activities and is targeted by PHD3 under normoxic conditions [304]. In doing so, HEXIM1 may prevent some of the growth-promoting effects of PHD3 on cancer cells while inhibiting HIF1α action. Although the basis for the opposite effects of HEXIM1 on PHD3 expression under normoxia and hypoxia is being further elucidated, the present study supports direct transcriptional regulation of PHD3 by HEXIM1. HDAC1 deacetylates HIF1α and is considered to be a positive regulator of HIF1α stability via direct interaction

[206]. However, the role of ARD1, the acetyltransferase that acetylates HIF1α, in the regulation of HIF1α is not clear, since both overexpression and silencing of ARD1 were shown to have no impact on the stability and on the mRNA levels of the downstream target genes of HIF1α [296, 297]. It has been proposed that these observations may be due to the expression of MTA1 (metastasis-associated protein 1)

[206]. It has been reported that MTA1 is up-regulated under hypoxic conditions in breast cells [206]. MTA1 then interacts with HDAC1, and the HDAC1–MTA1 complex interacts with the ODD domain of HIF1α. As a result, HIF1α is deacetylated,

137 thereby blocking the degradation of the protein [206, 305]. MTA1 inhibited

ARD1-induced HIF1α degradation, and MTA1 expression levels were closely associated with ARD1 function [206]. In cell lines that express low levels of MTA1,

ARD1- induced HIF1α degradation was significant, whereas ARD1 did not function in other cell lines, such as MCF7 and MDAMB-231 cells, that express high levels of

MTA1 in response to hypoxia. Thus cells with high levels of MTA1 may represent an instance where other modes of regulation of HIF1α are needed. MTA1-induced deacetylation may counteract PHD3- induced hydroxylation and prevent HIF1α ubiquitination and degradation. We determined that HEXIM1 is another determining factor in the acetylation of HIF1α, even in cells that express high levels of MTA1

[206, 306], by attenuating the interaction between HEXIM1 and HDAC1. It was recently reported that HDAC1 may be involved in the stabilization of HIF1α [306].

However, this finding is in contrast with reports that HDAC1 inhibitors attenuate

HIF1α levels and signaling, tumorigenesis and angiogenesis [307-309]. Finally, it is possible that the interaction between HEXIM1 and HDAC1 regulates

HIF1α-mediated transcription through histone deacetylation. However, the enhancement of HIF1α transcriptional activity has been reported to involve displacement of the HDACs, such as HDAC1, from the HIF1α transcriptional complex [310, 311]. HIF1α is overexpressed in many human cancers and activates

138 transcription of genes involved in crucial features of cancer biology, including angiogenesis, cell survival, glucose metabolism and invasiveness, thus representing an attractive target for a selective cancer therapy [223]. The present study provides new insight into how HIF1α can be inhibited in breast cancer cells. Inhibition of these factors by HEXIM1 reveals aspects of HEXIM1 mechanism of action not previously identified as it does not rely on HEXIM1 inhibition of the transcriptional elongation machinery. Our studies also support HEXIM1 inhibition of another HIF1α-regulated cellular process, cell invasion. Consistent with the proposed role of HSulf-1 in hypoxia-mediated cell invasion, HEXIM1 attenuated HIF1α inhibition of HSulf-1 expression. VEGF, SDF-1 and CXCR4 are all HIF1α target genes that are down-regulated by HEXIM1 and provide a mechanistic basis for our observed effects of HEXIM1 not only on the primary tumor, but also on the tumor microenvironment.

We recently reported that HEXIM1 inhibited the recruitment of BMDCs to primary tumors [126]. HEXIM1 inhibition of hypoxia induced SDF-1 and CXCR4 expression provides a mechanism for HEXIM1 to indirectly act on BMDCs and to suppress metastatic cancer. Intratumoral hypoxia leads to the recruitment of BMDCs, endothelial and pericyte progenitors, tumor-associated macrophages, immature monocytic cells and myeloid cells [312]. Current evidence suggests a promotional role of BMDCs on the existing blood vessels rather than de novo neovascularization

139 in tumors. These cells produce different pro-angiogenic factors and constitute an adaptive mechanism of resistance to angiogenic inhibitors under a low oxygen context.

VEGF and SDF-1 are essential in recruiting bone-marrow-derived myelomonocytic cells to tumors [313]. BMDCs express the SDF-1 receptor CXCR4 and are recruited to hypoxic tissue by cell tropism to SDF-1 [314].

Acknowledgement

We thank Dr Hung Ying Kao (Department of Biochemistry, CaseWesternReserve

University, Cleveland, OH, U.S.A.) for the HDAC1 expression vector and Dr Greg

Semenza (Department of Pediatrics, Johns Hopkins School of Medicine, Baltimore,

MD, U.S.A.) for the HIF-1α expression vector.

140 A.

B.

C.

D.

141 Figure V-1. HEXIM1 destabilizes HIF1α protein

(A) MCF7 cells were transfected with the expression vector for control miRNA or

HEXIM1 miRNA. Shown are representative Western blots (left-hand panels, representative of five experiments) or RT–PCR (right panel, representative of three experiments) indicating HIF1α levels under normoxia and hypoxia (0.5% oxygen). (B)

MDA-MB-231 cells were transfected with control vector or expression vector for FLAG (FL)–HEXIM1. Shown are representative Western blots (left-hand panel, representative of four experiments) or RT–PCR (right-hand panel, representative of three experiments) indicating HIF1α levels under normoxia and hypoxia. (C) Cells were transfected with expression vector for HEXIM1 or empty vector. The cells were incubated under hypoxia for 8 h as indicated. At the end of treatment, 100 μM cycloheximide (CHX) was added to the medium for the indicated time periods. The expression of HIF1α and GAPDH were analysed by Western blot analysis. The density of the HIF1α protein band was determined using an image analysis system.

The values were normalized to GAPDH and expressed as the percentage change relative to zero time. Graphs show means+−S.E.M. from three experiments. *P <0.05 compared with control transfected cells with no CHX. Statistical significance was based on one-way ANOVA. (D) HIF2α levels were analysed by Western blot analyses of cell extracts from MCF7 cells transfected with expression vector for

142 control miRNA or HEXIM1 miRNA and subjected to hypoxia. The results are representative of two experiments. Molecular masses in kDa are shown next to the

Western blots.

143 A.

B. C.

D.

E.

144 Figure V-2. HEXIM1 interacted with HIF1α and down-regulation of HEXIM1 resulted in decreased levels of hydroxylated and ubiquitinated HIF1α and attenuated the HIF1α–pVHL interaction

(A) Upper panels, MCF7 cells and MDA-MB-231 cells were subjected to hypoxia (8 h). Lysates were immunoprecipitated (IP) using antibodies against HIF1α or HEXIM1 and analyzed for co-immunoprecipitating proteins by Western blotting using an anti-HEXIM1 antibody. Normal rabbit immunoglobulin was used as a specificity control. Input lanes represent 25%of the total protein. MCF7 and MDA-MB-231 panels represent five and three experiments respectively. Lower panel, in vitro translated and [35S] methionine-labelled HIF1α was incubated with GST or GST–

HEXIM1 bound to Sepharose. Input lane represents 10% of the total volume of in vitro translated product used in each reaction. Panels represent eight experiments. (B)

Lysates from control miRNA or HEXIM1 miRNA transfected MCF7 cells were immunoprecipitated using an anti-HIF1α antibody and analysed by Western blotting using the indicated anti-(pan-ubiquitinated HIF1α ) (Ub- HIF1α ) antibody. Cells were subjected to hypoxia and MG132 (10 μM) treatments to accumulate ubiquitinated

HIF1α. Panels are representative of three experiments. (C) Western blot analyses of

HIF1α hydroxylated at Pro564 (P564OH) and total HIF1α in lysates from hypoxia-treated control miRNA or HEXIM1 miRNA transfected MCF7 cells. Panels

145 are representative of six experiments. (D) Control miRNA or HEXIM1 miRNA transfected MCF7 cells were subjected to hypoxia and MG132 (10 μM) treatments.

Lysates were immunoprecipitated using an anti- HIF1α antibody and analysed for co-immunoprecipitating proteins by Western blotting using an anti-pVHL antibody.

Normal rabbit immunoglobulin was used as a specificity control. Input lanes represent

25% of the total protein. Panels are representative of three experiments. (E) pVHL-deficient RCC4 or wild-type pVHL transfected RCC4 cells were transfected with control vector or expression vector for FLAG (FL)–HEXIM1.

The cells were incubated under hypoxia for 8 h as indicated. The expression of

HEXIM1, HIF1α and GAPDH were analysed by Western blot analyses. Panels are representative of four experiments. Molecular masses in kDa are shown next to the

Western blots.

146 A.

B.

C.

D.

147 Figure V-3. HEXIM1 up-regulated expression and interaction with PHD3

(A) Left-hand panel, ChIP analyses of MCF7 cell lysates immunoprecipitated with antibodies against HEXIM1 or control non-specific rabbit immunoglobin, followed by

PCR amplification of the −3790/−3677 region of the PHD3 gene. Panels are representative of three experiments. Right-hand panel, PHD3 mRNA levels in control and FLAG (FL)–HEXIM1-transfected MDA-MB-231 cells were quantified and normalized to GAPDH. Panels are representative of three experiments. (B)

Representative Western blots indicating regulation of PHD3 by HEXIM1 in hypoxia-treated MCF7 cells and MDA-MB-231 cells. Panels are representative of three experiments. Also shown is a representative Western blot of PHD3 expression in tumor lysates from control and doxycycline-treated PyMT/MMTV/HEXIM1 mice.

Blots were probed with an anti-GAPDH antibody as a loading control. Panels represent five mice per group [+−doxycycline (DOX)]. (C) MCF7 cells and

MDA-MB-231 cells were subjected to hypoxia treatment (8 h). Lysates were immunoprecipitated (IP) using the antibodies indicated and analysed for co-immunoprecipitating proteins by Western blotting using an anti-PHD3 antibody.

Normal rabbit immunoglobulin was used as a specificity control. Input lanes represent

25% of the total protein. Panels are representative of three experiments. (D)

Expression of HIF1α under hypoxic conditions in control miRNA- and HEXIM1

148 miRNA-transfected MCF7 cells and in the absence or presence of HA-tagged PHD3 were analysed by Western blot analyses. Panels are representative of at least three experiments. Molecular masses in kDa are shown next to the Western blots.

149 A.

B.

C.

150 Figure V-4. Down-regulation of HEXIM1 resulted in an enhanced HIF1α

HDAC1 interaction and deacetylation of HIF1α

(A) Control and HEXIM1miRNA-transfected MCF7 cells were subjected to hypoxia

(8 h). Lysates were immunoprecipitated (IP) using an anti- HIF1α antibody and analysed by Western blotting using the indicated anti-pan-acetyl antibody. Panels are representative of two experiments. (B) MCF7 cells were subjected to hypoxia (8 h).

Lysates were immunoprecipitated using the indicated antibodies and analysed for co-immunoprecipitating proteins by Western blotting using the indicated antibodies.

Normal rabbit immunoglobulin was used as a specificity control. Input lanes represent

25% of the total protein. Panels are representative of three experiments. (C) In vitro translated and indicated [35S] methionine-labelled proteins were incubated with the indicated GST-fusion proteins bound to Sepharose. The Input lane represents 10% of the total volume of in vitro translated product used in each reaction. Panels are representative of eight experiments. Molecular masses in kDa are shown next to the

Western blots.

151 A.

B.

C.

