Cyclin D1: Mechanism and Consequence of Androgen Receptor Co-Repressor Activity in Prostatic Adenocarcinoma

UNIVERSITY OF CINCINNATI

Date: April 1st, 2004

I, ______Christin E. Petre______, hereby submit this work as part of the requirements for the degree of: Doctor of Philosophy in: Cell and Molecular Biology It is entitled: Cyclin D1: Mechanism and Consequence of Androgen Receptor Co-repressor Activity in Prostatic Adenocarcinoma

This work and its defense approved by:

Chair: Karen E. Knudsen, Ph.D. Sohaib Khan, Ph.D. Kenji Fukasawa, Ph.D. Alvaro Puga, Ph.D. Linda Parysek, Ph.D. J. Alan Diehl, Ph.D. Robert Hennigan, Ph.D.

CYCLIN D1: MECHANISM AND CONSEQUENCE OF ANDROGEN RECEPTOR CO-REPRESSOR ACTIVITY IN PROSTATIC ADENOCARCINOMA

A dissertation submitted to the

Division of Research and Advanced Studies of the University of Cincinnati

in partial fulfillment of the requirements for the degree of

DOCTORATE OF PHILOSOPHY (Ph.D.)

in the Department of Cell Biology, Neurobiology, and Anatomy of the College of Medicine

2004

Christin E. Petre

B.A., Miami University, 2000

Committee Chair: Karen E. Knudsen, Ph.D.

ABSTRACT

It is increasingly evident that androgen receptor (AR) regulation plays a critical

role in the development and progression of prostate cancer. We show that induction of

cyclin D1 occurs upon ligand stimulation in androgen dependent prostatic

adenocarcinoma (LNCaP) cells. Such induction results in the formation of active cyclin

dependent kinase (CDK)-cyclin D1 complexes, phosphorylation of the rentinoblastoma

tumor suppressor protein, and concomitant cell cycle progression. In addition, we find

that cyclin D1 harbors a second cell cycle independent function responsible for restraining AR transactivation. We illustrate that cyclin D1 co-repressor activity is

extremely potent, inhibiting receptor transactivation independently of cellular

background, promoter context, co-activator over expression, ligand/non-ligand activators, and cancer predisposing AR mutations/polymorphisms. Cyclin D1 binds directly to the AR N-terminus, hindering its ligand dependent transactivation functions

(AF-1 and AF-2). The co-repressor activity of cyclin D1 likely involves two distinct mechanisms, including the recruitment of histone deactylase (HDAC) activity and inhibition of AR N- to C- terminal interactions. These data put forth the hypothesis that cyclin D1 is a negative feedback inhibitor of the AR. Supporting this model, over expression of cyclin D1 in LNCaP (prostate cancer) cells leads to marked abrogation of cell cycle progression. Further investigation into the region(s) of cyclin D1 responsible for AR co-repression reveal a conserved central portion of the protein required for both receptor binding and inhibition. The central domain itself, elicits many of the characteristics of the wild type protein, functioning to bind HDAC3, inhibit AR transactivation, and abrogate cell cycle progression in LNCaP cells. We show that the central domain of cyclin D1 is also required for co-repression of thyroid hormone

receptor beta-1 transactivation, suggesting the existence of a conserved nuclear

receptor repressor motif within this region. Surprisingly, the central domain is dispensable for estrogen receptor alpha co-activation, suggesting that the co-activator

and co-repressor functions of cyclin D1 are distinct and could be specifically targeted.

Together, these studies provide the first in-depth analysis of AR co-repressor function,

identifying the mechanism of cyclin D1 action and potentially leading to the discovery of

novel therapeutic targets for the treatment of prostate cancer.

ACKNOWLEDGEMENTS

I must thank, and feel privileged to thank, my friends and family. Without your support and encouragement I would never have overcome the obstacles that inevitably every graduate student encounters on their journey. To my fiancé, Phil, you have been an honest and motivating presence in the toughest of times, always expressing confidence in my abilities and never allowing me to admit defeat. You have been the light at the end of my day, a constant reminder that there is life to experience outside the world of science. I only hope that I can be as much a support in your work as you have been in mine.

Thank you also to my fellow graduate students, the administration, and the professors of the Cell Biology Department. Each with your own individuality, you have truly joined to create a ‘departmental family’, adding to my success through your friendship, support, and advice. In addition, the members of the Knudsen laboratories, both past and present have made these four years enjoyable.

Most importantly, I thank my advisor, Dr. Karen Knudsen. I decided to come to UC because I thought I would receive the best possible training as a graduate student; I have no regrets. Karen, you have taught me more than I could ever have expected. You are truly an amazing woman, capable of successfully juggling numerous responsibilities both at home and in the office. To Karen’s husband, Dr. Erik Knudsen, I thank you for your resourcefulness and for showing me an extremely no-nonsense view of the scientific world. Together you have taught me, nurtured me, and most importantly, you have allowed me to grow scientifically and personally. You have shaped my science and have truly become a part of my life.

This dissertation is dedicated to my Grandpa, Donald Beecher, who, like many other prostate cancer survivors I've met in these past few years, will never give up hope that we, as researchers, will find better ways to combat this devastating disease. It is the strength and attitude of individuals like you that drive all of us to be successful.

TABLE OF CONTENTS

Page List of Tables and Figures…………………..………..……………………...…….……2

Chapter I: Introduction……………………..………………………………………………….……..5 A. The evolution of prostate cancer…..……..……………..……...….…….5 B. The androgen receptor…….…..…..………………………………..…….6 C. Androgen receptor co-modulators...………...………..………….….…..8 D. AR deregulation in androgen independent prostate cancer……...….10 E. The dual actions of cyclin D1……………….……………………….…..12 F. Introductory conclusions and hypothesis……………..…………….…16 G. References……………………………………………….…….…………17

Chapter II-V: Results……………………………………………….….…………...24-124

Chapter II: Cyclin D1: Mechanism and Consequence of Androgen Receptor Co-repressor Activity………………...………...…..……….24 A. Abstract………………………………………..…….….………………….25 B. Introduction..………………..…………………..…………..……………..25 C. Experimental Procedures……………….………………………………..26 D. Results……………………………….……………………………………..27 E. Discussion.………………………………………...…..…………………..31 F. References………………………..………………………………………..33

Chapter III: Specificity of Cyclin D1 for Androgen Receptor Regulation….34 A. Abstract……………………………..………………………………...……35 B. Introduction…………………………………………………………………35 C. Materials and Methods...………………………………………………….36 D. Results……………………………………..……………….……………....37 E. Discussion…………………………………………………….…………....41 F. References…………………………………………………….…………...44

Chapter IV: Cyclin D1 Binding to the Androgen Receptor NH2-terminal Domain Inhibits AF2 Association and Reveals Dual Roles for AR Co-repression……………………………………………………………….46 A. Abstract……………...……………………………………………………..47 B. Introduction……………………………..……………………..….……….48 C. Materials and Methods………………………………………...…………52 D. Results………………………………..……………………………………54 E. Discussion……………………………...………………………………….62 F. References……………………………………………………..………….68 G. Figures & Legends………………………………………………………..72

1 Chapter V: The Central Domain of Cyclin D1 Elicits Nuclear Receptor Co-repressor Activity..………………………… ……….………. ……..84 A. Abstract………………………………………..……….……………..….85 B. Introduction…………………………………..………….……………….86 C. Experimental Procedures………………..……………….…..………..89 D. Results…………………………………..…….…………………….…...94 E. Discussion……………..……………..……………….....……….……105 F. References…………………………..………….……………….……..111 H. Figures & Legends………………..……..……………………………113

Chapter VI: Summary and Conclusions……………….….…………………....125 A. Characterization of cyclin D1 action………………………….....….125 B. Mechanism of cyclin D1 co-repressor activity………….………….130 C. Cyclin D1 as a transcriptional regulator………………………..…..133 D. Summary………...…………………………………………………….135 E. References…………………………………………………………….136

Chapter VII: Ongoing/Future Directions…………………………………….….141 A. Cyclin D1 transcript b…………………………………………….…..141 B. Identification of AR-cyclin D1 complex members………………….142 C. Therpuetic Development and Testing………………………………143 D. References……………………………………………….…………....145

LIST OF TABLES AND FIGURES

Figure Page

Chapter I: Introduction Fig.1 The androgen receptor…………………………………………………..…..6 Fig. 2 Cyclin D1 is a key regulatory component of the cell cycle.…………….12 Fig. 3 Cyclin D1 is induced by androgen in LNCaP cells……………………...12 Fig. 4 The structural organization of cyclin D1……………………………….…13 Fig. 5 Cyclin D1 is a CDK independent AR co-repressor…………………..….15 Fig. 6 Cyclin D1 is a negative feedback inhibitor of the AR…………………....16

Chapter II: Cyclin D1: mechanism and consequence of androgen receptor co-repressor activity Fig. 1 The LxxLL motif is dispensable for AR inhibition…………………….….27 Fig. 2 Nuclear cyclin D1 inhibits androgen receptor transactivation………….28

2 Fig. 3 Cyclin D1 binds the N-terminus of the androgen receptor ……..……29 Fig. 4 Cyclin D1 binds AR5 but fails to inhibit the constitutively active Androgen receptor…………………………………………………..…...29 Fig. 5 Cyclin D1 is dominant to androgen receptor co-activator function …30 Fig. 6 Cyclin D1 action is partially abrogated through HDAC inhibition……31 Fig. 7 Ectopic cyclin D1 abrogates cell cycle progression in androgen-dependent prostatic adenocarcinoma cells …..…………..31

Chapter III: Specificity of cyclin D1 action Fig. 1 Cyclin D1 inhibits endogenous and ectopic AR activity in LNCaP cells……………………………………………………………….37 Fig. 2 Cyclin D1 is refractory to cellular context……………………………....38 Fig. 3 Cyclin D1 demonstrates repressor activity across multiple AR target promoters……………………………………………………………...…..39 Fig. 4 NH2-terminal phosphorylation sites are not required for cyclin D1 action…..………………………………………………………………..…40 Fig. 5 Contracted or expanded AR polyglutamine polymorphisms are sensitive to cyclin D1……………………………….……..….………….40 Fig. 6 Cyclin D1 abrogates the function of clinically relevant prostate cancer- derived AR alleles…………………………………………………….….41 Fig. 7 Cyclin D1 prevents AR transcription stimulated by natural, nonconventional, and non-ligand (IL-6) activators………………..…..42

Chapter IV: Cyclin D1 binding to the androgen receptor NH2-terminal domain inhibits AF2 association and reveals dual roles for AR co-repression Fig. 1 Cyclin D1 binds to and inhibits the AR in a dose dependent manner.76 Fig. 2 Cyclin D1 interacts with the amino terminus of AR……………..….….77 Fig. 3 Cyclin D1 binds preferentially to the first 34 amino acids of the AR…78 Fig. 4 Cyclin D1 inhibits the NTD/AF2 interaction in the mammalian two-hybrid assay…….……………………………………………………80 Fig. 5 The FxxLF motif is required for full cyclin D1 binding ……..………....81 Fig. 6 Cyclin D1 regulated NTD/AF2 interaction through the FxxLF motif....82 Fig. 7 Cyclin D1 is recruited to AF2 through its LxxLL domain to selectively activate AF2 function……………………………………………………..83

Chapter V: The central domain of cyclin D1 elicits nuclear receptor co-repressor activity Fig. 1 Cyclin D1-∆XMN fails to bind the AR both in vivo and in vitro….…..118 Fig. 2 Cyclin D1-∆XMN demonstrates compromised AR co-repressor activity………………………………………………………………….…119 Fig. 3 The central domain is required in its entirety to bind and regulate AR activity….………………………………………………..………………..120 Fig. 4 Structural analysis of cyclin D1 and the AR N-terminus……...……...121 Fig. 5 The cyclin D1 central domain is sufficient for AR binding and Inhibition…………...…………………………………………………...... 122 Fig. 6 Cyclin D1 nuclear receptor co-repressor activity is distinct from its co-activator function…………………………………………….……….123

Chapter VI: Summary and Conclusions Fig. 1 Cyclin D1 uses two distinct mechanisms………………………………….132 Fig. 2 Proposed model demonstrating the distinct transcriptional regulator functions of cyclin D1………………………………………………………..133 Fig. 3 Internal deletion of cyclin D1 results in cell cycle abnormalities……...…134 Fig. 4 Summary of cyclin D1 action in prostate cancer………………………….136

Chapter VII: Ongoing/Future Directions Fig. 1 Cyclin D1 polymorphism at nucleotide 870 causes the production of two distinct splice variants mRNAs….…………………………….……141 Fig. 2 Reporter assay to detect cyclin D1-B co-repressor activity……….……..142

4 TABLE OF CONTENTS

Page List of Tables and Figures…………………..………..……………………...…….……2

Chapter II-V: Results……………………………………………….….…………...24-124

LIST OF TABLES AND FIGURES

Figure Page

4 Chapter I:

Introduction

The evolution of prostate cancer

It is estimated that in this year alone, over 230,000 American men will be

diagnosed with prostate cancer (1). The prostate is a highly structured glandular organ, which in healthy individuals functions primarily to produce and secrete seminal fluid. In early forms of cancer (diagnosed as prostatic intraepithelial neoplasia; PIN), the ducts of the prostate become disordered and neoplastic (2). Further progression of this disease leads to epithelial cell invasion of the surrounding stroma, causing inappropriate organ growth and the onset of patient symptoms. If diagnosed while confined to the margins of the prostate, this cancer is curable by prostatectomy or radiation seed implantation

(3). However, metastatic disease is often present upon patient diagnosis and requires an alternative form of therapy based upon the initial androgen dependent nature of such disseminated cancers (2, 4, 5). For these patients, ablation of androgen action through pharmaceutical (anti-androgen) and/or surgical (orchiectomy) means results in temporary tumor remission (24-36 month median time) (2). Unfortunately, subsets of cancer cells ultimately arise, which are refractory to therapy and survive in the absence of androgen. To date, no curative regimen has been identified for these androgen independent tumors, as prostate cancers are highly resistant to standard chemotherapeutic and radiation treatments (6). For this reason prostate cancer is currently the second leading cause of male cancer deaths in the United States (1).

Clearly, a better understanding of the mechanisms governing androgen dependence is

5 needed in order to identify and therapeutically target pathways disrupted in advanced stage disease.

The androgen receptor

Androgen manifests its biological activity through the androgen receptor (AR), a member of the steroid hormone family of nuclear receptors (NRs) (7, 8). These proteins constitute the largest group of eukaryotic transcription factors identified to date, regulating a wide variety of physiological processes including embryonic development, homeostasis, and cellular differentiation (9). NRs are conventionally viewed as ligand- inducible factors, yet a subset of these proteins, termed orphan receptors, have no defined ligand. Like most NRs, the AR can be roughly divided into three functional segments: a unique N-terminus followed by highly conserved DNA- and ligand-binding domains (Figure 1) (7, 10). Heat-shock proteins hold the receptor inactive until ligand binding triggers a conformational change, allowing the AR to escape sequestration, and translocate to the nucleus (11). Once inside the nucleus, two conserved zinc fingers within the receptor DNA-binding domain (DBD) mediate promoter recognition and homodimer formation (12-14). AR-regulated genes are recognized through interaction

FxxLF WxxLF NLS 1919AF-1 AF-5 AF-2 DBD/ N-terminus Hinge LBD (Gln) (Gly) n n

Figure 1.The structural organization of the androgen receptor (AR), including transactivation functions (AF-1, AF-2, and AF-5), nuclear localization signal (NLS), and N- to C-terminal interaction domains (FxxLF and WxxLF). The polymorphic AR glutamine and glycine repeats are designated (Gln)n and (Gly)n, respectively.

6 of the receptor DBD with androgen responsive DNA elements (AREs) on target promoters (14-16). Since AR homodimers bind DNA in anti-parallel fashion, AREs often occur as inverted repeats of the consensus sequence GGTACA (13). Transcription of

AR target genes is dependent upon the activity of internal transactivation functions (e.g.

