i

NFIC REGULATES AR MEDIATED

TRANSCRIPTION BY MULTIPLE MECHANISMS

YANG CHONG

B.Eng. (Hons.), NTU

A THESIS SUBMITTED

FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

NUS GRADUATE SCHOOL FOR INTEGRATIVE SCIENCES

AND ENGINEERING

NATIONAL UNIVERSITY OF SINGAPORE

2013

ii

DECLARATION

8th

November 2013

iii

ACKNOWLEDGEMENTS

First and foremost, I would like to express my sincere thanks to my Ph.D

supervisors, Dr. Yong Eu Leong and Dr. Edwin Cheung, for taking me as their

student for my Ph.D study. Dr. Edwin Cheung who has been a dedicated mentor

and provided valuable guidance and encouragement to me throughout the past

four years of my Ph.D studies, therefore I owe him the deepest gratitude. I would

also greatly appreciate the collaborative opportunities that he had provided me

with great research scientists within and outside of GIS. I also want to thank my

TAC members, Dr Yu Qiang, Dr Paul Yen and the annual GIS graduate seminar

committee for their insightful advice and opinions on the future directions to this

project.

I would like to give thanks to my colleagues who offered help on this work.

Special thanks to Dr Chng Kern Rei for his efforts in preliminary characterization

of NFIC in LNCaP cell line. I also would like to thank Dr Tan Peck Yean for the

generation of AR and FoxA1 ChIP-Seq datasets that facilitated my study on

NFIC. Thanks to Mr. Chang Cheng Wei for his extensive efforts in guiding our

group in the usage of computational programs so that we can be self-sufficient in

performing basic analyses on whole genome dataset.

iv

I wish to acknowledge Dr. Pan You Fu, Dr. Liu Mei Hui, Dr. Chuah Shin Chet,

Dr Tan Sikee and Dr. Sung Ying Ying for their valuable advice and insights

during the course of my study. It is a great pleasure to have met my present and

former colleagues in GIS who have always been encouraging and made my stay

enjoyable. I would like to convey thanks to the GTB sequencing group in GIS for

their commitment in generating high quality sequencing data.

More importantly, I would like to give thanks to the NUS Graduate School of

Integrative Science and Engineering (NGS). Without the financial support and

sponsorship from NGS, I wouldn’t be able to conduct the research for this thesis.

Last but not least, I would like to send my heartfelt gratitude to my parents

overseas for their caring and spiritual support throughout the four year of my

Ph.D studies.

.

v

TABLE OF CONTENT

DECLARATION .................................................................................................... II

ACKNOWLEDGEMENTS .................................................................................. III

TABLE OF CONTENT ......................................................................................... V

SUMMARY ....................................................................................................... VIII

LIST OF TABLES ................................................................................................. X

LIST OF FIGURES .............................................................................................. XI

ABBREVIATIONS ........................................................................................... XIII

PUBLICATION ...................................................................................................... II

CHAPTER 1 INTRODUCTION ............................................................................ 1

1.1 Prostate development and prostate cancer .....................................................1

1.2 Androgen and Androgen receptor .................................................................4

1.3 AR structure and its mode of action in prostate cancer .................................5

1.4 AR interacting proteins in AR mediated transcription ................................10

1.5 Decoding AR cistromes in prostate cancer cells .........................................11

1.6 Well Characterized AR collaborative factors ..............................................14

1.6.1 FoxA1 .................................................................................................14

1.6.2 GATA2 and Oct1 ...............................................................................17

1.6.3 NKX3.1 .............................................................................................18

1.6.4 ETS family of transcription factors ...................................................18

1.7 Pioneering factors as potential drug targets ................................................19

1.8 NFI family of transcription factors ..............................................................24

vi

1.9 NFIC CCAAT binding transcription factor .................................................28

1.10 Aim of the study ........................................................................................29

CHAPTER 2. MATERIALS AND METHODS .................................................. 30

2.1 Reagents and antibodies ..............................................................................30

2.2 Cell culture ..................................................................................................30

2.3 Cryogenic preservation and recovery of cells .............................................31

2.4 Western Blot Analysis .................................................................................31

2.5 RNA Isolation and Real-time Reverse Transcription reaction (RT-qPCR) 32

2.6 Chromatin Immunoprecipitation assays .....................................................33

2.7 Short interfering RNA (siRNA) studies ......................................................34

2.8 ChIP-Sequencing .........................................................................................34

2.9 Co-immunoprecipitation (Co-IP) ................................................................35

2.10 Microarray expression profiling ...............................................................35

2.11 Gene Ontology (GO) Analysis ..................................................................36

CHAPTER 3. RESULTS ...................................................................................... 37

3.1 NFIC as a novel AR collaborative factor ...................................................37

3.1.1 Identification of NFIC as a potential AR cofactor in LNCaP prostate

cancer cells .........................................................................................................37

3.1.2 Generation of NFIC binding cistromes in LNCaP prostate cancer cell

line ......................................................................................................................41

3.1.3 NFIC cistromes overlap with AR binding events ..............................44

3.1.4 NFIC mediates AR transcriptional activity ........................................49

3.2 Characterization and Validation of NFIC as pioneering factor to AR ........51

3.2.1 NFIC binding events at ARBS are independent of androgen

stimulation ..........................................................................................................51

3.2.2 NFIC and AR do not regulate each other ...........................................56

3.2.3 NFIC mediate AR dependent transcription ........................................59

vii

3.3 FoxA1 form co-regulatory complex with NFIC in Mediating AR

Transcriptional Regulation.....................................................................................64

3.4 NFIC has multiple mechanisms in distinct ARBS defined by FoxA1 ........79

3.4.1 FoxA1 regulated a large number of genes also regulated by NFIC ...79

3.4.1 NFIC plays the role of pioneering factor instead of FoxA1 at FoxA1

independent ARBS .............................................................................................82

3.4.2 NFIC collaborate with FoxA1 to contribute to AR transcription at

FoxA1 pioneered ARBS.....................................................................................95

DISCUSSION ..................................................................................................... 108

CONCLUSION, FUTURE DIRECTIONS AND PERSPECTIVES.................. 113

BIBLIOGRAPHY ............................................................................................... 116

APPENDIX I ...................................................................................................... 125

APPENDIX II ..................................................................................................... 130

APPENDIX III .................................................................................................... 132

viii

SUMMARY

Androgen receptor (AR) has been shown to play a key role in mediating the

transcriptional regulatory network directing prostate cancer initiation,

development and progression. To gain insights into the AR-dependent

transcriptional regulatory network, we utilized the ChIP-Seq approach and

generated genome-wide binding maps of AR in the LNCaP prostate cancer cell

line. From our bioinformatic analyses of AR binding sites (ARBS), we found an

enrichment of motifs belonging to the NFI family. We demonstrated that a

member of the NFI family, NFIC, shown to be over-expressed in clinical prostate

tumors, overlaps with a large fraction of ARBS across the genome. Our analysis

revealed that the recruitment of NFIC was independent of androgen treatment,

indicating NFIC as a potential pioneer factor. In addition, we showed that NFIC

was required for the global modulation of AR-dependent target genes including

KLK2, KLK3, and TMPRSS2. Interestingly, we found another well characterized

AR pioneering factor, FoxA1, to be recruited to the ARBS associated with these

genes. However, the expressions of these genes were FoxA1 independent,

suggesting NFIC was the more important pioneering factor at these ARBS. We

hypothesized that NFIC and FoxA1 could co-pioneer AR and went on to

demonstrate that NFIC and FoxA1 played distinct roles globally for different

classes of AR modulated genes and their associated ARBS. Collectively, our

ix

study suggests NFIC is a novel pioneering factor that is complementary to FoxA1

in AR-mediated transcription.

x

LIST OF TABLES

Table 1. Summary of ChIP-seq analysis ............................................ 66

xi

LIST OF FIGURES

Figure 1.1 H&E stain of micrograph showing prostate gland disorders ............................. 3

Figure 1.2 Schematic diagram featuring structural domains of the two isoforms

(AR-A and AR-B) of the human androgen receptor. .......................................................... 7

Figure 1.3 AR mode of action in prostate cancer ............................................................... 9

Figure 1.4 Model of AR coregulatory network ................................................................ 13

Figure 1.5 Diagram showing the model for FoxA1 mediated reprogramming of

distinct classes of AR enhancers in the LNCaP prostate cancer cells. ............................. 16

Figure 1.6 Cell type specific FoxA1 recruitment recognized by the lineage-

specific H3K4 methylation distributions .......................................................................... 22

Figure 1.7 Domains and alternative splicing of vertebrate NFI genes. ............................ 27

Figure 3.1 NFI motifs are highly enriched in AR binding peaks...................................... 40

Figure 3.2 Generation of NFIC ChIP-seq library in LNCaP cells .................................... 43

Figure 3.3 AR and NFIC are co-localized in LNCaP cells ............................................... 48

Figure 3.4 Ligand dependent response at ARBS requires AR and NFIC ......................... 50

Figure 3.5 NFIC genome wide recruitment is ligand independent ................................... 55

Figure 3.6 NFIC and AR do not regulate each other ........................................................ 58

Figure 3.7 NFIC regulates AR dependent genes ............................................................. 63

Figure 3.8 FoxA1 is a potential interplayer of AR/NFIC network ................................... 69

Figure 3.9 FoxA1 binding is highly enriched in AR/NFIC overlapping regions ............. 72

Figure 3.10 NFIC and FoxA1 pioneer AR recruitment at NFIC unique and FoxA1

unique ARBS respectively ................................................................................................ 76

Figure 3.11 Schematic diagram illustrating how NFIC and FoxA1 coordinate AR

transcriptional activity at distinct classes of ARBS. ......................................................... 78

Figure 3.12 NFIC regulate a substantial amount of FoxA1 regulated genes .................... 81

Figure 3.13 NFIC plays pioneering factor of AR at AR model genes instead of

FoxA1 ............................................................................................................................... 86

Figure 3.14 NFIC pioneers AR mediated transcription instead of FoxA1 at FoxA1

independent/occupied ARBS ............................................................................................ 91

xii

Figure 3.15 NFIC modulates chromatin accessibility to facilitate AR mediated

transcription instead of FoxA1 at FoxA1 independent/occupied ARBS .......................... 94

Figure 3.16 NFIC and FoxA1 collaborate at several FoxA1/NFIC co-pioneered

genes ................................................................................................................................. 99

Figure 3.17 NFIC and FoxA1 collaborate at additional FoxA1/NFIC co-pioneered

genes ............................................................................................................................... 104

Figure 3.18 Both NFIC and FoxA1 modulate chromatin accessibility and

contribute to pioneering AR mediated transcription at FoxA1/NFIC co-pioneered

regions ............................................................................................................................. 107

Figure 4.1 NFIC have no effect on H3K4me1/2 marks .................................................. 112

xiii

ABBREVIATIONS

AF Transactivation Function

Amp Ampicillin

AP-1 Activator Protein 1

AR Androgen receptor

ARBS Androgen receptor binding site

ARE Androgen response element

BPH Benign prostatic hyperplasia

BSA Bovine serum albumin

CARM1 Co-activator-associated arginine methyltransferase

C/EBP CCAAT/enhancer-binding protein

CCAT Control based ChIP-Seq Analysis Tool

CCND1 Cyclin D1

cDNA Complementary DNA

CD-FBS Charcoal-dextran treated FBS

ChIA-PET Chromatin Interaction Analysis-Paired End DiTag

ChIP Chromatin immunoprecipitation

ChIP-Seq ChIP-Sequencing

xiv

co-IP co-immunoprecipitation

cRNA Complementary RNA

CTCF CCCTC-binding factor

DBD DNA binding domain

DHT dihydrotestosterone

DMEM Dulbecco’s modified Eagle’s medium

DMSO Dimethyl sulfoxide

DNA Deoxyribonucleic acid

E2 17β-estradiol

EGF Epidermal growth factor

ER Estrogen receptor

ERBS ERα binding sites

ERE Estrogen response element

ETOH Ethanol

FAIRE Formaldehyde assisted isolation of regulatory elements

FBS Fetal bovine serum

FDR False discovery rate

FKH Forkhead

FoxA1BS FoxA1 binding sites

GAPDH Glyceraldehyde-3-phosphate dehydrogenase

GO Gene ontology

xv

GRO-Seq Global Nuclear Run-On and Sequencing

HAT Histone acetyltransferase

HDAC Histone deacetylase

HNF3α Hepatocyte nuclear factor 3α

JMJD JuMonJi domain-containing histone demethylase

LSD1 Lysine specific demethylase 1

MEME Multiple EM for Motif Elicitation

M-MLV RT Moloney Murine Leukemia Virus Reverse Transcriptase

NR Nuclear hormone receptor

NR3C4 Nuclear Receptor Subfamily 3, Group C, Member 4

Oct Octamer-binding transcription factor

PCR Polymerase chain reaction

PIC preinitiation complex

PKA Protein kinase A

PR Progesterone receptor

qPCR quantitative PCR

RNA Pol RNA polymerase

RNA-Seq RNA-Sequencing

SDS Sodium Dodecyl sulfate

SDS-PAGE SDS polyacrylamide gel

siRNA Short interfering RNA

xvi

SRC Steroid receptor co-activator

SWI/SNF Switching/sucrose non-fermenting

TSS Transcription start site

ii

PUBLICATION

Chng KR, Chang CW, Tan SK, Yang C, Hong SZ, Sng NY, Cheung E (2012) A

transcriptional repressor co-regulatory network governing androgen response in prostate

cancers. The EMBO journal 31: 2810-2823

1

CHAPTER 1 INTRODUCTION

1.1 Prostate development and prostate cancer

The prostate belongs to the male reproductive system and is an accessory gland

slightly larger than the size of a walnut. The prostate gland functions as secreting

prostatic fluid that usually constitutes more than half of the volume of the semen.

Prostate development is dependent upon the integration of steroid hormone

stimulations. A multitude of genes and signaling pathways has been shown to

regulate single or multiple steps prostate development, which includes epithelial

budding, the elongation of the duct, branching morphogenesis, and cellular

differentiation (Powers and Marker, 2013). Almost all prostatic acini developed

by outgrowths from the urethral epithelium and develop and grow into the

mesenchyme tissues in adjacent areas. The prostatic glandular epithelium

differentiates from the endodermal cells, while the associated mesenchyme of the

prostate differentiates into the dense stroma and the prostatic smooth muscle. In

the development of male reproductive systems, the prostate gland composed of

several glandular and non-glandular components are usually developed by the 9th

week of embryonic growth, formed by the condensation of mesenchyme, urethra

and Wolffian ducts.

2

The disorder of the prostate gland consists of Prostatitis, Benign prostatic

hyperplasia and finally leads to prostate cancer (Figure 1.1).

Prostatitis is defined as inflammation of the prostate gland. This sort of prostatic

disorder mainly comprises of four different forms, namesly acute prostatitis ,

chronic bacterial prostatitis, chronic non-bacterial prostatitis, and Category IV

prostatitis that is relatively uncommon in the general population, also classified as

a type of leukocytosis. Each type of prostatitis has different causes and outcomes.

Benign prostatic hyperplasia (BPH) happens when the prostate enlarges and

results in difficulty in urination. BPH often occurs in older men. The cause of

BPH is not well established, but potential factors that may contribute to BPH

include DHT. DHT is thought to play a potential role in the development in BPH,

by inducing cell growth in the prostate. Older men who stops producing DHT are

observed not to develop BPH(Isaacs and Coffey, 1989).

Prostate cancer is one of the most common types of cancers affecting aged men in

developed countries and has been considered the second leading cause of cancer

related death in US males. The most common form of prostate cancer is

adenocarcinoma. At the beginning cancer growth is only limited in the interior of

the prostate, and at certain stage the prostate cancer cells may progress to a more

advanced state, by metastasizing from the prostate and invade into other various

organs of the body, especially the bones and lymph nodes (Coffey and Pienta,

1987). The metastasized tumors are usually lethal.

3

Figure 1.1 H&E stain of micrograph showing prostate gland disorders

The H&E stain graph of prostatitis on the top panel and prostate adenocarcinoma on

the bottom panel (Isaacs and Coffey, 1989).

4

1.2 Androgen and Androgen receptor

Testosterone is a steroid hormone primarily derived from the testis. It plays key

physiological roles ranging from development of male reproductive tissues such

as prostate and epididymis (Desjardins, 1978; Cunha et al., 1987) to the

acquisition or promotion of the secondary male sexual characteristics during

puberty (Mooradian et al., 1987). Apart from the male-associated phenotypes,

androgens also regulate development or functions of other tissues and organs, for

instance, the bone and muscle tissues (Finkelstein et al., 1987), hair follicles on

the skin (Messenger, 1993), sebaceous gland (Deplewski and Rosenfield, 2000),

and adipose tissues (Schroeder et al., 2004).