152 Figure V-5. HEXIM1-regulated expressions of HIF1α-regulated genes

Hypoxia-induced changes in VEGF , SDF-1, CXCR4 and LOXL2 mRNA levels in (A) control miRNA- and HEXIM1 miRNA-transfected MCF7 cells or (B) control and

FLAG (Fl) – HEXIM1-transfected MDA-MB-231 cells were quantified and normalized to GAPDH. Panels are representative of three experiments. (C) SDF-1 and CXCR4 protein levels in control miRNA- and HEXIM1miRNA-transfected

MCF7 cells were quantified and normalized to GAPDH. Histograms show the means+− S.E.M. from three experiments. *P <0.05 compared with control transfected cells with the same treatment (for 8 h). Panels are representative of three experiments.

153 A.

B.

Figure V-6. HEXIM1 inhibited hypoxia-induced invasion of MDA-MB-231 cells

(A) MDA-MB-231 cells were transfected with expression vector for the same amount of HEXIM1 or empty vector, in the absence or presence of increasing levels of expression vectors for HDAC1 or HIF1α Cells were then subjected to hypoxia (0.5% oxygen). Invasion of cells through MatrigelTM was assessed in the Transwell invasion assay. *P <0.05 compared with cells transfected with HEXIM1 expression vector only (i.e. 1:0). Panels are representative of five experiments. (B)

Hypoxia-induced changes in HSulf-1 mRNA levels in control miRNA- and HEXIM1 miRNA-transfected MCF7 cells were quantified and normalized to GAPDH.

Histograms are means+− S.E.M. from three experiments. *P <0.05 compared with control transfected cells with the same treatment.

154 CHAPTER VI

Summary, Discussion and Future Directions

Summary and Discussion

Here we report that HEXIM1 regulates HIF1α post-translationally in both

ER-positive and ER-negative breast cancer cells, whereas HIF2α is not regulated by

HEXIM1 (Figure V-1D). Direct interaction between HIF1α and HEXIM1 was also demonstrated by endogenous CO-IP and GST-pull down assays. Down-regulation of

HEXIM1 reduces the levels of both ubiqutinated and hydroxylated HIF1α (Figure

V-2B, C). The basis for these observations is that HIF1α-VHL interaction is attenuated upon downregulation of HEXIM1. Over-expression of HEXIM1 reduces

HIF1α protein only in the presence of VHL suggesting that HEXIM1 regulation of

HIF1α is VHL-dependent (Figure V-2D, E). It is important to note that HEXIM1 regulated HIF1α under hypoxic conditions.

We also reported on modulation of PHD3 protein expression by HEXIM1 (Figure

V-3), with HEXIM1 having opposite effects on PHD3 levels under normoxic and hypoxic conditions. These findings suggest that HEXIM1 cooperates with distinct transcriptional complexes to regulate PHD3 under these different conditions. The transcriptional regulation of PHD3 by HEXIM1 may be HIF1α-independent because the HEXIM1 binding site on PHD3 is distinct from the HIF1α-binding site and

155 HEXIM1 regulates PHD3 expression even under normoxic conditions (Figure V-3B).

Up-regulation of HIF1α resulted in up-regulation of PHD3 in hypoxia and downregulation of HEXIM1 also resulted in enhanced up-regulation of PHD3 in hypoxia.

Interaction among HIF1α, HEXIM1 and PHD3 were revealed by endogenous

Co-IP (Figure V-3C). Over-expression of PHD3 down-regulated HIF1α in the presence, but not in the absence of HEXIM1, suggesting that regulation of HIF1α by

PHD3 requires HEXIM1 (Figure V-3D). We hypothesize that HEXIM1 and PHD3 cooperate to destabilize HIF1α in hypoxic condition.

Re-expression of HEXIM1 in MMTV-PyMT mouse model of breast cancer not only inhibited metastasis but also resulted in decreased expression of factors involved in vascularization [126], which can be correlated with increased PHD3 expression.

This result indicates that HEXIM1 may function as a tumor suppressor and is involved in modulation of vascularization. The involvement of PHD3 in inhibition of tumor vascularization and metastasis by HEXIM1 warrants further investigation.

Acetylation is another post-translational modification of HIF1α protein.

Down-regulation of HEXIM1 resulted in reduced levels of acetylated-HIF1α (Figure

V-4A). Interaction between HDAC1, HIF1α and HEXIM1 was also demonstrated using endogenous CO-IP and GST-pull down assays. Competition between HDAC1

156 and HEXIM1 for binding to HIF1α was also revealed by GST-pull down assays

(Figure V-4B, C).

Investigation of mRNA level of HIF1α target genes, such as VEGF, SDF1, CXCR4

and Loxl2, revealed that their mRNA levels are significantly up-regulated under

hypoxia when HEXIM1 is downregulated. Similar results were also shown for

CXCR4 and SDF1 protein levels (Figure V-5).

It has been reported that HIF1α down-regulates endosulfatases 1 (Hsulf-1)

expression and activates bFGF2 pathway, which promotes cell migration and invasion

[261]. Reduced mRNA levels of Hsulf-1 were observed in cells wherein HEXIM1 has

been downregulated (Figure V-6B). Based on these findings, regulation of cell

invasion by HEXIM1 was studied. Increased expression of HDAC1 and HIF1α

attenuated the inhibitory effect of HEXIM1 on cell invasive ability (Figure V-6A).

Based on the results, our working model is presented in Figure VI-1. However,

there are some issues that need to be further addressed.

(1) Why would HEXIM1 need two mechanisms to regulate HIF1α protein?

It was previously reported that MTA1 is up-regulated in breast cancer cells in

hypoxic conditions and helps to stabilize HIF1α by recruiting HDAC1 to overcome

the effects of ARD1 (arrest-defective protein 1) [206]. Cells with high levels of

MTA1, such as MCF7 and MDA-MB-231 cells [315], may need an alternative mode

157 of regulation of HIF1α degradation. PHD3-HEXIM1 regulation of HIF1α degradation reported herein occurs in hypoxic conditions. Thus this regulation may counteract the impaired activity of ARD1 in these cells. However, hydroxylation activity of PHD3 is not efficient under low oxygen density. Having HEXIM1 regulate

HIF1α through competition with HDAC1 for HIF1α binding can counteract the inefficiency of PHD3-guided hydroxylation reaction in hypoxic conditions.

(2) The role of PHD3 in hypoxic conditions

Based on previous studies, PHD2 is the predominant enzyme responsible for

HIF1α degradation in normoxic conditions [169, 171, 180]. PHD1 and PHD3 have been shown to promote HIF1α degradation by catalyzing prolyl hydroxylation reaction in vitro. Both PHD2 and PHD3 are hypoxia inducible proteins suggesting negative feedback regulation of HIF1α. However, hypoxic induction of PHD3 is more obvious than the other two paralogues [180]. It was reported that among three PHD isoforms, PHD3 retains the highest hydroxylase activity during prolonged hypoxia

[177]. PHD3 appears to have lower oxygen Km than PHD2, suggesting that the affinity of PHD3 for oxygen is higher than the affinity of PHD2 for oxygen [316].

Most importantly, clinical data indicate that PHD3 expression in breast tumor cells is correlated with good prognosis [263]. Together there studies support the role of PHD3 as an essential enzyme that regulates HIF1α stability in hypoxic condition. Thus,

158 HEXIM1 destabilizes HIF1α in hypoxic condition partly though regulation of PHD3 expression and interaction with PHD3, and may play a critical role in the inhibitory effect of HEXIM1 on tumor angiogenesis and metastasis.

(3) Multiple roles of PHD3 in cells

The induction of PHDs in hypoxia ensures the rapid degradation of HIF1α upon re-oxygenation. Another explanation is that increased levels of PHDs can compensate for the impaired activity of PHDs at low oxygen density. However, in addition to its role promoting HIF1α degradation [177, 180, 316], there are reports on the role of

PHD3 in enhancing HIF1α transcriptional activity [276, 317].

Glucose is metabolized into PEP, which is converted to pyruvate by pyruvate kinase. Pyruvate is further metabolized into acetyl-CoA by pyruvate dehydrogenase and enters the TCA cycle [318]. There are two isoforms of pyruvate kinase, PKM1 and PKM2. PKM2 converts glucose to pyruvate and promotes the Warburg effect as a result pyruvate is converted into lactate and enters aerobic glycolysis (Figure IV-4)

[318]. PKM2 is found to be up-regulated in many cancer cell types [318]. It was reported that PHD3 hydroxylates PKM2 and promotes the PKM2-HIF1α interaction, which in turn promotes HIF1α transcriptional activity in Hela cells [276]. This result indicates that PHD3 may enhance cancer cell survival by modulating glycolytic metabolism. Controversially, PHD3 also functions to promote apoptosis during

159 normoxia [266] and the activation of PHD3 mRNA level is decreased in prostate and breast carcinoma cell lines [319]. The seeming paradox indicates that the functions of

PHD3 may depend on its interacting partners in different cell types. We have reported here that HEXIM1 provides a platform for PHD3 to interact with, hydroxylate and destabilize HIF1α in breast cancer cells. Perhaps PKM2-PHD3 effect is dominant in breast cancer cells, and increased expression of HEXIM1 can overcome

PKM2-PHD3-promoted HIF1α transcriptional activity in breast cancer cells.

(4) The role of HDAC (Histone deacetylases) in regulating HIF1α

There are many members in HDAC family (Table 1) which can modulate HIF1α stability. However, the mechanisms behind HDAC regulation of HIF1α stability are not clear.

HDACs can be divided into four groups. Class I has four members, 1, 2, 3 and 8, which are localized in the nucleus. Class II consists of six members, 4, 5, 6, 7, 9 and

10, which are localized both in the nucleus and cytoplasm. Class III consists of sirtuins (1-7), which was shown to repress HIF1α transcriptional activity. Class IV consists of HDAC11-related enzymes, which exhibit Class I and II characteristics

[320]. HDAC1, 2 and 3 are hypoxia-inducible and HDAC1 and 3 were shown to directly bind to HIF1α ODD domain to stabilize it and promote its transcriptional activity. This phenomenon appears to be cell-type specific [321]. HDAC7 has been

160 reported to interact with FIH in normoxia whereas HDAC7 translocates into nucleus to stabilize HIF1α in hypoxia [322]. Loss of HDAC7 induces morphological changes in and migration of endothelial cells [307]. HDAC4 and 6 were also demonstrated to promote HIF1α stability and transcriptional activity in a proteasome-dependent manner in RCC cells [210].

Targeting VEGF or VEGF signaling pathway effectively prunes tumor vessels and results in intratumor hypoxia. These result in resistance to radiation therapy, chemotherapy and antiangiogeneic therapies. Thus, HDAC inhibitors appear to be an alternative therapeutic target. However, the physiological role of HDACs is broad, implying that the side effects of targeting HDACs might be numerous. Although the

HDAC inhibitors currently in clinical trials exhibit promising effects by attenuating tumorigenesis and angiogenesis, none of HDAC inhibitors have shown selectivity in inhibiting individual HDAC members [320, 321]. We reported here that HEXIM1 competes with HDAC1 binding to HIF1α, which increases the acetylation status of

HIF1α and results in HIF1α degradation. Due to the fact that HDAC family have broad functions in cells [323] and specific HDAC inhibitor for each member has not been discovered at present [323], increased expression of HEXIM1 in cancer cells would be an alternative strategy in cancer therapy. There are several ways to increase

HEXIM1 expression in cancer cells that are discussed below.