AF-1, AF-2 and AF-5). The AR differs from other NRs as its dominant ligand dependent

transactivation function, AF-1, lies within the receptor N-terminus (amino acids 141-338)

(17). AF-1 activity is highly dependent upon the association of the AR N- and C-termini

(18, 19). In addition, the AF-1 domain is the preferential binding site for several AR co-

activators (e.g. SRC-1) (20, 21). Traditionally, NR function relies primarily upon the C-

terminal AF-2 domain, yet in the AR this transactivation function is relatively weak (7,

22, 23). AF-2 constitutes a small portion of the receptor ligand-binding domain and is

regulated by the conformational positioning of a critical helix within this domain,

designated as helix 12. In the presence of ligand, helix 12 folds over the ligand-binding

pocket, exposing key residues required for co-activator binding to AF-2 (24). In the

absence of androgen or the presence of receptor antagonists (e.g. hydroxyflutamide)

helix 12 is hypothesized to adapt a second conformation, inhibiting co-activator binding

and fostering co-repressor interaction with the AR C-terminus (25). Interestingly, the C-

terminus itself is thought to serve as a co-repressor of AF-5 function (26). The AF-5

transactivation domain resides within the N-terminus of the AR and is regulated

specifically by the Rho A signal transduction pathway (27, 28). Specifically,

downstream Rho effectors (PRK1 and PRK2) appear to directly bind the AR N-terminus,

stimulating co-activator activity and increasing AF-5 transactivation potential (28). In

addition, ectopic expression of activated PRK isoforms enhances AR N- to C-terminal

7 interactions, indicating a role for AF-5 in AR intramolecular interactions (28). Such

internal contacts are required for full AR activity and prevent the dissociation of ligand

from the AR C-terminal ligand binding pocket (18, 22, 29, 30). N- to C-terminal

interactions are mediated through co-activator-like binding motifs within the AR N-

terminus including 23FxxLF27, 179LxxIL183, and 432WxxLF436 (21, 31). Of these, 23FxxLF27 is the predominant mediator of N- to C-terminal AR interaction, binding to the AF-2 hydrophobic cleft with high affinity (21, 31). Disruption of intramolecular AR interactions is the mechanism of action of several known AR co-repressors (e.g. p53 and SMRT)

(32, 33). However, in the agonist-bound conformation, AR co-activators (e.g. SRC-1,

ARIP3) are hypothesized to facilitate N- and C-terminal interactions and enhance AF-1 activity (18, 19).

Androgen receptor co-modulators

Two major classes of co-activators mediate AR transactivation. Type 1 co- activators serve as scaffolds to recruit basal transcriptional machinery and/or facilitate chromatin remodeling. DRIP/TRAP mediator complexes represent one form of type 1 co-activator, serving as a bridge between the AR and required transcriptional machinery for the recruitment of RNA polymerase to target gene promoters (34). Modification of the chromatin structure surrounding such promoters is dependent upon the activity of both SWI/SNF chromatin remodeling complexes and histone acetyl transferase (HAT) containing co-activators (35). SWI/SNFs are multi-subunit chromatin remodeling complexes containing one interchangeable core ATPase (either BRM or BRG-1) (36,

37). Using the energy generated through ATP hydrolysis, SWI/SNF complexes perturb

8 nucleosome conformation and regulate gene transcription by exposing, or locally

condensing short regions of DNA (38-43). Similar to the unraveling action of SWI/SNF

complexes, HAT co-activators disrupt nucleosome-DNA interactions through the

acetylation of lysine residues located on core histone tails (35, 44-47). The HAT activity of SRC-1, SRC-2/TIF2/GRIP1, SRC-3, p/CAF, and CBP/p300 is well-documented and

known to influence the transactivation of numerous AR regulated gene promoters (48-

52). Type 2 co-activators do not rely upon enzymatic activity to enhance AR

transactivation. Instead, these co-modulators function to enhance the stability of

activated AR, preventing the dissociation of bound ligand (reviewed in: 45).

Conventionally, type 2 AR co-activator activity is measured by mammalian-2-hybrid

assays, where N- to C-terminal interactions drive reporter gene transactivation. In this

fashion, ARIP3 and c-jun were originally identified as type 2 AR co-activators (19, 53).

In addition some type 1 co-activators (SRC-1, SRC-2/TIF-2/GRIP-1, and CBP/p300)

possess additional type 2 activities (18, 21). Interaction of both co-activator classes

with the AR frequently requires a conserved protein-protein interacting motif known as

an LxxLL domain or NR box. This consensus sequence of amino acids is similar to the

AR FxxLF motif and is thought to interact with the exposed C-terminal AR hydrophobic

groove following ligand binding and helix 12 repositioning (54, 55). However, high

affinity SRC-1 and SRC-2/TIF2/GRIP1 binding to the AR N-terminal AF-1 domain

occurs independently of LxxLL, suggesting a secondary interaction motif is utilized

predominately by these co-activators (20, 21).

Although numerous AR co-activators have been identified, far fewer co-

repressors are known. NR interaction with these repressor proteins conventionally

9 requires an LxxLL-like motif (consensus: L/IXXIIXXXL) known as a nuclear receptor interaction domain (ID) or CoRNR box (56-61). Binding of co-repressor IDs to the AR is hypothesized to thwart receptor activity via occupancy of potential co-activator docking

sites and the N-terminal FxxLF motif required for receptor N- to C- terminal interactions

(62). In addition, the NR co-repressors, N-CoR and SMRT, oppose type I co-activator

function through the recruitment of histone deacetylase activity (HDAC) to the receptor

complex, thus preventing access of RNA polymerase to target gene promoters (63).

Other AR co-repressors such as calreticulin and p53 oppose the activity of type 2 co-

activators, respectively preventing receptor-DNA and N- to C- terminal interactions (33,

64). We previously identified cyclin D1 is a potent inhibitor of AR transactivation whose

mechanism of action is analyzed and discussed in this dissertation (65).

AR deregulation in androgen independent prostate cancer

Circumvention of androgen dependence in late stage prostate cancer is almost

invariably mediated through the de-regulation of AR activity. Multiple mechanisms by

which AR regulation is disrupted are observed in androgen independent prostate

cancers. These include AR and co-activator gene amplifications shown to reduce the

hormonal requirement of the tumor, allowing survival in the presence of reduced levels

of serum androgens (reviewed in: 2). Indeed, AR amplification is observed in

microarray experiments wherein the mRNA profile of hormone sensitive tumor cells is

compared to that of their hormone refractory counterparts (66). Further studies

corroborate these findings, showing a lower requirement for serum androgen and

resistance to anti-androgen therapies (bicalutamide) in AR over expressing cell lines

10 (66). In addition, ectopic expression of dominant-negative or shRNA (short hair pin

RNA) AR constructs reduced androgen independent tumor growth in vivo, establishing that inappropriate AR activity can confer hormone-refractory growth (66). AR mutation within the ligand-binding pocket is also frequent and results in receptor activation in the presence of non-canonical ligands such as progesterone, 17β-estradiol, corticosteroids, and anti-androgens (reviewed in: 67). The observation that metastatic prostate tumors are surrounded by high levels of growth factors and cytokines lead to the discovery that the activation of such signaling pathways may trigger ligand-independent AR transactivation (68-72). Although the mechanism by which cytokine and growth factor signaling induces AR activity has yet to be solidified, it is thought that MAP kinase and

JAK/STAT activation leads to receptor phosphorylation (70, 73, 74). For ERα, such phosphorylation appears to increase ligand-independent receptor transactivation (75,

76). However, mutation of AR phosphorylation sites does not reduce receptor transactivation, leaving many unanswered questions concerning the role of non-ligand signaling in androgen independency (77). However, AR polymorphisms may play a significant role in receptor transactivation potential. Specifically, the AR N-terminus harbors a highly polymorphic stretch of glutamine residues known to influence receptor activity (see Figure 1) (10). In African Americans, who have the highest risk of developing prostate cancer, this repeat is extremely short (<19 residues) and AR activity greatly enhanced (78, 79). By contrast, individuals with androgen insensitivity syndromes often harbor receptors with expanded (40+ residues) polyglutamine tracts

(80, 81). Together, AR mutation, amplification and polymorphisms appear to play a significant role in prostate cancer relapse following anti-androgen treatment. In order to

11 combat such advanced stage disease novel therapies must be developed that target the

AR while taking into account these compensatory mechanisms.

The dual actions of cyclin D1 Work published by Knudsen et al. have previously showed that cyclin D1 is a

major target of AR activity during

Growth Factors mitogenesis (82). Cyclin D1 is the E2F RB D1 CDK4/6 downstream target of numerous mitogenic E CDK2 pathways and plays a significant role in the

P RB P P P regulation of gene transcription. As a cell E2F A cycle modulator, cyclin D1 regulates gene CDK2 repression by the retinoblastoma tumor Figure 2. Cyclin D1 is a key regulatory component of the mammalian cell cycle induced by mitogenic signals. When bound suppressor, RB (75). Phosphorylation of to CDK4, cyclin D1 phosphorylates the retinoblastoma tumor suppressor, RB, to drive RB by active cyclin dependent kinase cell cycle progression. (CDK) 4/6-cyclin D1 complexes releases bound E2F transcription factors leading to the induction of genes Figure 3. Cyclin D1 expression is induced by androgen in LNCaP required for cell cycle progression cells. LNCaP (androgen- dependent prostate into S-phase (Figure 2) (83). The cancer) cells were treated CDK4 - for 72 hours with complete serum (FBS), steroid- regulation of such CDK activity is Cyclin D - depleted serum (CDT), or CDT supplemented with Cyclin E - two-fold: requiring cyclin interaction 0.1nM DHT. Lysates were immunoblotted for Cyclin A - and subsequent phosphorylation CDK4, or the G1 cyclins, ppRB { as indicated to show cell pRB cycle progression. This by CDK activating kinases (CAKs) Data is the property of the 1 2 3 Journal of Biological (reviewed in: 84). Opposing the Chemistry, © 1999 (82)

12 action of functional cyclin D1-CDK 4/6 complexes are the CDK inhibitors, p16INK4a, p15

INK4b, p18 INK4c and p19 INK4d (85). Binding of inhibitor to CDK4/6 blocks the formation of active CDK4/6-cyclin D1 complexes and prevents RB phosphorylation, thus inducing cell cycle arrest (85). We previously showed that cyclin D1 levels are induced by AR activity in androgen dependent prostate cancer cells (LNCaP) (82). In the absence of hormone, these cells express low levels of cyclin D1 and undergo cell cycle arrest

(Figure 3; compare lanes 1 and 2). Addition of androgen (dihydrotestosterone; DHT) stimulates cyclin D1 expression leading to the phosphorylation of RB and subsequent de-repression of S-phase promoting genes including cyclin A (compare lanes 2 and 3).

Thus, cyclin D1 is a key regulator of androgen-dependent cell cycle progression.

Although cyclin D1 has yet to be crystallized, its structure can be inferred from that of highly homologous viral (V-, K- and M-cyclin) and mammalian cyclins (cyclin A and H) (Figure 4) (86-89). Within the N-terminal region of cyclin D1 lies a conserved cyclin box motif, consisting of five sequential alpha helices (90). Sequences within this region mediate the association of cyclins with their kinase partners. Specifically, cyclin

D1 is known to require residue K114 for interaction with CDK4/6 (91). Mutation of this residue (K114 to E) prevents the association of cyclin D1 with CDKs (91). Upstream of the cyclin box is the RB interacting motif (LLCCE) of which residues C7 and C8 appear

LLCCE K114 HLH LxxLL 1295 Cyclin Box PEST

Figure 4. The structural organization of cyclin D1, including RB (LLCCE), CDK (K114), and co- activator (LxxLL) interaction sites. The helix-loop-helix (HLH) domain is hypothesized to contain sequences necessary for CAK recognition of the CDK-cyclin D1 complex.

13 to be critical (92). Few functions are linked to the region outside of the cyclin D1 N-

terminus, however association of CAK with the CDK-cyclin complex requires residues

within the cyclin box as well as the short linker domain (a.a. 142-152) central to the

protein (93). Following this linker domain is a second set of five alpha helicies whose

function remains largely undefined. An acidic PEST (proline, glutamic acid, serine,

threonine) domain within the C-terminus was once perceived to regulate cyclin D1

turnover (94). However, it is now clear that cyclin D1 degradation is directly linked to its

cellular localization. Specifically, T286 plays a major role in cyclin D1 localization, as

mutation of this residue (T286A) prevents nuclear export and subsequent protein

proteolysis thus resulting in inappropriate mitogenesis (95, 96). In addition, cyclin D1

nuclear accumulation is regulated though association with the CDK inhibitors p21cip1 and p27kip1, which prevent CRM1-mediated nuclear export of T286 phosphorylated cyclin D1 (97). De-regulation of these pathways leading to inappropriate cyclin D1 kinase activity is often associated with tumor formation, yet cyclin D1 over expression is rarely noted in prostate cancers (98). Over expression of cyclin D1 only holds prognostic value when other downstream cell cycle targets, such as cyclin A, are also over expressed, thus indicating an overall increase in proliferation (99, 100).

More recently, a second cell cycle independent role for cyclin D1 has been identified. Independent of RB and CDK4/6 association, cyclin D1 serves as a transcriptional co-regulator. In fact, examination of cyclin D1 over-expressing mammary carcinoma (MCF-7) cells reveals that CDK independent mechanisms are directly linked to cyclin D1-mediated tumorigenesis (101). This observation supports previous reports that cyclin D1 over expression in breast cancer is linked to decreased patient survival

14 A. B. 7

6 6 0.1 % ETOH 0.1 % ETOH 5 5 0.1 nM R1881 0.1 nM R1881 4 4 3 3 2 2 1 1 Relative LUCactivity 0 Relative LUCactivity 0 pSG5-AR 1.0 1.0 1.0 1.0 pSG5-AR 1.0 1.0 1.0 1.0 Cyclin D1 - 1.0 3.0 5.0 Cyclin D1 - 3.0 - - D1-KE - - 3.0 - D1-GH - - - 3.0

Figure 5. Cyclin D1 is a CDK independent AR co-repressor. A and B, CV1 cells were transfected using the BES/calcium phosphate protocol with wild type AR, the PSA-luciferase reporter gene, CMV-β- galactosidase, and cyclin D1. DNA ratios utilized are indicated. Cyclin D1-KE and GH alleles fail to bind CDK4 and RB, respectively. 4-6 hours post-transfection, cells were stimulated with 0.1 nM R1881 (a androgen analog), or ethanol (EtOH) vehicle for 16 hours. Cells were then harvested, lysed, and luciferase and β-galactosidase activity monitored. Luciferase activity was normalized to β - galactosidase and AR transactivation in the absence of hormone (EtOH) set to “1”. Bars represent the average fold induction over vehicle (ethanol) treated samples. Data shown represent at least three independent experiments wherein error bars correspond to the standard deviation. (Data is the property of Cancer Research, © 1999) (65)

rates and that cyclin D1 serves as a CDK- and ligand-independent co-activator of

estrogen receptor α (ERα) activity (102, 103). It is already known that an LxxLL motif in

the cyclin D1 C-terminus recruits NR co-activators (P/CAF and SRC-1) to the ERα in a

ligand independent fashion (103-106). In contrast, we, as well as others, report that

cyclin D1 functions to limit AR transactivation (65, 107). In addition, further

investigations into the CDK independent roles of cyclin D1 have revealed that it serves

as a potent co-repressor of numerous transcription factors including Sp1, thyroid

hormone receptor beta-1 (TRβ1), PPARγ, v-myb, MyoD and STAT3 (reviewed in: 108).

Although few studies have examined the mechanism by which cyclin D1 represses

gene transcription, the binding of HDAC3 to cyclin D1 is implicated in the regulation of

TRβ1 (109). We observed that in CV1 cells, which contain no endogenous receptor,

ectopic expression of AR and cyclin D1 wild type or cell cycle defective mutants reveals

a concentration dependent decrease in receptor activity on the PSA-luciferase reporter

15 construct in the presence of co-repressor (Figure 5A and B) (65). To characterize the

mechanism and consequence of these initial findings with regards to prostatic

adenocarinoma formation and progression is the focus of the work herein.

Introductory Conclusions and Hypothesis

Together, the previous findings from our Androgen laboratory showed that the role of cyclin D1 in ARAR androgen-mediated growth is two fold. First, the

CycD1 + CDK4 induction and stabilization of cyclin D1 protein

P P RB RB CycA CDK2 P + CycA levels is essential for mitogenesis in LNCaP cells Cell Cycle Progression Fig. 6 Cyclin D1 is a negative feedback (82). Secondly, cyclin D1 serves as a potent inhibitor of the AR. Activated AR, stimulates cyclin D1 expression to drive inhibitor of AR transactivation (65). These two cell cycle progression. Excess cyclin D1, independent of its association with findings put forth a model wherein cyclin D1 CDK4, then feeds back to inhibit further inappropriate AR activity. serves as a negative feedback inhibitor of AR

activity (Figure 6). Thus, in androgen dependent cells, cyclin D1 serves as an initiator

of cell cycle progression, but also limits inappropriate growth through down regulation of

AR activity following mitogenic stimuli. We hypothesize that Cyclin D1 modulates AR

activity through a mechanism that is regulated in response to androgens.