The androgenic effect in a wide array of tissues is usually carried out by androgen

after it’s metabolized into 5-dihydrotestosterone (DHT) (Randall, 1994), which is

a more potent form of androgen. In the majority of androgen targeted tissues,

DHT binds to androgen receptor to form a regulatory complex to regulate target

genes, thus maintaining proper function of these vital organs (Liao and Fang,

1969). Androgen receptor plays a vital role both in the development of normal

prostate gland and prostate carcinoma, and is expressed throughout prostate

cancer progression and it remains expressed in many prostate tumors resistant to

androgen ablation therapy (Agoulnik and Weigel, 2008). Functional AR is

expressed through different stages of prostate carcinogenesis, including PIN and

early carcinoma, advanced carcinoma, and eventually in metastatic carcinoma.

Androgens are found to regulate the proliferation of epithelial cells, leading to the

5

adenocarcinoma, therefore androgens that modulate AR activity may result in

uncontrolled cell proliferation in the form of cancer.

1.3 AR structure and its mode of action in prostate cancer

The AR is also known as NR3C4 (nuclear receptor subfamily 3, group C, member

4). It is a nuclear transcription factor and is most closely related to the

progesterone receptor (Raudrant and Rabe, 2003). Androgen receptor carries out

its function as a DNA binding transcription factor by regulating gene expressions

(Mooradian et al., 1987).

The genomic structure of AR has been highly conserved throughout mammalian

evolution. The AR gene is located on the X-chromosome at position Xq11-12. It

spans around 90 kb and contains eight exons that code for nearly more than 900

amino acids within a 10.6 kb mRNA. Two isoforms of AR has been identified.

The less predominant isoform A is 87 kDa with its N-terminus truncated resulting

from in vitro proteolysis, while predominant isoform B is full length with 110

kDa mass (Wilson and McPhaul, 1994). Similar to many other steroid receptors,

the AR is modular in structure and consists of distinct functional domains known

as the amino-terminal domain transactivation domain (NTD), DNA binding

domain (DBD), carboxyl terminal ligand-binding domain (LBD) (Brinkmann et

al., 1989; Tsai and O'Malley, 1994). The DBD and LBD are connected by a hinge

region (Figure 1.2).

6

LBD, DBD and the N-terminal of the hinge segment are highly conserved during

evolution, while the NTD sequence is mostly variable. The NTD consists of

homopolymeric repeats including polyglutamine and polyglycine and contains the

major transactivation domain which is termed as activation function (AF)-1

responsible for transcriptional activity once it is bound by ligand. It also contains

activation function 5 (AF-5) that is responsible for the activity without ligand

binding. The LBD has specific motifs for ligand binding, as well as activation

function-2 (AF-2) domain responsible for agonist induced activity. AF-2 in AR

mainly binds preferably to N-terminal FXXFL motif but not FXXFL motifs that

are mostly found in many coactivators (He et al., 2002; He and Wilson, 2003).

Furthermore, the LBD contribute to not only the tethering to heat shock proteins

(HSPs), but also AR homo- and hetero-dimerization. The DBD consists of two

zinc-finger motifs. The DNA recognition specificity is determined by the first

zinc finger of the DBD, and the second zinc finger of DBD is mainly responsible

for the stabilization of the DNA and AR complex, as well as AR dimerization. In

addition, the DBD also mediates the nuclear localization of AR (Suzuki and Ito,

1999).

7

Figure 1.2 Schematic diagram featuring structural domains of the two

isoforms (AR-A and AR-B) of the human androgen receptor.

The numbers above the two structural isoforms indicates the number of amino

acid residues separating the domains starting from the N-terminus on the left to C-

terminus on the right. AR contains three functional domains, namely NTD (exon

A/B), DBD (exon C), and LBD (exon E to F). NTD and LBD are mainly

responsible for AR transcriptional activities, and LBD have additional functions

such as associating AR to HSPs and dimerization with DBD. DBD is required for

recognizing and binding AR to the ARE on chromatin. (Suzuki and Ito, 1999)

8

AR mode of action in mediating transcriptional regulation in prostate cancer in

highly ligand induced. In the absence of androgen stimulation, AR exists in the

cytoplasm as a complex with the HSPs as chaperones (Figure 1.3). Once bound

by ligand such as DHT converted from testosterone, AR undergoes

conformational change that triggers the subsequent dissociation of itself from

HSPs. This turns on the subsequent phosphorylation and homodimerization of AR,

leading to the translocation of dimerized AR into the nuleus where a serial of

events occurs. After AR moves into the nuleus, it recognizes and docks itself to

the canonical ARE (Chang et al., 1995; Gelmann, 2002; Harris et al., 2009). The

binding of AR to the ARE will subsequently facilitate the AR complex to recruit

other interacting proteins to mediate the downstream transcriptional regulation.

These recruited proteins include general transcriptional machinery such as RNA

PoII, chromatin remodeling complexes and transcriptional coactivators and

corepressors (Heinlein and Chang, 2004; Heemers and Tindall, 2007; Harris et al.,

2009).

9

Figure 1.3 AR mode of action in prostate cancer

After testosterone released from plasma proteins is converted to DHT by 5-

reductatse, AR becomes bound by DHT and HSP is dissociated caused by DHT

binding, leading to AR dimerization. The dimerized AR subsequently translocated

into the nucleus and binds to ARE, which facilitates recruitment of co-activators

and general transcription machinery, to regulate transcriptional activities (Chen et

al., 2008).

10

1.4 AR interacting proteins in AR mediated transcription

AR plays a central role in the transcriptional regulation in prostate cancer, but it

does not function alone. The androgen regulation in prostate cancer cell lines

often involves the formation of diverse AR transcriptional complexes, and this

regulatory network mainly consists of components of the general transcriptional

machinery, AR coactivators and corepressors, and specific transcription factors

(Shang et al., 2002; Heemers and Tindall, 2007).

AR may regulate its downstream transcription by regulating both transcriptional

initiation and elongation. General transcription factors to enhance AR

transcription involve the formation of preinitiation complex (PIC) to promote the

recruitment of RNA PoII to the AR target gene promoters. Besides the direct

contact of AR with general transcription factors, AR has been shown to interact

with RNA PoII directly by the subunit RPB2 (Lee et al., 2003).

Up to now a huge number of nuclear receptor coregulators of AR have been

identified. They can be broadly classified as AR coactivators and corepressors,

but they can also be classified based on the particular cellular activities they have

to regulate transcription. These cellular activities may include components of the

chromatin remodeling complex were shown to alter DNA topology and associated

with loss of canonical nucleosomal ladder. For instance, the switching / sucrose

nonfermenting (SWI/SNF) complex were shown to stimulate AR activity in an

ATP dependent manner (Marshall et al., 2003). Histone modifiers affect AR

transcriptional activity by modifying histone residues by acetylation, methylation,

phosphorylation, ubiquitination, and glycosylation. Such modifications would

11

result in changes of the net charge of the nucleosome and thereby changes in the

conformation of the histone. For example, histone acetylase (HAT) are commonly

asscociated with transcriptional activation while the histone deacetylase (HDAC)

recruitment render a more compact chromatin structure and transcriptional

repression (Xu and O'Malley, 2002). Histone methylation, however, can render

both transcriptional activation and repression depending on the position of the

histone residues modified. For instance, H3K4 methylation is associated with

active genes, while H3K9 and H3K27 methylation often indicate repressed state

of genes (Mellor, 2006). Histone demethylase such as lysine specific demethylase

1 (LSD1) and JMJ domain-containing family of histone demethylases (JMJD1A

and JMJD2C) have recently drew attention to their role in demethylating mono, di

and trimethylated H3K9, resulting in activation of AR dependent transcription

(Metzger et al., 2005; Yamane et al., 2006; Wissmann et al., 2007). And

Coactivator-associated arginine methyltransferase (CARM-1) is one of the well

studied histone methyltransferases (HMTs) to interact with SRC coactivators and

is classified as a secondary coactivator (Chen et al., 1999).

1.5 Decoding AR cistromes in prostate cancer cells

Early studies on transcriptional regulation by AR were largely gene and site

specific, with limited number of candidates (Cleutjens et al., 1996). However, AR

signaling involves hundreds of targets genes and the regulation by AR appears to

have high differential expression kinetics as well based on time course expression

profiling (Tan et al., 2012). Therefore the development of techniques such as

12

Chromatin Immunoprecipitation coupled with microarray (ChIP-chip) and

Chromatin Immunoprecipitation coupled with massively parallel sequencing

(ChIP-seq) has provided much more comprehensive insights into the studies of

genome wide AR cistromes and transcriptional activities. By combining ChIP-on

chip data with gene expression profile, transcription factors such as FoxA1, Oct1

and GATA2 have been identified as AR collaborators that control androgen-

mediated transcription (Wang et al., 2007). For instance, FoxA1 is a well studied

pioneering factor of AR in prostate cancer (Gao et al., 2003), and its role in

defining distinct classes of AR binding programs has recently been discovered

(Sahu et al., 2011; Wang et al., 2011). The general model for the AR

transcriptional complex is illustrated in Figure 1.4. I would like to elaborate on

several well characterized AR collaborative factors in the following chapters.

13

Figure 1.4 Model of AR coregulatory network

Pioneering factors such as FoxA1 and GATA2 preoccupy chromatin to modulate

chromatin structures prior to hormone stimulation. Upon DHT stimulation, AR is

recruited to distal enhancers and further recruits a hierarchical assembly of co-

activator complexes including p160, Oct1, HATs and HMTs, general transcription

machinery, PoII to ARE to start downstream transcription.

14

1.6 Well Characterized AR collaborative factors

AR-mediated transcription is composed of a wide array of coordinated events,

such as chromatin stucture remodeling, epigenetic signature modifications,

chromosomal looping events and polymerase tracking (Wang et al., 2005). AR

collaborative factors regulate AR transcriptional activities with diverse functions

on AR and regulating its transcriptional output, therefore are often overexpressed

and contribute to carcinogenesis in prostate cancer (Heinlein and Chang, 2004).

1.6.1 FoxA1

FoxA1 is short for Forkhead box protein A1. It is also known as hepatocyte

nuclear factor 3-alpha (HNF-3α) and is encoded by FOXA1 gene. FoxA1 belongs

to the FKH family of DNA binding transcription factor and plays key roles in the

development of various organs including liver, pancreas, breast and prostate (Lee

et al., 2005). In the early years it was shown to be involved in the regulation of

several ER target genes such as TFF-1 (Beck et al., 1999) as well as interacting

with ERα. Later studies proved that FoxA1 is required for modulating chromatin

structure, thereby facilitating ERα recruitment to mediate ER transcriptional

activities and this is the first study to establish FoxA1 as a pioneering factor of

ERα (Carroll et al., 2005). Furthermore, FoxA1 was also implicated in prostate

cancer development and progression by interacting with AR and regulate AR

target genes (Gao et al., 2003), thus a lot of studies emerged to focus on its role in

AR mediated transcription (Jia et al., 2008; Lupien et al., 2008). FoxA1 has

recently been shown to possess a cell type-specific functions which primarily rely

15

on its differential recruitment to chromatin at enhancers, subsequently leads to

chromatin structure changes and functional interplay with lineage-specific

collaborative factors. The differential recruitment of FoxA1 in prostate and breast

cancer is defined by the distribution of mono- and dimethylated H3 lysine 4

residues. Therefore FoxA1 translates epigenetic signatures into changes in

chromatin states to establish cell type specific transcriptional programs.

Most recently, studies have investigated the genome wide redistribution of ARBS

after depleting FoxA1 in prostate cancer cell lines and discovered that FoxA1

defines three distinct classes of ARBS, which are lost, conserved or gained after

its depletion. The AR redistributed binding program is nicely correlated with

similar patterns in the redistribution of gene expression profiling (Sahu et al.,

2011; Wang et al., 2011) (Figure 1.5). These studies provided a more

comprehensive global picture on FoxA1 regulation in prostate cancer, that apart

from FoxA1 pioneered ARBS, there are also ARBS that is independent on FoxA1

regulation, and thereby eliciting needs for new studies to explore what other

pioneering factor might come into play to regulate the AR mediated transcription

in these particular ARBS.

16

Figure 1.5 Diagram showing the model for FoxA1 mediated reprogramming

of distinct classes of AR enhancers in the LNCaP prostate cancer cells.

In the lost AR binding program, FoxA1 faciliates AR recruitment to ARE in

relatively open chromatin regions. Nucleosome remodelling is not induced upon

AR binding to ARE in this class of enhancers. In the conserved AR program, AR

recruitment to ARE is independent of FoxA1, and the binding events induce

nucleosome remodelling. In the third type of program in which AR binding is

gained after FoxA1 depletion, FoxA1 inhibits AR binding, in spite of the presence

of strong AREs. In all the above three classes of AR binding programs, eRNAs

were produced or increased following AR binding to AREs (Wang et al., 2011).

17

1.6.2 GATA2 and Oct1

By integrating Chromatin Immunoprecipitation with tiled oligonucleotide

Microarrays, previous studies have identified transcription factors such as

GATA2 and Oct1 to be cooperating in mediating AR transcription. They were

shown to form a regulatory hierarchy with AR to regulate androgen dependent

gene expression and prostate cancer progression (Wang et al., 2007).

Even earlier studies have discovered that GATA family members have the ability

to open compact chromatin structures via zinc finger motifs (Boyes et al., 1998),

therefore it may also serve the role of pioneering factor that modulate chromatin

to facilitate AR binding. However, Oct1 is not required for AR recruitment to

PSA and TMPRSS2, indicating Oct1 functions downstream to GATA2 and

cooperate to work together with AR. Silencing GATA2 and Oct1 results in a

significant decrease in the androgen dependent cell proliferation.

A later study utilized ChIP-microarray technique to map ARBS and histone H3

acetylation loci in C4-2B prostate cancer cell line, and manage to find large

portion of ARBS to be marked by AcH3. From the analysis of the ARBS

sequence motifs, they also discovered FoxA1, GATA2 to be AR coregulators.

Adopting transient siRNA knockout experiments, they discovered that these

coregulators of AR have diverse effects on the expression level of several AR

target genes such as KLK2, KLK3, TMPRSS2 and FKBP5 (Jia et al., 2008).

18

1.6.3 NKX3.1

NKX belongs to a subfamily of Homeobox Genes which were shown to be

aberrantly expressed in cancers. Human NKX3-1 is largely confined within the

prostate gland (Bieberich et al., 1996). The implication fo NKX3-1 in prostate

cancer development was addressed in recent studies. NKX homeodomain has

been observed to be enriched in AR cistrome analysis. NKX3-1 is the first known

marker for prostate epithelial differentiation. Studies have shown that NKX3-1

collaborate with AR at several AR target prostate cancer model genes such as

PSA, TMPRSS2 and FKBP5, and it regulate AR in a feed forward manner (Tan et

al., 2012).

Functional analysis has shown that NKX3-1 collaborate with AR and pioneering

factor such as FoxA1 to promote prostate cancer survival.

1.6.4 ETS family of transcription factors

The ETS family comprises of several transcription factors containing highly

conserved ETS DNA binding domain (Karim et al., 1990) . The ETS family of

transcription factors plays important roles in regulating diverse cellular processes

such as cell proliferation, cell differentiation, angiogenesis and apoptosis (Sevilla et

al., 1999). ETS transcription factors could function both as coactivator or corepressor

depending on the cellular context (Sharrocks et al., 1997; Sharrocks, 2001) . The

function of this family of transcription factors in cancer was notably found to be

through gene fusion with another protein and in the context of prostate cancer,

various papers have reported the discovery of recurrent ETS fusion genes (Tomlins et

19

al., 2008). Among all gene fusions of ETS family members such as ETV1, ETV4 and

ETV5 occurring in prostate cancer, the most common type of occurring variant of

ETS gene fusion in prostate cancer is the TMPRSS2-ERG fusion gene (Kumar-Sinha

et al., 2008) . From AR ChIP-on ChIP studies in prostate cancer cell line, the ETS

motif was found to be enriched at AR-bound promoter regions, and members such as

ETV1 and ERG were subsequently found to be collaborative factors of AR (Massie et

al., 2007). Interestingly, while ETS-1 and ETV1 were found to be AR coactivators in

a feed forward manner (Shin et al., 2009), ERG, on the other hand, was recently

implicated to play the role of AR corepressor in androgen dependent transcriptions

(Sun et al., 2008).

Genome-wide cistrome studies of AR and ERG in VCaP prostate cancer cells have

shown that ERG suppresses androgen dependent transcription as well as epithelial

differentiation (Yu et al., 2010). A recent study has investigated further into the

mechanism of the transcriptional collaboration between AR and ERG in ERG

overexpressed prostate cancer cells (Chng et al., 2012). They assigned functional

roles of various corepressors such as ERG, HDACs and EZH2 in mediating

metastasis in a way that they cooperate in a coregulatory network centered by AR to

attain optimal androgen signaling for prostate cancer progression and development.