161 Overall, we discovered a novel mechanism for regulation of HIF1α stability: (1)

HEXIM1 up-regulates the expression of and interacts with PHD3; (2) HEXIM1 interacts with HIF1α and increase HIF1α ubiquitination by enhancing HIF1α-pVHL interaction; (3) HEXIM1 inhibits HDAC1-HIF1α interaction and increases HIF1α acetylation; (4) HEXIM1 inhibits the expression of HIF1α target genes; and (5)

HEXIM1 inhibits hypoxia- induced breast cancer cell invasion. The tumor suppressor role of HEXIM1 is supported by our findings that the expression of HEXIM1 is attenuated in more aggressive tumor stages. Overall, we conclude that HEXIM1 down-regulates HIF1α in hypoxia through two mechanisms: i) by serving as a scaffold protein for PHD3 and HIF1α thereby promoting the hydroxylation of HIF1α; and ii) by competing with HDAC1 for HIF1α binding, thereby promoting the acetylation of HIF1α by unidentified acetyl-transferase in human breast tumor cells

(Figure VI-1).

Future Directions

(1) Investigation of the opposite effect of HEXIM1 on PHD3 expression.

We observed that increased expression of HEXIM1 enhances PHD3 expression in hypoxic-treated MDA-MB-231 cells and in MMTV-PyMT breast cancer mouse model. Modulation of PHD3 expression by HEXIM1 was also observed in normoxia indicating that HIF1α-independent regulation is involved. Distinct binding sites of

162 HEXIM1 to PHD3 regulatory regions further strengthen this assumption. However,

HEXIM1 reduces PHD3 expression in normoxia-treated MDA-MB-231 cells and in normoxic and hypoxic MCF7 cells. Distinct transcriptional complexes on HEXIM1 may account for this opposite result. To identify distinct transcriptional complexes, we will use the following cell line models: MCF7 (control and HEXIM1 miR transfected) and MDA-MB-231 (control and flag-HEXIM1 transfected) exposed to normoxia or hypoxia. Next, immunoprecipitation of HEXIM1 will be performed using nuclear extracts from these cells. The immunoprecipitated complex will be analyzed by two-dimensional-electrophoresis and mass spectrometry/proteomics methods to identify HEXIM1 interacting proteins. Proteins that differentially interact with HEXIM1 under the different conditions described above will be selected for further study. Down-regulation of the selected proteins in MCF7 and MDA-MB-231 cells will be performed and PHD3 expression will be examined by western blot.

Proteins that alter PHD3 expression will be selected as further candidates. Serial re-CHIP assays will be performed by using antibodies for HEXIM1 and candidate

HEXIM1 interacting proteins. Recruitment to PHD3 promoter/enhancer region will be examined.

Previous reports indicate that ubiquitin ligase Siah2 down-regulates PHD3 expression by ubiquitination of PHD3 in hypoxic condition, which results in the

163 stabilization of HIF1α. Moreover, the Akt pathway promotes Siah2 activity by phosphorylating Siah2. This phosphorylation is also stimulated by hypoxia [324, 325].

We reported here that HEXIM1 interacts with PHD3 under hypoxia. Our studies in progress revealed that HEXIM1 may also be involved in the regulation of Akt pathway in breast cancer cells. Whether Siah2 and/or Akt is involved in HEXIM1 regulation of PHD3 will be investigated further.

(2) Validation of our working model in vivo

Decreased expression of PHD3 has been reported to be correlated with larger breast tumor size, poor differentiation and increased proliferation [263]. We reported that HEXIM1 cooperates with PHD3 either by regulating PHD3 expression or by interacting with PHD3, which results in the degradation of HIF1α in breast cancer cell lines. To recapitulate this phenomenon in vivo, we will evaluate HIF1α protein levels in tumors dissected from PyMT/MMTV/HEXIM1 triple transgenic mice with or without PHD3 inhibitor [326]. HEXIM1 expression will be induced using doxycyline.

Due to the cooperation of HEXIM1 and PHD3 to destabilize HIF1α, we expect that the relative amount of HIF1α protein will be the most abundant in the absence of both

PHD3 and HEXIM1 expression model; be the medium in either non-HEXIM1 or under PHD3 inhibitor treated-model and be the least under expression of both

HEXIM1 and PHD3.

164 Concluding remarks

Due to the role of HEXIM1 as a tumor suppressor, methods to up-regulate HEXIM1 expression in prostate cancer and breast cancer should be developed. There are several ways to increase HEXIM1 expression in the clinical setting, either through

HMBA or derivatives of HMBA. The clinical use of HMBA has been hampered by the dose limiting toxicity, thrombocytopenia or reduction in blood platelet levels

[327]. Other serious adverse effects of HMBA include neurotoxicity and metabolic acidosis. Our lab has shown that introduction of HMBA into tumor tissues using

PLGA (poly lactic-co-glycolicacid, a FDA-approved injectable polymer), resulted in decreased breast cancer cell metastasis without the thrombocytopenia [126]. This is due to more localized delivery and increased exposure time of HMBA in tumor tissues. In addition we have developed more potent HMBA derivatives that may have less adverse effects and are alternatives to increase HEXIM1 expression in tumor cells

[328].

Overall, our studies demonstrated a role for HEXIM1 in prostate and breast cancer

(Figure VI-2). Thus, we conclude that HEXIM1 is a promising therapeutic target for treating metastatic cancers.

165

Figure VI-1. Hypothesis models. In hypoxia, HEXIM1 down-regulates HIF1α (a) by serving as a scaffolding protein for PHD3 and HIF1α thereby promoting the hydroxylation of HIF1α; (b) by competing with HDAC1 for binding to HIF1α thereby promoting the acetylation of HIF1α

166

Figure VI-2. Overall model for HEXIM1 in prostate cancer and breast cancer.

HEXIM1 inhibits cell proliferation by inhibiting ERα activity. HEXIM1 inhibits angiogenesis, cell homing, migration and invasion through regulating HIF1α protein expression as well as inhibiting 67LR localization. HEXIM1 prevents CRPC by interacting with AR and preventing AR from executing CRPC transcriptional program.

To do so HEXIM1 interacts with KDM5B, induce its recruitment to AR target genes, inhibit H3K4 methylation and FOXA1 recruitment, and represses AR mediated transcription.

167

Table 1. Nuclear Receptor Family. Class I: PR (Progesterone receptor), ERα, β

(Estrogen Receptor), GR (Glucocorticoid receptor), AR (Androgen Receptor), MR

(Mineralocorticoid receptor); Class II: VDR (Vitamin D receptor), TR (Thyroid

Receptor), RARα, β, γ, (Retinoic acid Receptor), PPARα, β/δ,γ (Peroxisome

Proliferators-activated Receptors); Class III: SF-1 (Steroidogenic factor 1), RXR

(Retinoid X Receptor), ERR (Estrogen-related receptor), ROR (retinoid-related orphan receptors), GPR (GPCR orphan Receptors).

168

Table 2. The regulation of HIF1α activity by HDACs.

L. Ellis et aL/ Cancer Letters 280 (2009) 145-153

169 BIBLIOGRAPHY

1. Rove, K.O. and E.D. Crawford, Dutasteride: novel milestones in prostate cancer chemoprevention. Drugs Today (Barc). 47(2): p. 135-44. 2. Siegel, R., D. Naishadham, and A. Jemal, Cancer statistics, 2013. CA Cancer J Clin. 63(1): p. 11-30. 3. Cancer, N.C.C.f., Prostate Cancer: diagnosis and treatment. National Collaborating Centre for Cancer, 2008. 4. Feldman, B.J. and D. Feldman, The development of androgen-independent prostate cancer. Nat Rev Cancer, 2001. 1(1): p. 34-45. 5. Gregory, C.W., et al., A mechanism for androgen receptor-mediated prostate cancer recurrence after androgen deprivation therapy. Cancer Res, 2001. 61(11): p. 4315-9. 6. Heinlein, C.A. and C. Chang, Androgen receptor in prostate cancer. Endocr Rev, 2004. 25(2): p. 276-308. 7. Wang, Q., et al., Androgen receptor regulates a distinct transcription program in androgen-independent prostate cancer. Cell, 2009. 138(2): p. 245-56. 8. Gao, W., C.E. Bohl, and J.T. Dalton, Chemistry and structural biology of androgen receptor. Chem Rev, 2005. 105(9): p. 3352-70. 9. McKenna, N.J., R.B. Lanz, and B.W. O'Malley, Nuclear receptor coregulators: cellular and molecular biology. Endocr Rev, 1999. 20(3): p. 321-44. 10. Olefsky, J.M., Nuclear receptor minireview series. J Biol Chem, 2001. 276(40): p. 36863-4. 11. Giguere, V., Orphan nuclear receptors: from gene to function. Endocr Rev, 1999. 20(5): p. 689-725. 12. Mangelsdorf, D.J., et al., The nuclear receptor superfamily: the second decade. Cell, 1995. 83(6): p. 835-9. 13. Sablin, E.P., et al., Structural basis for ligand-independent activation of the orphan nuclear receptor LRH-1. Mol Cell, 2003. 11(6): p. 1575-85. 14. Levoye, A., et al., Do orphan G-protein-coupled receptors have ligand-independent functions? New insights from receptor heterodimers. EMBO Rep, 2006. 7(11): p. 1094-8. 15. Blumberg, B. and R.M. Evans, Orphan nuclear receptors--new ligands and new possibilities. Genes Dev, 1998. 12(20): p. 3149-55. 16. McEwan, I.J., Molecular mechanisms of androgen receptor-mediated gene regulation: structure-function analysis of the AF-1 domain. Endocr Relat Cancer, 2004. 11(2): p. 281-93. 17. Kaku, N., et al., Characterization of nuclear import of the domain-specific androgen receptor in association with the importin alpha/beta and Ran-guanosine 5'-triphosphate systems. Endocrinology, 2008. 149(8): p. 3960-9. 18. Lonergan, P.E. and D.J. Tindall, Androgen receptor signaling in prostate cancer development and progression. J Carcinog. 10: p. 20. 19. Gelmann, E.P., Molecular biology of the androgen receptor. J Clin Oncol, 2002. 20(13): p. 3001-15. 20. Nettles, K.W. and G.L. Greene, Ligand control of coregulator recruitment to nuclear receptors. Annu Rev Physiol, 2005. 67: p. 309-33. 21. He, B., et al., The FXXLF motif mediates androgen receptor-specific interactions with coregulators. J Biol Chem, 2002. 277(12): p. 10226-35. 22. Bennett, N., et al., Androgen receptor and caveolin-1 in prostate cancer. IUBMB Life, 2009. 61(10): p. 961-70. 23. Heinlein, C.A. and C. Chang, The roles of androgen receptors and androgen-binding proteins in nongenomic androgen actions. Mol Endocrinol, 2002. 16(10): p. 2181-7.