16 References

1. American Cancer Society, I. Cancer Facts and Figures 2004. pp. 1-60. Atlanta, GA, 2004. 2. Feldman, B. J. and Feldman, D. The development of androgen-independent prostate cancer. Nat Rev Cancer, 1: 34-45, 2001. 3. Ragde, H., Grado, G. L., Nadir, B., and Elgamal, A. A. Modern prostate brachytherapy. CA Cancer J Clin, 50: 380-393, 2000. 4. Denmeade, S. R. and Isaacs, J. T. A history of prostate cancer treatment. Nat Rev Cancer, 2: 389-396, 2002. 5. Taplin, M. E. and Ho, S. M. Clinical review 134: The endocrinology of prostate cancer. J Clin Endocrinol Metab, 86: 3467-3477, 2001. 6. Culine, S. and Droz, J. P. Chemotherapy in advanced androgen-independent prostate cancer 1990-1999: a decade of progress? Ann Oncol, 11: 1523-1530, 2000. 7. Brinkmann, A. O., Blok, L. J., de Ruiter, P. E., Doesburg, P., Steketee, K., Berrevoets, C. A., and Trapman, J. Mechanisms of androgen receptor activation and function. J Steroid Biochem Mol Biol, 69: 307-313, 1999. 8. Chang, C. S., Kokontis, J., and Liao, S. T. Molecular cloning of human and rat complementary DNA encoding androgen receptors. Science, 240: 324-326, 1988. 9. Gronemeyer, H. and Moras, D. Nuclear receptors. How to finger DNA. Nature, 375: 190- 191, 1995. 10. Jenster, G., van der Korput, J. A., Trapman, J., and Brinkmann, A. O. Functional domains of the human androgen receptor. J Steroid Biochem Mol Biol, 41: 671-675, 1992. 11. Pratt, W. B. and Toft, D. O. Steroid receptor interactions with heat shock protein and immunophilin chaperones. Endocr Rev, 18: 306-360, 1997. 12. Kasper, S., Rennie, P. S., Bruchovsky, N., Sheppard, P. C., Cheng, H., Lin, L., Shiu, R. P., Snoek, R., and Matusik, R. J. Cooperative binding of androgen receptors to two DNA sequences is required for androgen induction of the probasin gene. J Biol Chem, 269: 31763-31769, 1994. 13. Schoenmakers, E., Alen, P., Verrijdt, G., Peeters, B., Verhoeven, G., Rombauts, W., and Claessens, F. Differential DNA binding by the androgen and glucocorticoid receptors involves the second Zn-finger and a C-terminal extension of the DNA-binding domains. Biochem J, 341 ( Pt 3): 515-521, 1999. 14. Claessens, F., Verrijdt, G., Schoenmakers, E., Haelens, A., Peeters, B., Verhoeven, G., and Rombauts, W. Selective DNA binding by the androgen receptor as a mechanism for hormone-specific gene regulation. J Steroid Biochem Mol Biol, 76: 23-30, 2001. 15. Zhou, Z. X., Sar, M., Simental, J. A., Lane, M. V., and Wilson, E. M. A ligand-dependent bipartite nuclear targeting signal in the human androgen receptor. Requirement for the DNA-binding domain and modulation by NH2-terminal and carboxyl-terminal sequences. J Biol Chem, 269: 13115-13123, 1994. 16. Jenster, G., Trapman, J., and Brinkmann, A. O. Nuclear import of the human androgen receptor. Biochem J, 293 ( Pt 3): 761-768, 1993. 17. Jenster, G., van der Korput, H. A., van Vroonhoven, C., van der Kwast, T. H., Trapman, J., and Brinkmann, A. O. Domains of the human androgen receptor involved in steroid binding, transcriptional activation, and subcellular localization. Mol Endocrinol, 5: 1396- 1404, 1991.

17 18. Ikonen, T., Palvimo, J. J., and Janne, O. A. Interaction between the amino- and carboxyl- terminal regions of the rat androgen receptor modulates transcriptional activity and is influenced by nuclear receptor coactivators. J Biol Chem, 272: 29821-29828, 1997. 19. Moilanen, A. M., Karvonen, U., Poukka, H., Yan, W., Toppari, J., Janne, O. A., and Palvimo, J. J. A testis-specific androgen receptor coregulator that belongs to a novel family of nuclear proteins. J Biol Chem, 274: 3700-3704, 1999. 20. Bevan, C. L., Hoare, S., Claessens, F., Heery, D. M., and Parker, M. G. The AF1 and AF2 domains of the androgen receptor interact with distinct regions of SRC1. Mol Cell Biol, 19: 8383-8392, 1999. 21. Alen, P., Claessens, F., Verhoeven, G., Rombauts, W., and Peeters, B. The androgen receptor amino-terminal domain plays a key role in p160 coactivator-stimulated gene transcription. Mol Cell Biol, 19: 6085-6097, 1999. 22. Berrevoets, C. A., Doesburg, P., Steketee, K., Trapman, J., and Brinkmann, A. O. Functional interactions of the AF-2 activation domain core region of the human androgen receptor with the amino-terminal domain and with the transcriptional coactivator TIF2 (transcriptional intermediary factor2). Mol Endocrinol, 12: 1172-1183, 1998. 23. Hong, H., Kohli, K., Trivedi, A., Johnson, D. L., and Stallcup, M. R. GRIP1, a novel mouse protein that serves as a transcriptional coactivator in yeast for the hormone binding domains of steroid receptors. Proc Natl Acad Sci U S A, 93: 4948-4952, 1996. 24. Matias, P. M., Donner, P., Coelho, R., Thomaz, M., Peixoto, C., Macedo, S., Otto, N., Joschko, S., Scholz, P., Wegg, A., Basler, S., Schafer, M., Egner, U., and Carrondo, M. A. Structural evidence for ligand specificity in the binding domain of the human androgen receptor. Implications for pathogenic gene mutations. J Biol Chem, 275: 26164- 26171, 2000. 25. Nichols, M., Rientjes, J. M., and Stewart, A. F. Different positioning of the ligand- binding domain helix 12 and the F domain of the estrogen receptor accounts for functional differences between agonists and antagonists. Embo J, 17: 765-773, 1998. 26. Jenster, G., van der Korput, H. A., Trapman, J., and Brinkmann, A. O. Identification of two transcription activation units in the N-terminal domain of the human androgen receptor. J Biol Chem, 270: 7341-7346, 1995. 27. Muller, J. M., Metzger, E., Greschik, H., Bosserhoff, A. K., Mercep, L., Buettner, R., and Schule, R. The transcriptional coactivator FHL2 transmits Rho signals from the cell membrane into the nucleus. Embo J, 21: 736-748, 2002. 28. Metzger, E., Muller, J. M., Ferrari, S., Buettner, R., and Schule, R. A novel inducible transactivation domain in the androgen receptor: implications for PRK in prostate cancer. Embo J, 22: 270-280, 2003. 29. Langley, E., Kemppainen, J. A., and Wilson, E. M. Intermolecular NH2-/carboxyl- terminal interactions in androgen receptor dimerization revealed by mutations that cause androgen insensitivity. J Biol Chem, 273: 92-101, 1998. 30. Doesburg, P., Kuil, C. W., Berrevoets, C. A., Steketee, K., Faber, P. W., Mulder, E., Brinkmann, A. O., and Trapman, J. Functional in vivo interaction between the amino- terminal, transactivation domain and the ligand binding domain of the androgen receptor. Biochemistry, 36: 1052-1064, 1997. 31. He, B., Kemppainen, J. A., and Wilson, E. M. FXXLF and WXXLF sequences mediate the NH2-terminal interaction with the ligand binding domain of the androgen receptor. J Biol Chem, 275: 22986-22994, 2000.

18 32. Liao, G., Chen, L. Y., Zhang, A., Godavarthy, A., Xia, F., Ghosh, J. C., Li, H., and Chen, J. D. Regulation of androgen receptor activity by the nuclear receptor corepressor SMRT. J Biol Chem, 278: 5052-5061, 2003. 33. Shenk, J. L., Fisher, C. J., Chen, S. Y., Zhou, X. F., Tillman, K., and Shemshedini, L. p53 represses androgen-induced transactivation of prostate-specific antigen by disrupting hAR amino- to carboxyl-terminal interaction. J Biol Chem, 276: 38472-38479, 2001. 34. Wang, Q., Sharma, D., Ren, Y., and Fondell, J. D. A coregulatory role for the TRAP- mediator complex in androgen receptor-mediated gene expression. J Biol Chem, 277: 42852-42858, 2002. 35. Glass, C. K. and Rosenfeld, M. G. The coregulator exchange in transcriptional functions of nuclear receptors. Genes Dev, 14: 121-141, 2000. 36. Tsukiyama, T. and Wu, C. Purification and properties of an ATP-dependent nucleosome remodeling factor. Cell, 83: 1011-1020, 1995. 37. Khavari, P. A., Peterson, C. L., Tamkun, J. W., Mendel, D. B., and Crabtree, G. R. BRG1 contains a conserved domain of the SWI2/SNF2 family necessary for normal mitotic growth and transcription. Nature, 366: 170-174, 1993. 38. Tyler, J. K. and Kadonaga, J. T. The "dark side" of chromatin remodeling: repressive effects on transcription. Cell, 99: 443-446, 1999. 39. Muchardt, C. and Yaniv, M. When the SWI/SNF complex remodels...the cell cycle. Oncogene, 20: 3067-3075, 2001. 40. Narlikar, G. J., Fan, H. Y., and Kingston, R. E. Cooperation between complexes that regulate chromatin structure and transcription. Cell, 108: 475-487, 2002. 41. King, I. F. and Kingston, R. E. Specifying transcription. Nature, 414: 858-861, 2001. 42. Muchardt, C. and Yaniv, M. ATP-dependent chromatin remodelling: SWI/SNF and Co. are on the job. J Mol Biol, 293: 187-198, 1999. 43. Klochendler-Yeivin, A., Fiette, L., Barra, J., Muchardt, C., Babinet, C., and Yaniv, M. The murine SNF5/INI1 chromatin remodeling factor is essential for embryonic development and tumor suppression. EMBO Rep, 1: 500-506, 2000. 44. McKenna, N. J., Nawaz, Z., Tsai, S. Y., Tsai, M. J., and O'Malley, B. W. Distinct steady- state nuclear receptor coregulator complexes exist in vivo. Proc Natl Acad Sci U S A, 95: 11697-11702, 1998. 45. Heinlein, C. A. and Chang, C. Androgen receptor (AR) coregulators: an overview. Endocr Rev, 23: 175-200, 2002. 46. Collingwood, T. N., Urnov, F. D., and Wolffe, A. P. Nuclear receptors: coactivators, corepressors and chromatin remodeling in the control of transcription. J Mol Endocrinol, 23: 255-275, 1999. 47. Young, J. C. and Hartl, F. U. Chaperones and transcriptional regulation by nuclear receptors. Nat Struct Biol, 9: 640-642, 2002. 48. Ogryzko, V. V., Schiltz, R. L., Russanova, V., Howard, B. H., and Nakatani, Y. The transcriptional coactivators p300 and CBP are histone acetyltransferases. Cell, 87: 953- 959, 1996. 49. Yang, X. J., Ogryzko, V. V., Nishikawa, J., Howard, B. H., and Nakatani, Y. A p300/CBP-associated factor that competes with the adenoviral oncoprotein E1A. Nature, 382: 319-324, 1996.

19 50. Voegel, J. J., Heine, M. J., Zechel, C., Chambon, P., and Gronemeyer, H. TIF2, a 160 kDa transcriptional mediator for the ligand-dependent activation function AF-2 of nuclear receptors. Embo J, 15: 3667-3675, 1996. 51. Spencer, T. E., Jenster, G., Burcin, M. M., Allis, C. D., Zhou, J., Mizzen, C. A., McKenna, N. J., Onate, S. A., Tsai, S. Y., Tsai, M. J., and O'Malley, B. W. Steroid receptor coactivator-1 is a histone acetyltransferase. Nature, 389: 194-198, 1997. 52. Chen, H., Lin, R. J., Schiltz, R. L., Chakravarti, D., Nash, A., Nagy, L., Privalsky, M. L., Nakatani, Y., and Evans, R. M. Nuclear receptor coactivator ACTR is a novel histone acetyltransferase and forms a multimeric activation complex with P/CAF and CBP/p300. Cell, 90: 569-580, 1997. 53. Bubulya, A., Zhou, X. F., Shen, X. Q., Fisher, C. J., and Shemshedini, L. c-Jun targets amino terminus of androgen receptor in regulating androgen-responsive transcription. Endocrine, 13: 55-62, 2000. 54. Darimont, B. D., Wagner, R. L., Apriletti, J. W., Stallcup, M. R., Kushner, P. J., Baxter, J. D., Fletterick, R. J., and Yamamoto, K. R. Structure and specificity of nuclear receptor- coactivator interactions. Genes Dev, 12: 3343-3356, 1998. 55. McInerney, E. M., Rose, D. W., Flynn, S. E., Westin, S., Mullen, T. M., Krones, A., Inostroza, J., Torchia, J., Nolte, R. T., Assa-Munt, N., Milburn, M. V., Glass, C. K., and Rosenfeld, M. G. Determinants of coactivator LXXLL motif specificity in nuclear receptor transcriptional activation. Genes Dev, 12: 3357-3368, 1998. 56. Nagy, L., Kao, H. Y., Love, J. D., Li, C., Banayo, E., Gooch, J. T., Krishna, V., Chatterjee, K., Evans, R. M., and Schwabe, J. W. Mechanism of corepressor binding and release from nuclear hormone receptors. Genes Dev, 13: 3209-3216, 1999. 57. Perissi, V., Staszewski, L. M., McInerney, E. M., Kurokawa, R., Krones, A., Rose, D. W., Lambert, M. H., Milburn, M. V., Glass, C. K., and Rosenfeld, M. G. Molecular determinants of nuclear receptor-corepressor interaction. Genes Dev, 13: 3198-3208, 1999. 58. Hu, X. and Lazar, M. A. The CoRNR motif controls the recruitment of corepressors by nuclear hormone receptors. Nature, 402: 93-96, 1999. 59. Webb, P., Anderson, C. M., Valentine, C., Nguyen, P., Marimuthu, A., West, B. L., Baxter, J. D., and Kushner, P. J. The nuclear receptor corepressor (N-CoR) contains three isoleucine motifs (I/LXXII) that serve as receptor interaction domains (IDs). Mol Endocrinol, 14: 1976-1985, 2000. 60. Cohen, R. N., Putney, A., Wondisford, F. E., and Hollenberg, A. N. The nuclear corepressors recognize distinct nuclear receptor complexes. Mol Endocrinol, 14: 900- 914, 2000. 61. Cohen, R. N., Brzostek, S., Kim, B., Chorev, M., Wondisford, F. E., and Hollenberg, A. N. The specificity of interactions between nuclear hormone receptors and corepressors is mediated by distinct amino acid sequences within the interacting domains. Mol Endocrinol, 15: 1049-1061, 2001. 62. Zhang, J., Hu, X., and Lazar, M. A. A novel role for helix 12 of retinoid X receptor in regulating repression. Mol Cell Biol, 19: 6448-6457, 1999. 63. Jones, P. L. and Shi, Y. B. N-CoR-HDAC corepressor complexes: roles in transcriptional regulation by nuclear hormone receptors. Curr Top Microbiol Immunol, 274: 237-268, 2003.