1.7 Pioneering factors as potential drug targets

Pioneer factors are an emerging class of transcription factors that contribute to the

cell-type-specific transcriptional programs and facilitate transcription factors

binding to enhancers by modulating chromatin accessibility (Zaret and Carroll,

2011; Jozwik and Carroll, 2012). Pioneering factors are usually prebound to

20

condensed chromatin in a hormone independent manner, and its modulation of the

chromatin accessibility via various molecular mechanisms, they are able to

facilitate recruitment of other transcription factors to turn on the subsequent

transcription. Recent studies on nuclear receptors have identified a variety of

pioneering factors in different classes of cell lines. Besides the role of FoxA1 as

pioneering factor of AR in prostate cancer, it has also been shown to pioneer the

Estrogen Receptor (ER) mediated transcription in breast cancer, by facilitating

ERα recruitment and modulating chromatin accessibility at the regulatory sites of

ERα target genes (Carroll et al., 2005; Lupien et al., 2008; Eeckhoute et al., 2009)

(Figure 1.6).

PBX1 has been shown to be a novel pioneering factor for the ER-mediated

transcriptional response that drives aggressive tumors in breast cancer. PBX1

translate specific epigenetic signatures into an open chromatin state that guides

the recruitment of ERα and its regulated transcriptional programs are highly

associated with poor prognosis in breast cancer patient (Magnani et al., 2011).

Ap2γ is another novel pioneering factor and it has recently been shown to play a

role in ERα recruitment and mediating long range chromatin interactions in

response to estrogen stimulation in breast cancer. It also has interplay with FoxA1

in mediating ER dependent transcription as well as long range chromatin

interactions (Tan et al., 2011).

Another pioneering factor GATA3 has been shown to act upstream of FoxA1 to

directly affect ER enhancer accessibility in breast cancer cells, and its depletion

21

resulted in a global redistribution of active histone marks such as H3K4me1 and

H3K27Ac (Theodorou et al., 2013).

Due to the importance of pioneer factors in mediating transcription in hormone

dependent cancer, they constitute attractive therapeutic drug targets for cancer

treatment. Examples such as FOXA1 is found to be play the role as pioneering

factor to both ERα mediated transcription and tumor growth in breast cancer and

AR mediated transcription and cancer progression in prostate cancer, therefore the

development of a specific inhibitor to FOXA1 would provide a useful clinical

utility for the treatment of ER+ hormone-resistant breast cancer and AR

dependent prostate cancer.

22

Figure 1.6 Cell type specific FoxA1 recruitment recognized by the lineage-

specific H3K4 methylation distributions

FoxA1 recruitment occurs in a cell-specific manner. The FoxA1 are directed to

bind to its respective response elements recognized by active enhancers marked

by lineage specific H3K4me1 and H3K4me2. In contrast, H3K9 methylation

marks silenced enhancer regions and those repressed loci are usually devoid of

FoxA1 binding. FoxA1 are found to bind and actively modulate the associated

chromatin to a more relaxed state, thereby facilitating lineage-specific

23

transcription factor such as ERα, AR or other collaborative factors recruitment to

the more exposed binding sites in response to hormone stimulation to establish

transcriptional competency (Eeckhoute et al., 2009).

24

1.8 NFI family of transcription factors

Our study utilized genomic approaches to identify and characterize novel

collaborative factor of AR. By adopting ChIP-seq technology (Schmidt et al.,

2009), we generated AR binding patterns across the genome in LNCaP prostate

cancer cells. Apart from the other well characterized AR cofactor motifs such as

Forkhead, ETS, NKX, GATA, another motif that is highly enriched motifs in the

AR cistrome belongs to the NFI family of DNA-binding proteins. This family of

transcription factors consists of four members, namely NFI-A, NFI-B, NFI-C and

NFI-X (Qian et al., 1995). They are known to function in both viral DNA

replication and in the regulation of gene expression (Gronostajski, 2000; Perez-

Casellas et al., 2009) (Figure 1.7).

NFI proteins have modular structures. The N-terminal domains of NFI proteins

mediate DNA binding and dimerzation, and the C-terminal domains usually

regulate transcriptional activation and/or repression (Mermod et al., 1989;

Bandyopadhyay and Gronostajski, 1994). NFI binding sites have been found in

many mammalian tissues, including brain, liver, lung and mammary gland,

pituitary, retina and various other tissues (Cereghini et al., 1987; Courtois et al.,

1990; Watson et al., 1991; Furlong et al., 1996; Bachurski et al., 1997; Bedford et

al., 1998). In most of these tissues, the function of NFI proteins has been indicated

to be important for gene expressions. The NFI family of transcription factors

usually binds as a either hetero- or homodimer with a preferred binding sequence

as a palindrome containing two half sites TTGGCANNTGCCAA on duplex DNA

(Gronostajski et al., 1985). NFI proteins are found to be either repressing or

25

activating gene expression in a gene specific manner (Furlong et al., 1996;

Johansson et al., 2003). It has been recently shown to be involved in long rang

regulation of gene expression by forming chromatin barrier that blocks the

propagation of heterochromatin structures (Esnault et al., 2009).

Studies also showed that the NFI proteins have dual role in hormone-mediated

transactivation of mouse mammary tumor virus (MMTV) promoter. NFI family

of transcription factors are found to play a role in both transcriptional activation

and nucleosome remodeling via the formation and maintenance of a preset

chromatin structure in the MMTV promoter region in mouse (Chikhirzhina et al.,

2008).

In situ hybridization assays in mouse indicated that the four NFI gene members

are expressed in a relatively unique but widely overlapping pattern during

embryogenesis, suggesting that it’s due to the differential expression of these

genes that lead to differential expression of the target proteins that regulate

development (Chaudhry et al., 1997).

NFI proteins also have implications in cancer from their role in the control of cell

growth. The aberrant fusion between the C-terminus of NFIB and HMGIC leads

to pleomorphic adenomas (Geurts et al., 1998). However, the overexpression of

NFIX has opposing role in cancer development by suppressing oncogenic

susceptibility in mink lung epithelial cells (Sun et al., 1998). The early findings of

the role of NFI proteins in DNA replication might be related to the influence of

the altered expression of NFI proteins on cell proliferation in cancer development.

The current models for investigating the role of the four members in cancer are

26

mainly systems that are depleted or overexpressing the specific isoforms of NFI

protein family members (Gronostajski, 2000).

27

Figure 1.7 Domains and alternative splicing of vertebrate NFI genes.

The top figure illustrates the general feature of NFI proteins and the below shows

the alternatively spliced products of the four NFI genes from vertebrates, NFIA,

NFIB, NFIC and NFIX. The chromosomal locations of each individual member

are indicated below the name of the member. NFI proteins are generally

composed of the N-terminal DNA-binding and dimerization domain and the C-

terminal regions including the proline-rich transactivation domain (Gronostajski,

2000).

28

1.9 NFIC CCAAT binding transcription factor

Based on the clinical transcriptome studies available in Oncomine database, we

narrowed down to NFIC which is overepressed in the metastatic prostate tumor

samples (Lapointe et al., 2004; Yu et al., 2004).

Up till now there have been relatively limited studies of NFIC in prostate cancer,

and most studies on the biological functions of NFIC have been performed on nfic

knock out mice, which showed that NFIC played an essential role for odontogenic

cell proliferation and odontoblast differentiation during tooth root development

(Lee et al., 2009). Nfic-deficient mice exhibited normal molar crown formation

but aberrant odontoblast differentiation during root formation (Steele-Perkins et

al., 2003). Mice that lack NFIC develop abnormal roots and lose their teeth,

which shares similar symptoms to the radicular dentin dysplasia I in humans.

Studies found significant increases in the several mRNA expressions in NFIC

depleted cells, suggesting that NFIC might regulate odontoblast cell migration

and differentiation, and thus play a role in the root development (Kim et al., 2009).

One study on the transcriptional regulatory function of NFI proteins based on the

model of NFI-dependent mouse mammary tumor virus (MMTV) promoter

indicated that the expression of NFIC or NFIX, but not NFIA or NFIB proteins,

represses glucocorticoid induction of the MMTV promoter in HeLa cells. The

repressive regulatory functions by NFIC appears to be highly cell type specific

that such regulation is observed in HeLa and COS-1 cells but not 293 or JEG-3

cells (Chaudhry et al., 1999).

29

The role of NFIC in cancer development has limited implication up to date. A

study of ERα centric regulatory network in breast cancer suggest that NFIC

negatively regulates the cyclin D1 (CCND1) oncogene expression, therefore its

expression is negatively correlated with breast cancer proliferation (Eeckhoute et

al., 2006). However, the role of NFIC in the context of AR in prostate cancer has

not yet been established.

1.10 Aim of the study

In this study, we integrate molecular and computational approach to decode AR

and NFIC cistrome in prostate cancer cells. By combining ChIP-sequencing with

microarray gene expression profiling, we endeavor to find out the role of NFIC in

AR mediated transcriptional network. We established NFIC as a novel pioneering

factor of AR and functionally dissect its role as a pioneering factor in relation to

another pioneering factor, FoxA1, in mediating AR transcriptional regulations.

We found huge amount of AR and NFIC overlapping cistromes also converge

FoxA1 binding events, suggesting that FoxA1 comes into play with AR and NFIC

regulatory network. Lastly, we discovered that NFIC mediates AR transcriptional

regulation at distinct classes of ARBS defined by their dependency on FoxA1,

and at these sites NFIC and FoxA1 differentially modulates chromatin

accessibility, thereby potentiated the role of NFIC in the prostate cancer

development.

30

CHAPTER 2. MATERIALS AND METHODS

2.1 Reagents and antibodies

The hormone dihydrotestosterone (DHT) was purchased from Tokyo Chemical

Industry. The antibodies we used for ChIP-assay were purchased from the

following companies: anti-AR (sc-816 and sc-815x), anti-NFIC (sc-5567), normal

rabbit IgG (sc-2027), normal goat IgG (sc-75 2028) from Santa Cruz, and anti-

FoxA1 (ab5089) from Abcam.

The antibodies used for western blot analysis were shown as follows. Primary

antibodies used were: anti-AR (AR441) from Labvision, anti-NFIC (sc-5567)

from Santa Cruz; anti- FoxA1 (ab5089) and anti-βactin (ab3280-500) from

Abcam. Secondary antibodies used were: donkey anti-goat IgG HRP (sc-2033)

from Santa Cruz; ECL anti-rabbit IgG horseradish peroxidase (HRP) linked whole

antibody (NA934V) and ECL anti-mouse IgG HRP linked whole antibody (NA931V)

from Amersham.

2.2 Cell culture

LNCaP cells were purchased from American Type Culture Collection (ATCC).

LNCaP were maintained in RPMI Medium 1640 (RPMI) with 10% fetal bovine

31

serum (FBS), 1 mM sodium pyruvate, 100 units/ml penicillin and together with

30 μg/ml gentamycin. LNCaP cells were maintained in hormone deprived phenol

red free RPMI medium for at least 3 days followed by DHT or ethanol treatment.

2.3 Cryogenic preservation and recovery of cells

Early passages of LNCaP cells were trypsinized and spun down at 1,000 rpm for 3

mins to remove the trypsin. Cell pellet was resuspended in culture media which was

further supplemented with 45% FBS and 5% dimethyl sulfoxide (DMSO) (Sigma),

and then aliquoted into 2 ml Cryobank Vials (Thermo Scientific, Nunc). The vials

were placed into a pre-cooled isopropanol-filled container for slow freezing at -80˚C.

After freezing overnight, the vials were transferred to a liquid nitrogen tank for long

term storage. During recovery, the cells were thawed quickly at 37˚C and transferred

to T75 flask for culture. The medium was changed the next day to remove DMSO.

2.4 Western Blot Analysis

After LNCaP cells were stimulated with DHT for 2 hours, we use Triton X lysis

buffer (0.25% Triton X-100, 1mM EDTA, 10 mM TrisHCl, 150 mM NaCl) to

lyse the cell membrane. Total protein lysates were collected and protein

concentration was subsequently quantified using Pierce BCA Protein Assay kit

(Thermoscientific). Then we prepared 20 to 30 ug of protein lysates of each

sample and mixed them with 5x SDS loading buffer, followed by denaturing at

99˚C for 5 mins. After that, the protein lysates were resolved on 10% SDS

polyacrylamide gel (SDS-PAGE) and subsequently transferred onto the

32

nitrocellulose membrane (GE Healthcare). The transferred membrane was

blocked with 5% milk for 1 hr at room temperature before it was incubated with

corresponding primary antibodies and secondary antibodies at an optimized duration

and temperature. For signal detection, ECL plus Western Blotting Detection System

(Amersham) was sprayed onto the surface of the membrane and subsequently

incubated in dark for 5 minutes before the membrane was exposed to films.

2.5 RNA Isolation and Real-time Reverse Transcription reaction (RT-qPCR)

LNCaP cells were pretreated with ETOH or 10 nM DHT for 12 hrs, followed by

extraction of total cellular RNA using TRI® Reagent (Sigma). After precipitated

with 75% ETOH, chloroform was added to facilitate RNA isolation. The

extracted RNA was then purified using the Invitrogen PureLinkTM RNA Mini

Kit according to the manufacturer’s protocol. After measuring RNA quality and

concentration using Nanodrop, 1 μg RNA was obtained to undergo reverse

transcription using Oligo (dT) primer (Promega), dNTP Mix (Fermentas),

Moloney Murine Leukemia Virus Reverse Transcriptase (M-MLV RT) (Promega)

and M-MLV RT 5X Buffer (Promega). After cDNA are synthesized, real-time

qPCR was subsequently performed by adding the KAPA SYBR FAST qPCR kit.

CT values were analyzed with the ABI 7900HT Fast Real-Time PCR System to

obtain relative expression. The sequences for real-time qPCR primers of cDNA

are listed in Appendix I. The experiments were repeated at least three times to

obtain the gene expression profiles.

33

2.6 Chromatin Immunoprecipitation assays

LNCaP cells were treated with 100nM DHT or EtOH for 2 hours before the cells

were fixed with 1% formaldehyde (Sigma-Aldrich, Missouri). After cell pellets were

collected, they were subsequently lysed with SDS lysis buffer with cocktail

Proteinase Inhibitor and underwent sonication to shear the genomic DNA into lengths

of 500-1000 bp. From the sonicated samples, 30 uL was aliquoted and kept as input.

The rest of the sheared chromatin was then precleared with normal rabbit/goat IgG

(Santa Cruz Biotechnology, Santa Cruz) and sepharose beads A or A/G (Zymed, San

Francisco) for at least 2 hours in 4°C. It’s followed by incubation overnight for

complete immunoprecipitation after adding the specific antibodies and sepharose

beads. On the next day, the beads were washed with four rounds of washing buffers

to remove non-specific bindings. Then the protein bound DNA complex was

subjected to elution and subsequent reverse-crosslinking at 65°C overnight to

completely isolate DNA fragment with the protein complex. The DNA was then

purified with QIAquick spin PCR purification kit (Qiagen, California). Real-time

quantitative Polymerase Chain Reaction (q-PCR) was carried out by using the KAPA

SYBR FAST qPCR kit. Amplified DNA products were subsequently detected using

the ABI 7900HT Fast Real-Time PCR System. The relative binding profile was

determined by calculating the ratio of immunoprecipitated DNA over the input.

Real-time qPCR primer sequences for ChIP assays are listed in Supplementary

Table S. All ChIP experiments were performed at least three times.

34

2.7 Short interfering RNA (siRNA) studies

We transfected LNCaP cells with 100 nM siRNA against NFIC, AR, or FoxA1

purchased from Dharmacon or negative control siRNA from 1stBase via

Lipofectamine™ RNAiMAX Transfection Reagent (Invitrogen). Cells continued

to be incubated for 72 hr. After that, cells were treated with vehicle or 10 nM

DHT for another 12 hr followed by collected for ChIP, Real-time qPCR and

western blot analyses.

2.8 ChIP-Sequencing

After quantification using Pico-green ds DNA assay kit (Invitrogen) , 5ng of

immunoprecipitated DNA were obtained and used for ChIP-Seq library construction.

The ChIP-Seq DNA sample Prep Kit (Illumina) was utilized for library preparation

with some minor modifications. The ChIP DNA was made to undergo end-repair

followed by adapter ligation. Platinum Pfx DNA polymerase (Invitrogen) was then

used to amplify the ChIP-DNA. Electrophoresis of the amplified DNA products on a

2% agarose gel was subsequently performed. Detection of the amplified DNA

products was done by staining the gel with SYBR® Green I Nucleic Acid Gel Stain

(Invitrogen). DNA products of size 200-300bp were extracted by gel excision and

purified. Confirmation of the size and quality of the extracted purified ChIP DNA

was done using bioanalyzer prior sequencing on the Solexa platform. The sequencing

depth for each library was at least 10million tags. ChIP-Seq read tags were aligned to

the reference human genome (UCSC, hg18). The binding peaks were determined

with CCAT (Xu et al., 2010) normalized against input .