170 24. Matsumoto, T., et al., The androgen receptor in health and disease. Annu Rev Physiol. 75: p. 201-24. 25. http://en.wikipedia.org/wiki/Prostate. 26. Stenman, U.H., et al., Prostate-specific antigen. Semin Cancer Biol, 1999. 9(2): p. 83-93. 27. http://en.wikipedia.org/wiki/Hypothalamus. 28. http://www.austincc.edu/apreview/PhysText/Reproductive.html. 29. http://en.wikipedia.org/wiki/File:Steroidogenesis.svg. 30. Nickel, J.C., Prostatitis: myths and realities. Urology, 1998. 51(3): p. 362-6. 31. http://www.nlm.nih.gov/medlineplus/ency/article/000524.htm. 32. http://en.wikipedia.org/wiki/Benign_prostatic_hyperplasia. 33. Geller, J., et al., Comparison of androgen metabolites in benign prostatic hypertrophy (BPH) and normal prostate. J Clin Endocrinol Metab, 1976. 43(3): p. 686-8. 34. Thorpe, A. and D. Neal, Benign prostatic hyperplasia. Lancet, 2003. 361(9366): p. 1359-67. 35. Luo, J., et al., Human prostate cancer and benign prostatic hyperplasia: molecular dissection by gene expression profiling. Cancer Res, 2001. 61(12): p. 4683-8. 36. Debes, J.D. and D.J. Tindall, Mechanisms of androgen-refractory prostate cancer. N Engl J Med, 2004. 351(15): p. 1488-90. 37. Zhu, M.L. and N. Kyprianou, Androgen receptor and growth factor signaling cross-talk in prostate cancer cells. Endocr Relat Cancer, 2008. 15(4): p. 841-9. 38. Bernardo, G.M. and R.A. Keri, FOXA1: a transcription factor with parallel functions in development and cancer. Biosci Rep. 32(2): p. 113-30. 39. Sahu, B., et al., Dual role of FoxA1 in androgen receptor binding to chromatin, androgen signalling and prostate cancer. Embo J. 30(19): p. 3962-76. 40. Lupien, M., et al., FoxA1 translates epigenetic signatures into enhancer-driven lineage-specific transcription. Cell, 2008. 132(6): p. 958-70. 41. Eeckhoute, J., et al., Cell-type selective chromatin remodeling defines the active subset of FOXA1-bound enhancers. Genome Res, 2009. 19(3): p. 372-80. 42. Schulz, W.A. and M.J. Hoffmann, Epigenetic mechanisms in the biology of prostate cancer. Semin Cancer Biol, 2009. 19(3): p. 172-80. 43. http://www.dnatestingcentre.com/PSA-kit.htm. 44. Pereira MartinsI, A., et al., Performance of PSA and of PSA density in the diagnosis of prostate carcinoma. Acta Cirúrgica Brasileira, 2002. 17: p. 7-11. 45. http://www.uspreventiveservicestaskforce.org/prostatecancerscreening/pro statefinalrs.htm. 46. Thompson, I.M., et al., Prevalence of prostate cancer among men with a prostate-specific antigen level < or =4.0 ng per milliliter. N Engl J Med, 2004. 350(22): p. 2239-46. 47. http://prostate-cancer.org/the-gleason-score-a-significant-biologic-manifest ation-of-prostate-cancer-aggressiveness-on-biopsy/. 48. Nacusi, L.P. and D.J. Tindall, Androgen receptor abnormalities in castration-recurrent prostate cancer. Expert Rev Endocrinol Metab, 2009. 4(5): p. 417-422. 49. Shiota, M., et al., Androgen receptor cofactors in prostate cancer: potential therapeutic targets of castration-resistant prostate cancer. Curr Cancer Drug Targets. 11(7): p. 870-81.

171 50. Cheng, H., et al., Short hairpin RNA knockdown of the androgen receptor attenuates ligand-independent activation and delays tumor progression. Cancer Res, 2006. 66(21): p. 10613-20. 51. Knudsen, K.E. and T.M. Penning, Partners in crime: deregulation of AR activity and androgen synthesis in prostate cancer. Trends Endocrinol Metab. 21(5): p. 315-24. 52. Nadiminty, N. and A.C. Gao, Mechanisms of persistent activation of the androgen receptor in CRPC: recent advances and future perspectives. World J Urol. 30(3): p. 287-95. 53. Han, G., et al., Mutation of the androgen receptor causes oncogenic transformation of the prostate. Proc Natl Acad Sci U S A, 2005. 102(4): p. 1151-6. 54. Steinkamp, M.P., et al., Treatment-dependent androgen receptor mutations in prostate cancer exploit multiple mechanisms to evade therapy. Cancer Res, 2009. 69(10): p. 4434-42. 55. Sharma, V., et al., A novel Arg615Ser mutation of androgen receptor DNA-binding domain in three 46,XY sisters with complete androgen insensitivity syndrome and bilateral inguinal hernia. Fertil Steril. 95(2): p. 804 e19-21. 56. Korpal, M., et al., An F876L mutation in androgen receptor confers genetic and phenotypic resistance to MDV3100 (enzalutamide). Cancer Discov. 3(9): p. 1030-43. 57. Tindal, D.J., Recent Advances in Prostate Cancer: Basic Science Discoveries and Clinical Advances. 2011. 58. Dehm, S.M. and D.J. Tindall, Alternatively spliced androgen receptor variants. Endocr Relat Cancer. 18(5): p. R183-96. 59. Watson, P.A., et al., Constitutively active androgen receptor splice variants expressed in castration-resistant prostate cancer require full-length androgen receptor. Proc Natl Acad Sci U S A. 107(39): p. 16759-65. 60. Koivisto, P., et al., Androgen receptor gene amplification: a possible molecular mechanism for androgen deprivation therapy failure in prostate cancer. Cancer Res, 1997. 57(2): p. 314-9. 61. Heinlein, C.A. and C. Chang, Androgen receptor (AR) coregulators: an overview. Endocr Rev, 2002. 23(2): p. 175-200. 62. Heemers, H.V. and D.J. Tindall, Androgen receptor (AR) coregulators: a diversity of functions converging on and regulating the AR transcriptional complex. Endocr Rev, 2007. 28(7): p. 778-808. 63. Goetz, M.P., et al., The Hsp90 chaperone complex as a novel target for cancer therapy. Ann Oncol, 2003. 14(8): p. 1169-76. 64. Lamoureux, F., et al., A novel HSP90 inhibitor delays castrate-resistant prostate cancer without altering serum PSA levels and inhibits osteoclastogenesis. Clin Cancer Res. 17(8): p. 2301-13. 65. Chen, H., et al., Nuclear receptor coactivator ACTR is a novel histone acetyltransferase and forms a multimeric activation complex with P/CAF and CBP/p300. Cell, 1997. 90(3): p. 569-80. 66. Kim, Y.R., et al., Differential CARM1 expression in prostate and colorectal cancers. BMC Cancer. 10: p. 197. 67. Hurtado, A., et al., FOXA1 is a key determinant of estrogen receptor function and endocrine response. Nat Genet. 43(1): p. 27-33. 68. Baniahmad, A., Nuclear hormone receptor co-repressors. J Steroid Biochem Mol Biol, 2005. 93(2-5): p. 89-97. 69. Ketchart, W., et al., HEXIM1 is a critical determinant of the response to tamoxifen. Oncogene. 30(33): p. 3563-9. 70. Bitting, R.L. and A.J. Armstrong, Targeting the PI3K/Akt/mTOR pathway in castration-resistant prostate cancer. Endocr Relat Cancer. 20(3): p. R83-99.

172 71. Shaw, R.J. and L.C. Cantley, Ras, PI(3)K and mTOR signalling controls tumour cell growth. Nature, 2006. 441(7092): p. 424-30. 72. Wegiel, B., et al., Molecular pathways in the progression of hormone-independent and metastatic prostate cancer. Curr Cancer Drug Targets. 10(4): p. 392-401. 73. van der Steen, T., D.J. Tindall, and H. Huang, Posttranslational modification of the androgen receptor in prostate cancer. Int J Mol Sci. 14(7): p. 14833-59. 74. Gioeli, D., et al., Androgen receptor phosphorylation. Regulation and identification of the phosphorylation sites. J Biol Chem, 2002. 277(32): p. 29304-14. 75. Sherwood, E.R., et al., Epidermal growth factor receptor activation in androgen-independent but not androgen-stimulated growth of human prostatic carcinoma cells. Br J Cancer, 1998. 77(6): p. 855-61. 76. Mahajan, N.P., et al., Activated Cdc42-associated kinase Ack1 promotes prostate cancer progression via androgen receptor tyrosine phosphorylation. Proc Natl Acad Sci U S A, 2007. 104(20): p. 8438-43. 77. Seton-Rogers, S., Resistance is (hopefully) futile. NATURE REVIEWS/Cancer, 2009. 9. 78. Xu, K., et al., Regulation of androgen receptor transcriptional activity and specificity by RNF6-induced ubiquitination. Cancer Cell, 2009. 15(4): p. 270-82. 79. Kaikkonen, S., et al., SUMO-specific protease 1 (SENP1) reverses the hormone-augmented SUMOylation of androgen receptor and modulates gene responses in prostate cancer cells. Mol Endocrinol, 2009. 23(3): p. 292-307. 80. Wang, H., et al., Purification and functional characterization of a histone H3-lysine 4-specific methyltransferase. Mol Cell, 2001. 8(6): p. 1207-17. 81. Ellinger, J., et al., Global levels of histone modifications predict prostate cancer recurrence. Prostate. 70(1): p. 61-9. 82. Hublitz, P., M. Albert, and A.H. Peters, Mechanisms of transcriptional repression by histone lysine methylation. Int J Dev Biol, 2009. 53(2-3): p. 335-54. 83. Blair, L.P. and Q. Yan, Epigenetic mechanisms in commonly occurring cancers. DNA Cell Biol. 31 Suppl 1: p. S49-61. 84. Schulz, W.A. and J. Hatina, Epigenetics of prostate cancer: beyond DNA methylation. J Cell Mol Med, 2006. 10(1): p. 100-25. 85. Kooistra, S.M. and K. Helin, Molecular mechanisms and potential functions of histone demethylases. Nat Rev Mol Cell Biol. 13(5): p. 297-311. 86. Yamane, K., et al., PLU-1 is an H3K4 demethylase involved in transcriptional repression and breast cancer cell proliferation. Mol Cell, 2007. 25(6): p. 801-12. 87. Breast Cancer Research 2010. BioMed Central, 2010. 12. 88. Scibetta, A.G., et al., Functional analysis of the transcription repressor PLU-1/JARID1B. Mol Cell Biol, 2007. 27(20): p. 7220-35. 89. Roesch, A., et al., RBP2-H1/JARID1B is a transcriptional regulator with a tumor suppressive potential in melanoma cells. Int J Cancer, 2008. 122(5): p. 1047-57. 90. Li, Q., et al., Binding of the JmjC demethylase JARID1B to LSD1/NuRD suppresses angiogenesis and metastasis in breast cancer cells by repressing chemokine CCL14. Cancer Res. 71(21): p. 6899-908. 91. Chow, J.P., R.Y. Poon, and H.T. Ma, Inhibitory phosphorylation of cyclin-dependent kinase 1 as a compensatory mechanism for mitosis exit. Mol Cell Biol. 31(7): p. 1478-91. 92. Chen, Z., et al., Phospho-MED1-enhanced UBE2C locus looping drives castration-resistant prostate cancer growth. Embo J. 30(12): p. 2405-19.