20 64. Dedhar, S., Rennie, P. S., Shago, M., Hagesteijn, C. Y., Yang, H., Filmus, J., Hawley, R. G., Bruchovsky, N., Cheng, H., Matusik, R. J., and et al. Inhibition of nuclear hormone receptor activity by calreticulin. Nature, 367: 480-483, 1994. 65. Knudsen, K. E., Cavenee, W. K., and Arden, K. C. D-type cyclins complex with the androgen receptor and inhibit its transcriptional transactivation ability. Cancer Res, 59: 2297-2301, 1999. 66. Chen, C. D., Welsbie, D. S., Tran, C., Baek, S. H., Chen, R., Vessella, R., Rosenfeld, M. G., and Sawyers, C. L. Molecular determinants of resistance to antiandrogen therapy. Nat Med, 10: 33-39, 2004. 67. Buchanan, G., Greenberg, N. M., Scher, H. I., Harris, J. M., Marshall, V. R., and Tilley, W. D. Collocation of androgen receptor gene mutations in prostate cancer. Clin Cancer Res, 7: 1273-1281, 2001. 68. Giri, D., Ozen, M., and Ittmann, M. Interleukin-6 is an autocrine growth factor in human prostate cancer. Am J Pathol, 159: 2159-2165, 2001. 69. Drachenberg, D. E., Elgamal, A. A., Rowbotham, R., Peterson, M., and Murphy, G. P. Circulating levels of interleukin-6 in patients with hormone refractory prostate cancer. Prostate, 41: 127-133, 1999. 70. Hobisch, A., Eder, I. E., Putz, T., Horninger, W., Bartsch, G., Klocker, H., and Culig, Z. Interleukin-6 regulates prostate-specific protein expression in prostate carcinoma cells by activation of the androgen receptor. Cancer Res, 58: 4640-4645, 1998. 71. Culig, Z., Hobisch, A., Cronauer, M. V., Radmayr, C., Trapman, J., Hittmair, A., Bartsch, G., and Klocker, H. Androgen receptor activation in prostatic tumor cell lines by insulin- like growth factor-I, keratinocyte growth factor, and epidermal growth factor. Cancer Res, 54: 5474-5478, 1994. 72. Russell, P. J., Bennett, S., and Stricker, P. Growth factor involvement in progression of prostate cancer. Clin Chem, 44: 705-723, 1998. 73. Yeh, S., Lin, H. K., Kang, H. Y., Thin, T. H., Lin, M. F., and Chang, C. From HER2/Neu signal cascade to androgen receptor and its coactivators: a novel pathway by induction of androgen target genes through MAP kinase in prostate cancer cells. Proc Natl Acad Sci U S A, 96: 5458-5463, 1999. 74. Ueda, T., Mawji, N. R., Bruchovsky, N., and Sadar, M. D. Ligand-independent activation of the androgen receptor by interleukin-6 and the role of steroid receptor coactivator-1 in prostate cancer cells. J Biol Chem, 277: 38087-38094, 2002. 75. Kato, J., Matsushime, H., Hiebert, S. W., Ewen, M. E., and Sherr, C. J. Direct binding of cyclin D to the retinoblastoma gene product (pRb) and pRb phosphorylation by the cyclin D-dependent kinase CDK4. Genes Dev, 7: 331-342, 1993. 76. Bunone, G., Briand, P. A., Miksicek, R. J., and Picard, D. Activation of the unliganded estrogen receptor by EGF involves the MAP kinase pathway and direct phosphorylation. Embo J, 15: 2174-2183, 1996. 77. Gioeli, D., Ficarro, S. B., Kwiek, J. J., Aaronson, D., Hancock, M., Catling, A. D., White, F. M., Christian, R. E., Settlage, R. E., Shabanowitz, J., Hunt, D. F., and Weber, M. J. Androgen receptor phosphorylation. Regulation and identification of the phosphorylation sites. J Biol Chem, 277: 29304-29314, 2002. 78. Giovannucci, E., Platz, E. A., Stampfer, M. J., Chan, A., Krithivas, K., Kawachi, I., Willett, W. C., and Kantoff, P. W. The CAG repeat within the androgen receptor gene and benign prostatic hyperplasia. Urology, 53: 121-125, 1999.

21 79. Hakimi, J. M., Schoenberg, M. P., Rondinelli, R. H., Piantadosi, S., and Barrack, E. R. Androgen receptor variants with short glutamine or glycine repeats may identify unique subpopulations of men with prostate cancer. Clin Cancer Res, 3: 1599-1608, 1997. 80. La Spada, A. R., Wilson, E. M., Lubahn, D. B., Harding, A. E., and Fischbeck, K. H. Androgen receptor gene mutations in X-linked spinal and bulbar muscular atrophy. Nature, 352: 77-79, 1991. 81. Orr, H. T., Chung, M. Y., Banfi, S., Kwiatkowski, T. J., Jr., Servadio, A., Beaudet, A. L., McCall, A. E., Duvick, L. A., Ranum, L. P., and Zoghbi, H. Y. Expansion of an unstable trinucleotide CAG repeat in spinocerebellar ataxia type 1. Nat Genet, 4: 221-226, 1993. 82. Knudsen, K. E., Arden, K. C., and Cavenee, W. K. Multiple G1 regulatory elements control the androgen-dependent proliferation of prostatic carcinoma cells. J Biol Chem, 273: 20213-20222, 1998. 83. Sherr, C. J. G1 phase progression: cycling on cue. Cell, 79: 551-555, 1994. 84. Morgan, D. O. Principles of CDK regulation. Nature, 374: 131-134, 1995. 85. Sherr, C. J. and Roberts, J. M. Inhibitors of mammalian G1 cyclin-dependent kinases. Genes Dev, 9: 1149-1163, 1995. 86. Card, G. L., Knowles, P., Laman, H., Jones, N., and McDonald, N. Q. Crystal structure of a gamma-herpesvirus cyclin-cdk complex. Embo J, 19: 2877-2888, 2000. 87. Schulze-Gahmen, U., Jung, J. U., and Kim, S. H. Crystal structure of a viral cyclin, a positive regulator of cyclin-dependent kinase 6. Structure Fold Des, 7: 245-254, 1999. 88. Brown, N. R., Noble, M. E., Endicott, J. A., Garman, E. F., Wakatsuki, S., Mitchell, E., Rasmussen, B., Hunt, T., and Johnson, L. N. The crystal structure of cyclin A. Structure, 3: 1235-1247, 1995. 89. Kim, K. K., Chamberlin, H. M., Morgan, D. O., and Kim, S. H. Three-dimensional structure of human cyclin H, a positive regulator of the CDK-activating kinase. Nat Struct Biol, 3: 849-855, 1996. 90. Jeffrey, P. D., Russo, A. A., Polyak, K., Gibbs, E., Hurwitz, J., Massague, J., and Pavletich, N. P. Mechanism of CDK activation revealed by the structure of a cyclinA- CDK2 complex. Nature, 376: 313-320, 1995. 91. Hinds, P. W., Dowdy, S. F., Eaton, E. N., Arnold, A., and Weinberg, R. A. Function of a human cyclin gene as an oncogene. Proc Natl Acad Sci U S A, 91: 709-713, 1994. 92. Dowdy, S. F., Hinds, P. W., Louie, K., Reed, S. I., Arnold, A., and Weinberg, R. A. Physical interaction of the retinoblastoma protein with human D cyclins. Cell, 73: 499- 511, 1993. 93. Diehl, J. A. and Sherr, C. J. A dominant-negative cyclin D1 mutant prevents nuclear import of cyclin-dependent kinase 4 (CDK4) and its phosphorylation by CDK-activating kinase. Mol Cell Biol, 17: 7362-7374, 1997. 94. Rogers, S., Wells, R., and Rechsteiner, M. Amino acid sequences common to rapidly degraded proteins: the PEST hypothesis. Science, 234: 364-368, 1986. 95. Diehl, J. A., Zindy, F., and Sherr, C. J. Inhibition of cyclin D1 phosphorylation on threonine-286 prevents its rapid degradation via the ubiquitin-proteasome pathway. Genes Dev, 11: 957-972, 1997. 96. Diehl, J. A., Cheng, M., Roussel, M. F., and Sherr, C. J. Glycogen synthase kinase-3beta regulates cyclin D1 proteolysis and subcellular localization. Genes Dev, 12: 3499-3511, 1998.

22 97. Alt, J. R., Gladden, A. B., and Diehl, J. A. p21(Cip1) Promotes cyclin D1 nuclear accumulation via direct inhibition of nuclear export. J Biol Chem, 277: 8517-8523, 2002. 98. Gumbiner, L. M., Gumerlock, P. H., Mack, P. C., Chi, S. G., deVere White, R. W., Mohler, J. L., Pretlow, T. G., and Tricoli, J. V. Overexpression of cyclin D1 is rare in human prostate carcinoma. Prostate, 38: 40-45, 1999. 99. Aaltomaa, S., Eskelinen, M., and Lipponen, P. Expression of cyclin A and D proteins in prostate cancer and their relation to clinopathological variables and patient survival. Prostate, 38: 175-182, 1999. 100. Maddison, L. A., Huss, W. J., Barrios, R. M., and Greenberg, N. M. Differential expression of cell cycle regulatory molecules and evidence for a "cyclin switch" during progression of prostate cancer. Prostate, 58: 335-344, 2004. 101. Lamb, J., Ramaswamy, S., Ford, H. L., Contreras, B., Martinez, R. V., Kittrell, F. S., Zahnow, C. A., Patterson, N., Golub, T. R., and Ewen, M. E. A mechanism of cyclin D1 action encoded in the patterns of gene expression in human cancer. Cell, 114: 323-334, 2003. 102. Schuuring, E., Verhoeven, E., van Tinteren, H., Peterse, J. L., Nunnink, B., Thunnissen, F. B., Devilee, P., Cornelisse, C. J., van de Vijver, M. J., Mooi, W. J., and et al. Amplification of genes within the chromosome 11q13 region is indicative of poor prognosis in patients with operable breast cancer. Cancer Res, 52: 5229-5234, 1992. 103. Zwijsen, R. M., Wientjens, E., Klompmaker, R., van der Sman, J., Bernards, R., and Michalides, R. J. CDK-independent activation of estrogen receptor by cyclin D1. Cell, 88: 405-415, 1997. 104. McMahon, C., Suthiphongchai, T., DiRenzo, J., and Ewen, M. E. P/CAF associates with cyclin D1 and potentiates its activation of the estrogen receptor. Proc Natl Acad Sci U S A, 96: 5382-5387, 1999. 105. Neuman, E., Ladha, M. H., Lin, N., Upton, T. M., Miller, S. J., DiRenzo, J., Pestell, R. G., Hinds, P. W., Dowdy, S. F., Brown, M., and Ewen, M. E. Cyclin D1 stimulation of estrogen receptor transcriptional activity independent of cdk4. Mol Cell Biol, 17: 5338- 5347, 1997. 106. Zwijsen, R. M., Buckle, R. S., Hijmans, E. M., Loomans, C. J., and Bernards, R. Ligand- independent recruitment of steroid receptor coactivators to estrogen receptor by cyclin D1. Genes Dev, 12: 3488-3498, 1998. 107. Reutens, A. T., Fu, M., Wang, C., Albanese, C., McPhaul, M. J., Sun, Z., Balk, S. P., Janne, O. A., Palvimo, J. J., and Pestell, R. G. Cyclin D1 binds the androgen receptor and regulates hormone-dependent signaling in a p300/CBP-associated factor (P/CAF)- dependent manner. Mol Endocrinol, 15: 797-811, 2001. 108. Coqueret, O. Linking cyclins to transcriptional control. Gene, 299: 35-55, 2002. 109. Lin, H. M., Zhao, L., and Cheng, S. Y. Cyclin D1 Is a Ligand-independent Co-repressor for Thyroid Hormone Receptors. J Biol Chem, 277: 28733-28741, 2002.

Chapter II:

Cyclin D1: Mechanism and Consequence of AR Co- Repressor Activity

The text of chapter is a reprint of the material as it appears in the Journal of Biological Chemistry (277:2207-2215, 2002). In this publication I was the primary researcher and author. Dr. Karen Knudsen directed and supervised the research that forms this chapter.

24 THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 277, No. 3, Issue of January 18, pp. 2207–2215, 2002 © 2002 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A. Cyclin D1: Mechanism and Consequence of Androgen Receptor Co-repressor Activity*

Received for publication, July 9, 2001, and in revised form, October 11, 2001 Published, JBC Papers in Press, November 19, 2001, DOI 10.1074/jbc.M106399200

Christin E. Petre‡§, Yelena B. Wetherill‡¶, Mark Danielsenʈ, and Karen E. Knudsen‡** From the ‡Department of Cell Biology, the University of Cincinnati College of Medicine, Cincinnati, Ohio 45267-0521 and the ʈDepartment of Biochemistry and Molecular Biology, Georgetown University Medical Center, Washington, D. C. 20007

Androgen receptor regulation is pivotal for prostate Development, growth, and survival of prostatic epithelia are growth and development. Activation of the receptor is dependent on serum androgen, which exerts its biological effect dictated by association with androgen (ligand) and through the AR (4–6). It has been suggested that aberrant or through interaction with co-activators and co-repres- inappropriate activation of the AR facilitates the formation of sors. We have shown previously that cyclin D1 functions prostate hyperplasias or neoplasias. Given the importance of as a co-repressor to inhibit ligand-dependent androgen the AR in prostate cancer progression, it is critical to determine receptor activation. We demonstrate that cyclin D1 di- its precise mode of regulation. rectly binds the N terminus of the androgen receptor The AR is expressed at high levels in prostatic epithelial cells and that this interaction is independent of ligand. Fur- and is activated in this cell type by dihydrotestosterone (DHT), thermore, we show that the interaction occurs in the a high affinity ligand for the AR (1, 7). Prior to DHT binding, nucleus and does not require the LXXLL motif of cyclin the AR exists diffusely throughout the cytoplasm and nucleus D1. Although two distinct transactivation domains exist of the cell and is held inactive through the interaction of spe- in the N terminus (AF-1 and AF-5), the data shown sup- cific heat shock proteins (8, 9). Upon ligand binding, heat shock port the hypothesis that cyclin D1 targets the AF-1 transactivation function. The constitutively active AF-5 proteins are displaced. The receptor then dimerizes, translo- domain was refractory to cyclin D1 inhibition. By con- cates to the nucleus, and modulates transcription from discrete trast, cyclin D1 completely abolished androgen receptor DNA sequences, termed androgen-responsive elements (10– activity, even in the presence of potent androgen recep- 12). From androgen-responsive elements, the receptor is capa- tor co-activators. This action of cyclin D1 at least par- ble of mediating transactivation and potential transcriptional tially required de-acetylase activity. Finally, we show repression (13). Although activation of the AR is clearly re- that transient, ectopic expression of cyclin D1 results in quired for prostate proliferation, its critical transcription tar- reduced cell cycle progression in androgen-dependent gets have yet to be identified. The best known target of the AR LNCaP cells independent of CDK4 association. Collec- is prostate-specific antigen (PSA), whose expression levels are tively, our data support a model wherein cyclin D1 has a monitored clinically to diagnose aberrant prostate growth mitogenic (CDK4-dependent) function and an anti-mito- (14–16). genic function (dependent on regulation of the AF-1 do- Like all nuclear receptors, the AR is loosely divided into main) that can collectively control the rate of androgen- three functional domains as follows: a C-terminal ligand bind- dependent cellular proliferation. These findings ing domain, a DNA binding domain, and a variable N-terminal provide insight into the non-cell cycle functions of cy- region (17, 18). Although in other nuclear receptors the C- clin D1 and provide the impetus to study its pleiotropic terminal ligand binding domain contains a potent transactiva- effects in androgen-dependent cells, especially prostatic tion function, the C-terminal transactivation domain of the AR adenocarcinomas. (AF-2) is relatively weak (18–20). The AR is unique in that two additional transactivation functions exist in the N terminus, termed AF-1 and AF-5 (21, 22). Like AF-2, the AF-1 transacti- The androgen receptor (AR)1 is a 110-kDa ligand-dependent vation function is dependent on ligand binding to the receptor transcription factor whose mis-regulation is implicated in the (2). By contrast, AF-5 transactivation is constitutively active formation and progression of prostatic adenocarcinoma (1–3). (21, 22). In the absence of ligand, it is hypothesized that the C terminus folds in such a way as to inhibit N-terminal transac- * This work was supported in part by grants from the National tivation domains (2, 19, 23–25). Deletion of the C terminus Institutes of Health and American Cancer Society (to K. E. K.). The permits ligand-independent AF-5 activation, supporting this costs of publication of this article were defrayed in part by the payment hypothesis (21). In the full-length receptor it is suggested that of page charges. This article must therefore be hereby marked “adver- ligand binding fosters a conformational change that promotes tisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. interaction of the N-terminal and C-terminal transactivation § Supported by the University of Cincinnati Distinguished Graduate functions (25). Apart from these regulatory mechanisms, Assistantship award. emerging evidence makes clear that meaningful activation of ¶ Supported by NCI Training Grant ES07250-13 from the National transcription also requires the recruitment of transcriptional Institutes of Health. ** To whom correspondence should be addressed. Tel.: 513-558-7371; co-modifiers (13, 26, 27). Fax: 513-558-4454; E-mail: [email protected]. The AR is known to interact with a series of co-activators, 1 The abbreviations used are: AR, androgen receptor; DHT, dihy- and a smaller subset of co-repressors (13, 26, 27). Interaction drotestosterone; PSA, prostate-specific antigen; HAT, histone acetyla- with these regulatory proteins shows some specificity with tion; HDAC, histone deacetylase; FBS, fetal bovine serum; GFP, green regard to transactivation functions. For example, TIF-1 and fluorescent protein; TSA, trichostatin A; BrdUrd, bromodeoxyuridine; RB, retinoblastoma; ER, estrogen receptor; BES, N,N-bis(2-hydroxy- GRIP2 are known to act on the C-terminal AF-2 domain, ethyl)-2-aminoethanesulfonic acid. whereas SRC1 also interacts with and activates the N-terminal