35

2.9 Co-immunoprecipitation (Co-IP)

LNCaP cells were maintained in full serum RPMI medium for the Co-IP

experiments. After cells were seeded for at least three days, they were

subsequently trypsinized and lysed to obtain whole cell lysate. This is followed by

pre-clearing process of the cell lysate with Protein A-Agarose beads (Roche

Applied Science) at 4°C for 4 h. Before the IP process, we collected an aliquot of

around 30 ul and kept it at -80°C as input for the final western blot analysis. After

pre-clearing, we add 5 mg of anti-NFI (H-300, sc-5567) to the supernatant and

incubated the mixture on the roller overnight at 4°C. The next morning we added

30ul Roche beads into the incubated mixture and continued incubation for another

1.5 h at 4°C. After that we washed the mixture with TBS for four times

consecutively. Finally, we removed supernatant and heated the beads at 99°C for

5 min followed by elution with SDS loading buffer for the subsequent western

blot analysis.

2.10 Microarray expression profiling

Total RNA was extracted and purified from LNCaP cells, under the NC and NFIC

depleted or FoxA1 depleted conditions. The cells were pre-treated with ETOH or

10 nM DHT for the 12 hours. Then the Invitrogen PureLinkTM RNA Mini Kit

was used according to the manufacturer’s protocol. RNA from three biological

replicates was then processed to cRNA using the Illumina® TotalPrepTM-96

RNA Amplification Kit (Ambion) according to the manufacturer’s instructions.

36

Subsequently direct hybridization assay was performed with the HumanRef-8 v2

Expression BeadChip Kit and scanning the chip array using the BeadArray

Reader (Illumina). The BeadStudio software was subsequently used to process the

image data and GeneSpring GX11 software was further implemented to analyze

the expression data.

2.11 Gene Ontology (GO) Analysis

GO was performed using the IPA (Ingenuity Pathway Analysis) online software

on NFIC depleted microarray. NFIC activated and repressed DHT responsive

genes were analyzed respectively. Threshold for DHT responsiveness is 1.3 fold

change and cut off for NFIC dependency is 1.2 fold change.

37

CHAPTER 3. RESULTS

3.1 NFIC as a novel AR collaborative factor

3.1.1 Identification of NFIC as a potential AR cofactor in LNCaP prostate

cancer cells

To gain insight into the AR coregulatory network in mediating androgen

dependent transcription in prostate cancer and to discover novel AR collaborators

in prostate cancer, we first performed AR ChIP-Seq in the androgen-dependent

prostate cancer cell line, LNCaP to obtain a genome wide map of AR binding

events (Tan et al., 2012). AR ChIP-seq was performed at 2 hour after DHT

treatment. We managed to obtain 18,117 AR peaks in the ETOH treated condition

and 75,296 peaks in DHT treated condition based on FDR 0.05.

Next we used our in house motif discovery algorithm called Center of

Distribution (CENDIST) to analyze the ARBS sequences. CENTDIST program is

designed based on an algorithm that scans across the genome wide binding sites

and rank motifs of transcription factors available in TRANSFAC database

according to the enrichment of center distribution around the binding events of the

ChIP-seq dataset (Zhang et al., 2011). By inputting obtained AR peaks into

CENDIST, we discovered significant enrichment of several transcription factor

family motifs including Forkhead, ETS, and NKX as previously reported (Figure

1A). In addition, we also found NFI motif among the top ten most highly enriched

38

motifs derived from this analysis, indicating members of NFI family of

transcription factors might play a potential role in the AR mediated transcription.

After examining the Oncomine database that deposit cancer transcriptome studies,

we found that NFIC, one of the NFI family of transcription factors, to be over-

expressed in metastatic prostate carcinoma with respect to normal prostate tumor

in several clinical prostate tumor transcriptome studies (Lapointe et al., 2004; Yu

et al., 2004). This indicates that NFIC might play a role in prostate cancer

progression, especially in more advanced prostate cancer types (Figure 3.1).

39

A

B

40

Figure 3.1 NFI motifs are highly enriched in AR binding peaks

(A) CentDist motif enrichment analysis of the top co-occupying transcription

factor family motifs from AR ChIP-seq peaks under the condition of DHT

stimulation. The motif of families of transcription factors are ranked based on the

score of the distribution of the top member within the transcription factor family

respectively. (B) Boxplot comparing the transcript expression level of NFIC in

prostate adenocarcinoma and metastasis prostate tumor samples based on clinical

studies available in the Oncomine database of transcriptomes (Yu et al. study).

The differential gene expression data is centered on the median of expression

levels and plotted in a log2 scale.

41

3.1.2 Generation of NFIC binding cistromes in LNCaP prostate cancer cell line

We next examined the potential role of NFIC as an AR collaborative factor. By

performing NFIC ChIP-qPCR on known binding sites associated with AR model

target genes such as KLK3, KLK2, and TMPRSS2, we observed NFIC and AR

are co-localized at these loci. Based on these findings we would like to expand the

scope for the NFIC binding events across the genome by performing genome

wide ChIP sequencing of NFIC in LNCaP prostate cancer cell line under the same

2 hours hormone stimulation condition as AR ChIP-seq. Adopting FDR less than

0.05 in the in-house CCAT peak calling algorithm (Xu et al., 2010), we identified

89,377 and 82,645 NFIC binding sites in LNCaP cells prior to and after DHT

treatment, respectively. De novo motif analysis of NFIC binding sites was

subsequently performed and it revealed an enriched binding motif that is highly

similar to the NFI motif in the TRANSFAC database (Figure 3.2).

42

(A)

(B)

43

Figure 3.2 Generation of NFIC ChIP-seq library in LNCaP cells

(A) ChIP-qPCR of AR and NFIC binding profile at the PSA enhancer, KLK2

enhancer and promoter regions, and two ARBS around TMPRSS2. The data

represents the mean ± SEM of at least 3 independent assays. (B) Logos showing

the canonical ARE available from the Transfac database in the top panel and

derived via MEME algorithm in the bottom panel.

44

3.1.3 NFIC cistromes overlap with AR binding events

We observed that NFIC shares huge proportion of overlapping binding events

between DHT and ethanol treated conditions, which in turn also overlaps

considerably with ARBS (Figure 3.3 A and B).

The previous ChIP-qPCR of AR and NFIC recruitment on several AR model

genes can further be visualized on the USCS genome browser from the binding

peaks of the AR and NFIC ChIP-seq datasets. The co-localization of AR and

NFIC at PSA enhancer, KLK2 enhancer and promoter, and two sites associated

with TMPRSS2 can be observed (Figure 3.3 E).

We also investigated the genomic distribution of AR and NFIC binding events.

Most of our AR binding sites are located distances away from the transcriptional

start sites (TSS), while the NFIC binding events occurs both at distal and

proximal regions relative to the TSS. Interestingly, the AR and NFIC co-localized

binding sites exhibits similar patterns to most of the AR binding sites, that

majority of these binding events are found far away from the TSS (Figure 3.3 D).

This suggested that the collaboration between NFIC and AR most probably

occurs at enhancers of the genes to modulate downstream transcription.

To investigate whether NFIC interact with AR endogenously in LNCaP prostate

cancer cells, Co-immunoprecipitation experiment was subsequently performed.

Cell lysates were incubated with NFIC antibody and we pulled down the potential

protein complexes using specifically designed beads. After western blot analysis

45

using AR antibody, we managed to observe a weak interaction between AR and

NFIC, as compared to the intensity after negative control IgG pull down. This

provided support that NFIC not only co-localize with AR, but also physically

interact with AR, possibly by forming a transcriptional complex (Figure 3.3 C).

However, the Co-IP experiments need to be optimized to obtain a better

visualization of the endogenous protein-protein interactions between AR and

NFIC.

46

(A)

(B)

(C)

58743

66092

16553

AR DHT

NFIC DHT

AR/NFIC overlapping

47

(D)

(E)

48

Figure 3.3 AR and NFIC are co-localized in LNCaP cells

(A) Venn diagram illustrating the overlap between AR binding peaks under DHT

treatment and NFIC ChIP-seq peaks in both ETOH and DHT treated conditions.

(B) Percentage of AR unique, NFIC unique and AR/NFIC shared binding events.

(C) Co-immunoprecipitation experiments showing the interaction between AR

and NFIC. (D) Genome-wide distribution of AR and NFIC binding sites with

respect to the transcription start sites (TSSs) of RefSeq genes. (E) UCSC Genome

browser (Hg18) featuring representative androgen dependent AR model genes,

KLK3 (top left panel), KLK2 (top right panel) and TMPRSS2 (bottom panel), co-

occupied by AR and NFIC. The binding tag intensities are indicated on the right

y-axes.

49

3.1.4 NFIC mediates AR transcriptional activity

To confirm the effect that NFIC had on AR-mediated transcription is at least, in

part, attributed to NFIC binding on the cis-regulatory elements, we cloned the two

AR binding sites (C7, C8), that were shown to recruit NFIC in vivo on androgen

stimulation (Figure 3.4) respectively into a luciferase reporter plasmid. The

predicted AR and NFIC binding motif of each of the AR binding site was then

distorted via mutagenesis separately prior to performing reporter assays.

Androgen-induced reporter activities of both binding sites were decreased when

the AR or NFIC binding motifs were abolished (Figure 3.4), pointing to the

importance of NFIC occupancy in androgenic transcriptional response.

50

Figure 3.4 Ligand dependent response at ARBS requires AR and NFIC

The top panel is the schematic diagram illustrating the reporter constructs used in the

transient transfection analysis. The Bottom panel shows the reporter assay after

LNCaP cells were transfected with reporter constructs. Prior to transfection, LNCaP

cells were treated with or without DHT for 24 hrs. C7 and C8 ARBS full mutated

ARE and NFIC motifs were cloned into pGL4-TATA and assessed for their

luciferase reporter activity. Empty vector pGL4-TATA was integrated as a negative

control. All results represent the average of 3 independent experiments ± SEM.

51

3.2 Characterization and Validation of NFIC as pioneering factor to AR

3.2.1 NFIC binding events at ARBS are independent of androgen stimulation

We moved on to globally analyze the NFIC binding events. To validate the NFIC

ChIP-seq datasets, we randomly selected 5 high tag intensity, 5 medium tag

intensity and 10 low tag intensity NFIC binding sites. The ChIP-qPCR validation

of these binding events shows the NFIC recruitment happens to be independent of

DHT treatment for all 20 binding sites tested (Figure 3.5 A).

The genome wide binding profile of AR and NFIC showed that AR binding is

minimal prior to DHT treatment in both AR unique and AR and NFIC co-

localized regions, while the occupancy of NFIC at both NFIC unique and AR and

NFIC overlapping regions is comparable before and after androgen stimulation.

This indicates the binding of NFIC to its binding sites is independent of androgen

stimulation. The scatter plot showing the distribution of the tag intensity of each

NFIC binding sites available in either ETOH or DHT treated ChIP-seq library

exhibits a high correlation between the two datasets. This provided additional

evidence that NFIC binding events are indeed ligand independent (Figure 3.5 B

and C).

Together with the substantial overlap between AR and NFIC binding events after

androgen stimulation, our data suggested that NFIC might play a role as a pioneer

factor of AR that first binds to ARBS prior to DHT treatment and subsequently

enhances the transcriptional competency of AR. One of the most well studied

52

pioneer factors to AR in prostate cancer cells is FoxA1, which remodels the

chromatin accessibility rendering it competent for other transcription factor

binding (Lupien et al., 2008; Magnani et al., 2011). Hence it would be intriguing

to explore whether NFIC also possesses the ability of pioneering the

transcriptional activity of AR in the context of prostate cancer.

53

(A)

54

(B)

(C)

55

Figure 3.5 NFIC genome wide recruitment is ligand independent

(A) ChIP-qPCR of NFIC occupancy at the 5 high tag intensity binding sites, 5

medium intensity binding sites and 10 low intensity binding sites randomly

picked from the NFIC ChIP-seq dataset. The data represents the mean ± SEM of

at least 3 independent assays. (B) Comparison between average tag densities of

binding sites after peak calling using CCAT in both ethanol and DHT treated

NFIC ChIP-seq binding maps that also have ARBS. (C) Scatter plot showing the

tag intensity of NFIC binding events in both DHT-treated and ETOH-treated

datasets.

56

3.2.2 NFIC and AR do not regulate each other

To investigate the role of NFIC as pioneering factor to AR, we adopted transient

siRNA knockdown approach to transfect into the LNCaP cells to examine the

subsequent effect of NFIC depletion on AR recruitment and androgen dependent

gene expressions. It would be important to find out whether there’s a mutual

regulation between AR and NFIC in the first place. After knocking down AR and

NFIC respectively, we observed that the depletion of NFIC has no appreciable

effect on AR expression both in mRNA level and protein level, indicating AR is

not a direct target of NFIC. On the other hand, NFIC expression was unaffected

by AR knockdown or DHT stimulation, confirming that NFIC is not an AR-

regulated gene (Fig 3.6).

57

(A) (B)

(C)

58

Figure 3.6 NFIC and AR do not regulate each other

(A and B) Effect of NFIC silencing on AR mRNA expression level. Starved

LNCaP cells were transfected with either negative control siRNA or siRNA

agains NFIC, followed by DHT or ETOH treatment for 6 hours. Total RNA was

then isolated and quantified using real-time RT-qPCR with the primers of AR,

NFIC and GAPDH genes. Subequently all mRNA expressions were normalized

against GAPDH expression levels. The data presented are relative expression to

GAPDH ± SEM of at least 3 independent experiments. (C) Western blot showing

the protein levels of AR and NFIC as well as loading control β-actin after

transfected with siRNA against NFIC and AR respectively.

59

3.2.3 NFIC mediate AR dependent transcription

Next, we sought to determine the requirement for NFIC in AR mediated

transcriptional response. AR and NFIC were observed to be co-localized at

several AR target genes such as PSA, KLK2, and TMPRSS2. NFIC depletion

resulted in a significant reduction of the expression of these AR model target

genes (Figure 3.7 A) as well as the AR recruitment to its binding sites associated

with these genes (Figure 3.7 B), suggesting that NFIC acts upstream of AR

recruitment, thereby promoting AR signaling. Above results provide support that

NFIC plays a role as an AR collaborative factor in regulating AR target gene

expression

To obtain more candidate genes regulated by NFIC for further analysis, we

performed genome wide microarray gene expression analysis under the condition

of NFIC depletion on LNCaP cells with or without DHT treatment (Fig 3.7 C).

Our analysis revealed 1868 DHT responsive genes by adopting 1.3 fold change

cut-off in comparison of the expression profile prior and after DHT treatment.

Interestingly, the depletion of NFIC affected the expression of a significant

number of DHT-induced (402) and DHT-repressed genes (158), indicating that

NFIC plays the role of either a transcriptional coactivator or corepressor to

mediate AR-mediated transcription. The expression profiling array also provides a

candidate pool of NFIC target genes in the context of androgen regulation. Taken

together, our results potentiated the role of NFIC as a pioneer factor of AR that

acts upstream and subsequently mediates AR transcriptional regulation.

60

To further analyze the genome wide microarray profiling data after NFIC

silencing, we use IPA gene ontology to investigate the cellular pathways and

processes involved in NFIC regulation of AR dependent transcription. NFIC

activated and repressed gene sets were both analyzed by IPA. Cancer related

cellular process involved in NFIC regulation include cellular movement, cell

death, cell proliferation and cell development. This further proves that NFIC plays

a role in androgen mediated prostate cancer growth and development (Figure 3.7

D).

61

(A)

(B)

62

(C)

(D)

63

Figure 3.7 NFIC regulates AR dependent genes

(A) Effect of NFIC silencing on PSA, TMPRSS2, KLK2 mRNA expression

levels. LNCaP cells were maintained in hormone depleted medium followed by

transfection with either control negative control siRNA or siRNA targeting

against NFIC prior to stimulation with or without 10 nM of DHT for 12 hr. Total

RNA was isolated and amplified with real-time RT-qPCR primers for PSA,

TMPRSS2 and KLK2. mRNA expression levels were normalized against

GAPDH. (B) ChIP-qPCR of AR occupancy after NFIC silencing at the PSA,

KLK2, and TMPRSS2. (C) Microarray gene expression profiling was performed

on LNCaP cells transfected with control or NFIC siRNA and stimulated with

ETOH as vehicle or DHT for 12 hours. The heatmap represents all DHT-

regulated genes and fold change in expression is indicated below. All results

represent the average of 3 independent experiments ±S.E.M. (D). Gene Ontology

using IPA showing cellular process involved via the regulation of the NFIC

activated and repressed genes.