173 93. Wicha, M.S., PSA lo and behold: prostate cancer stem cells. Cell Stem Cell. 10(5): p. 482-3. 94. Qin, J., et al., The PSA(-/lo) prostate cancer cell population harbors self-renewing long-term tumor-propagating cells that resist castration. Cell Stem Cell. 10(5): p. 556-69. 95. James R. Marthick, A.F.H.a.J.L.D., Integrins as Determinants of Genetic Susceptibility, Tumour Behaviour and Their Potential as Therapeutic Targets. Prostate Cancer - From Bench to Bedside, 2011. 96. Mohler, M.L., et al., Androgen receptor antagonists: a patent review (2008-2011). Expert Opin Ther Pat. 22(5): p. 541-65. 97. Garner, C., Uses of GnRH agonists. J Obstet Gynecol Neonatal Nurs, 1994. 23(7): p. 563-70. 98. Balk, S., Androgen metabolism in progression to androgen-independent prostate cancer. Beth Israel Deaconess Medical Center, Inc. Boston, MA 02215, 2011. 99. Audenet, F., et al., [CYP17A1 inhibitors in prostate cancer: mechanisms of action independent of the androgenic pathway]. Prog Urol. 23 Suppl 1: p. S9-15. 100. Culig, Z., et al., Switch from antagonist to agonist of the androgen receptor bicalutamide is associated with prostate tumour progression in a new model system. Br J Cancer, 1999. 81(2): p. 242-51. 101. Rosenfeld, M.G., V.V. Lunyak, and C.K. Glass, Sensors and signals: a coactivator/corepressor/epigenetic code for integrating signal-dependent programs of transcriptional response. Genes Dev, 2006. 20(11): p. 1405-28. 102. Masiello, D., et al., Bicalutamide functions as an androgen receptor antagonist by assembly of a transcriptionally inactive receptor. J Biol Chem, 2002. 277(29): p. 26321-6. 103. Yuan, X. and S.P. Balk, Mechanisms mediating androgen receptor reactivation after castration. Urol Oncol, 2009. 27(1): p. 36-41. 104. Osguthorpe, D.J. and A.T. Hagler, Mechanism of androgen receptor antagonism by bicalutamide in the treatment of prostate cancer. Biochemistry. 50(19): p. 4105-13. 105. Kusuhara, M., et al., Cloning of Hexamethylene-bis-acetamide-inducible Transcript, HEXIM1, in Human Vascular Smooth Muscle Cells. Biomed Res, 1999. 20(5): p. 273-279. 106. Li, Q., et al., Analysis of the large inactive P-TEFb complex indicates that it contains one 7SK molecule, a dimer of HEXIM1 or HEXIM2, and two P-TEFb molecules containing Cdk9 phosphorylated at threonine 186. J Biol Chem, 2005. 280(31): p. 28819-26. 107. Michels, A.A., et al., Binding of the 7SK snRNA turns the HEXIM1 protein into a P-TEFb (CDK9/cyclin T) inhibitor. Embo J, 2004. 23(13): p. 2608-19. 108. Yik, J.H., et al., Inhibition of P-TEFb (CDK9/Cyclin T) kinase and RNA polymerase II transcription by the coordinated actions of HEXIM1 and 7SK snRNA. Mol Cell, 2003. 12(4): p. 971-82. 109. Van Lint, C., S. Bouchat, and A. Marcello, HIV-1 transcription and latency: an update. Retrovirology. 10: p. 67. 110. Fraldi, A., et al., Inhibition of Tat activity by the HEXIM1 protein. Retrovirology, 2005. 2: p. 42. 111. Barboric, M., et al., Tat competes with HEXIM1 to increase the active pool of P-TEFb for HIV-1 transcription. Nucleic Acids Res, 2007. 35(6): p. 2003-12. 112. Mascareno, E.J., et al., Hexim-1 modulates androgen receptor and the TGF-beta signaling during the progression of prostate cancer. Prostate. 72(9): p. 1035-44. 113. Gordon, V., et al., CDK9 regulates AR promoter selectivity and cell growth through serine 81 phosphorylation. Mol Endocrinol. 24(12): p. 2267-80.

174 114. Wang, C., et al., mCOPA: analysis of heterogeneous features in cancer expression data. J Clin Bioinforma. 2(1): p. 22. 115. Dehm, S.M. and D.J. Tindall, Androgen receptor structural and functional elements: role and regulation in prostate cancer. Mol Endocrinol, 2007. 21(12): p. 2855-63. 116. Devlin, H.L. and M. Mudryj, Progression of prostate cancer: multiple pathways to androgen independence. Cancer Lett, 2009. 274(2): p. 177-86. 117. Wittmann, B.M., et al., The breast cell growth inhibitor, estrogen down regulated gene 1, modulates a novel functional interaction between estrogen receptor alpha and transcriptional elongation factor cyclin T1. Oncogene, 2005. 24(36): p. 5576-88. 118. Garriga, J. and X. Grana, Cellular control of gene expression by T-type cyclin/CDK9 complexes. Gene, 2004. 337: p. 15-23. 119. Ogba, N., et al., HEXIM1 regulates 17beta-estradiol/estrogen receptor-alpha-mediated expression of cyclin D1 in mammary cells via modulation of P-TEFb. Cancer Res, 2008. 68(17): p. 7015-24. 120. Ogba, N., et al., HEXIM1 modulates vascular endothelial growth factor expression and function in breast epithelial cells and mammary gland. Oncogene. 29(25): p. 3639-49. 121. Wang Q, et al., Androgen receptor regulates a distinct transcription program in androgen-independent prostate cancer. Cell, 2009. 138: p. 245-256. 122. Thalmann, G.N., et al., Androgen-independent cancer progression and bone metastasis in the LNCaP model of human prostate cancer. Cancer Res, 1994. 54(10): p. 2577-81. 123. Song, K., et al., DHT selectively reverses Smad3-mediated/TGF-beta-induced responses through transcriptional down-regulation of Smad3 in prostate epithelial cells. Mol Endocrinol. 24(10): p. 2019-29. 124. Montano, M.M., et al., An estrogen receptor-selective coregulator that potentiates the effectiveness of antiestrogens and represses the activity of estrogens. Proc Natl Acad Sci U S A, 1999. 96(12): p. 6947-52. 125. Wittmann, B.M., N. Wang, and M.M. Montano, Identification of a novel inhibitor of breast cell growth that is down-regulated by estrogens and decreased in breast tumors. Cancer Res, 2003. 63(16): p. 5151-8. 126. Ketchart, W., et al., Inhibition of metastasis by HEXIM1 through effects on cell invasion and angiogenesis. Oncogene. 32(33): p. 3829-39. 127. Montano, M.M., et al., Mutation of the HEXIM1 gene results in defects during heart and vascular development partly through downregulation of vascular endothelial growth factor. Circ Res, 2008. 102(4): p. 415-22. 128. Ross, R.W., et al., Efficacy of androgen deprivation therapy (ADT) in patients with advanced prostate cancer: association between Gleason score, prostate-specific antigen level, and prior ADT exposure with duration of ADT effect. Cancer, 2008. 112(6): p. 1247-53. 129. Shibata, Y., et al., Impact of pre-treatment prostate tissue androgen content on the prediction of castration-resistant prostate cancer development in patients treated with primary androgen deprivation therapy. Andrology. 1(3): p. 505-11. 130. Tomlins, S.A., et al., Integrative molecular concept modeling of prostate cancer progression. Nat Genet, 2007. 39(1): p. 41-51. 131. Varambally, S., et al., Integrative genomic and proteomic analysis of prostate cancer reveals signatures of metastatic progression. Cancer Cell, 2005. 8(5): p. 393-406. 132. Yu, Y.P., et al., Gene expression alterations in prostate cancer predicting tumor aggression and preceding development of malignancy. J Clin Oncol, 2004. 22(14): p. 2790-9.

175 133. Xiang, Y., et al., JARID1B is a histone H3 lysine 4 demethylase up-regulated in prostate cancer. Proc Natl Acad Sci U S A, 2007. 104(49): p. 19226-31. 134. Vicent, G.P., et al., Four enzymes cooperate to displace histone H1 during the first minute of hormonal gene activation. Genes Dev. 25(8): p. 845-62. 135. Chen, Z., et al., Phospho-MED1-enhanced UBE2C locus looping drives castration-resistant prostate cancer growth. EMBO J, 2011. 30(12): p. 2405-19. 136. Cannata, D.H., A. Kirschenbaum, and A.C. Levine, Androgen deprivation therapy as primary treatment for prostate cancer. J Clin Endocrinol Metab. 97(2): p. 360-5. 137. Kirby, M., C. Hirst, and E.D. Crawford, Characterising the castration-resistant prostate cancer population: a systematic review. Int J Clin Pract. 65(11): p. 1180-92. 138. Basu, S. and D.J. Tindall, Androgen action in prostate cancer. Horm Cancer. 1(5): p. 223-8. 139. Lee, D.K. and C. Chang, Molecular communication between androgen receptor and general transcription machinery. J Steroid Biochem Mol Biol, 2003. 84(1): p. 41-9. 140. Lee, D.K., H.O. Duan, and C. Chang, Androgen receptor interacts with the positive elongation factor P-TEFb and enhances the efficiency of transcriptional elongation. J Biol Chem, 2001. 276(13): p. 9978-84. 141. Byers, S.A., et al., HEXIM2, a HEXIM1-related protein, regulates positive transcription elongation factor b through association with 7SK. J Biol Chem, 2005. 280(16): p. 16360-7. 142. Dou, Y., et al., Physical association and coordinate function of the H3 K4 methyltransferase MLL1 and the H4 K16 acetyltransferase MOF. Cell, 2005. 121(6): p. 873-85. 143. Pekowska, A., et al., H3K4 tri-methylation provides an epigenetic signature of active enhancers. Embo J. 30(20): p. 4198-210. 144. Pirngruber, J., et al., CDK9 directs H2B monoubiquitination and controls replication-dependent histone mRNA 3'-end processing. EMBO Rep, 2009. 10(8): p. 894-900. 145. Pirngruber, J., A. Shchebet, and S.A. Johnsen, Insights into the function of the human P-TEFb component CDK9 in the regulation of chromatin modifications and co-transcriptional mRNA processing. Cell Cycle, 2009. 8(22): p. 3636-42. 146. Zippo, A., et al., Histone crosstalk between H3S10ph and H4K16ac generates a histone code that mediates transcription elongation. Cell, 2009. 138(6): p. 1122-36. 147. Chandrasekharan, M.B., F. Huang, and Z.W. Sun, Histone H2B ubiquitination and beyond: Regulation of nucleosome stability, chromatin dynamics and the trans-histone H3 methylation. Epigenetics. 5(6): p. 460-8. 148. Nakanishi, S., et al., Histone H2BK123 monoubiquitination is the critical determinant for H3K4 and H3K79 trimethylation by COMPASS and Dot1. J Cell Biol, 2009. 186(3): p. 371-7. 149. Roesch, A., et al., Re-expression of the retinoblastoma-binding protein 2-homolog 1 reveals tumor-suppressive functions in highly metastatic melanoma cells. J Invest Dermatol, 2006. 126(8): p. 1850-9. 150. Roesch, A., et al., Retinoblastoma-binding protein 2-homolog 1: a retinoblastoma-binding protein downregulated in malignant melanomas. Mod Pathol, 2005. 18(9): p. 1249-57. 151. Mulholland, D.J., et al., Cell autonomous role of PTEN in regulating castration-resistant prostate cancer growth. Cancer Cell. 19(6): p. 792-804. 152. Wei, Y., et al., CDK1-dependent phosphorylation of EZH2 suppresses methylation of H3K27 and promotes osteogenic differentiation of human mesenchymal stem cells. Nat Cell Biol. 13(1): p. 87-94.