This paper is available on line at http://www.jbc.org 2207

25 2208 Cyclin D1 Inhibits Androgen Receptor Activity

transactivation functions (19, 20, 28, 29). The net result of LNCaP cells were transfected with the indicated plasmids using Fu- co-activator recruitment is to stimulate transcription of target GENE6 transfection reagent (Roche Molecular Biochemicals), in ac- genes, at least in part through the modification of histones. cordance with the manufacturer’s recommended protocol. Post-transfection, CV1 and C33A cells were allowed to recover for a period of 5–6 Several known co-activators (e.g. P/CAF, p300, and SRC-1) h and then supplemented with 0.1 nM dihydrotestosterone (DHT; Sig- contain inherent histone acetylation activity (HAT) (30–32). ma), 1 nM R1881, or 0.1% ethanol vehicle (ETOH) for 18 h. Following This enzymatic activity is critical because histone acetylation is stimulation, cells were harvested, and luciferase activity was quantified thought to relax chromatin, thereby facilitating DNA unwind- using the Promega luciferase assay kit (Promega, Madison, WI). ␤-Ga- ing required for gene transcription (33, 34). Co-activators that lactosidase activity was used as an internal control for transfection lack HAT activity (e.g. ARA70) are speculated to function by efficiency and measured by classic colorimetric assay or Galacto-Star protocols (Tropix, Bedford, MA). recruiting HATs to the promoter complex (35). Plasmids—The pSG5AR wild type androgen receptor expression Counter-balancing the effect of these co-activators are the plasmid and pSG5-ARA70 were kindly provided by Dr. Chawnshang co-repressors. As might be expected, these proteins are thought Chang (University of Rochester, Rochester, NY) (35). The pAR5 consti- to either harbor intrinsic histone deacetylase (HDAC) activity tutively active androgen receptor construct was the gift of Dr. Albert O. or to recruit HDACs to the receptor complex (33, 34). The AR is Brinkmann (Erasmus University Rotterdam, Rotterdam, The speculated to interact with general HDAC recruiting proteins, Netherlands) (47). The pSG5-AR-T877A biologically relevant androgen receptor mutation construct was kindly provided by Dr. David Feldman such as NcoR and SMRT (36, 37). The mechanisms governing (Stanford University School of Medicine, Stanford, CA). The CMV-␤- these interactions have yet to be fully explored. galactosidase construct was the gift of Dr. Jean Wang (University of Intriguingly, we have previously identified cyclin D1 as a California at San Diego, La Jolla, CA). The PSA61LUC reporter was co-modifier of AR activity (38). Cyclin D1 has an important role kindly provided by Dr. Kitty Cleutjens (Erasmus Universiteit (48)) and in cell cycle control as a required component of the CDK4 contains 6.1 kb of the human PSA promoter. The pRc/CMV-cyclin D1 kinase complex (39). Cyclin D1 binds CDK4 directly and ini- and pRc/CMV-cyclin D1-LALA constructs were gifts of Dr. R. Bernards (The Netherlands Cancer Institute, Amsterdam, The Netherlands) (45). tiates CDK4-mediated phosphorylation of the retinoblastoma The pFlex-D1-T286A nuclear cyclin D1 expression plasmid was the gift tumor suppressor protein, RB (40, 41). Because inactivation of Dr. Alan Diehl (University of Nebraska Medical Center, Omaha, NE) RB is required for cell cycle progression, cyclin D1 is requisite (49). The pCR3.1 hSRC-1A construct was the gift of Dr. B. W. O’Malley for cellular proliferation (42). Outside of this role in the cell (Baylor College of Medicine, Houston, TX) (50). Plasmid encoding H2B- cycle, cyclin D1 is known to bind estrogen receptor (ER)-␣ and GFP was a gift of Dr. G. Wahl (Salk Institute, La Jolla, CA). pCMVp300 activate its transactivation function in the absence of ligand was the gift of Dr. T. Kouzarides (Wellcome/CRC Institute). CMV-P/ CAF was the generous gift of Dr. S. Y. R. Dent (M. D. Anderson Cancer (43, 44). This function of cyclin D1 requires its LXXLL motif Center, Houston, TX). The pGEMAR⌬C, wtAR-pGEM, pAR1–501, and is thought to involve the recruitment of SRC-1 (45). In pAR1–661, pAR506–918, and pAR623–918 have been described previ- contrast, we have demonstrated previously (38) an additional ously (51). The CMV-CD44 construct was the generous gift of Dr. L. role for cyclin D1. We showed that cyclin D1 interacts with the Sherman (University of Cincinnati, Cincinnati, OH). pGEX 3ϫ cyclin AR in vivo and inhibits its transactivation potential, without D1-GST was provided by Dr. M. Roussel (St. Jude Children’s Hospital, affecting AR expression. This inhibition is independent of Memphis, TN). The cyclin D1-KE construct was provided by Dr. R. Weinberg (Whitehead Institute for Biomedical Research, Cambridge, CDK4 and thus also independent of the role of cyclin D1 in RB MA). The green fluorescent protein (GFP) encoding plasmid, Green phosphorylation. lantern 2, was supplied from Invitrogen. We demonstrate that cyclin D1 binds with high affinity to Immunoprecipitation and Immunoblots—Cells transfected as de- the N terminus of the AR and inhibits the AF-1 transactivation scribed above were pelleted and lysed in a NETN (20 mM Tris (pH 8.0), function. We show that the inhibitory action of cyclin D1 is 100 mM NaCl, 1 mM EDTA (pH 8.0), and 0.5% Nonidet P-40) solution ␮ dominant to both HAT and non-HAT co-activators and that the containing 1 mM phenylmethylsulfonyl fluoride, 10 g/ml 1,10-phenanthroline, 10 ␮g/ml aprotinin, 10 ␮g/ml leupeptin, 10 mM sodium fluo- complete action of cyclin D1 requires regulation of acetylation. ride, 1 mM sodium vanadate, and 60 mM ␤-glycerophosphate. Lysates Interestingly, ectopic expression of cyclin D1 or a mutant inca- were subjected to brief sonication and clarified by centrifugation. For pable of binding to CDK4 reduced the proliferative index of an immunoblots, equal protein was loaded and subjected to SDS-PAGE. androgen-dependent prostatic cell line (LNCaP), indicating For co-immunoprecipitations, equal amounts of protein were immuno- that cyclin D1-induced AR inhibition acts to limit cellular pro- precipitated with antibodies directed against cyclin D1 (NeoMarkers, liferation in this cell type. These data put forth the hypothesis Fremont, CA), AR (Santa Cruz Biotechnology, Santa Cruz, CA), E1A (Santa Cruz Biotechnology), or MDM2 (Santa Cruz Biotechnology) an- that cyclin D1 has dual roles in androgen-dependent cell types tibodies. Precipitates were recovered through incubation with protein as follows: one that is CDK4-dependent and initiates cell cycle A-Sepharose beads (Amersham Biosciences). Input, bound, and flow- progression and another that is independent of CDK4 and through fractions were obtained from each reaction and subjected to counter-balances its mitogenic activity in androgen-dependent SDS-PAGE. Proteins were transferred to Immobilon (Millipore Corp., cells. Thus, our results classify cyclin D1 as a potent AR co- Bedford, MA) and immunoblotted for the indicated proteins. Antisera repressor and provide the impetus to study deregulation of the against GFP were purchased from Roche Molecular Biochemicals. Horseradish peroxidase-conjugated protein A (Bio-Rad) and enhanced cyclin D1/AR interaction in prostate carcinoma. chemiluminescence enhancer (PerkinElmer Life Sciences) were used to visualize proteins. EXPERIMENTAL PROCEDURES Binding Assays—35S-Labeled wild type and mutant androgen recep- Cell Culture and Treatment—CV1, C33A, and LNCaP cell lines were tor proteins were generated using the TnT-coupled Reticulocyte Lysate

obtained from ATCC and maintained in a 5% CO2 incubator. CV1 and system for in vitro transcription and translation (Promega), in accord- C33A cells were cultured in Dulbecco’s modified Eagle’s medium sup- ance with the manufacturer’s recommended protocol. NEG-772 Easytag plemented with 10% heat-inactivated fetal bovine serum (FBS; Atlanta Express Protein Labeling Mix was utilized for 35S-protein labeling Biologicals, Norcross, GA), 2 mML-glutamine, and 100 units/ml peni- (PerkinElmer Life Sciences). Reactions were performed in a total vol- cillin/streptomycin (Mediatech, Herndon, VA). For CV1 and C33A re- ume of 50 ␮l. GST-cyclin D1 was transformed into BL21 bacteria as porter assays, 10% charcoal dextran-treated FBS (HyClone Laborato- specified by Novagen (Madison, WI). Cyclin D1 expression was induced ries, Logan, UT) and phenol red-free Dulbecco’s modified Eagle’s through 1 mM isopropyl-1-thio-␤-D-galactopyranoside addition for a pe- medium were utilized. LNCaP cells were maintained in Iscove’s modi- riod of 3–5 h. Bacteria were harvested by centrifugation and resus- fied Eagle’s medium supplemented with 5% heat-inactivated FBS, 2 mM pended in 4 ml of NET-N plus 0.003% Sarkosyl, phenylmethylsulfonyl L-glutamine, and 100 units/ml penicillin/streptomycin. The HDAC in- fluoride, and protease inhibitors (NET-N ϩ SPP). The resuspended hibitor trichostatin A was obtained from Sigma. pellets were sonicated and then supplemented with 1% Triton X-100. Transfection and Transcriptional Reporter Assays—CV1 or C33A Lysed pellets were subjected to clarification. A portion of the remaining cells were transfected in the absence of androgen with the indicated lysate was run on a SDS-PAGE gel and Coomassie-stained to verify plasmid constructs using the BES/calcium phosphate protocol (46). expression (data not shown). From the total lysate 3 ml were removed

26 Cyclin D1 Inhibits Androgen Receptor Activity 2209

FIG.1.The LXXLL motif is dispensable for AR inhibition. A, CV1 cells were cultured in charcoal dextran-treated serum that contains growth factors but is devoid of steroids. Cells were subsequently co-transfected with constructs encoding ␤-galactosidase (CMV-␤gal, 1.0 ␮g), wild type AR (pSG5-AR, 1.5 ␮g), 6.1 kb of the human PSA promoter linked to a luciferase reporter (PSA61LUC; 2 ␮g), and either wild type, mutant cyclin D1 (cyclinD1-LALA), or empty vector (pCDNA3; 4.5 ␮g). Following transfection, cells were stimulated with either R1881 (solid bars) or 0.1% ethanol vehicle (ETOH, striped bars) as indicated. After 16 h, cells were harvested, and luciferase activity was measured. These values were then normalized against ␤-galactosidase activity. Experiments were performed at least in triplicate, and vehicle-treated AR transfection values were set to 1. Bars represent mean induction, and error bars indicate S.D. B, experiments were performed as in A, but 0.1 nM DHT was utilized to stimulate ligand-dependent AR activation (upper panel). Lysates from parallel experiments wherein 1.0 ␮g of the H2B-GFP plasmid was transfected in place of CMV-␤gal were subject to SDS-PAGE and immunoblotting (lower panel). C, CV1 cells were transfected with pSG5AR and cyclin D1-LALA at a 1:1 ratio. 48 h post-transfection, cells were lysed, and clarified extracts were used for immunoprecipitations (IP) with anti-AR, anti-cyclin D1, anti-MDM2, or anti-E1A antibodies. Protein antibody complexes were recovered through binding to protein A-Sepharose beads and washed extensively with NETN. The bound fraction was then boiled and protein content resolved on a 10% SDS-PAGE gel. Resolved proteins were transferred to an Immobilon membrane and probed with antibodies to recognize either the cyclin D1 (left panel) or the AR (right panel). and added to 50 ␮l of glutathione-agarose beads, which were incubated ability of cyclin D1 to inhibit AR activity may also rely on this for3hat4°C. Beads were then washed 6 times with 1 ml of NET- interaction site. To test this hypothesis, we employed a char- NϩSPP. In vitro translated proteins were incubated with the beads for acterized mutant of cyclin D1 (cyclin D1-LALA) that is defec- 1.5hat4°C with rotation. Beads were subsequently washed 5 times with 1 ml of NET-N ϩ SPP. Input (4% of reaction), flow-through (3% of tive in the LXXLL motif and lacks the ability to modulate ER initial unbound fraction), and bound (10% of bound proteins prior to function (45). CV1 cells, which express no endogenous AR (data washing) were subjected to 10% SDS-PAGE. Following electrophoresis, not shown), were co-transfected in the absence of steroid hor- the gel was incubated in Fluoro-Hance (Research Products Interna- mones with plasmids encoding the human AR (pSG5-AR), tional Corp., Mount Prospect, IL), as specified by the manufacturer. PSA61LUC (a reporter for the AR containing a 6.1-kb fragment Proteins were detected via autoradiography. Immunofluorescence—LNCaP cells were seeded in 6-well dishes on of the endogenous prostate-specific antigen promoter), and ei- ␤ poly-L-lysine-coated glass coverslips. On the following day, cells were ther wild type cyclin D1 or cyclin D1-LALA. CMV- -gal was transfected using FuGENE6 Transfection Reagent (Roche Molecular included in all transfections for reporter assay, as a control for Biochemicals) as described in the manufacturer’s protocol. 4 ␮gof transfection efficiency. In parallel experiments, androgen re- wild type pRc/CMV-cyclin D1, cyclin D1-KE, or pcDNA3 (Invitrogen ceptor expression was monitored via immunoblot and com- Carlsbad, CA) constructs were transfected overnight along with 1 ␮gof pared with co-transfected GFP. Subsequent to transfection, H2B-GFP. Following transfection, the media were replaced, and cells were permitted to express the constructs for 18 h. Subsequently, bro- cells were treated with either 1 nM R1881 (Fig. 1A), 0.1 nM DHT modeoxyuridine (BrdUrd, Amersham Biosciences) was added for a 16-h (Fig. 1B), or 0.1% ethanol vehicle. Treatment was continued for labeling period. Coverslips were fixed in 3.7% formaldehyde at room 16 h at which time cells were harvested, lysed, and monitored temperature for 15 min and then washed in phosphate-buffered saline. for either luciferase and ␤-galactosidase activity or for AR and BrdUrd incorporation was determined by indirect immunofluorescence GFP expression. Relative luciferase activity is shown. In ac- as previously described (52). Experiments were performed in duplicate and at least 200 transfected cells were tallied per coverslip. cordance with previous observations (38), R1881 up-regulated AR activity ϳ5-fold during the stimulation period (Fig. 1A), RESULTS whereas the physiological concentration of DHT stimulated AR Cyclin D1 Regulation of the Androgen Receptor Is Independ- activity ϳ10–12-fold above vehicle treatment (Fig. 1B, upper ent of the LXXLL Motif—We have shown previously that cyclin panel). Transfection of multiple constructs did not affect AR D1 does not affect AR expression but binds to the AR and expression (Fig. 1B, lower panel), but in both instances, wild inhibits its ligand-dependent transactivation (38). However, type cyclin D1 and cyclin D1-LALA reduced its activity in the the mechanism underlying this inhibitory action of cyclin D1 presence of ligand to basal levels (Fig. 1, A and B). These data was not determined. Because cyclin D1 modulates ER-␣ activ- indicate that the LXXLL motif of cyclin D1 is dispensable for ity through its LXXLL motif (43–45), we hypothesized that the inhibition of ligand-dependent AR transactivation potential.