64

3.3 FoxA1 form co-regulatory complex with NFIC in Mediating AR

Transcriptional Regulation

So as to reveal a more comprehensive mechanism of the AR and NFIC regulatory

network, we decided to investigate additional potential players. To achieve this,

we analyzed NFIC cistrome that also contains ARBS via CENTDIST motif

analysis program to discover motifs that are also centered around AR and NFIC.

As expected, both NFI and ARE motifs were highly ranked at these binding sites,

and interestingly, the second highly enriched motif revealed by CENDIST is

Forkhead motif (Figure 3.8 A). This indicates that NFIC and AR might form a

transcriptional regulatory network with factors that belong to the Forkhead family.

Since the role of FoxA1 as a pioneering factor to mediate AR gene transcription

has been recently studied and reported (Lupien et al., 2008; Serandour et al.,

2011), we would like to investigate the potential role of FoxA1 in the AR and

NFIC regulatory network.

Collaborations between multiple pioneer factors to mediate the transcriptional

competency of regulatory elements have been reported. For instance, FoxA1 has

been shown to cooperate with AP- occupied by

both factors there is a functional interplay between AP-2γ and FoxA1 to mediate

ER transcription and long range interaction(Tan et al., 2011).To study the

potential interplay among AR, NFIC and FoxA1, we utilized the FoxA1 ChIP-Seq

dataset in LNCaP cells from the previous study by our group (Tan et al., 2012).

Integrating the DHT stimulated AR, NFIC and FoxA1 ChIP-Seq datasets, we

65

discovered a huge proportion of AR and NFIC overlapping binding sites that are

also bound by FoxA1 (Figure 3.8 B).

66

Table 1 Summary of the unique tag counts and total number of peaks called

using FDR 0.05 for each ChIP-seq library

Library ID

Description

Sequencing Depth

No. of peaks called

SCS800

AR ETOH

17,413,241

18,117

SCS801 AR DHT 13,252,823 75,296

CHL005 NFIC ETOH 918,0715 89377

CHL006 NFIC DHT 994,3604 82645

CHL001 FOXA1 ETOH 13,462,620 79,975

CHL002 FOXA1 DHT 8,677,867 61,309

67

(A)

(B)

68

(C)

69

Figure 3.8 FoxA1 is a potential interplayer of AR/NFIC network

(A) CentDist output of the top co- occurring transcription factor family motifs

from DHT stimulated AR and NFIC overlapping ChIP-seq peaks. (B) Venn

diagram illustrating the overlap between DHT stimulated AR, NFIC and FoxA1

ChIP-seq peaks. (C) Screenshots of AR, NFIC and FoxA1 ChIP-seq peaks

around the KLK3, KLK2 and TMPRSS2.

70

In agreement with previous studies, by examining the heatmap of genome wide

FoxA1 binding profile, similar to the binding profile of NFIC, the pattern of

FoxA1 binding is generally independent of androgen stimulation in our study, and

the majority of AR and NFIC co-localized sites also contain substantial amount of

FoxA1 occupancy (Figure 3.9A). Furthermore, by comparing the tag intensity

distribution of FoxA1 binding peaks on AR unique, NFIC unique and AR/NFIC

overlapping regions, we found that FoxA1 binding events are more enriched at

AR and NFIC co-localized regions than AR and NFIC uniquely occupied regions

(Figure 3.9B). All these results suggested that FoxA1 possibly forms a

transcriptional network with AR and NFIC.

71

(A)

(B)

72

Figure 3.9 FoxA1 binding is highly enriched in AR/NFIC overlapping regions

(A) Heatmap of AR (red), NFIC (green) and FoxA1 (purple). ChIP-seq tag

intensity sorted according to NFIC (DHT) tag intensity (top-bottom: highest to

lowest) and centered on AR (DHT) peaks in a 2 kb window. 1-3: Groups of

common or unique NFIC binding sites identified, namely (1) AR unique; (2)

NFIC unique; (3) AR/NFIC overlapping

73

Before depleting FoxA1 for the subsequent analysis, we also noted that silencing

FoxA1 or NFIC did not affect the expression of the other; neither did it have any

appreciable effect on AR protein level (Figure 3.10A). Then based on the

available genome wide ChIP-seq datasets of AR, NFIC and FoxA1, we randomly

selected ARBS co-localized with FoxA1 only, as well as ARBS that are co-

occupied uniquely by NFIC, and examined the AR recruitment at these sites after

introducing siRNA against NFIC and FoxA1 respectively. Not surprisingly, AR

recruitment was regulated by FoxA1 at ARBS void of NFIC occupancy, while

NFIC regulated AR recruitment at sites bound by only NFIC itself (Figure 3.10B).

FAIRE-qPCR was subsequently performed to investigate the role of NFIC and

FoxA1 on modeling chromatin openness in different classes of ARBS that could

contribute to AR recruitment and subsequent transcription. Under normal

conditions a relatively high FAIRE signal detected by qPCR suggests that the

chromatin state at specific loci is relatively open in state. From the FAIRE qPCR

results, it has been revealed that at NFIC unique ARBS, FAIRE signal is

compromised upon NFIC depletion, but not after FoxA1 knockdown, while for

FoxA1 unique ARBS it is FoxA1 that modulates chromatin state instead of NFIC

(Figure 3.10 C&D).

74

(A)

(B)

75

(C)

(D)

76

Figure 3.10 NFIC and FoxA1 pioneer AR recruitment at NFIC unique and

FoxA1 unique ARBS respectively

(A) Western blot showing the effect of NFIC, FoxA1 silencing on AR, NFIC and

FoxA1 as well as loading control β-actin protein levels. (B) ChIP-qPCR of AR

occupancy after FoxA1 silencing at the PSA, KLK2, and TMPRSS2. (C)

Heatmap showing the AR ChIP-qPCR experiment after NFIC or FoxA1 silencing

on randomly selected NFIC unique ARBS and FoxA1 unique ARBS. The

intensity of the relative enrichment of % input is indicated below. All relative

enrichments of individual ARBS have been normalized against NC ETOH treated

condition. All ChIP qPCR experiments were repeated three times. (C) Box plot

showing the FAIRE qPCR experiments after NFIC or FoxA1 silencing on NFIC

unique ARBS. All relative FAIRE signal enrichments have been normalized

against NC ETOH treated condition. (D) Box plot showing the FAIRE qPCR

experiments after NFIC or FoxA1 silencing on FoxA1 unique ARBS. All relative

FAIRE signal enrichments have been normalized against NC ETOH treated

condition. All FAIRE qPCR experiments were repeated at least three times.

77

It is not to our surprise to find out patterns of NFIC and FoxA1 functioning as

pioneering factor to AR respectively at those ARBS not co-occupied by both

factors. Previous studies have indicated the role of FoxA1 to modulate chromatin

accessibility and direct AR recruitment to respective ARBS it preoccupies

(Eeckhoute et al., 2009). It is interesting to discover the role of NFIC as

pioneering factor to AR, acting in a similar manner to FoxA1 at ARBS occupied

by AR and NFIC only. This suggests NFIC could function independently as a

pioneering factor to AR and gives rise to questions about what the mechanism is

like at FoxA1 and NFIC co-bound regions.

We summarize our above results so far, that these results indicate that NFIC and

FoxA1 function as pioneering factor to facilitate AR recruitment to NFIC and

FoxA1 uniquely occupied ARBS respectively in an independent manner. And

whether there is collaboration between NFIC and FoxA1 in mediating AR

dependent transcription at ARBS shared by both pioneering factors remains to be

investigated.

78

(A)

(B)

Figure 3.11 Schematic diagram illustrating how NFIC and FoxA1 coordinate

AR transcriptional activity at distinct classes of ARBS.

(A) At NFIC unique ARBS, NFIC plays the role of AR pioneering factor and

facilitate the AR recruitment. (B) At FoxA1 unique ARBS, FoxA1 plays the role

of AR pioneering factor and facilitate the AR recruitment.

79

3.4 NFIC has multiple mechanisms in distinct ARBS defined by FoxA1

3.4.1 FoxA1 regulated a large number of genes also regulated by NFIC

To obtain a global view of the coregulatory functions of NFIC and FoxA1 on AR

transcription, we next performed genome wide microarray expression profiling in

LNCaP cell line after depleting FoxA1. FoxA1 depletion was performed in

accordance to NFIC knockdown conditions. Cells were transfected with negative

control siRNA as well as siRNA against FoxA1 for 72 hours, followed by

treatment with or without DHT for 12 hours after the total RNA have been

collected.

We adopted 1.3 fold changes for the DHT responsiveness and 1.3 fold changes as

threshold for FoxA1 dependency. As a result, we observed a striking difference

between the expression patterns of FoxA1 regulated genes before and after FoxA1

depletion.

Comparing the NFIC regulated genes with the FoxA1 regulated genes, we found

that around 42% of NFIC regulated genes are also regulated by FoxA1 and the

rest 58% of genes are regulated by NFIC without interference by FoxA1.

Therefore these datasets provided candidates for the subsequent analysis of the

role of NFIC in regulating AR transcription in the presence of the absence of

FoxA1 regulation.

80

(A)

(B)

81

Figure 3.12 NFIC regulate a substantial amount of FoxA1 regulated genes

(A) Microarray gene expression profiling was performed on LNCaP cells

transfected with control or FoxA1 siRNA and stimulated with ETOH as vehicle or

DHT for 12 hours. The heatmap represents all DHT-regulated genes and fold

change in expression is indicated below. All results represent the average of 3

independent experiments ±S.E.M. (B) Vann Diagram showing the overlap

between NFIC regulated genes and FoxA1 regulated genes obtained from

knockdown microarray expression profiling of NFIC and FoxA1.

82

3.4.1 NFIC plays the role of pioneering factor instead of FoxA1 at FoxA1

independent ARBS

Next we sought to explore the regulatory roles of NFIC and FoxA1 on NFIC and

FoxA1 co-localized ARBS. From the expression microarray and the ChIP-Seq

datasets of AR, NFIC and FoxA1, we selected several androgen responsive and

NFIC regulated genes associated with ARBS occupied by both NFIC and FoxA1.

By performing the expression microarray after FoxA1 depletion under similar

condition to NFIC knockdown microarray, we compared the gene list to the NFIC

regulated gene sets , and found out that some of the NFIC regulated genes are also

regulated by FoxA1, while another substantial group of NFIC regulated genes are

independent on FoxA1 depletion.

Based on this, we further classified the ARBS associated with these genes into

FoxA1 independent and pioneered regions. And this is consistent with recent

studies that revealed distinct AR binding programs defined by FoxA1 in LNCaP

prostate cancer cell line. FoxA1 depletion elicits extensive redistribution of AR

binding events and this is correlated with changes in androgen-dependent gene

expressions (Sahu et al., 2011; Wang et al., 2011).

Although AR, NFIC and FoxA1 co-occupy around a few well characterized AR

model genes such as PSA, KLK2 and TMPRSS2 (Figure 4A), none require

FoxA1 for gene activation (Figure 4B), but were found to be regulated by NFIC

(Figure 2D). Additionally, depletion of FoxA1 did not affect AR recruitment to

ARBS associated with these genes (Figure 4C), suggesting NFIC plays a more

83

important role as pioneering factor to AR at these AR model gene associated

ARBS. Although they were occupied by FoxA1 as well, FoxA1 does not appear

to play any regulatory role. Next we carried out knock down of NFIC or FoxA1

and checked the subsequent binding of either factor, but both seem not to be

required for the recruitment of the other (Figure 4D and 4E), suggesting NFIC and

FoxA1 do not have mutual dependency at these ARBS despite their co-occupancy.

84

(A)

(B)

85

(C)

(D)

86

Figure 3.13 NFIC plays pioneering factor of AR at AR model genes instead of

FoxA1

(A) Effect of FoxA1 silencing on PSA, TMPRSS2, KLK2 expression levels.

LNCaP cells were maintained in hormone depleted medium followed by

transfection with either control negative control siRNA or siRNA targeting

against FoxA1 prior to stimulation with or without 10 nM of DHT for 12 hr. Total

RNA was isolated and amplified with real-time RT-qPCR primers for PSA,

TMPRSS2 and KLK2. mRNA expression levels were normalized against

GAPDH. (B) ChIP-qPCR of AR occupancy after FoxA1 silencing at the PSA,

KLK2, and TMPRSS2. (C) ChIP-qPCR of FoxA1 occupancy after NFIC

silencing at the PSA, KLK2, and TMPRSS2. (D) ChIP-qPCR of NFIC occupancy

after FoxA1 silencing at the PSA, KLK2, and TMPRSS2. All results represent the

mean ± SEM of at least 3 independent assays.

87

To provide support to our hypothesis on the mechanism of NFIC in mediating AR

transcription alone at FoxA1 independent loci, additional FoxA1 independent

ARBS have been revealed from gene expression profiling and ChIP assay

validation, namely, AADAT, MBOAT2, NDRG1, SGK3 and SEC24D. As can be

observed from Figure 3.14, these genes were all NFIC regulated genes but

independent of FoxA1 regulation with patterns pretty similar to the few AR model

genes. Depletion of neither NFIC nor FoxA1 affected the occupancy of one

another on these loci, indicating NFIC is pioneering factor of AR where FoxA1

binds to chromatin without contributing to the activation of AR transcriptional

programs.

88

(A)

(B)

89

(C)

(D)

90

(E)

(F)

91

Figure 3.14 NFIC pioneers AR mediated transcription instead of FoxA1 at

FoxA1 independent/occupied ARBS

(A) Effect of NFIC silencing on AADAT, MBOAT2, NDRG1, SGK3, and

SEC24D mRNA expression levels. LNCaP cells were maintained in hormone

depleted medium followed by transfection with either control negative control

siRNA or siRNA targeting against NFIC prior to stimulation with or without 10

nM of DHT for 12 hr. Total RNA was isolated and amplified with real-time RT-

qPCR primers for AADAT, MBOAT2, NDRG1, SGK3, and SEC24D. mRNA

expression levels were normalized against GAPDH. (B) Effect of FoxA1

silencing on AADAT, MBOAT2, NDRG1, SGK3, SEC24D expression levels. (C)

ChIP-qPCR of AR occupancy after NFIC silencing at the AADAT, MBOAT2,

NDRG1, SGK3, and SEC24D. (D) ChIP-qPCR of AR occupancy after FoxA1

silencing at the AADAT, MBOAT2, NDRG1, SGK3, and SEC24D. (E) ChIP-

qPCR of FoxA1 occupancy after NFIC silencing at the AADAT, MBOAT2,

NDRG1, SGK3, and SEC24D. (F) ChIP-qPCR of NFIC occupancy after FoxA1

silencing at the AADAT, MBOAT2, NDRG1, SGK3, and SEC24D. All results

represent the mean ± SEM of at least 3 independent assays.

92

In order to find out the role of NFIC and FoxA1 in modulating chromatin

accessibility at FoxA1 independent co-occupied ARBS, FAIRE qPCR

experiments were subsequently performed after silencing of NFIC and FoxA1

respectively. FAIRE qPCR provided further evidence for the effect of both factors

to AR recruitment and transcription, that NFIC plays a more important role than

FoxA1 for opening the chromatin structures at FoxA1 independent/NFIC

pioneered ARBS (Figure 6A).

Above results suggest that the co-presence of several transcription factors does

not necessarily indicate all are functioning in transcriptional regulation. AR

transcription is mainly mediated by key regulators including NFIC in the context

of FoxA1 independent ARBS. The schematic diagram depicting the mechanism

of NFIC and FoxA1 in mediating AR transcription at FoxA1 independent ARBS

is shown in Figure 3.15, that FoxA1 and NFIC are both present and co-occupy the

chromatin prior to DHT stimulation. However, FoxA1 only binds to chromatin as

a redundant factor without contributing to the remodeling of chromatin state.

NFIC instead modulates chromatin accessibility that subsequently facilitates AR

recruitment to the ARBS after DHT stimulation, and enhances AR dependent

gene transcription such as PSA, KLK2, SGK3, MBOAT2, AADAT and so on.

93

(A)

(B)

94

Figure 3.15 NFIC modulates chromatin accessibility to facilitate AR

mediated transcription instead of FoxA1 at FoxA1 independent/occupied

ARBS

(A) Box plot showing the FAIRE qPCR experiments after NFIC or FoxA1

silencing on FoxA1 occupied/independent ARBS. All FAIRE signal enrichments

are normalized against NC ETOH treated conditions. (B) Schematic diagram

illustrating how NFIC and FoxA1 coordinate AR transcriptional activity at

NFIC/FoxA1 overlapping ARBS where AR recruitment and gene regulation is

independent of FoxA1, it is NFIC that plays the role of AR pioneering factor.

95

3.4.2 NFIC collaborate with FoxA1 to contribute to AR transcription at FoxA1

pioneered ARBS

Besides the set of genes regulated by NFIC alone without interference by FoxA1,

we also would like to examine the mechanism of NFIC and FoxA1 coregulation

at NFIC and FoxA1 co-pioneered regions. Since microarray analysis of NFIC

regulated gene set and FoxA1 regulated gene set suggest 42% of NFIC regulated

genes are also regulated by FoxA1, and there’s a striking enrichment of Forkhead

motif in the NFIC and AR binding peaks, we propose there’s a collaborative

mechanism of NFIC and FoxA1 in regulating a subset of AR target genes.