176 153. Chng, K.R., et al., A transcriptional repressor co-regulatory network governing androgen response in prostate cancers. Embo J. 31(12): p. 2810-23. 154. Wang, H., et al., CCI-779 inhibits cell-cycle G2-M progression and invasion of castration-resistant prostate cancer via attenuation of UBE2C transcription and mRNA stability. Cancer Res. 71(14): p. 4866-76. 155. Shimizu, N., et al., HEXIM1 forms a transcriptionally abortive complex with glucocorticoid receptor without involving 7SK RNA and positive transcription elongation factor b. Proc Natl Acad Sci U S A, 2005. 102(24): p. 8555-60. 156. Yeh, I.J., et al., HEXIM1 Plays a Critical Role in the Inhibition of the Androgen Receptor by Antiandrogens. Biochem J. 157. Arora, V.K., et al., Glucocorticoid receptor confers resistance to antiandrogens by bypassing androgen receptor blockade. Cell. 155(6): p. 1309-22. 158. Donovan, L., et al., Hypoxia--implications for pharmaceutical developments. Sleep Breath. 14(4): p. 291-8. 159. Kambiz Gilany*, M.V., Hypoxia: a Review. Journal of Paramedical Sciences (JPS), 2010. 1(2). 160. Peers, C., The G. L. Brown Prize Lecture. Hypoxic regulation of ion channel function and expression. Exp Physiol, 2002. 87(4): p. 413-22. 161. Lopez-Barneo, J., et al., Regulation of oxygen sensing by ion channels. J Appl Physiol (1985), 2004. 96(3): p. 1187-95; discussion 1170-2. 162. Dehne, N. and B. Brune, Sensors, Transmitters, and Targets in Mitochondrial Oxygen Shortage-A Hypoxia-Inducible Factor Relay Story. Antioxid Redox Signal. 163. Guzy, R.D. and P.T. Schumacker, Oxygen sensing by mitochondria at complex III: the paradox of increased reactive oxygen species during hypoxia. Exp Physiol, 2006. 91(5): p. 807-19. 164. Hagen, T., Oxygen versus Reactive Oxygen in the Regulation of HIF-1alpha: The Balance Tips. Biochem Res Int. 2012: p. 436981. 165. Ryter, S.W., et al., Heme oxygenase/carbon monoxide signaling pathways: regulation and functional significance. Mol Cell Biochem, 2002. 234-235(1-2): p. 249-63. 166. Myllyharju, J. and E. Schipani, Extracellular matrix genes as hypoxia-inducible targets. Cell Tissue Res. 339(1): p. 19-29. 167. Gu, Y.Z., et al., Molecular characterization and chromosomal localization of a third alpha-class hypoxia inducible factor subunit, HIF3alpha. Gene Expr, 1998. 7(3): p. 205-13. 168. Mabjeesh, N.J. and S. Amir, Hypoxia-inducible factor (HIF) in human tumorigenesis. Histol Histopathol, 2007. 22(5): p. 559-72. 169. Metzen, E. and P.J. Ratcliffe, HIF hydroxylation and cellular oxygen sensing. Biol Chem, 2004. 385(3-4): p. 223-30. 170. Brahimi-Horn, C., N. Mazure, and J. Pouyssegur, Signalling via the hypoxia-inducible factor-1alpha requires multiple posttranslational modifications. Cell Signal, 2005. 17(1): p. 1-9. 171. Semenza, G.L., Hydroxylation of HIF-1: oxygen sensing at the molecular level. Physiology (Bethesda), 2004. 19: p. 176-82. 172. Huang, L.E. and H.F. Bunn, Hypoxia-inducible factor and its biomedical relevance. J Biol Chem, 2003. 278(22): p. 19575-8. 173. Pugh, C.W. and P.J. Ratcliffe, The von Hippel-Lindau tumor suppressor, hypoxia-inducible factor-1 (HIF-1) degradation, and cancer pathogenesis. Semin Cancer Biol, 2003. 13(1): p. 83-9. 174. Tanimoto, K., et al., Mechanism of regulation of the hypoxia-inducible factor-1 alpha by the von Hippel-Lindau tumor suppressor protein. Embo J, 2000. 19(16): p. 4298-309.

177 175. Yu, F., et al., Dynamic, site-specific interaction of hypoxia-inducible factor-1alpha with the von Hippel-Lindau tumor suppressor protein. Cancer Res, 2001. 61(10): p. 4136-42. 176. Chun, Y.S., M.S. Kim, and J.W. Park, Oxygen-dependent and -independent regulation of HIF-1alpha. J Korean Med Sci, 2002. 17(5): p. 581-8. 177. Stiehl, D.P., et al., Increased prolyl 4-hydroxylase domain proteins compensate for decreased oxygen levels. Evidence for an autoregulatory oxygen-sensing system. J Biol Chem, 2006. 281(33): p. 23482-91. 178. Henze, A.T., et al., Prolyl hydroxylases 2 and 3 act in gliomas as protective negative feedback regulators of hypoxia-inducible factors. Cancer Res. 70(1): p. 357-66. 179. Berra, E., et al., HIF prolyl-hydroxylase 2 is the key oxygen sensor setting low steady-state levels of HIF-1alpha in normoxia. Embo J, 2003. 22(16): p. 4082-90. 180. Appelhoff, R.J., et al., Differential function of the prolyl hydroxylases PHD1, PHD2, and PHD3 in the regulation of hypoxia-inducible factor. J Biol Chem, 2004. 279(37): p. 38458-65. 181. Chan, D.A., et al., Coordinate regulation of the oxygen-dependent degradation domains of hypoxia-inducible factor 1 alpha. Mol Cell Biol, 2005. 25(15): p. 6415-26. 182. Schofield, C.J. and P.J. Ratcliffe, Oxygen sensing by HIF hydroxylases. Nat Rev Mol Cell Biol, 2004. 5(5): p. 343-54. 183. Mahon, P.C., K. Hirota, and G.L. Semenza, FIH-1: a novel protein that interacts with HIF-1alpha and VHL to mediate repression of HIF-1 transcriptional activity. Genes Dev, 2001. 15(20): p. 2675-86. 184. Ravi, R., et al., Regulation of tumor angiogenesis by p53-induced degradation of hypoxia-inducible factor 1alpha. Genes Dev, 2000. 14(1): p. 34-44. 185. Bae, M.K., et al., Jab1 interacts directly with HIF-1alpha and regulates its stability. J Biol Chem, 2002. 277(1): p. 9-12. 186. Koh, M.Y., B.G. Darnay, and G. Powis, Hypoxia-associated factor, a novel E3-ubiquitin ligase, binds and ubiquitinates hypoxia-inducible factor 1alpha, leading to its oxygen-independent degradation. Mol Cell Biol, 2008. 28(23): p. 7081-95. 187. http://www.pearsonhighered.com/mathews/ch07/c07mobhp.htm. 188. Ke, Q. and M. Costa, Hypoxia-inducible factor-1 (HIF-1). Mol Pharmacol, 2006. 70(5): p. 1469-80. 189. Metzen, E., et al., Nitric oxide impairs normoxic degradation of HIF-1alpha by inhibition of prolyl hydroxylases. Mol Biol Cell, 2003. 14(8): p. 3470-81. 190. Hagen, T., et al., Redistribution of intracellular oxygen in hypoxia by nitric oxide: effect on HIF1alpha. Science, 2003. 302(5652): p. 1975-8. 191. Semenza, G.L., Intratumoral hypoxia, radiation resistance, and HIF-1. Cancer Cell, 2004. 5(5): p. 405-6. 192. Bardos, J.I., N.M. Chau, and M. Ashcroft, Growth factor-mediated induction of HDM2 positively regulates hypoxia-inducible factor 1alpha expression. Mol Cell Biol, 2004. 24(7): p. 2905-14. 193. Fukuda, R., et al., Insulin-like growth factor 1 induces hypoxia-inducible factor 1-mediated vascular endothelial growth factor expression, which is dependent on MAP kinase and phosphatidylinositol 3-kinase signaling in colon cancer cells. J Biol Chem, 2002. 277(41): p. 38205-11. 194. Laughner, E., et al., HER2 (neu) signaling increases the rate of hypoxia-inducible factor 1alpha (HIF-1alpha) synthesis: novel mechanism for HIF-1-mediated vascular endothelial growth factor expression. Mol Cell Biol, 2001. 21(12): p. 3995-4004.

178 195. Liu, R., et al., Mechanism of cancer cell adaptation to metabolic stress: proteomics identification of a novel thyroid hormone-mediated gastric carcinogenic signaling pathway. Mol Cell Proteomics, 2009. 8(1): p. 70-85. 196. Kelly, T.J., et al., A hypoxia-induced positive feedback loop promotes hypoxia-inducible factor 1alpha stability through miR-210 suppression of glycerol-3-phosphate dehydrogenase 1-like. Mol Cell Biol. 31(13): p. 2696-706. 197. Katschinski, D.M., et al., Interaction of the PAS B domain with HSP90 accelerates hypoxia-inducible factor-1alpha stabilization. Cell Physiol Biochem, 2004. 14(4-6): p. 351-60. 198. Bohonowych, J.E., et al., Comparative analysis of novel and conventional Hsp90 inhibitors on HIF activity and angiogenic potential in clear cell renal cell carcinoma: implications for clinical evaluation. BMC Cancer. 11: p. 520. 199. Gogate, S.S., et al., Tonicity enhancer binding protein (TonEBP) and hypoxia-inducible factor (HIF) coordinate heat shock protein 70 (Hsp70) expression in hypoxic nucleus pulposus cells: role of Hsp70 in HIF-1alpha degradation. J Bone Miner Res. 27(5): p. 1106-17. 200. Millson, P.W.P.a.S.H., Mechanisms of Resistance to Hsp90 Inhibitor Drugs: A Complex Mosaic Emerges. Pharmaceuticals, 2011. 4: p. 1400-1422. 201. Abdollahi, H., et al., The role of hypoxia in stem cell differentiation and therapeutics. J Surg Res. 165(1): p. 112-7. 202. Liu, Y.V., et al., RACK1 competes with HSP90 for binding to HIF-1alpha and is required for O(2)-independent and HSP90 inhibitor-induced degradation of HIF-1alpha. Mol Cell, 2007. 25(2): p. 207-17. 203. Moll, U.M. and N. Slade, p63 and p73: roles in development and tumor formation. Mol Cancer Res, 2004. 2(7): p. 371-86. 204. Montagner, M., et al., SHARP1 suppresses breast cancer metastasis by promoting degradation of hypoxia-inducible factors. Nature. 487(7407): p. 380-4. 205. Piccolo, S., E. Enzo, and M. Montagner, p63, Sharp1, and HIFs: master regulators of metastasis in triple-negative breast cancer. Cancer Res. 73(16): p. 4978-81. 206. Yoo, Y.G., G. Kong, and M.O. Lee, Metastasis-associated protein 1 enhances stability of hypoxia-inducible factor-1alpha protein by recruiting histone deacetylase 1. Embo J, 2006. 25(6): p. 1231-41. 207. Arnesen, T., et al., Interaction between HIF-1 alpha (ODD) and hARD1 does not induce acetylation and destabilization of HIF-1 alpha. FEBS Lett, 2005. 579(28): p. 6428-32. 208. Pelicci, M.a., HIF1a and ARD1: enemies, friends or neither? Nature Reviews Cancer, 2006. 6. 209. Lim, J.H., et al., Sirtuin 1 modulates cellular responses to hypoxia by deacetylating hypoxia-inducible factor 1alpha. Mol Cell. 38(6): p. 864-78. 210. Qian, D.Z., et al., Class II histone deacetylases are associated with VHL-independent regulation of hypoxia-inducible factor 1 alpha. Cancer Res, 2006. 66(17): p. 8814-21. 211. Kang, X., et al., PIASy stimulates HIF1alpha SUMOylation and negatively regulates HIF1alpha activity in response to hypoxia. Oncogene. 29(41): p. 5568-78. 212. Cai, Q., et al., Hypoxia inactivates the VHL tumor suppressor through PIASy-mediated SUMO modification. PLoS One. 5(3): p. e9720. 213. Cheng, J., et al., SUMO-specific protease 1 is essential for stabilization of HIF1alpha during hypoxia. Cell, 2007. 131(3): p. 584-95. 214. Zhang, Y., et al., Negative regulation of HDM2 to attenuate p53 degradation by ribosomal protein L26. Nucleic Acids Res. 38(19): p. 6544-54.