27 2210 Cyclin D1 Inhibits Androgen Receptor Activity

These findings predict that the cyclin D1-LALA mutant would retain binding to the AR. To verify this hypothesis, CV1 cells were transfected at equal ratios with expression plasmid for the wild type AR and cyclin D1-LALA. Transfected cells were lysed, and protein complexes were recovered using antisera generated against cyclin D1, AR, or negative controls (E1A or MDM2). Immunoprecipitated complexes were subjected to SDS-PAGE and immunoblotted for either the AR or cyclin D1. As shown in Fig. 1C, complexes immunoprecipitated using anti-cyclin D1 contained both cyclin D1-LALA and the AR (lanes 4 and 7, respectively). Conversely, complexes immunoprecipitated using anti-AR antisera contained both the AR and cyclin D1-LALA (lanes 2 and 8). Lysates immunoprecipitated using antisera against E1A or MDM2 failed to recover either cyclin D1 or the AR (lanes 3 and 6). Thus, cyclin D1-LALA retains its ability to bind AR. In addition, these data indicate that the mechanism by which cyclin D1 modulates AR activity is distinct from its ability to regulate the estrogen receptor. Nuclear Cyclin D1 Inhibits Androgen Receptor Activity— Cyclin D1 localization is known to transition from the nucleus to the cytoplasm throughout the cell cycle and in response to mitogen stimulation (53, 54). In a similar fashion, AR localization between nucleus and cytoplasm changes due to an external proliferative signal, androgen (ligand) (10, 11). Because it is possible that cyclin D1 inhibits AR activity through cytoplasmic sequestration, we employed an exclusively nuclear cyclin FIG.2.Nuclear cyclin D1 inhibits androgen receptor transac- D1 mutant protein, cyclin D1-T286A (49). This well character- tivation. CV1 cells were co-transfected with reporters and wild type or nuclear cyclin D1 T286A as in Fig. 1A, with the ratios indicated. AR ized mutant form of cyclin D1 cannot be phosphorylated by activity in the absence of ligand was set to 1. Experiments were per- GSK-3␤ or associate with CRM1 and therefore is not exported formed at least in triplicate (upper panel). Lysates from parallel exper- from the nucleus. By using this construct we assessed the iments wherein 1.0 ␮g of the H2B-GFP plasmid was transfected in ␤ ability of nuclear cyclin D1 to diminish AR transactivation place of CMV- gal were subject to SDS-PAGE and immunoblotting (lower panel). through reporter assay. The nuclear cyclin D1 construct was transfected into CV1 cells along with wild type AR and reporter constructs as in Fig. 1. Transfected cells were subsequently onstrated previously (38) cyclin D1 binding to AR only in vivo, stimulated with physiological levels of DHT (0.1 nM) for a via co-immunoprecipitation experiments. The in vitro binding period of 16 h. As seen in Fig. 2 (upper panel), cyclin D1-T286A studies shown here demonstrate that cyclin D1 binding to the retained AR inhibitory activity, reducing ligand-dependent AR AR does not require ligand and occurs in the absence of steroid transactivation to basal levels, comparable with that seen with receptor accessory factors. By using N- and C-terminal trunca- wild type cyclin D1. Transfection of cyclin D1 constructs did not tions of the androgen receptor, it was evident that N-terminal affect androgen receptor expression as shown by immunoblot AR proteins (amino acids 1–661 and 1–502) bound strongly to (Fig. 2, lower panel). These results demonstrate that cyclin D1 GST-cyclin D1-agarose (compare lanes 7–9 and 10–12, respec- has the ability to repress AR transactivation without prevent- tively). By contrast, C-terminal fragments of the AR (amino ing AR expression levels, nuclear translocation, or previous acids 506–918 and 623–918) failed to bind cyclin D1 (compare cytoplasmic complex formation. lanes 13–18). These data demonstrate that a cyclin D1-binding Cyclin D1 Interacts with the N-terminal Transactivation site resides within amino acids 1–502 of the AR. Functions—Upon activation and DNA binding, the AR has the AF-5 Activity Is Refractory to Cyclin D1-mediated Inhibi- ability to modulate transcription of target genes. The specific tion—The N terminus of the AR contains two independent form of transcriptional modification depends on the recruit- transactivation domains, AF-1 and AF-5 (21, 22). The AF-1 ment of co-activators or co-repressors (13, 27). To determine the function is ligand-dependent and requires the presence of the C mechanism by which cyclin D1 functions as a co-repressor for terminus to invoke transactivation potential (2). In contrast, ligand-dependent AR transactivation, we first mapped the site the AF-5 transactivation function is ligand-independent and of cyclin D1 interaction. Initially, interaction sites were deter- does not require the presence of the C terminus. Thus, deletion mined using GST pull-down experiments (Fig. 3). Immobilized of the C terminus results in a constitutively active receptor, GST-cyclin D1 was recovered on glutathione-agarose. A series dependent solely on the AF-5 activity (21). This mutant recep- of previously characterized AR deletion constructs (51) were tor, AR5, has been characterized previously to exhibit consti- utilized to generate [35S]methionine-labeled proteins via in tutive and yet diminished transactivational potential. Indeed, vitro transcription/translation. Plasmid encoding the cell sur- in our system, AR5 demonstrated a ligand-independent activ- face protein CD44 was also subjected to in vitro transcription/ ity that was ϳ3–4-fold lower than the ligand-induced wild type translation as a negative control for the binding assay. In vitro receptor (data not shown). Because our findings implicated a translated proteins were incubated with immobilized GST-cy- binding region for cyclin D1 within the AR5 construct (amino clin D1, washed, and collected. Input, bound, and flow-through acids 1–627), we tested the ability of cyclin D1 to abrogate AR5 fractions were resolved by SDS-PAGE, and proteins were de- transactivation. Strikingly, whereas AR5 expression levels re- tected by autoradiography. As shown in Fig. 3, in vitro trans- mained constant, co-expression of cyclin D1 failed to inhibit the lated CD44 was not retained by GST-cyclin D1-agarose (com- AF-5 transactivation function (Fig. 4A, upper panel). This find- pare lanes 1–3), whereas wild type AR remained bound to the ing was also recapitulated in reporter assays performed in column (lanes 4–6). This finding is important, as we had dem- C33A cells (data not shown). As with the wild type AR, expres-

28 Cyclin D1 Inhibits Androgen Receptor Activity 2211

FIG.3.Cyclin D1 binds the N terminus of the androgen receptor. GST-conjugated wild type cyclin D1 was expressed in E. coli and purified. Purification was verified by SDS-PAGE (data not shown). GST-cyclin D1 was bound to glutathione-agarose beads, washed, and subsequently incubated with [35S]methionine-labeled full-length or truncated AR proteins or CD44 control (as shown) at 4 °C with rotation for 1.5 h. GST-cyclin D1 beads were subsequently washed 5 times in NETN supplemented with 0.003% Sarkosyl. Column input, flow-through, and bound were subjected to SDS-PAGE and detected via autoradiography.

generated against the AR. As can be seen in Fig. 4B, AR5 efficiently co-precipitates cyclin D1 (lane 4), in contrast to control antisera (lane 3). Thus, cyclin D1 binds and yet fails to inhibit constitutive AF-5 transactivation. Taken together, these data suggest that cyclin D1 acts through the AF-1 function to repress ligand-dependent transactivation. Cyclin D1 Is Dominant to Known AR Co-activators—Because our data implicated the AF-1 transactivation function as the site of cyclin D1 co-repressor activity, we began by examining the effect of cyclin D1 on general co-activators known to activate N-terminal AR transactivation domains (SRC-1) or harbor intrinsic HAT activity (P/CAF, p300) (19, 20, 28, 29). (Fig. 5, A–C). For these experiments, CV1 cells were transfected with plasmid encoding the wild type AR, PSA61LUC reporter, and specific N-terminal co-activators in the presence or absence of cyclin D1. Although each of these molecules has been shown to super-activate ligand-dependent AR transactivation, some of these initial characterizations were performed with relatively high ratios of AR to co-activator (up to 1:8 ratio). Because cyclin D1 demonstrates efficient AR inhibition at 1:3 ratios with the AR (or less (38)), each co-activator was tested for receptor activation at a 1:3 ratio. As shown in Fig. 5, A–C, DHT stimulated AR-dependent PSA61LUC activity ϳ12-fold. Co-expression of SRC-1, p300, and P/CAF at a 1:3 ratio enhances DHT- stimulated AR transactivation between 17- and 27-fold over cells treated with ethanol vehicle alone. These results clearly demonstrate that enhanced activation occurs even at relatively low receptor to co-activator ratios. Strikingly, equal expression FIG.4.Cyclin D1 binds AR5 but fails to inhibit the constitu- of cyclin D1 (1:1 ratio with co-activator) reduced activation of tively active androgen receptor. A, CV1 cells were co-transfected the PSA reporter to basal (non-ligand dependent) levels. Im- with reporters as in Fig. 1A with expression plasmid for AR5 (a known munoblots demonstrate that AR expression was not affected by constitutively active form of the receptor) in the presence or absence of cyclin D1, with the plasmid ratios indicated. Cells were treated with 0.1 the transfection of multiple constructs, as compared with the nM DHT or ethanol vehicle, lysed, and relative luciferase activity meas- GFP control (Fig. 5E, compare lanes 1–7). These data demon- ured. Constitutive AR5 activity in the presence of vehicle alone was set strate the dominance of cyclin D1 as a co-repressor against to 100% (upper panel). Reporter assays were performed at least in co-activators with intrinsic HAT activity. duplicate. Lysates from parallel experiments in C33A wherein 1.0 ␮gof the H2B-GFP plasmid was transfected in place of CMV-␤gal were In addition, we tested the ability of cyclin D1 to inhibit subject to SDS-PAGE and immunoblotting (lower panel). B, CV1 cells ARA70, a co-activator specific to the AR that binds the C cultured in the presence of steroid were transfected with pAR5 and/or terminus and lacks HAT activity (35). Although not a strong cyclin D1 at a 1:1 ratio. As in Fig. 1B, cells were lysed, and protein was co-activator in all systems (55), ARA70 enhanced AR activity extracted and immunoprecipitated with anti-AR, anti-cyclin D1, and ϳ anti-E1A antibodies. Complexes were recovered on protein A-Sepharose on the PSA promoter to 32-fold above vehicle-treated cells, as beads, washed, and then subjected to SDS-PAGE. Resolved proteins compared with 12-fold activation by DHT alone (Fig. 5D). Co- were transferred to Immobilon and immunoblotted for cyclin D1. expression of cyclin D1 once again demonstrated its potency in inhibiting AR transactivation potential, reducing activity to sion levels of AR5 were unchanged by cyclin D1 co-expression basal levels when transfected at equal ratios with ARA70. (Fig. 4A, bottom panel). Although ARA70 binds the C terminus of the AR, it is known To determine the ability of cyclin D1 to bind AR5, in vivo that interaction of this region with the N terminus is important co-immunoprecipitation experiments were performed. Briefly, for full activity (19, 25, 56). It is perhaps this action that is CV1 cells were co-transfected with expression plasmids for blocked by cyclin D1 binding. Taken together, these data dem- cyclin D1 and AR5. Post-transfection, cells were harvested, onstrate cyclin D1 co-repressor activity is dominant to both lysed, and subjected to co-immunoprecipitation using antisera HAT and HAT recruiting co-activators.

29 2212 Cyclin D1 Inhibits Androgen Receptor Activity

FIG.5.Cyclin D1 is dominant to androgen receptor co-activator function. CV1 cells were transfected as in Fig. 1A in the presence or absence of co-activator (A, SRC1; B, P/CAF; C, p300; and D, ARA70) and/or cyclin D1, at the relative plasmid ratios shown. Cells were stimulated, harvested, and monitored for ␤-galactosidase activity as in Fig. 1A. AR activity in the presence of ethanol vehicle (ETOH) was set to 1 (upper panel). Data bars represent mean relative luciferase activity, and error bars indicate S.D. E, lysates from parallel experiments wherein 1.0 ␮gofthe H2B-GFP plasmid was transfected in place of CMV-␤gal were subject to SDS-PAGE and immunoblotting.

Cyclin D1 Inhibition Is Partially Dependent on Histone 6, 50 nM TSA partially reversed the ability of cyclin D1 to Deacetylase Activity—Because cyclin D1 binds the N terminus inhibit AR transactivation (12-fold reduction in cyclin D1 re- of AR and inhibits transactivation even in the presence of pression). These data demonstrate that cyclin D1 is partially co-activators, we hypothesized that cyclin D1 action may be dependent on deacetylation to inhibit AR activity. mediated through changes in chromatin acetylation. In gen- Cyclin D1 Inhibits Androgen-dependent Proliferation—To- eral, co-activators harbor inherent histone acetylase activity gether, the data shown demonstrate the potency of cyclin D1 as (e.g. SRC1, p300, and P/CAF) or recruit histone acetylases to a co-repressor of the androgen receptor. Androgen receptor target sites (13, 27, 33). Similarly, many described co-repres- activity is required for proliferation of prostate cells and of sors harbor HDAC activity (57); no such function has been early prostatic adenocarcinomas. This dependence is utilized ascribed to cyclin D1. To test the hypothesis that cyclin D1 clinically to treat prostate cancer, wherein AR antagonists (e.g. action is dependent upon chromatin state, we examined the bicalutamide) are commonly used as therapeutic agents (60). effect of TSA, a specific inhibitor of HDACs (58), to abrogate Given the strength of AR inhibition we observed, we tested the cyclin D1 function. It has been shown previously that TSA possibility that transient overexpression of cyclin D1 in AR-de- induces AR activity in the presence of ligand (59). For these pendent prostatic carcinoma cells would impact proliferation. experiments, cells were transfected as usual with AR and For these studies, androgen-dependent LNCaP cells were uti-

PSA61LUC in the presence or absence of cyclin D1. After lized, which arrest in G1 in response to androgen withdrawal transfection, cells were treated with either ethanol or DHT in (52). These cells express a tumor-derived endogenous AR (AR- the presence or absence of TSA. First, a dose-response curve T877A) that is androgen-responsive and whose activation is was performed to determine the level of TSA required to max- required for cell cycle progression (61–63). Initially, reporter imally activate AR activity in our system, because TSA can assays were carried out, which demonstrated that AR-T877A is have nonspecific effects on transcription. In our system maxi- inhibited by cyclin D1 in a manner comparable with wild type mal induction was achieved at a dose of 50 nM TSA (data not AR, shown in Fig. 7A (upper panel). This observation was not shown). Therefore, this concentration was chosen to study the due a decrease in AR-T877A expression, as levels remained ability of TSA to reverse cyclin D1 repression. As shown in Fig. constant even in the presence of cyclin D1 constructs (Fig. 7A,

30 Cyclin D1 Inhibits Androgen Receptor Activity 2213

FIG.6.Cyclin D1 action is partially abrogated through HDAC inhibition. CV1 cells were transfected as in Fig. 1A, with the relative plasmid ratios as indicated. Following recovery, cells were treated with 50 nM TSA and either 0.1 nM DHT or ethanol. Cells were incubated 16 h under these conditions and then harvested and monitored for ␤-galactosidase and luciferase activity. AR activity in the absence of ligand was set to 1. Experiments were performed in triplicate. lower panel). To determine the consequence of this inhibition, LNCaP cells were then transiently transfected with expression plasmids encoding cyclin D1 constructs, and tested for cell cycle progression via BrdUrd incorporation. Histone H2B-coupled GFP was co-transfected with each of the cyclin D1 constructs such that transfection positive cells could easily be identified. As shown in Fig. 7B, cells transfected with wild type cyclin D1 demonstrated a reduced BrdUrd incorporation as compared with cells transfected with parental vector. Similar results were observed with the cyclin D1-KE allele, which cannot bind CDK4 (44) but still inhibited AR activity. These data provide evidence that cyclin D1 harbors a distinct anti-mitogenic function in androgen-dependent prostatic adenocarcinoma cells, via antagonizing AR-mediated mitogenic signaling.

DISCUSSION In this report, we show that cyclin D1 is a critical negative regulator of the AR. We demonstrate that cyclin D1 functions in the nucleus, independent of its LXXLL motif, to bind the N terminus of the AR and inhibit transactivation. This repressive function of cyclin D1 is likely manifested through an inhibitory effect on the AF-1 domain. Cyclin D1-mediated repression is dominant to both HAT and HAT recruiting co-activators and is dependent on deacetylase activity. The inhibition of AR activity by cyclin D1 has profound anti-mitogenic effects on androgen- dependent prostatic carcinoma cells, underscoring the importance of relative cyclin D1 levels on the proliferative status of androgen-dependent cell types. Nuclear Cyclin D1 Inhibits AR Activity through Interaction with the N Terminus—It is well documented that cyclin D1 FIG.7. Ectopic cyclin D1 abrogates cell cycle progression in harbors functions independent of its role in the cell cycle. androgen-dependent prostatic adenocarcinoma cells. A, CV1 Specifically, cyclin D1 is known to bind the ER␣ and induce cells were transfected with the PSA61LUC reporter, T877A mutant AR, ligand-dependent transactivation (43, 44). This function of cy- and/or cyclin D1 as in Fig. 1A, with the relative plasmid ratios shown. Cells were stimulated, harvested, and monitored for both luciferase and clin D1 requires a leucine-rich “LXXLL”-like motif, which ␤-galactosidase activity as in Fig. 1A. Experiments were performed in serves as a bridging factor between the receptor and steroid triplicate, and averages are shown (upper panel). Lysates from parallel receptor co-activators (45). Mutation of the leucine-rich region experiments wherein 1.0 ␮g of the H2B-GFP plasmid was transfected in in cyclin D1 (cyclin D1-LALA, mutation of leucines 254 and 255 place of CMV-␤gal were subject to SDS-PAGE and immunoblotting to alanine) results in a protein that binds the estrogen receptor (lower panel). B, LNCaP cells containing the endogenous T877A mutant androgen receptor were transfected with either parental vector, wild but fails to recruit co-activators. As a result, this protein acts as type cyclin D1, or the non-CDK binding allele of cyclin D1 (cyclin a dominant negative for estrogen receptor activation. In con- D1-KE) and histone H2B-coupled GFP as described under “Experimen- trast to the estrogen receptor, cyclin D1 strongly inhibits ligand- tal Procedures.” Cells were pulsed with BrdUrd for a period of 16 h, and dependent AR activation (38). However, the mechanism under- incorporation was monitored via indirect immunofluorescence. Trans- fected (GFP-positive) were scored for percent BrdUrd incorporation. lying this inhibition was unknown. We postulated that this Experiments were performed at least in duplicate, and averages are same motif of cyclin D1 was likely required for AR regulation, shown. perhaps serving to compete for co-activator binding or inhibit co-activator activity. Surprisingly, this motif is dispensable for Cyclin D1 and the AR both cycle between the nucleus and both AR binding and for inhibition of AR activity (Fig. 1). These cytoplasm in response to cellular cues. Specifically, cyclin D1 is data indicate that cyclin D1 modifies AR activity through exported from the nucleus in response to anti-mitogenic sig- mechanisms distinct from ER-␣ regulation. nals, and nuclear export is initiated via GSK-3␤-mediated