Genes such as SASH1, LRIG1, AFF3, SLC2A12, PNMA1, SEPP1 and DEGS1

are found to be FoxA1 and NFIC co-pioneered genes selected from the gene

expression profiling dataset and ChIP validation. The cis-regulatory elements of

these genes were found to be co-localized by AR, NFIC and FoxA1 and co-

regulated by NFIC and FoxA1. By knocking down NFIC and FoxA1 respectively,

we are able to observe a drop in expression of these genes to different extents

(Figure 5A, 5B). In accordance to the expression profiling of these gene

candidates, we subsequently performed AR ChIP under the condition of NFIC

and FoxA1 depletion respectively. And it has been found that recruitment of AR

is reduced after depleting either NFIC or FoxA1 (Figure 5C, 5D).

Interestingly, the mutual dependency between NFIC and FoxA1 for the

recruitment of each other exhibits distinct patterns for different genes. Based on

the mutual knockdown ChIP, genes such as SASH1, AFF3, SLC2A12 and

96

PNMA1 shows that FoxA1 depletion resulted in a significant decrease of NFIC

occupancy, while knocking down NFIC did not affect the binding of FoxA1,

indicating at these loci FoxA1 acts upstream of NFIC and is required for NFIC to

bind to chromatin, and both collaborate as AR pioneering factor (Figure 5E and

5F). However, NFIC and FoxA1 did not show mutual dependency near LRIG1,

SEPP1 and DEGS1 (Figure 5F), and it’s likely that both factors function

independently and both contribute to pioneering AR transcription.

97

(A)

(B)

98

(C)

(D)

(E)

99

Figure 3.16 NFIC and FoxA1 collaborate at several FoxA1/NFIC co-

pioneered genes

(A) Effect of NFIC silencing on SASH1, LRIG1, AFF3, SLC2A12, PNMA1,

SEPP1, and DEGS1 mRNA expression levels. LNCaP cells were maintained in

hormone depleted medium followed by transfection with either control negative

control siRNA or siRNA targeting against NFIC prior to stimulation with or

without 10 nM of DHT for 12 hr. Total RNA was isolated and amplified with

real-time RT-qPCR primers for SASH1, LRIG1, AFF3, SLC2A12, PNMA1,

SEPP1, and DEGS1. mRNA expression levels were normalized against GAPDH.

(B) Effect of FoxA1 silencing on SASH1, LRIG1, AFF3, SLC2A12, PNMA1,

SEPP1, and DEGS1 expression levels. (C) ChIP-qPCR of AR occupancy after

NFIC and FoxA1 silencing at the SASH1, LRIG1, AFF3, SLC2A12, PNMA1,

SEPP1, and DEGS1. (D) ChIP-qPCR of FoxA1 occupancy after NFIC silencing

at the SASH1, LRIG1, AFF3, SLC2A12, PNMA1, SEPP1, and DEGS1. (E)

ChIP-qPCR of NFIC occupancy after FoxA1 silencing at the SASH1, LRIG1,

AFF3, SLC2A12, PNMA1, SEPP1, and DEGS1. All results represent the mean ±

SEM of at least 3 independent assays.

100

Similar patterns can be observed on additional FoxA1/NFIC dependent ARBS

such as LPAR3, PEX10, CORO2A, DNM1L and THRAP3. These additional

candidates are also selected from the genome wide microarray gene expression

profiling of FoxA1 and NFIC respectively. Among these candidates, LPAR3 has

been reported as pioneered by FoxA1 in previous studies (Wang et al., 2011).

Expression of these genes are regulated by both NFIC and FoxA1, and the

decrease in mRNA expression after silencing either factor is correlated to a

compromise in the AR recruitment after either factor is knocked-down in the

ChIP q-PCR experiments. Mutual knockdown ChIP also shows distinct patterns

in these set of candidates and further classifies them into two subtypes. PEX10,

LPAR3 and CORO2A shows a consistent decrease in NFIC recruitment after

FoxA1 silencing, suggest FoxA1 act upstream of NFIC at this class of ARBS,

while this pattern is not observed at DNM1L and THRAP3. Silencing NFIC,

however, has no appreciable effect on FoxA1 recruitment, suggesting FoxA1

selectively act as AR pioneering factor upstream of NFIC regulation, and both

collaborate to mediate AR dependent gene transcription.

101

(A)

(B)

102

(C)

(D)

103

(E)

(F)

104

Figure 3.17 NFIC and FoxA1 collaborate at additional FoxA1/NFIC co-

pioneered genes

(A) Effect of NFIC silencing on LPAR3, PEX10, CORO2A, DNM1L and

THRAP3 mRNA expression levels. LNCaP cells were maintained in hormone

depleted medium followed by transfection with either control negative control

siRNA or siRNA targeting against NFIC prior to stimulation with or without 10

nM of DHT for 12 hr. Total RNA was isolated and amplified with real-time RT-

qPCR primers for LPAR3, PEX10, CORO2A, DNM1L and THRAP3. mRNA

expression levels were normalized against GAPDH. (B) Effect of FoxA1

silencing on LPAR3, PEX10, CORO2A, DNM1L and THRAP3 expression levels.

(C) ChIP-qPCR of AR occupancy after NFIC silencing at the LPAR3, PEX10,

CORO2A, DNM1L and THRAP3. (D) ChIP-qPCR of AR occupancy after

FoxA1 silencing at the LPAR3, PEX10, CORO2A, DNM1L and THRAP3. (E)

ChIP-qPCR of FoxA1 occupancy after NFIC silencing at the LPAR3, PEX10,

CORO2A, DNM1L and THRAP3. (F) ChIP-qPCR of NFIC occupancy after

FoxA1 silencing at the LPAR3, PEX10, CORO2A, DNM1L and THRAP3. All

results represent the mean ± SEM of at least 3 independent assays.

105

We would like to examine whether both FoxA1 and NFIC contribute to the

remodeling of chromatin accessibility as well and subsequently performed FAIRE

experiments. As expected, FAIRE qPCR results suggest that both NFIC and

FoxA1 contribute to modulating chromatin accessibility at FoxA1/NFIC

pioneered ARBS (Figure 6A). Above results indicated a hierarchy of the AR

transcriptional regulation by NFIC and FoxA1, found at the NFIC and FoxA1 co-

pioneered ARBS.

The schematic diagram illustrating the role of NFIC and FoxA1 at NFIC/FoxA1

co-pioneered regions are shown in Figure 3.18. Both NFIC and FoxA1 contribute

to the modulation of chromatin accessibility in this type of ARBS. The chromatin

remodeling function of NFIC and FoxA1 facilitates subsequent AR recruitment

after DHT stimulation, therefore promotes the NFIC/FoxA1 co-pioneered genes

such as PEX10, LPAR3, AFF3, SASH1, PNMA1 and so on.

Taken together, the interplay between NFIC and FoxA1 has multiple mechanisms

and exhibits distinct patterns at different classes of ARBS.

106

(A)

(B)

107

Figure 3.18 Both NFIC and FoxA1 modulate chromatin accessibility and

contribute to pioneering AR mediated transcription at FoxA1/NFIC co-

pioneered regions

(A) Box plot showing the FAIRE qPCR experiments after NFIC or FoxA1

silencing on FoxA1 occupied/dependent ARBS. All FAIRE signal enrichments

are normalized against NC ETOH treated conditions. (B) Schematic diagram

illustrating how NFIC and FoxA1 coordinate AR transcriptional activity at

NFIC/FoxA1 overlapping ARBS where AR recruitment and gene regulation is co-

pioneered by NFIC and FoxA1, there is selective collaboration between NFIC and

FoxA1 to mediate AR dependent transcription at these ARBS.

108

DISCUSSION

AR has been shown to play a critical role in prostate development and

carcinogenesis, and it functions in both androgen-dependent and hormone-

refractory prostate cancer (Takayama et al., 2007; Wang et al., 2007). AR-

mediated transcription in prostate cancer involves the recruitment of the AR

together with general transcriptional machinery, collaborative factors,

coactivators, and corepressors in coordinated temporal and spatial fashion (Beato

and Sanchez-Pacheco, 1996; Heinlein and Chang, 2002; Heemers and Tindall,

2007; Chng et al., 2012). By analyzing the AR occupied region sequences

coupled with gene expression profiling, previous studies have identified a

network of AR transcriptional collaborative factors such as FoxA1, CEBPβ, Oct1

and GATA2 (Wang et al., 2007; Jia et al., 2008). Pioneering factors is an

emerging class of collaborative factors that are able to modulate condensed

chromatin to render the binding of additional transcription factors (Jozwik and

Carroll, 2012). Their role in hormone-dependent cancers has been implicated

recently. For instance, AP2γ and PBX1 mediate ER transcription in ER-positive

breast cancer (Magnani et al., 2011; Tan et al., 2011), and FoxA1 defines distinct

classes of AR binding programs in AR-positive prostate cancer (Sahu et al., 2011;

Wang et al., 2011).

109

In our study, chromatin immunoprecipitation coupled to massively parallel

sequencing (ChIP-Seq) allowed us to identify the genome-wide AR binding

cistromes in LNCaP prostate cancer cell line. And by combining molecular,

genomic, and bioinformatic approaches, we have identified NFIC as a novel

collaborative factor in the androgen signaling pathway that is critical for the AR

mediated gene transcription.

We showed that NFIC is globally recruited across the genome prior to androgen

stimulation and mediates AR binding to the cis-regulatory elements that drive

transcription of AR target genes, such as PSA, KLK2, and TMPRSS2.

Interestingly, the expressions of these AR model target genes have been shown to

be independent of FoxA1 regulation based on FoxA1 knockdown studies (Jia et

al., 2008; Sahu et al., 2011). Silencing FoxA1 in LNCaP prostate cancer cells

results in differentiating AR binding programs into distinct classes based on

subsequent AR occupancy and expression profiles, and FoxA1 was found to

function distinctively at these ARBS (Sahu et al., 2011; Wang et al., 2011). Those

well known AR target genes, such as PSA, KLK2 and TMPRSS2, happen to be

occupied by FoxA1, but neither their expression nor the recruitment of AR to sites

associated with these genes are affected by FoxA1 depletion. In such cases, AR

recruitment and chromatin structure might have been modulated by factors other

than FoxA1, and our study suggested that NFIC might be one of the regulators at

these FoxA1 independent ARBS.

Hence it intrigued us to examine the role of NFIC at the distinct ARBS

defined by FoxA1. Surprisingly, we found that NFIC has distinct roles in

110

regulating the AR transcriptional regulation at different classes of ARBS

classified by FoxA1 occupancy and functions. Our observation can be divided

into the following classes.

(1).For NFIC unique ARBS that are void of FoxA1 occupancy, NFIC serves as

the pioneering factor to facilitate AR recruitment at these ARBS. (2).For FoxA1

unique ARBS not occupied by NFIC, it is FoxA1 that plays the role as AR

pioneering factor to guide the binding of AR. (3). For NFIC and FoxA1 co-

occupied regions, these ARBS can be further divided into FoxA1 dependent and

independent sites. For FoxA1 independent regions, FoxA1 are found to occupy

the ARBS but does not contribute to AR transcriptional regulation, and it is NFIC

that plays the role of AR pioneering factor. At this class of ARBS, NFIC

facilitates the recruitment of AR, mediates downstream target gene expression

and contributes to the chromatin accessibility. While for FoxA1 dependent

regions, we found NFIC selectively collaborates with FoxA1, possibly by forming

a pioneering complex and both contributes to AR recruitment and subsequent

regulation of AR target genes, as well as modulating chromatin openness. In this

category, however, there are sites that exhibit no dependency between NFIC and

FoxA1 on each other, possibly suggesting both factors can also function as AR

pioneering factor independently at certain ARBS.

Studies on pioneering factors have also investigated various histone marks that

are possibly modulated by these factors in order to render better chromatin

accessibility. For instance, Forkhead and GATA family members have been

implicated to be able to open compacted chromatin by disrupting histone bindings

111

(Cirillo et al., 2002). FoxA1 has been predominantly found in H3K4me2 rich and

H3K9me2 poor regions in prostate cancer cells, and it is able to modulate

H3k4me2 in a genome wide manner and facilitate AR recruitment after androgen

stimulation (Sahu et al., 2011). Another pioneering factor GATA3 has been

shown to directly affect ER enhancer accessibility in breast cancer cells, and its

depletion resulted in a global redistribution of active histone marks such as

H3K4me1 and H3K27Ac (Theodorou et al., 2013). However, from the site

specific histone mark ChIP assays, we did not observe a significant change in

H3K4me1/me2 after depleting NFIC (Figure 4.1), although the chromatin

accessibility determined by FAIRE assays shows NFIC has certain impact. This

might be due to the possibility that NFIC modulates chromatin accessibility via

other histone marks at these ARBS.

Collectively, our data suggests that NFIC is one essential component of AR

transcriptional complex, and mediates the recruitment of AR and the transcription

of AR target genes in prostate cancer.

112

(A)

(B)

Figure 4.1 NFIC have no effect on H3K4me1/2 marks

(A) Boxplot showing the recruitment of H3K4me1 at 15 ARBS. Individual ChIP

assays have been repeated at least three times. (B) Boxplot showing the

recruitment of H3K4me2 at 15 ARBS. Individual ChIP assays have been repeated

at least three times.

113

CONCLUSION, FUTURE DIRECTIONS AND PERSPECTIVES

In the current study based on AR, NFIC ChIP-seq, we have identified a novel AR

cofactor belonging to the NFI family of transcription factors. From the genomic

analysis of the binding profile of NFIC, we further established NFIC as a novel

pioneering factor of AR that preoccupy the chromatin prior to DHT treatment and

facilities chromatin remodeling. Based on molecular characterization, we

demonstrated that NFIC is required for AR recruitment and the regulation of AR

target gene transcription. Motif analysis of the AR and NFIC co-occupied regions

reveals that FoxA1 might come into play in the NFIC and AR regulatory network.

From the expression profiling datasets, we selected genes differentially regulated

by FoxA1 and NFIC, and functionally dissect the role of NFIC as a pioneering

factor in relation to another pioneering factor, FoxA1, in mediating AR

transcriptional regulations. We discovered that NFIC mediates AR transcriptional

regulation at distinct classes of ARBS defined by their dependency on FoxA1,

and at these sites NFIC and FoxA1 differentially modulates chromatin

accessibility, thereby potentiated the role of NFIC in the prostate cancer

development.

In order to obtain a more comprehensive of picture of the pioneering factor role of

NFIC in mediating AR transcription, generating a genome wide AR ChIP-seq

library under the condition of NFIC depletion would provide important

114

information. However, due to the technical difficulty such as knockdown

efficiency happens to be the major challenge in this study. In future study of the

current project, it would be important to use a more efficient siRNA against NFIC

without disturbing AR expression levels to generate the genome wide binding

profile of AR, thus provide evidence for NFIC pioneering regulations globally.

Based on the oncomine database that deposit available clinical transcriptome

studies, we found that NFIC is overexpressed in many advanced prostate

carcinoma tissues relative to primary site (Lapointe et al., 2004; Yu et al., 2004).

This suggests that NFIC might play a role in the prostate cancer progressions. Our

current cell model LNCaP is an androgen responsive human prostate

adenocarcinoma cell line derived from the left supraclavicular lymph node

metastasis. It has been well characterized and chosen to be cell model for AR

positive prostate cancer studies based on its androgen responsiveness. However,

in terms of invasiveness, LNCaP is not very aggressive as observed in many

invasion assays, therefore to fully understand the role of NFIC in the progression

of the more advanced prostate cancer, characterization of other more invasive cell

lines might provide deeper insight.

Studies on pioneering factors of AR in the context of prostate cancer have

provided insight into the genome wide redistribution of the AR binding programs

after FoxA1 depletion (Sahu et al., 2011; Wang et al., 2011). They also left gaps

of knowledge for future studies to fill. For example, those ARBS independent of

FoxA1 regulation include several well characterized AR model genes such as

PSA, KLK2 and TMPRSS2, which are frequently implicated in the AR dependent

115

prostate cancer studies. If these model genes are not regulated by FoxA1, what

factor might contribute to the pioneering of AR regulation of these genes? Our

current study has partially filled this gap of knowledge by providing an

interplayer NFIC that pioneers the transcription of several FoxA1 independent

genes at ARBS co-occupied by NFIC and FoxA1. Since previous studies in

pioneering factors have assigned certain roles of epigenetic modulation to several

pioneer factors, our study also investigated the potential histone marks that might

be modulated by NFIC to facilitate chromatin remodeling at those ARBS it

pioneers, but so far we have not established a detailed epigenetic mechanism

regulated by NFIC. More histone marks such as other active histone methylation

and acetylation could be analyzed to understand how NFIC contributes to the AR

transcriptional regulation in the epigenetic point of view. It would also be

interesting to investigate the role of FoxA1 at those ARBS it occupies without

contributing to the regulation of the target gene.