179 215. Sodhi, A., et al., The Kaposi's sarcoma-associated herpes virus G protein-coupled receptor up-regulates vascular endothelial growth factor expression and secretion through mitogen-activated protein kinase and p38 pathways acting on hypoxia-inducible factor 1alpha. Cancer Res, 2000. 60(17): p. 4873-80. 216. Yasinska, I.M. and V.V. Sumbayev, S-nitrosation of Cys-800 of HIF-1alpha protein activates its interaction with p300 and stimulates its transcriptional activity. FEBS Lett, 2003. 549(1-3): p. 105-9. 217. Haase, V.H., Regulation of erythropoiesis by hypoxia-inducible factors. Blood Rev. 27(1): p. 41-53. 218. Gustafsson, M.V., et al., Hypoxia requires notch signaling to maintain the undifferentiated cell state. Dev Cell, 2005. 9(5): p. 617-28. 219. Greer, S.N., et al., The updated biology of hypoxia-inducible factor. Embo J. 31(11): p. 2448-60. 220. Johnson, E.A., HIF takes it up a notch. Sci Signal. 4(181): p. pe33. 221. Covello, K.L., et al., HIF-2alpha regulates Oct-4: effects of hypoxia on stem cell function, embryonic development, and tumor growth. Genes Dev, 2006. 20(5): p. 557-70. 222. Gordan, J.D., et al., HIF-2alpha promotes hypoxic cell proliferation by enhancing c-myc transcriptional activity. Cancer Cell, 2007. 11(4): p. 335-47. 223. Semenza, G.L., Targeting HIF-1 for cancer therapy. Nat Rev Cancer, 2003. 3(10): p. 721-32. 224. Bartrons, R. and J. Caro, Hypoxia, glucose metabolism and the Warburg's effect. J Bioenerg Biomembr, 2007. 39(3): p. 223-9. 225. Rabiya Majeed, A.H., Yasrib Qurishi, Asif Khurshid Qazi, Aashiq Hussain, Mudassier Ahmed, Rauf Ahmad Najar, Javeed Ahmad Bhat, Shashank Kumar Singh and Ajit Kumar Saxena, Therapeutic Targeting of Cancer Cell Metabolism: Role of Metabolic Enzymes, Oncogenes and Tumor Suppressor Genes. J Cancer Sci Ther, 2012. 4(9): p. 281-291. 226. Byrne, A.M., D.J. Bouchier-Hayes, and J.H. Harmey, Angiogenic and cell survival functions of vascular endothelial growth factor (VEGF). J Cell Mol Med, 2005. 9(4): p. 777-94. 227. Hoeben, A., et al., Vascular endothelial growth factor and angiogenesis. Pharmacol Rev, 2004. 56(4): p. 549-80. 228. Ferrara, N., H.P. Gerber, and J. LeCouter, The biology of VEGF and its receptors. Nat Med, 2003. 9(6): p. 669-76. 229. Kaur, B., et al., Hypoxia and the hypoxia-inducible-factor pathway in glioma growth and angiogenesis. Neuro Oncol, 2005. 7(2): p. 134-53. 230. Piret, J.P., et al., Is HIF-1alpha a pro- or an anti-apoptotic protein? Biochem Pharmacol, 2002. 64(5-6): p. 889-92. 231. Carmeliet, P., et al., Role of HIF-1alpha in hypoxia-mediated apoptosis, cell proliferation and tumour angiogenesis. Nature, 1998. 394(6692): p. 485-90. 232. Goda, N., S.J. Dozier, and R.S. Johnson, HIF-1 in cell cycle regulation, apoptosis, and tumor progression. Antioxid Redox Signal, 2003. 5(4): p. 467-73. 233. Cory, S. and J.M. Adams, The Bcl2 family: regulators of the cellular life-or-death switch. Nat Rev Cancer, 2002. 2(9): p. 647-56. 234. Guo, K., et al., Hypoxia induces the expression of the pro-apoptotic gene BNIP3. Cell Death Differ, 2001. 8(4): p. 367-76. 235. Kim, J.Y. and J.H. Park, ROS-dependent caspase-9 activation in hypoxic cell death. FEBS Lett, 2003. 549(1-3): p. 94-8. 236. Favaro, E., et al., Hypoxia inducible factor-1alpha inactivation unveils a link between tumor cell metabolism and hypoxia-induced cell death. Am J Pathol, 2008. 173(4): p. 1186-201.

180 237. Ahmed, A., Regulation of p53--dependent cell death responses in normoxia and hypoxia. University College London, 2011. 238. Brahimi-Horn, C. and J. Pouyssegur, The role of the hypoxia-inducible factor in tumor metabolism growth and invasion. Bull Cancer, 2006. 93(8): p. E73-80. 239. Rankin, E.B. and A.J. Giaccia, The role of hypoxia-inducible factors in tumorigenesis. Cell Death Differ, 2008. 15(4): p. 678-85. 240. Nguyen, D.X., P.D. Bos, and J. Massague, Metastasis: from dissemination to organ-specific colonization. Nat Rev Cancer, 2009. 9(4): p. 274-84. 241. Valastyan, S. and R.A. Weinberg, Tumor metastasis: molecular insights and evolving paradigms. Cell. 147(2): p. 275-92. 242. Paget, S., The distribution of secondary growths in cancer of the breast. 1889. Cancer Metastasis Rev, 1989. 8(2): p. 98-101. 243. Fidler, I.J., The pathogenesis of cancer metastasis: the 'seed and soil' hypothesis revisited. Nat Rev Cancer, 2003. 3(6): p. 453-8. 244. Buijs, J.T., K.R. Stayrook, and T.A. Guise, The role of TGF-beta in bone metastasis: novel therapeutic perspectives. Bonekey Rep. 1: p. 96. 245. Weigelt, B., J.L. Peterse, and L.J. van 't Veer, Breast cancer metastasis: markers and models. Nat Rev Cancer, 2005. 5(8): p. 591-602. 246. Shipitsin, M., et al., Molecular definition of breast tumor heterogeneity. Cancer Cell, 2007. 11(3): p. 259-73. 247. Nishida, N., et al., Angiogenesis in cancer. Vasc Health Risk Manag, 2006. 2(3): p. 213-9. 248. Liao, D. and R.S. Johnson, Hypoxia: a key regulator of angiogenesis in cancer. Cancer Metastasis Rev, 2007. 26(2): p. 281-90. 249. Morello, F., A. Perino, and E. Hirsch, Phosphoinositide 3-kinase signalling in the vascular system. Cardiovasc Res, 2009. 82(2): p. 261-71. 250. Arcaro, A. and A.S. Guerreiro, The phosphoinositide 3-kinase pathway in human cancer: genetic alterations and therapeutic implications. Curr Genomics, 2007. 8(5): p. 271-306. 251. Odemis, V., et al., CXCR7 is an active component of SDF-1 signalling in astrocytes and Schwann cells. J Cell Sci. 123(Pt 7): p. 1081-8. 252. Liekens, S., D. Schols, and S. Hatse, CXCL12-CXCR4 axis in angiogenesis, metastasis and stem cell mobilization. Curr Pharm Des. 16(35): p. 3903-20. 253. Kucia, M., et al., CXCR4-SDF-1 signalling, locomotion, chemotaxis and adhesion. J Mol Histol, 2004. 35(3): p. 233-45. 254. Teicher, B.A. and S.P. Fricker, CXCL12 (SDF-1)/CXCR4 pathway in cancer. Clin Cancer Res. 16(11): p. 2927-31. 255. Nervi, B., et al., Chemosensitization of acute myeloid leukemia (AML) following mobilization by the CXCR4 antagonist AMD3100. Blood, 2009. 113(24): p. 6206-14. 256. Kaiser, P.K., Verteporfin photodynamic therapy and anti-angiogenic drugs: potential for combination therapy in exudative age-related macular degeneration. Curr Med Res Opin, 2007. 23(3): p. 477-87. 257. Grepin, R. and G. Pages, Molecular mechanisms of resistance to tumour anti-angiogenic strategies. J Oncol. 2010: p. 835680. 258. Yeh, I.J., et al., HEXIM1 down-regulates hypoxia-inducible factor-1alpha protein stability. Biochem J. 456(2): p. 195-204. 259. Uchimura, K., et al., HSulf-2, an extracellular endoglucosamine-6-sulfatase, selectively mobilizes heparin-bound growth factors and chemokines: effects on VEGF, FGF-1, and SDF-1. BMC Biochem, 2006. 7: p. 2. 260. Khurana, A., et al., Role of heparan sulfatases in ovarian and breast cancer. Am J Cancer Res. 3(1): p. 34-45. 261. Khurana, A., et al., HSulf-1 modulates FGF2- and hypoxia-mediated migration and invasion of breast cancer cells. Cancer Res. 71(6): p. 2152-61.

181 262. Zhang, Q., et al., Control of cyclin D1 and breast tumorigenesis by the EglN2 prolyl hydroxylase. Cancer Cell, 2009. 16(5): p. 413-24. 263. Peurala, E., et al., Expressions of individual PHDs associate with good prognostic factors and increased proliferation in breast cancer patients. Breast Cancer Res Treat. 133(1): p. 179-88. 264. Liu, Y., et al., Prolyl hydroxylase 3 interacts with Bcl-2 to regulate doxorubicin-induced apoptosis in H9c2 cells. Biochem Biophys Res Commun. 401(2): p. 231-7. 265. Rantanen, K., The dual role of HIF hydroxylase PHD3 in cancer cell survival. Turun Yliopiston Julkaisuja Annales Universitatis Turkuensis, 2012. 266. Walmsley, S.R., et al., Prolyl hydroxylase 3 (PHD3) is essential for hypoxic regulation of neutrophilic inflammation in humans and mice. J Clin Invest. 121(3): p. 1053-63. 267. Schietke, R., et al., The lysyl oxidases LOX and LOXL2 are necessary and sufficient to repress E-cadherin in hypoxia: insights into cellular transformation processes mediated by HIF-1. J Biol Chem. 285(9): p. 6658-69. 268. Barker, H.E., et al., LOXL2-mediated matrix remodeling in metastasis and mammary gland involution. Cancer Res. 71(5): p. 1561-72. 269. abcam, Anti-Pyruvate Kinase antibody (Biotin) ab34574. poduct datasheet. 270. Noguchi, T., H. Inoue, and T. Tanaka, The M1- and M2-type isozymes of rat pyruvate kinase are produced from the same gene by alternative RNA splicing. J Biol Chem, 1986. 261(29): p. 13807-12. 271. Ros, S. and A. Schulze, Balancing glycolytic flux: the role of 6-phosphofructo-2-kinase/fructose 2,6-bisphosphatases in cancer metabolism. Cancer Metab. 1(1): p. 8. 272. Gatenby, R.A. and R.J. Gillies, Why do cancers have high aerobic glycolysis? Nat Rev Cancer, 2004. 4(11): p. 891-9. 273. Harris, A.L., Hypoxia--a key regulatory factor in tumour growth. Nat Rev Cancer, 2002. 2(1): p. 38-47. 274. Christofk, H.R., et al., The M2 splice isoform of pyruvate kinase is important for cancer metabolism and tumour growth. Nature, 2008. 452(7184): p. 230-3. 275. Gupta, V. and R.N. Bamezai, Human pyruvate kinase M2: a multifunctional protein. Protein Sci. 19(11): p. 2031-44. 276. Luo, W., et al., Pyruvate kinase M2 is a PHD3-stimulated coactivator for hypoxia-inducible factor 1. Cell. 145(5): p. 732-44. 277. Kim, J.W., et al., HIF-1-mediated expression of pyruvate dehydrogenase kinase: a metabolic switch required for cellular adaptation to hypoxia. Cell Metab, 2006. 3(3): p. 177-85. 278. Zhang, H., et al., HIF-1 inhibits mitochondrial biogenesis and cellular respiration in VHL-deficient renal cell carcinoma by repression of C-MYC activity. Cancer Cell, 2007. 11(5): p. 407-20. 279. Gogvadze, V., S. Orrenius, and B. Zhivotovsky, Mitochondria in cancer cells: what is so special about them? Trends Cell Biol, 2008. 18(4): p. 165-73. 280. Xu, R.H., et al., Inhibition of glycolysis in cancer cells: a novel strategy to overcome drug resistance associated with mitochondrial respiratory defect and hypoxia. Cancer Res, 2005. 65(2): p. 613-21. 281. Yang, L.H., et al., Survival benefit of tamoxifen in estrogen receptor-negative and progesterone receptor-positive low grade breast cancer patients. J Breast Cancer. 15(3): p. 288-95. 282. Semenza, G.L., Hypoxia-inducible factor 1 (HIF-1) pathway. Sci STKE, 2007. 2007(407): p. cm8. 283. Lundgren, K., C. Holm, and G. Landberg, Hypoxia and breast cancer: prognostic and therapeutic implications. Cell Mol Life Sci, 2007. 64(24): p. 3233-47.