31 2214 Cyclin D1 Inhibits Androgen Receptor Activity phosphorylation and increased association with CRM1 (49, 54). terminus and C terminus of the AR is required for maximal Because the AR translocates to the nucleus after ligand binding transactivation potential (25). Therefore, recruitment of a co- (10, 11), it was possible that cyclin D1 inhibited AR activity activator to the C terminus may not overcome the action of a through cytoplasmic sequestration. To test this possibility we co-repressor on the N terminus. These results demonstrate the utilized cyclin D1-T286A, a constitutively nuclear allele that potency of cyclin D1 as repressor of the AR. cannot be phosphorylated by GSK-3␤ (49). This mutant re- Although co-repressors are thought to recruit HDAC activity tained full AR repressor activity (Fig. 2). Moreover, overexpres- (e.g. SMRT or NCoR) (67), cyclin D1 is not known to harbor this sion of cyclin D1 did not prevent ligand-dependent AR nuclear activity. However, the HDAC inhibitor TSA partially reversed translocation, as judged by indirect immunofluorescence (data the inhibitory action of cyclin D1 on the AR (Fig. 6). These data not shown). These observations demonstrate that nuclear cy- indicate that deacetylation may be involved in the co-repressor clin D1 is capable of AR inhibition. function and suggest that cyclin D1 may recruit an HDAC Mapping studies were used to identify the site of cyclin D1 molecule to the AR complex. Alternatively, it is possible that action. We identified the AR N terminus as a principal region of cyclin D1 action is dependent on de-acetylase activity that is cyclin D1 binding, between amino acids 1 and 502 (Figs. 3 and independent of histones. For example, it has been shown re- 4). We observed minimal cyclin D1 interaction with AR frag- cently that the AR can be acetylated by p300 and p300/cAMP ments encoding amino acids 506–918 and 623–918, as com- response element-binding protein and that mutation of lysines pared with N-terminal truncations (1–661 and 1–502) (Fig. 3). 632 and 633 (sites of p300-driven acetylation) results in de- However, while this manuscript was in preparation, another creased transactivation potential (68). Whereas it is possible report emerged demonstrating some binding of cyclin D1 to that cyclin D1 may recruit a deacetylase that acts directly on amino acids 633–668 (reported to be ϳ2-fold above back- the AR to reduce transcriptional activation, it is equally con- ground), which lies within the hinge region of the AR (64). ceivable that the required deacetylase acts on an intermediary Therefore, whereas a binding site for cyclin D1 may exist protein. Future investigations will distinguish among these within the AR hinge region, a predominant site of interaction possibilities. lies within the N-terminal transactivation domain. Our data Cyclin D1 Serves Dual Roles in Androgen-dependent Pros- indicate that cyclin D1 interacts with the N-terminal transac- tatic Adenocarcinoma Cells—The data shown demonstrate the tivation functions of the AR, AF-1, and AF-5. This finding is of potency of cyclin D1 as a co-repressor of the androgen receptor. importance, because critical co-activator proteins are known to In early prostatic adenocarcinoma cells, AR activity is required bind this region and modulate AR function. for proliferation (69, 70). We have shown that androgen induces Cyclin D1 Inhibition Is Dominant to Co-activators and In- cyclin D1 expression in prostatic adenocarcinoma cells as part volves Deacetylase Activity—Several co-activators have been of its mitogenic signal (52). Although androgen also induces shown to act through direct interaction and enhance ligand-de- CDK2/cyclin E activity (71), activation of cyclin D1 expression pendent AR activity. These co-activators function to bridge the and cyclin D1/CDK4 kinase activity is required for cell cycle receptors to the pre-initiation complex and facilitate transcrip- progression (39). Thus, a paradox exists for androgen-depend- tion. A subset of these harbor intrinsic HAT activity, thought to ent cells, wherein cyclin D1 harbors a mitogenic, androgen- loosen nucleosome structure and allow access of additional responsive function (induction of CDK4 activity) (52) and an transcription factors to DNA. However, HAT activity does not anti-mitogenic function (repression of AR activity) that occurs always predict co-activator function, as the recently identified independent of CDK4 (38). These observations suggest a model HBO1 protein harbors HAT activity but serves as a co-repres- wherein androgen stimulation induces cyclin D1 expression, sor for the AR through an N-terminal interaction (65). We CDK4 is activated, and cell cycle progression ensues. Indeed, examined the ability of known HAT-containing AR co-activa- after clonal selection, LNCaP cells that modestly overexpress tors (SRC-1, P/CAF, and p300) to relieve cyclin D1-mediated cyclin D1 exhibit a slight growth advantage (72). These proce- repression. Each co-activator enhanced ligand-dependent AR dures would likely select against any anti-mitogenic function of transactivation, as expected. However, cyclin D1 proved dom- cyclin D1. It is well documented that the main cell cycle target inant to all three activities and reduced AR transactivation of cyclin D-CDK4 complexes is the retinoblastoma tumor sup- potential to basal levels (Fig. 5). It has been suggested recently pressor protein, RB (42). After CDK4 mediated RB phosphoryl- that cyclin D1 may compete with P/CAF for AR binding and ation in early G1, cyclin D1 expression persists (39). Our data that excess co-activator expression may abrogate the repressor showing that ectopic expression of cyclin D1 or CDK4-refrac- function of cyclin D1, using the murine mammary tumor virus tory cyclin D1 actually inhibits cell cycle progression (Fig. 7) promoter as a readout and increased concentrations of ligand supports the model that after RB inactivation, androgen-in- Ϫ7 (10 M DHT) (64). In our experiments, it is noteworthy that duced cyclin D1 expression serves to negatively regulate AR the dominance of cyclin D1 over co-activator function did not activity and thereby limit the rate of future mitogenic activa- require excessive expression; in each experiment cyclin D1 was tion and cell cycle progression. co-transfected at 1:1 ratios with the co-activator in question. The balance of these opposing actions may contribute to the Moreover, experiments were performed using a physiological equivocal results observed in human prostate cancers upon target of the AR (the PSA promoter) and biologically relevant examination of cyclin D1 expression. Whereas some studies Ϫ10 levels of ligand (10 M DHT). These data show that cyclin D1 report relatively high frequency of cyclin D1 overexpression (up is dominant to the effect of known AR co-activators. to 30%) in prostatic adenocarcinomas (73, 74), others report Interestingly, the activity of a C-terminal binding co-activa- that this is a rare event (75) or that cyclin D1 overexpression tor, ARA70, was also abrogated by cyclin D1 action. Whereas does not correlate with tumor grade or progression (76). A ARA70 is controversial in its ability to stimulate AR transac- study of 213 patients also demonstrated no prognostic value for tivation (55), co-activators do show variant activation capacity cyclin D1 expression alone in prostate tumors (77). dependent on cell type and promoter analyzed (66). Clearly, In summary, we demonstrate that cyclin D1 is a critical ARA70 acts as an AR co-activator for PSA transactivation negative regulator of AR transactivation. These effects involve under the conditions utilized, and co-expression of cyclin D1 at a mechanism unrelated to the LXXLL motif and are dominant a 1:1 ratio abolished AR activity (Fig. 5D). Although somewhat to the known AR co-activators. This function of cyclin D1 ab- surprising, it is speculated that interaction between the N rogates AR signaling and androgen-dependent mitogenesis in

32 Cyclin D1 Inhibits Androgen Receptor Activity 2215 prostatic adenocarcinoma cells. These data demonstrate that Eisenman, R. N., Rose, D. W., Glass, C. K., and Rosenfeld, M. G. (1997) Nature 387, 43–48 cyclin D1 contains both mitogenic (CDK4 dependent) and anti- 37. Nagy, L., Kao, H. Y., Chakravarti, D., Lin, R. J., Hassig, C. A., Ayer, D. E., mitogenic (dependent on regulation of AF-1) interactions in Schreiber, S. L., and Evans, R. M. (1997) Cell 89, 373–380 androgen-dependent LNCaP cells. Given the importance of AR 38. Knudsen, K. E., Cavenee, W. K., and Arden, K. C. (1999) Cancer Res. 59, 2297–2301 regulation in the progression and treatment of prostatic ade- 39. Sherr, C. J. (1996) Science 274, 1672–1677 nocarcinomas, these data provide the impetus to delineate fur- 40. Mittnacht, S. (1998) Curr. Opin. Genet. & Dev. 8, 21–27 41. Harbour, J. W., and Dean, D. C. (2000) Nat. Cell Biol. 2, 65–67 ther the regulation of cyclin D1 activity in this tumor type. 42. Lukas, J., Pagano, M., Staskova, Z., Draetta, G., and Bartek, J. (1994) Onco- gene 9, 707–718 Acknowledgments—We thank Drs. A. Brinkmann, R. Bernards, C. 43. Zwijsen, R. M., Wientjens, E., Klompmaker, R., van der Sman, J., Bernards, Chang, K. Cleutjens, A. Diehl, D. Feldman, T. Kouzarides, B. O’Malley, R., and Michalides, R. J. (1997) Cell 88, 405–415 S. Y. R. Dent, M. Roussel, L. Sherman, J. Y. J. Wang, and R. Weinberg 44. Neuman, E., Ladha, M. H., Lin, N., Upton, T. M., Miller, S. J., DiRenzo, J., for the generous supply of reagents; Drs. W. Cavenee, K. Arden, S. Pestell, R. G., Hinds, P. W., Dowdy, S. F., Brown, M., and Ewen, M. E. Khan, and R. Hennigan for critical reading of the manuscript; and (1997) Mol. Cell. Biol. 17, 5338–5347 Dr. E. Knudsen for technical assistance and critical ongoing discussions. 45. Zwijsen, R. M., Buckle, R. S., Hijmans, E. M., Loomans, C. J., and Bernards, R. (1998) Genes Dev. 12, 3488–3498 REFERENCES 46. Chen, C., and Okayama, H. (1987) Mol. Cell. Biol. 7, 2745–2752 47. Jenster, G., de Ruiter, P. E., van der Korput, H. A., Kuiper, G. G., Trapman, J., 1. Trapman, J., and Brinkmann, A. O. (1996) Pathol. Res. Pract. 192, 752–760 and Brinkmann, A. O. (1994) Biochemistry 33, 14064–14072 2. Brinkmann, A. O., and Trapman, J. (2000) Adv. Pharmacol. 47, 317–341 48. Cleutjens, K. B., van der Korput, H. A., van Eekelen, C. C., van Rooij, H. C., 3. Brinkmann, A. O., Klaasen, P., Kuiper, G. G., van der Korput, J. A., Bolt, J., Faber, P. W., and Trapman, J. (1997) Mol. Endocrinol. 11, 148–161 de Boer, W., Smit, A., Faber, P. W., van Rooij, H. C., Geurts van Kessel, A., 49. Alt, J. R., Cleveland, J. L., Hannink, M., and Diehl, J. A. (2000) Genes Dev. 14, Vorrhorst, M. M., Mulder, E., and Trapman, J. (1989) Urol. Res. 17, 87–93 3102–3114 4. Isaacs, J. T. (1984) Prostate 5, 545–557 50. Rowan, B. G., Garrison, N., Weigel, N. L., and O’Malley, B. W. (2000) Mol. Cell. 5. Jenster, G. (1999) Semin. Oncol. 26, 407–421 Biol. 20, 8720–8730 6. Hayward, S. W., Rosen, M. A., and Cunha, G. R. (1997) Br. J. Urol. 79, Suppl. 51. Lu, J., and Danielsen, M. (1998) J. Biol. Chem. 273, 31528–31533 2, 18–26 52. Knudsen, K. E., Arden, K. C., and Cavenee, W. K. (1998) J. Biol. Chem. 273, 7. Chang, C. S., Kokontis, J., and Liao, S. T. (1988) Science 240, 324–326 20213–20222 8. Pratt, W. B., and Toft, D. O. (1997) Endocr. Rev. 18, 306–360 53. Baldin, V., Lukas, J., Marcote, M. J., Pagano, M., and Draetta, G. (1993) Genes 9. Kallio, P. J., Janne, O. A., and Palvimo, J. J. (1994) Endocrinology 134, Dev. 7, 812–821 998–1001 54. Diehl, J. A., Cheng, M., Roussel, M. F., and Sherr, C. J. (1998) Genes Dev. 12, 10. Zhou, Z. X., Sar, M., Simental, J. A., Lane, M. V., and Wilson, E. M. (1994) 3499–3511 J. Biol. Chem. 269, 13115–13123 55. Gao, T., Brantley, K., Bolu, E., and McPhaul, M. J. (1999) Mol. Endocrinol. 13, 11. Jenster, G., Trapman, J., and Brinkmann, A. O. (1993) Biochem. J. 293, 1645–1656 761–768 56. Karvonen, U., Kallio, P. J., Janne, O. A., and Palvimo, J. J. (1997) J. Biol. 12. Claessens, F., Verrijdt, G., Schoenmakers, E., Haelens, A., Peeters, B., Chem. 272, 15973–15979 Verhoeven, G., and Rombauts, W. (2001) J. Steroid Biochem. Mol. Biol. 76, 57. Xu, L., Glass, C. K., and Rosenfeld, M. G. (1999) Curr. Opin. Genet. & Dev. 9, 23–30 140–147 13. Jenster, G. (1998) Mol. Cell. Endocrinol. 143, 1–7 58. Yoshida, M., Horinouchi, S., and Beppu, T. (1995) BioEssays 17, 423–430 14. Pettaway, C. A. (1998) Tech. Urol. 4, 35–42 59. List, H. J., Smith, C. L., Rodriguez, O., Danielsen, M., and Riegel, A. T. (1999) 15. Cleutjens, K. B., van Eekelen, C. C., van der Korput, H. A., Brinkmann, A. O., Exp. Cell Res. 252, 471–478 and Trapman, J. (1996) J. Biol. Chem. 271, 6379–6388 60. Richie, J. P. (1999) Urology 54, 15–18 16. Nomura, A. M., and Kolonel, L. N. (1991) Epidemiol. Rev. 13, 200–227 61. Veldscholte, J., Ris-Stalpers, C., Kuiper, G. G., Jenster, G., Berrevoets, C., 17. Jenster, G., van der Korput, J. A., Trapman, J., and Brinkmann, A. O. (1992) Claassen, E., van Rooij, H. C., Trapman, J., Brinkmann, A. O., and Mulder, J. Steroid Biochem. Mol. Biol. 41, 671–675 E. (1990) Biochem. Biophys. Res. Commun. 173, 534–540 18. Brinkmann, A. O., Blok, L. J., de Ruiter, P. E., Doesburg, P., Steketee, K., Berrevoets, C. A., and Trapman, J. (1999) J. Steroid Biochem. Mol. Biol. 69, 62. Veldscholte, J., Berrevoets, C. A., Brinkmann, A. O., Grootegoed, J. A., and 307–313 Mulder, E. (1992) Biochemistry 31, 2393–2399 19. Berrevoets, C. A., Doesburg, P., Steketee, K., Trapman, J., and Brinkmann, 63. Veldscholte, J., Berrevoets, C. A., Ris-Stalpers, C., Kuiper, G. G., Jenster, G., A. O. (1998) Mol. Endocrinol. 12, 1172–1183 Trapman, J., Brinkmann, A. O., and Mulder, E. (1992) J. Steroid Biochem. 20. Hong, H., Kohli, K., Trivedi, A., Johnson, D. L., and Stallcup, M. R. (1996) Proc. Mol. Biol. 41, 665–669 Natl. Acad. Sci. U. S. A. 93, 4948–4952 64. Reutens, A. T., Fu, M., Wang, C., Albanese, C., McPhaul, M. J., Sun, Z., Balk, 21. Jenster, G., van der Korput, H. A., Trapman, J., and Brinkmann, A. O. (1995) S. P., Janne, O. A., Palvimo, J. J., and Pestell, R. G. (2001) Mol. Endocrinol. J. Biol. Chem. 270, 7341–7346 15, 797–811 22. Jenster, G., van der Korput, H. A., van Vroonhoven, C., van der Kwast, T. H., 65. Sharma, M., Zarnegar, M., Li, X., Lim, B., and Sun, Z. (2000) J. Biol. Chem. Trapman, J., and Brinkmann, A. O. (1991) Mol. Endocrinol. 5, 1396–1404 275, 35200–35208 23. Doesburg, P., Kuil, C. W., Berrevoets, C. A., Steketee, K., Faber, P. W., Mulder, 66. Kotaja, N., Aittomaki, S., Silvennoinen, O., Palvimo, J. J., and Janne, O. A. E., Brinkmann, A. O., and Trapman, J. (1997) Biochemistry 36, 1052–1064 (2000) Mol. Endocrinol. 14, 1986–2000 24. Ikonen, T., Palvimo, J. J., and Janne, O. A. (1997) J. Biol. Chem. 272, 67. Huang, E. Y., Zhang, J., Miska, E. A., Guenther, M. G., Kouzarides, T., and 29821–29828 Lazar, M. A. (2000) Genes Dev. 14, 45–54 25. Langley, E., Kemppainen, J. A., and Wilson, E. M. (1998) J. Biol. Chem. 273, 68. Fu, M., Wang, C., Reutens, A. T., Wang, J., Angeletti, R. H., Siconolfi-Baez, L., 92–101 Ogryzko, V., Avantaggiati, M. L., and Pestell, R. G. (2000) J. Biol. Chem. 26. Collingwood, T. N., Urnov, F. D., and Wolffe, A. P. (1999) J. Mol. Endocrinol. 275, 20853–20860 23, 255–275 69. Gao, J., and Isaacs, J. T. (1998) Cancer Res. 58, 3299–3306 27. Bevan, C., and Parker, M. (1999) Exp. Cell Res. 253, 349–356 70. Olapade-Olaopa, E. O., MacKay, E. H., Taub, N. A., Sandhu, D. P., Terry, T. R., 28. Voegel, J. J., Heine, M. J., Zechel, C., Chambon, P., and Gronemeyer, H. (1996) and Habib, F. K. (1999) Clin. Cancer Res. 5, 569–576 EMBO J. 15, 3667–3675 71. Fribourg, A. F., Knudsen, K. E., Strobeck, M. W., Lindhorst, C. M., and 29. Bevan, C. L., Hoare, S., Claessens, F., Heery, D. M., and Parker, M. G. (1999) Knudsen, E. S. (2000) Cell Growth Differ. 11, 361–372 Mol. Cell. Biol. 19, 8383–8392 72. Chen, Y., Martinez, L. A., LaCava, M., Coghlan, L., and Conti, C. J. (1998) 30. Spencer, T. E., Jenster, G., Burcin, M. M., Allis, C. D., Zhou, J., Mizzen, C. A., Oncogene 16, 1913–1920 McKenna, N. J., Onate, S. A., Tsai, S. Y., Tsai, M. J., and O’Malley, B. W. 73. Shiraishi, T., Watanabe, M., Muneyuki, T., Nakayama, T., Morita, J., Ito, H., (1997) Nature 389, 194–198 Kotake, T., and Yatani, R. (1998) Urol. Int. 61, 90–94 31. Ogryzko, V. V., Schiltz, R. L., Russanova, V., Howard, B. H., and Nakatani, Y. 74. Han, S. W., Rha, K. H., Lee, W. T., Mah, S. Y., Choi, H. K., and Choi, S. K. (1996) Cell 87, 953–959 (1998) Yonsei Med. J. 39, 13–19 32. Bannister, A. J., and Kouzarides, T. (1996) Nature 384, 641–643 75. Gumbiner, L. M., Gumerlock, P. H., Mack, P. C., Chi, S. G., deVere White, 33. Struhl, K. (1998) Genes Dev. 12, 599–606 R. W., Mohler, J. L., Pretlow, T. G., and Tricoli, J. V. (1999) Prostate 38, 34. Kouzarides, T. (1999) Curr. Opin. Genet. & Dev. 9, 40–48 40–45 35. Yeh, S., and Chang, C. (1996) Proc. Natl. Acad. Sci. U. S. A. 93, 5517–5521 76. Drobnjak, M., Osman, I., Scher, H. I., Fazzari, M., and Cordon-Cardo, C. (2000) 36. Heinzel, T., Lavinsky, R. M., Mullen, T. M., Soderstrom, M., Laherty, C. D., Clin. Cancer Res. 6, 1891–1895 Torchia, J., Yang, W. M., Brard, G., Ngo, S. D., Davie, J. R., Seto, E., 77. Aaltomaa, S., Eskelinen, M., and Lipponen, P. (1999) Prostate 38, 175–182