Taken together, our studies established NFIC as a novel pioneering factor to

regulate AR transcription and dissected the multiple mechanisms of NFIC to co-

regulate AR transcription with FoxA1 at ARBS either dependent or independent

on FoxA1 regulation. We are currently investigating the functional role of NFIC

in regulating prostate cancer progession, and hopefully would provide deeper

insight and evidence for NFIC to be considered a potential attractive candidate as

a therapeutic target to treat prostate cancer.

116

BIBLIOGRAPHY

Agoulnik, I. U. and N. L. Weigel (2008). "Androgen receptor coactivators and

prostate cancer." Adv Exp Med Biol 617: 245-255.

Bachurski, C. J., S. E. Kelly, et al. (1997). "Nuclear factor I family members

regulate the transcription of surfactant protein-C." J Biol Chem 272(52):

32759-32766.

Bandyopadhyay, S. and R. M. Gronostajski (1994). "Identification of a conserved

oxidation-sensitive cysteine residue in the NFI family of DNA-binding

proteins." J Biol Chem 269(47): 29949-29955.

Beato, M. and A. Sanchez-Pacheco (1996). "Interaction of steroid hormone

receptors with the transcription initiation complex." Endocr Rev 17(6):

587-609.

Beck, S., P. Sommer, et al. (1999). "Hepatocyte nuclear factor 3 (winged helix

domain) activates trefoil factor gene TFF1 through a binding motif

adjacent to the TATAA box." DNA Cell Biol 18(2): 157-164.

Bedford, F. K., D. Julius, et al. (1998). "Neuronal expression of the 5HT3

serotonin receptor gene requires nuclear factor 1 complexes." J Neurosci

18(16): 6186-6194.

Bieberich, C. J., K. Fujita, et al. (1996). "Prostate-specific and androgen-

dependent expression of a novel homeobox gene." J Biol Chem 271(50):

31779-31782.

Boyes, J., P. Byfield, et al. (1998). "Regulation of activity of the transcription

factor GATA-1 by acetylation." Nature 396(6711): 594-598.

Brinkmann, A. O., P. W. Faber, et al. (1989). "The human androgen receptor:

domain structure, genomic organization and regulation of expression." J

Steroid Biochem 34(1-6): 307-310.

117

Carroll, J. S., X. S. Liu, et al. (2005). "Chromosome-wide mapping of estrogen

receptor binding reveals long-range regulation requiring the forkhead

protein FoxA1." Cell 122(1): 33-43.

Cereghini, S., M. Raymondjean, et al. (1987). "Factors involved in control of

tissue-specific expression of albumin gene." Cell 50(4): 627-638.

Chang, C., A. Saltzman, et al. (1995). "Androgen receptor: an overview." Crit

Rev Eukaryot Gene Expr 5(2): 97-125.

Chaudhry, A. Z., G. E. Lyons, et al. (1997). "Expression patterns of the four

nuclear factor I genes during mouse embryogenesis indicate a potential

role in development." Dev Dyn 208(3): 313-325.

Chaudhry, A. Z., A. D. Vitullo, et al. (1999). "Nuclear factor I-mediated

repression of the mouse mammary tumor virus promoter is abrogated by

the coactivators p300/CBP and SRC-1." J Biol Chem 274(11): 7072-7081.

Chen, D., H. Ma, et al. (1999). "Regulation of transcription by a protein

methyltransferase." Science 284(5423): 2174-2177.

Chikhirzhina, G. I., R. I. Al'-Shekhadat, et al. (2008). "[Transcription factors of

the nuclear factor 1 (NF1) family. Role in chromatin remodelation]." Mol

Biol (Mosk) 42(3): 388-404.

Chng, K. R., C. W. Chang, et al. (2012). "A transcriptional repressor co-

regulatory network governing androgen response in prostate cancers."

EMBO J 31(12): 2810-2823.

Cirillo, L. A., F. R. Lin, et al. (2002). "Opening of compacted chromatin by early

developmental transcription factors HNF3 (FoxA) and GATA-4." Mol

Cell 9(2): 279-289.

Cleutjens, K. B., C. C. van Eekelen, et al. (1996). "Two androgen response

regions cooperate in steroid hormone regulated activity of the prostate-

specific antigen promoter." J Biol Chem 271(11): 6379-6388.

Coffey, D. S. and K. J. Pienta (1987). "New concepts in studying the control of

normal and cancer growth of the prostate." Prog Clin Biol Res 239: 1-73.

Courtois, S. J., D. A. Lafontaine, et al. (1990). "Nuclear factor-I and activator

protein-2 bind in a mutually exclusive way to overlapping promoter

118

sequences and trans-activate the human growth hormone gene." Nucleic

Acids Res 18(1): 57-64.

Cunha, G. R., A. A. Donjacour, et al. (1987). "The endocrinology and

developmental biology of the prostate." Endocr Rev 8(3): 338-362.

Deplewski, D. and R. L. Rosenfield (2000). "Role of hormones in pilosebaceous

unit development." Endocr Rev 21(4): 363-392.

Desjardins, C. (1978). "Endocrine regulation of reproductive development and

function in the male." J Anim Sci 47 Suppl 2: 56-79.

Eeckhoute, J., J. S. Carroll, et al. (2006). "A cell-type-specific transcriptional

network required for estrogen regulation of cyclin D1 and cell cycle

progression in breast cancer." Genes Dev 20(18): 2513-2526.

Eeckhoute, J., M. Lupien, et al. (2009). "Cell-type selective chromatin remodeling

defines the active subset of FOXA1-bound enhancers." Genome Res 19(3):

372-380.

Esnault, G., S. Majocchi, et al. (2009). "Transcription factor CTF1 acts as a

chromatin domain boundary that shields human telomeric genes from

silencing." Mol Cell Biol 29(9): 2409-2418.

Finkelstein, J. S., A. Klibanski, et al. (1987). "Osteoporosis in men with

idiopathic hypogonadotropic hypogonadism." Ann Intern Med 106(3):

354-361.

Furlong, E. E., T. Rein, et al. (1996). "YY1 and NF1 both activate the human p53

promoter by alternatively binding to a composite element, and YY1 and

E1A cooperate to amplify p53 promoter activity." Mol Cell Biol 16(10):

5933-5945.

Gao, N., J. Zhang, et al. (2003). "The role of hepatocyte nuclear factor-3 alpha

(Forkhead Box A1) and androgen receptor in transcriptional regulation of

prostatic genes." Mol Endocrinol 17(8): 1484-1507.

Gelmann, E. P. (2002). "Molecular biology of the androgen receptor." J Clin

Oncol 20(13): 3001-3015.

Geurts, J. M., E. F. Schoenmakers, et al. (1998). "Identification of NFIB as

recurrent translocation partner gene of HMGIC in pleomorphic

adenomas." Oncogene 16(7): 865-872.

119

Gronostajski, R. M. (2000). "Roles of the NFI/CTF gene family in transcription

and development." Gene 249(1-2): 31-45.

Gronostajski, R. M., S. Adhya, et al. (1985). "Site-specific DNA binding of

nuclear factor I: analyses of cellular binding sites." Mol Cell Biol 5(5):

964-971.

Harris, W. P., E. A. Mostaghel, et al. (2009). "Androgen deprivation therapy:

progress in understanding mechanisms of resistance and optimizing

androgen depletion." Nat Clin Pract Urol 6(2): 76-85.

He, B., J. T. Minges, et al. (2002). "The FXXLF motif mediates androgen

receptor-specific interactions with coregulators." J Biol Chem 277(12):

10226-10235.

He, B. and E. M. Wilson (2003). "Electrostatic modulation in steroid receptor

recruitment of LXXLL and FXXLF motifs." Mol Cell Biol 23(6): 2135-

2150.

Heemers, H. V. and D. J. Tindall (2007). "Androgen receptor (AR) coregulators:

a diversity of functions converging on and regulating the AR

transcriptional complex." Endocr Rev 28(7): 778-808.

Heinlein, C. A. and C. Chang (2002). "Androgen receptor (AR) coregulators: an

overview." Endocr Rev 23(2): 175-200.

Heinlein, C. A. and C. Chang (2004). "Androgen receptor in prostate cancer."

Endocr Rev 25(2): 276-308.

Isaacs, J. T. and D. S. Coffey (1989). "Etiology and disease process of benign

prostatic hyperplasia." Prostate Suppl 2: 33-50.

Jia, L., B. P. Berman, et al. (2008). "Genomic androgen receptor-occupied regions

with different functions, defined by histone acetylation, coregulators and

transcriptional capacity." PLoS One 3(11): e3645.

Johansson, E. M., M. Kannius-Janson, et al. (2003). "The p53 tumor suppressor

gene is regulated in vivo by nuclear factor 1-C2 in the mouse mammary

gland during pregnancy." Oncogene 22(38): 6061-6070.

Jozwik, K. M. and J. S. Carroll (2012). "Pioneer factors in hormone-dependent

cancers." Nat Rev Cancer 12(6): 381-385.

120

Karim, F. D., L. D. Urness, et al. (1990). "The ETS-domain: a new DNA-binding

motif that recognizes a purine-rich core DNA sequence." Genes Dev 4(9):

1451-1453.

Kim, M. Y., J. Reyna, et al. (2009). "Role of the transcription factor NFIC in

odontoblast gene expression." J Calif Dent Assoc 37(12): 875-881.

Kumar-Sinha, C., S. A. Tomlins, et al. (2008). "Recurrent gene fusions in prostate

cancer." Nat Rev Cancer 8(7): 497-511.

Lapointe, J., C. Li, et al. (2004). "Gene expression profiling identifies clinically

relevant subtypes of prostate cancer." Proc Natl Acad Sci U S A 101(3):

811-816.

Lee, C. S., J. R. Friedman, et al. (2005). "The initiation of liver development is

dependent on Foxa transcription factors." Nature 435(7044): 944-947.

Lee, D. K., M. Li, et al. (2003). "The second largest subunit of RNA polymerase

II interacts with and enhances transactivation of androgen receptor."

Biochem Biophys Res Commun 302(1): 162-169.

Lee, D. S., J. T. Park, et al. (2009). "Nuclear factor I-C is essential for

odontogenic cell proliferation and odontoblast differentiation during tooth

root development." J Biol Chem 284(25): 17293-17303.

Liao, S. and S. Fang (1969). "Receptor-proteims for androgens and the mode of

action of androgens on gene transcription in ventral prostate." Vitam

Horm 27: 17-90.

Lupien, M., J. Eeckhoute, et al. (2008). "FoxA1 translates epigenetic signatures

into enhancer-driven lineage-specific transcription." Cell 132(6): 958-970.

Magnani, L., E. B. Ballantyne, et al. (2011). "PBX1 genomic pioneer function

drives ERalpha signaling underlying progression in breast cancer." PLoS

Genet 7(11): e1002368.

Magnani, L., J. Eeckhoute, et al. (2011). "Pioneer factors: directing transcriptional

regulators within the chromatin environment." Trends Genet 27(11): 465-

474.

Marshall, T. W., K. A. Link, et al. (2003). "Differential requirement of SWI/SNF

for androgen receptor activity." J Biol Chem 278(33): 30605-30613.

121

Massie, C. E., B. Adryan, et al. (2007). "New androgen receptor genomic targets

show an interaction with the ETS1 transcription factor." EMBO Rep 8(9):

871-878.

Mellor, J. (2006). "Dynamic nucleosomes and gene transcription." Trends Genet

22(6): 320-329.

Mermod, N., E. A. O'Neill, et al. (1989). "The proline-rich transcriptional

activator of CTF/NF-I is distinct from the replication and DNA binding

domain." Cell 58(4): 741-753.

Messenger, A. G. (1993). "The control of hair growth: an overview." J Invest

Dermatol 101(1 Suppl): 4S-9S.

Metzger, E., M. Wissmann, et al. (2005). "LSD1 demethylates repressive histone

marks to promote androgen-receptor-dependent transcription." Nature

437(7057): 436-439.

Mooradian, A. D., J. E. Morley, et al. (1987). "Biological actions of androgens."

Endocr Rev 8(1): 1-28.

Perez-Casellas, L. A., X. Wang, et al. (2009). "Nuclear factor I transcription

factors regulate IGF binding protein 5 gene transcription in human

osteoblasts." Biochim Biophys Acta 1789(2): 78-87.

Powers, G. L. and P. C. Marker (2013). "Recent advances in prostate

development and links to prostatic diseases." Wiley Interdiscip Rev Syst

Biol Med 5(2): 243-256.

Qian, F., U. Kruse, et al. (1995). "Chromosomal localization of the four genes

(NFIA, B, C, and X) for the human transcription factor nuclear factor I by

FISH." Genomics 28(1): 66-73.

Randall, V. A. (1994). "Role of 5 alpha-reductase in health and disease."

Baillieres Clin Endocrinol Metab 8(2): 405-431.

Raudrant, D. and T. Rabe (2003). "Progestogens with antiandrogenic properties."

Drugs 63(5): 463-492.

Sahu, B., M. Laakso, et al. (2011). "Dual role of FoxA1 in androgen receptor

binding to chromatin, androgen signalling and prostate cancer." EMBO J

30(19): 3962-3976.

122

Schmidt, D., M. D. Wilson, et al. (2009). "ChIP-seq: using high-throughput

sequencing to discover protein-DNA interactions." Methods 48(3): 240-

248.

Schroeder, E. T., L. Zheng, et al. (2004). "Effects of androgen therapy on adipose

tissue and metabolism in older men." J Clin Endocrinol Metab 89(10):

4863-4872.

Serandour, A. A., S. Avner, et al. (2011). "Epigenetic switch involved in

activation of pioneer factor FOXA1-dependent enhancers." Genome Res

21(4): 555-565.

Sevilla, L., C. Aperlo, et al. (1999). "The Ets2 transcription factor inhibits

apoptosis induced by colony-stimulating factor 1 deprivation of

macrophages through a Bcl-xL-dependent mechanism." Mol Cell Biol

19(4): 2624-2634.

Shang, Y., M. Myers, et al. (2002). "Formation of the androgen receptor

transcription complex." Mol Cell 9(3): 601-610.

Sharrocks, A. D. (2001). "The ETS-domain transcription factor family." Nat Rev

Mol Cell Biol 2(11): 827-837.

Sharrocks, A. D., A. L. Brown, et al. (1997). "The ETS-domain transcription

factor family." Int J Biochem Cell Biol 29(12): 1371-1387.

Shin, S., T. D. Kim, et al. (2009). "Induction of prostatic intraepithelial neoplasia

and modulation of androgen receptor by ETS variant 1/ETS-related

protein 81." Cancer Res 69(20): 8102-8110.

Steele-Perkins, G., K. G. Butz, et al. (2003). "Essential role for NFI-C/CTF

transcription-replication factor in tooth root development." Mol Cell Biol

23(3): 1075-1084.

Sun, C., A. Dobi, et al. (2008). "TMPRSS2-ERG fusion, a common genomic

alteration in prostate cancer activates C-MYC and abrogates prostate

epithelial differentiation." Oncogene 27(40): 5348-5353.

Sun, P., P. Dong, et al. (1998). "p53-independent role of MDM2 in TGF-beta1

resistance." Science 282(5397): 2270-2272.

Suzuki, H. and H. Ito (1999). "Role of androgen receptor in prostate cancer."

Asian J Androl 1(3): 81-85.

123

Takayama, K., K. Kaneshiro, et al. (2007). "Identification of novel androgen

response genes in prostate cancer cells by coupling chromatin

immunoprecipitation and genomic microarray analysis." Oncogene 26(30):

4453-4463.

Tan, P. Y., C. W. Chang, et al. (2012). "Integration of regulatory networks by

NKX3-1 promotes androgen-dependent prostate cancer survival." Mol

Cell Biol 32(2): 399-414.

Tan, S. K., Z. H. Lin, et al. (2011). "AP-2gamma regulates oestrogen receptor-

mediated long-range chromatin interaction and gene transcription."

EMBO J 30(13): 2569-2581.

Theodorou, V., R. Stark, et al. (2013). "GATA3 acts upstream of FOXA1 in

mediating ESR1 binding by shaping enhancer accessibility." Genome Res

23(1): 12-22.

Tomlins, S. A., B. Laxman, et al. (2008). "Role of the TMPRSS2-ERG gene

fusion in prostate cancer." Neoplasia 10(2): 177-188.

Tsai, M. J. and B. W. O'Malley (1994). "Molecular mechanisms of action of

steroid/thyroid receptor superfamily members." Annu Rev Biochem 63:

451-486.