182 284. Dachs, G.U. and G.M. Tozer, Hypoxia modulated gene expression: angiogenesis, metastasis and therapeutic exploitation. Eur J Cancer, 2000. 36(13 Spec No): p. 1649-60. 285. Mohyeldin, A., T. Garzon-Muvdi, and A. Quinones-Hinojosa, Oxygen in stem cell biology: a critical component of the stem cell niche. Cell Stem Cell. 7(2): p. 150-61. 286. Saltz, L.B., et al., Bevacizumab in combination with oxaliplatin-based chemotherapy as first-line therapy in metastatic colorectal cancer: a randomized phase III study. J Clin Oncol, 2008. 26(12): p. 2013-9. 287. Brahimi-Horn, M.C., J. Chiche, and J. Pouyssegur, Hypoxia and cancer. J Mol Med (Berl), 2007. 85(12): p. 1301-7. 288. Mazure, N.M., et al., HIF-1: master and commander of the hypoxic world. A pharmacological approach to its regulation by siRNAs. Biochem Pharmacol, 2004. 68(6): p. 971-80. 289. Bruick, R.K. and S.L. McKnight, A conserved family of prolyl-4-hydroxylases that modify HIF. Science, 2001. 294(5545): p. 1337-40. 290. Epstein, A.C., et al., C. elegans EGL-9 and mammalian homologs define a family of dioxygenases that regulate HIF by prolyl hydroxylation. Cell, 2001. 107(1): p. 43-54. 291. Jeong, J.W., et al., Regulation and destabilization of HIF-1alpha by ARD1-mediated acetylation. Cell, 2002. 111(5): p. 709-20. 292. Masson, N., et al., Independent function of two destruction domains in hypoxia-inducible factor-alpha chains activated by prolyl hydroxylation. Embo J, 2001. 20(18): p. 5197-206. 293. Du, W., et al., Tumor-derived macrophage migration inhibitory factor promotes an autocrine loop that enhances renal cell carcinoma. Oncogene. 32(11): p. 1469-74. 294. Shen, C., et al., Roles of the HIF-1 hypoxia-inducible factor during hypoxia response in Caenorhabditis elegans. J Biol Chem, 2005. 280(21): p. 20580-8. 295. Chan, D.A., et al., Tumor vasculature is regulated by PHD2-mediated angiogenesis and bone marrow-derived cell recruitment. Cancer Cell, 2009. 15(6): p. 527-38. 296. Bilton, R., et al., Arrest-defective-1 protein, an acetyltransferase, does not alter stability of hypoxia-inducible factor (HIF)-1alpha and is not induced by hypoxia or HIF. J Biol Chem, 2005. 280(35): p. 31132-40. 297. Fisher, T.S., et al., Analysis of ARD1 function in hypoxia response using retroviral RNA interference. J Biol Chem, 2005. 280(18): p. 17749-57. 298. Cronin, P.A., J.H. Wang, and H.P. Redmond, Hypoxia increases the metastatic ability of breast cancer cells via upregulation of CXCR4. BMC Cancer. 10: p. 225. 299. Marlow, R., et al., SLITs suppress tumor growth in vivo by silencing Sdf1/Cxcr4 within breast epithelium. Cancer Res, 2008. 68(19): p. 7819-27. 300. Wong, C.C., et al., Hypoxia-inducible factor 1 is a master regulator of breast cancer metastatic niche formation. Proc Natl Acad Sci U S A. 108(39): p. 16369-74. 301. Marxsen, J.H., et al., Hypoxia-inducible factor-1 (HIF-1) promotes its degradation by induction of HIF-alpha-prolyl-4-hydroxylases. Biochem J, 2004. 381(Pt 3): p. 761-7. 302. Yan, M., et al., BRCA1 tumours correlate with a HIF-1alpha phenotype and have a poor prognosis through modulation of hydroxylase enzyme profile expression. Br J Cancer, 2009. 101(7): p. 1168-74. 303. Huang, K.T., et al., DNA methylation analysis of the HIF-1alpha prolyl hydroxylase domain genes PHD1, PHD2, PHD3 and the factor inhibiting HIF gene FIH in invasive breast carcinomas. Histopathology. 57(3): p. 451-60.

183 304. Anderson, K., et al., Regulation of cellular levels of Sprouty2 protein by prolyl hydroxylase domain and von Hippel-Lindau proteins. J Biol Chem. 286(49): p. 42027-36. 305. Yoo, Y.G., et al., Hepatitis B virus X protein induces the expression of MTA1 and HDAC1, which enhances hypoxia signaling in hepatocellular carcinoma cells. Oncogene, 2008. 27(24): p. 3405-13. 306. Geng, H., et al., HIF1alpha protein stability is increased by acetylation at lysine 709. J Biol Chem. 287(42): p. 35496-505. 307. Ellis, L., H. Hammers, and R. Pili, Targeting tumor angiogenesis with histone deacetylase inhibitors. Cancer Lett, 2009. 280(2): p. 145-53. 308. Kang, F.W., et al., Effects of trichostatin A on HIF-1alpha and VEGF expression in human tongue squamous cell carcinoma cells in vitro. Oncol Rep. 28(1): p. 193-9. 309. Naldini, A., et al., Downregulation of hypoxia-related responses by novel antitumor histone deacetylase inhibitors in MDAMB231 breast cancer cells. Anticancer Agents Med Chem. 12(4): p. 407-13. 310. Bodily, J.M., K.P. Mehta, and L.A. Laimins, Human papillomavirus E7 enhances hypoxia-inducible factor 1-mediated transcription by inhibiting binding of histone deacetylases. Cancer Res. 71(3): p. 1187-95. 311. Kato, A., et al., Induction of truncated form of tenascin-X (XB-S) through dissociation of HDAC1 from SP-1/HDAC1 complex in response to hypoxic conditions. Exp Cell Res, 2008. 314(14): p. 2661-73. 312. Seagroves, T.N., The Complexity of the HIF-1-Dependent Hypoxic Response in Breast Cancer Presents Multiple Avenues for Therapeutic Intervention. In Pharmaceutical Perspectives of Cancer Therapeutics, 2009: p. pp521-558. 313. Du, R., et al., HIF1alpha induces the recruitment of bone marrow-derived vascular modulatory cells to regulate tumor angiogenesis and invasion. Cancer Cell, 2008. 13(3): p. 206-20. 314. Ceradini, D.J., et al., Progenitor cell trafficking is regulated by hypoxic gradients through HIF-1 induction of SDF-1. Nat Med, 2004. 10(8): p. 858-64. 315. Jiang, Q., H. Zhang, and P. Zhang, ShRNA-mediated gene silencing of MTA1 influenced on protein expression of ER alpha, MMP-9, CyclinD1 and invasiveness, proliferation in breast cancer cell lines MDA-MB-231 and MCF-7 in vitro. J Exp Clin Cancer Res. 30: p. 60. 316. Minamishima, Y.A., et al., A feedback loop involving the Phd3 prolyl hydroxylase tunes the mammalian hypoxic response in vivo. Mol Cell Biol, 2009. 29(21): p. 5729-41. 317. Fujita, N., et al., Expression of prolyl hydroxylases (PHDs) is selectively controlled by HIF-1 and HIF-2 proteins in nucleus pulposus cells of the intervertebral disc: distinct roles of PHD2 and PHD3 proteins in controlling HIF-1alpha activity in hypoxia. J Biol Chem. 287(20): p. 16975-86. 318. Wong, N., J. De Melo, and D. Tang, PKM2, a Central Point of Regulation in Cancer Metabolism. Int J Cell Biol. 2013: p. 242513. 319. Place, T.L., et al., Aberrant promoter CpG methylation is a mechanism for impaired PHD3 expression in a diverse set of malignant cells. PLoS One. 6(1): p. e14617. 320. de Ruijter, A.J., et al., Histone deacetylases (HDACs): characterization of the classical HDAC family. Biochem J, 2003. 370(Pt 3): p. 737-49. 321. Dokmanovic, M., C. Clarke, and P.A. Marks, Histone deacetylase inhibitors: overview and perspectives. Mol Cancer Res, 2007. 5(10): p. 981-9. 322. Kato, H., S. Tamamizu-Kato, and F. Shibasaki, Histone deacetylase 7 associates with hypoxia-inducible factor 1alpha and increases transcriptional activity. J Biol Chem, 2004. 279(40): p. 41966-74.

184 323. Haberland, M., R.L. Montgomery, and E.N. Olson, The many roles of histone deacetylases in development and physiology: implications for disease and therapy. Nat Rev Genet, 2009. 10(1): p. 32-42. 324. Qi, J., et al., The ubiquitin ligase Siah2 regulates tumorigenesis and metastasis by HIF-dependent and -independent pathways. Proc Natl Acad Sci U S A, 2008. 105(43): p. 16713-8. 325. Nakayama, K., J. Qi, and Z. Ronai, The ubiquitin ligase Siah2 and the hypoxia response. Mol Cancer Res, 2009. 7(4): p. 443-51. 326. Takeda, K., et al., Regulation of adult erythropoiesis by prolyl hydroxylase domain proteins. Blood, 2008. 111(6): p. 3229-35. 327. Reuben, R.C., et al., A new group of potent inducers of differentiation in murine erythroleukemia cells. Proc Natl Acad Sci U S A, 1976. 73(3): p. 862-6. 328. Zhong, B., et al., Lead optimization of HMBA to develop potent HEXIM1 inducers. Bioorg Med Chem Lett. 24(5): p. 1410-3.

185