Wang, D., I. Garcia-Bassets, et al. (2011). "Reprogramming transcription by

distinct classes of enhancers functionally defined by eRNA." Nature

474(7351): 390-394.

Wang, Q., J. S. Carroll, et al. (2005). "Spatial and temporal recruitment of

androgen receptor and its coactivators involves chromosomal looping and

polymerase tracking." Mol Cell 19(5): 631-642.

Wang, Q., W. Li, et al. (2007). "A hierarchical network of transcription factors

governs androgen receptor-dependent prostate cancer growth." Mol Cell

27(3): 380-392.

Watson, C. J., K. E. Gordon, et al. (1991). "Interaction of DNA-binding proteins

with a milk protein gene promoter in vitro: identification of a mammary

gland-specific factor." Nucleic Acids Res 19(23): 6603-6610.

Wilson, C. M. and M. J. McPhaul (1994). "A and B forms of the androgen

receptor are present in human genital skin fibroblasts." Proc Natl Acad Sci

U S A 91(4): 1234-1238.

124

Wissmann, M., N. Yin, et al. (2007). "Cooperative demethylation by JMJD2C and

LSD1 promotes androgen receptor-dependent gene expression." Nat Cell

Biol 9(3): 347-353.

Xu, H., L. Handoko, et al. (2010). "A signal-noise model for significance analysis

of ChIP-seq with negative control." Bioinformatics 26(9): 1199-1204.

Xu, J. and B. W. O'Malley (2002). "Molecular mechanisms and cellular biology

of the steroid receptor coactivator (SRC) family in steroid receptor

function." Rev Endocr Metab Disord 3(3): 185-192.

Yamane, K., C. Toumazou, et al. (2006). "JHDM2A, a JmjC-containing H3K9

demethylase, facilitates transcription activation by androgen receptor."

Cell 125(3): 483-495.

Yu, J., R. S. Mani, et al. (2010). "An integrated network of androgen receptor,

polycomb, and TMPRSS2-ERG gene fusions in prostate cancer

progression." Cancer Cell 17(5): 443-454.

Yu, Y. P., D. Landsittel, et al. (2004). "Gene expression alterations in prostate

cancer predicting tumor aggression and preceding development of

malignancy." J Clin Oncol 22(14): 2790-2799.

Zaret, K. S. and J. S. Carroll (2011). "Pioneer transcription factors: establishing

competence for gene expression." Genes Dev 25(21): 2227-2241.

Zhang, Z., C. W. Chang, et al. (2011). "CENTDIST: discovery of co-associated

factors by motif distribution." Nucleic Acids Res 39(Web Server issue):

W391-399.

125

APPENDIX I

List of ChIP real-time quantitative PCR primers

NFIC_H1_F TGCTGCCAGCCAGTATGCT

NFIC_H1_R GTTGGCCTCTGTACCTGATTCTG

NFIC_H2_F TGCGGCCAGTGTCAACAA

NFIC_H2_R GGTAATCGGGCTATGGTAATGC

NFIC_H3_F ATGGCCATCCTGCGAAAC

NFIC_H3_R TCTGTGGCAATAACTGTCAGTATCG

NFIC_H4_F TGGAGCAGTCCAGGACATTG

NFIC_H4_R GAGGTCAGCGCCACAGTTG

NFIC_H5_F CTGGGTGTCAAGTTCATGTGTGT

NFIC_H5_R GCACAGCCTGGCTTGCA

NFIC_M1_F TCCTTGAGGGAAAGAACACACA

NFIC_M1_R CAGCACACTGGCTGCCAAT

NFIC_M2_F AAGGACAAACAGAACTGGACACAA

NFIC_M2_R GCTAGTAATTCCCAGACAGAAAACAA

NFIC_M3_F TTGAGGGACCTGCCTGGAT

NFIC_M3_R CTGCCCCACCCTCAACAG

NFIC_M4_F CAAGTTTAAACCAAAACCAGAAATATCTAC

NFIC_M4_R CCTCTTAAAAATAAACAGATGAGCATGT

NFIC_M5_F GAAGGATTAAAGGGTACATTATGTGAAAT

NFIC_M5_R CCCAGCCTCTACTTGACTTGTAAAA

NFIC_L1_F CCCACTGAACTGTGAGCTCTGT

NFIC_L1_R TGGGCACTCACGATTTAATAATAAAC

NFIC_L2_F TCCTCTGTGCCTTGCCTATACA

NFIC_L2_R AACTGGGCTCAATTCCAAACA

NFIC_L3_F GAGGGAACATGGGTTATACAAGCT

126

NFIC_L3_R AGCCAAGACCCAAGCAATCTT

NFIC_L4_F AGAGCCAGAGAAGTGACAAGGAA

NFIC_L4_R CCACTGTGCTTGGCTGAGTTC

NFIC_L5_F TCAGTCACCTCCCTCCAACAC

NFIC_L5_R GGCAGAACTAGTGATGTGCAGAA

NFIC_L6_F TGGTAGAGGTGGGATTGAAACC

NFIC_L6_R GAAGGAAGAGACAGCACAGTATGC

NFIC_L7_F CCCTGTGTGTGACCTCGATTG

NFIC_L7_R TTTTTGTAAGTGGGCTGTCCAA

NFIC_L8_F GATCCACCACAGCACAGGAA

NFIC_L8_R CCTTGCTGCTCCCAGTCATT

NFIC_L9_F ACAAGTAACTACTCCCTTGAAAATCAGA

NFIC_L9_R GCAAGCACCACAGGAATGAA

NFIC_L10_F TCGACCTACATGTGGCTGAATG

NFIC_L10_R CTCAACATGCAGCAGTCAGATG

AR/NFIC_1_F GCACCTGCTGCGCAATC

AR/NFIC_1_R TCGCTGGAGGCCCACAT

AR/NFIC_2_F GGGAGCACAAGCAGAGCTTCT

AR/NFIC_2_R GGGTGTTCTTAGGAGCTGGAAA

AR/NFIC_3_F GCTCCCTGTCTTCTCCCAGTCT

AR/NFIC_3_R AAGGACTCGAGCAGGGAGATT

AR/NFIC_4_F CTGGAGCTTGCACAATCGAA

AR/NFIC_4_R TGCCAAGGAGACAGCACCTA

AR/NFIC_5_F TGCTGCCAGCCAGTATGCT

AR/NFIC_5_R GTTGGCCTCTGTACCTGATTCTG

AR/NFIC_6_F CGCCACGGCCTAATCCT

AR/NFIC_6_R AAGGGAGAGGTCGGTTTATATAATCTC

AR/NFIC_7_F CAGGGATGCACGTTGATCAG

AR/NFIC_7_R GAGAGGCAGCTGCAAACTTGA

AR/NFIC_8_F GCTCCCTGTCTTCTCCCAGTCT

AR/NFIC_8_R AAGGACTCGAGCAGGGAGATT

127

AR/NFIC_9_F TGGAGCAGTCCAGGACATTG

AR/NFIC_9_R GAGGTCAGCGCCACAGTTG

AR/NFIC_10_F GCTCAGTCTGTCCCCAGTAGAAA

AR/NFIC_10_R CCAAGCTGGCGACTGCTT

AR/FOXA1_1_F TCTTGTGTATGTGAGTGCAGCATAG

AR/FOXA1_1_R GTCCAGTACCAAAATGTAAAGCTTCTT

AR/FOXA1_2_F TGGCAAGTATGGAAGCCTTTATC

AR/FOXA1_2_R CCGGTTCCCTGTTGAATTACA

AR/FOXA1_3_F ACAATGCCAAGTGTTGGTGA

AR/FOXA1_3_R TGGTGGGTGGATTATACCTCA

AR/FOXA1_4_F TTCCCTTCCTTTCTTTTCAAGATACT

AR/FOXA1_4_R GCTGCTTACATTTGGATTATAACTGTTG

AR/FOXA1_5_F CAGCAAAAGAAACTATCAACAGAGTGA

AR/FOXA1_5_R TGCATTGTTTGCAAATAGCTTCTAC

AR/FOXA1_6_F GCTCTCTGGGAGTTAGGTACTGTCA

AR/FOXA1_6_R GCAGGGATTTGGTCTAGTTTATTCA

AR/FOXA1_7_F GCCACTGTACCTTGGAGAGG

AR/FOXA1_7_R AGGAAGGCCATCTGGATTTT

AR/FOXA1_8_F TGC CAC CTT GTG TAT TTT TTG ATA C

AR/FOXA1_8_R TCA CGC CAC TGT GTA CAA TAC ATA GT

AR/FOXA1_9_F GTG TTA AGA GCT CTC ACA ATA CAA GGA

AR/FOXA1_9_R AAT GTC CCC ATT AAA CCA ATA AGG

AR/FOXA1_10_F GGG AAG AGA TTA GTG TTT TGA GAT GTT

AR/FOXA1_10_R GCA AAA GCA CTG GTA ATC TCA GTA AA

PSA enh_F TGGGACAACTTGCAAACCTG

PSA enh_R CCAGAGTAGGTCTGTTTTCAA

KLK2 enh_F GTTGAAAGCAGACCTACTCTGGA

KLK2 enh_R CTGGACCATCTTTTCAAGCAT

KLK2 prom_F GGGAATGCCTCCAGACTGAT

KLK2 prom_R CTTGCCCTGTTGGCACCTA

TMPRSS2_1_F CAAATGGCCACCTGGTGAA

128

TMPRSS2_1_R TTAAAGGGTACGGCAGGTACTCA

TMPRSS2_2_F GGCCCCATCTCCCAATATTG

TMPRSS2_2_R TCCCCCTAAAAGCATGTGTTG

PEX10_ChIP_F CACACAAAAAATTCTCATCATGTCACT

PEX10_ChIP_R CAGGTGGGCAGCCATTG

LPAR3_ChIP_F CTTGTTTGCTTGGCAATTCTAGTTAG

LPAR3_ChIP_R GCCAAAGGCCCCATGAA

DNM1L_chip_F CCAGGCTGGCATGATCTCA

DNM1L_chip_R GGCAGGAGAATCACTTAGAACACA

CORO2A_chip_F TTGATCAACAGTGGCAATCACA

CORO2A_chip_R AGGAGCAAAGCAAGCTGAAGA

THRAP3_chip_F GAAGAGCCTCCACTGGGAATG

THRAP3_chip_R ACCTTAAACTCTCCCCCAATTACA

AADAT_ChIP_F TGCACTGTAACTCAGCGAAAGG

AADAT_ChIP_R GCAGTGTGGTGCTGCTGAAC

MBOAT2_ChIP_F CCAGTGATGCACTTAGGATTATCAA

MBOAT2_ChIP_R GATGCCAGTGTAAACAAACCTAATGT

NDRG1_ChIP_F TGTTCTGCCTCATAACTGTTGTTTAA

NDRG1_ChIP_R GGTCTCAGAGCCAGTAAGCTGACT

SGK3_ChIP_F CACCGGGTGTGCCTTCTG

SGK3_ChIP_R GGGTGGTGGCTTGTCTTACAG

SEC24D_chip_F TCTCTTCCAGCACACTCTAAGCA

SEC24D_chip_R CCCCAATGCCAGTTTCATG

SASH1_chip_F CCGTTACCTTGGCCATGTCTA

SASH1_chip_R TCCCTCGCAGACAACGATCT

LRIG1_chip_F ATTCAGGTGGAGACTGGATAAAAAA

LRIG1_chip_R CCTGGCGGAGGCTGTTT

AFF3_chip_F CCAGCTGTTATTTACACTGCCTACA

AFF3_chip_R TGAACCCTTCAGCCTTTTGTG

SLC2A12_chip_F CCCACCCACTTGTGCTTGA

SLC2A12_chip_R CCAGAGGAGACCCAAACATTG

129

PNMA1_chip_F AAAATCTATCCGCGTTATCAGTAATCT

PNMA1_chip_R GGCCTGGCTCACAGGAAGT

FBXO28_chip_F GGGCAATAAAAAGGTACATGCTACA

FBXO28_chip_R GTTTCTGGCTTCTTTCATTTAGCATA

SEPP1_chip_F CCATTGTCTGCCTGTGGAACT

SEPP1_chip_R TGCTCACGGACCCAGGAA

DEGS1_chip_F GGGCTCCCGGAATTTGTC

DEGS1_chip_R GGCAGGGAACAATGTGTGATT

control_F CCTGGAGGGCTTGGAGATG

control_R GATCCTACGGCTGGCTGTGA

130

APPENDIX II

List of cDNA real-time quantitative PCR primers

AR_cDNA_F GTGTCACTATGGAGCTCTCACATGT

AR_cDNA_R GTTTCCCTTCAGCGGCTCTT

NFIC_cDNA_F GGACATGGAAGGAGGCATCTC

NFIC_cDNA_R GGGCTGTTGAATGGTGACTTG

PSA_cDNA_F TGTGTGCTGGACGCTGGA

PSA_cDNA_R CACTGCCCCATGACGTGAT

TMPRSS2_cDNA_F GGACAGTGTGCACCTCAAAGAC

TMPRSS2_cDNA_R TCCCACGAGGAAGGTCCC

KLK2_cDNA_F CTGTGTCAGCATGTGGGACCT

KLK2_cDNA_R CCATGATGTGATACCTTGAAGCA

PEX10_cDNA_F GGTTGAGCTGCTCTCAGATGTG

PEX10_cDNA_R AGGGTCTGGTAGCCTGCAAGT

LPAR3_cDNA_F CTCATGGCCTTCCTCATCAT

LPAR3_cDNA_R TACCACAAACGCCCCTAAGA

DNM1L_cDNA_F ATCCACTTGGTGGCCTTAACA

DNM1L_cDNA_R GGACGAGGACCAGTAGCATTTC

CORO2A_cDNA_F GAATGCCCCCCACCAAA

CORO2A_cDNA_R CTCCTGTTGCCGGTAGAACATC

THRAP3_cDNA_F TGTCCGGATGGACTCTTTTGAT

THRAP3_cDNA_R CTTGCGTTCCTGAGCCAATAA

AADAT_cDNA_F GAGAAACGGGCTGACATATTGAG

AADAT_cDNA_R CCACCAGCCAAGGAGATCAT

MBOAT2_cDNA_F CTGATTTTTCAGGCCCAATGA

MBOAT2_cDNA_R CATGAATTTCGCAAGCCAAA

NDRG1_cDNA_F CGGCATGAACCACAAAACC

131

NDRG1_cDNA_R TGATCTCCTGCATGTCCTCGTA

SGK3_cDNA_F GCTGGGCTGACCTTGTACAAA

SGK3_cDNA_R TCTGGTCCAGCCACATTAGGA

SEC24D_cDNA_F ACTGAAGACCATGCTGGAAAAAA

SEC24D_cDNA_R CACTCGAATTGCAGACGTCTCT

SASH1_cDNA_F CGCATGGCGATTCCAAGT

SASH1_cDNA_R CGAAATGCACCCAGGAGAGA

AFF3_cDNA_F CGGAGACCTCACGAAAGCA

AFF3_cDNA_R TCTGCGGCCGTGACTTGT

SLC2A12_cDNA_F TTTTTGAAATGGCTGTCCTTAGC

SLC2A12_cDNA_R CCAGGGCATTGGTCCTAGAC

PNMA1_cDNA_F CCACCCTTCTGCAGCTTTAGTG

PNMA1_cDNA_R GACCCCACAGGGACAAGAAA

FBXO28_cDNA_F CCCAAGGAGAGAGTCAGAAAGGA

FBXO28_cDNA_R TCTCAACACACGATAAATCTCATCAA

SEPP1_cDNA_F TGCTCTCTCACGACTCTCAAAGA

SEPP1_cDNA_R GCGATGGAGTTTCAACTGTTTTATC

DEGS1_cDNA_F CGGGAGATCCTGGCAAAGTA

DEGS1_cDNA_R CCAACTGGGTGAGAACCATCA

GAPDH_cDNA_F GGCCTCCAAGGAGTAAGACC

GAPDH_cDNA_R AGGGGAGATTCAGTGTGGTG

132

APPENDIX III

List of siRNA sense sequences

siNC 1st BASE Pte Ltd

sense 5'-rUrUrC rUrCrC rGrArA rCrGrU rGrUrC rArCrG rUTT-3'

antisense 5'-rArCrG rUrGrA rCrArC rGrUrU rCrGrG rArGrA rATT-3'

siNFIC Dharmacon ON-TARGETplus SMARTpool siRNA NM_005597

sense 5'-GGAGCAAGCGGCACAAAUC-3'

sense 5'-GAACAGGACCCAACUUCUC-3'

sense 5'-AGACAGAGAUGGACAAGUC-3'

sense 5'-GCUCAAAGAUCUUGUCUCG-3'

siFoxA1 Dharmacon

sense 5'-GAGAGAAAAAAUCAACAGC-3'

antisense 5'-GCUGUUGAUUUUUUCUCUC-3'.