Chapter III:

Specificity of Cyclin D1 for AR Regulation

The text of chapter is a reprint of the material as it appears in the Cancer Research (63:4903-4913, 2003). In this publication I was the primary researcher and author. Dr. Karen Knudsen directed and supervised the research that forms this chapter.

34 [CANCER RESEARCH 63, 4903–4913, August 15, 2003] Specificity of Cyclin D1 for Androgen Receptor Regulation1

Christin E. Petre-Draviam, Stephen L. Cook, Craig J. Burd, Thomas W. Marshall, Yelena B. Wetherill, and Karen E. Knudsen2 Department of Cell Biology, University of Cincinnati College of Medicine, Cincinnati, Ohio 45267-0521

ABSTRACT Prostatic epithelial and adenocarcinoma cells sense androgen through the AR,3 a member of the steroid hormone receptor super- Androgen receptor (AR) activity is required for prostate growth, dif- family of transcription factors (5, 6). The AR contains three functional ferentiation, and secretion. Deregulation of AR activity results in inap- domains, classified based upon their homology to other known nu- propriate mitogenic signaling and is thought to contribute both to the initiation and progression of prostate cancers. Cyclin D1 functions as a clear receptors: a NH2-terminal transactivation domain; a highly con- strong AR corepressor by directly interacting with and inhibiting receptor served DNA binding region; and a COOH-terminal ligand binding activity. However, the extent to which cyclin D1 functions to inhibit AR pocket (5, 7). The AR differs from other nuclear receptors in that its activity under conditions associated with cancer progression has not been NH2-terminal domain is the site of its major transactivation function, determined. We now demonstrate that cyclin D1 action is conserved in AF-1 (7). In addition, interaction between the NH2 and COOH- multiple tumor cell backgrounds, inhibiting AR-dependent gene activa- terminal regions of the AR is necessary for complete receptor activity tion in breast, bladder, and androgen-independent prostatic adenocarci- (8). Binding of androgens such as DHT to the AR causes the disso- noma cell lines. In androgen-dependent prostatic adenocarcinomas, cyclin ciation of heat shock proteins from the receptor and allows for its D1 effectively muted androgen-stimulated target gene expression in a manner analogous to dominant negative ARs. The ability of cyclin D1 to dimerization and translocation into the nucleus (9, 10). Within the inhibit AR activity was conserved with regard to target promoter, repress- nucleus, the AR binds to AREs located on target genes such as PSA, ing transactivation from mouse mammary tumor virus, probasin, and which is used clinically to monitor prostate cancer progression (11– prostate-specific antigen promoters. Inappropriate, nonligand AR activa- 13). The gene expression profile initiated by the AR results in a tion, postulated to act through regulation of receptor phosphorylation, diverse set of biological outcomes, including secretion, differentia- was also sensitive to cyclin D1 regulation. Moreover, we show that several tion, growth, and survival (11). The specificity of such biological phosphorylation site mutants of the AR were equally inhibited by cyclin outcomes is hypothesized to hinge upon the cellular environment and D1 as compared with the wild-type receptor. Given these data establishing availability of AR cofactors. Nevertheless, the precise gene targets the potency of cyclin D1-mediated repression, we evaluated the ability of cyclin D1 to inhibit tumor-derived AR alleles and polymorphisms associ- involved in these diverse functions remain largely undefined. ated with tumor progression and increased prostate cancer risk. We Intriguingly, in recurrent androgen independent prostate cancer, the demonstrate that the AR alleles and polymorphisms tested respond com- AR is expressed and inappropriately activated (i.e., in the absence of pletely to cyclin D1 corepressor activity. In addition, activation of a ligand; Ref. 2). This activation event is known to occur through common tumor-derived AR allele by 17␤-estradiol and progesterone was multiple mechanisms, including AR amplification (up to 30% of inhibited through ectopic expression of cyclin D1. Taken together, these recurrent tumors) and mutations within the AR itself, which allow data establish the potency of cyclin D1 as an AR corepressor and provide alternative steroids (e.g., 17␤-estradiol, progesterone) to serve as support for additional studies examining the efficacy of developing novel ligands (2). Also thought to contribute to the androgen-independent prostate cancer therapies for both androgen-dependent and -independent tumors. phenotype is indirect stimulation of the AR by growth factors and signal transduction pathways (reviewed in Ref. 14). Specifically, EGF, IGF-I, KGF, and IL-6 were previously demonstrated to induce INTRODUCTION AR activity in the absence of ligand and may synergize with low-level Treatment of nonorgan confined prostate cancer relies on its unique DHT to enhance AR action (15, 16). It has been hypothesized that requirement for androgen (1–3). The objective of prostate cancer activation of signal transduction pathways in response to cytokines therapy is to eliminate androgen action through bilateral orchiectomy and growth factors results in phosphorylation of the AR, thus provid- and/or through administration of antiandrogens. Such androgen abla- ing a potential mechanism by which receptor activity is modulated tion results in cell cycle arrest and cell death in prostatic adenocarci- (17). It is through these disparate pathways that the AR is thought to nomas and is highly effective because Ͼ80% of patients respond be inappropriately activated, facilitating proliferation and tumor pro- favorably to treatment (4). Unfortunately, median time to the forma- gression in the absence of canonical ligand. Thus, inhibition of AR tion of recurrent tumors is only 24–36 months with relapse occurring activity is a major goal of therapies used to treat both early and late in virtually 100% of treated patients (2). Cells of the recurrent tumors stage prostate cancers. proliferate in the absence of androgen, and no effective treatment We and others have previously shown that cyclin D1 is a potent currently exists for androgen-independent prostate cancer. Given the inhibitor of AR activity (18, 19). Although well characterized for its importance of androgen in prostate cancer formation and treatment, role in cell cycle transitions, cyclin D1 has been shown to harbor much emphasis has been placed on understanding the molecular multiple transcriptional functions independent of the cell cycle. mechanisms of androgen action. Through an LxxLL motif in its COOH terminus and independent of CDK association, cyclin D1 forms a trimeric complex with ER␣ and Received 12/4/02; revised 6/6/03; accepted 6/9/03. the steroid receptor coactivator, SRC-1, to enhance estrogen-respon- The costs of publication of this article were defrayed in part by the payment of page sive transcription (20, 21). Association of cyclin D1 with the AR charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 This work was supported by NIH Grant R01CA099996 (to K. E. K.) and the 3 The abbreviations used are: AR, androgen receptor; DHT, dihydrotestosterone; ARE, Department of Defense Grant DAMD17-02-1-0037 (to K. E. K.). C. E. P-D. and C. J. B. androgen-responsive element; PSA, prostate-specific antigen; EGF, epidermal growth are supported by the University Distinguished Graduate Fellowship (University of Cin- factor; IGF, insulin-like growth factor; KGF, keratinocyte growth factor; IL, interleukin; cinnati) and the Albert J. Ryan Foundation. CDK, cyclin-dependent kinase; ER, estrogen receptor; MMTV, mouse mammary tumor 2 To whom requests for reprints should be addressed, at Phone: (513) 558-7371; Fax: virus; CDT, charcoal dextran treated; GFP, green fluorescent protein; GST, glutathione (513) 558-4454; E-mail: [email protected]. S-transferase; GAPDH, glyceraldehyde-3-phosphate dehydrogenase. 4903

35 SPECIFICITY OF CYCLIN D1 ACTION

coactivator, P/CAF, also has been demonstrated to enhance ER- lian expression plasmids encoding tumor derived AR alleles were generously mediated transactivation and suggests a second, possibly cell type provided by Dr. Steven P. Balk (Beth Israel Hospital, Boston, MA; Ref. 31). specific, mechanism of cyclin D1 enhancement of ER activity (22). In Plasmid-encoding dominant negative AR (pSG5-AR⌬46-408) was supplied by addition, cyclin D1 has been demonstrated to possess the opposite Jorma Palvimo (University of Helsinki, Helsinki, Finland; Ref. 32). The effect, serving as a corepressor for many transcription factors, includ- pEGFP-C1-ARQ0 and pEGFP-C1-ARQ48 plasmids for the expression of ing v-Myb, STAT3, DMP1, the thyroid hormone receptor, and the AR polymorphic ARs were gifts of Dr. Michael Mancini (Baylor College of Medicine Houston, TX; Ref. 33). The pCMVhARSA650, pCM- (reviewed in Ref. 23). Using the PSA promoter as read-out, we VhARSA81,94, and pGEXAR1-173 expression plasmids were kindly pro- previously established that repression of the AR is CDK and LxxLL vided by Dr. Elizabeth Wilson (University of North Carolina School of independent, dominant to AR coactivators, and is mediated through Medicine, Chapel Hill, NC; Ref. 34). pCDNA3 empty vector was obtained direct, ligand-independent binding of cyclin D1 to the AR NH2 from Invitrogen. PBS3XERE-LUC and pCMV5-hER␣ were kindly provided terminus (18, 19, 24). Because cyclin D1 expression is induced by by Dr. Sohaib Khan (University of Cincinnati, Cincinnati, OH). Plasmid- androgen in androgen-dependent prostatic adenocarcinoma cells (24) encoding myc-tagged, wild-type cyclin E was the gift of Dr. Jim Roberts (Fred and represses receptor activity when overexpressed, the hypothesis Hutchinson Cancer Research Center, Seattle, WA). The MMTV-luciferase was put forth that cyclin D1 serves as a feedback inhibitor of the AR. reporter construct was obtained from Dr. Richard Pestell (Georgetown Uni- Indeed, this hypothesis was supported by our observation that andro- versity, Washington, DC). ARR2PB-LUC was constructed as described pre- gen-dependent prostatic adenocarcinoma cells (LNCaP) undergo a viously (35). decrease in cell cycle progression when expressing ectopic cyclin D1 Transfection and Transcriptional Reporter Assays. CV1, MCF-7, PC-3, or a mutant form, cyclin D1-KE, which fails to bind CDK4 and cannot and 22Rv1 cells were seeded for transfection in CDT serum, which lacks regulate cell cycle transitions but is competent to inhibit AR activity steroids but maintains growth factors. The N,N-bis(2-hydroxyethyl)-2-aminoethanesulfonic acid/calcium phosphate transfection protocol (36) was used for (19, 25). In addition to this finding, endogenous AR has been ob- transfection of pSG5-AR (0.5 ␮g), luciferase reporter (MMTV-LUC, served in complex with cyclin D1, and AR activity is reduced at the ARR2PB-LUC, or PSA61-LUC; 0.75 ␮g), RSV-cyclin D1 (1.5 ␮g), CMV- G1-S transition, wherein cyclin D1 levels are highest (26, 27). Taken ␤-galactosidase (0.5 ␮g), and empty vector (pCDNA3.1; to total of 4 ␮g/well together, these data demonstrate that cyclin D1 serves as a potent for a 6-well dish). After transfection, cells were allowed to recover for a period inhibitor of AR activity. of 5–6 h before stimulation with 0.1% ethanol vehicle, 0.1 nM DHT (Sigma), Given that bypass pathways activate the AR in recurrent adenocar- 0.1 nM testosterone, or 50 ng/ml IL-6 (ID Labs Biotechnology, London, cinomas, it is critical to determine whether cyclin D1 corepressor Ontario, Canada) for 22–24 h. For IL-6 assays, low serum (0.1% CDT) activity can be maintained under these conditions. Here, we deter- conditions were used during the recovery and stimulation periods. After mined the specificity of cyclin D1 action, with emphasis upon factors treatment, all cells were harvested and assayed for luciferase and ␤-galacto- that facilitate the transition of prostatic adenocarcinomas to androgen sidase activity as described previously (19). Reporter data represents at least independence. We now demonstrate that cyclin D1 maintains its three independent experiments. Appropriate Ps were obtained using ANOVA ability to repress AR activity in a wide variety of cellular backgrounds and Newman-Keuls Multiple Comparison post tests. Transfection of LNCaP cells was performed using Lipofectin reagent ac- including androgen-dependent and -independent prostate cancer cells. cording to the manufacturers’ protocol (Invitrogen, Carlsbad, CA). For LNCaP We also provide evidence that cyclin D1 regulation of AR activity transfections in 6-cm dishes, plasmid concentrations of 1.0 ␮g of pSG5-AR, spans multiple androgen-responsive promoters, inhibiting not only 1.0 ␮g of ARR2-LUC, 3.0 ␮g of RSV-cyclin D1, and 0.5 ␮g of CMV-␤- PSA but also MMTV and probasin promoters, indicating that the galactosidase were used and supplemented where necessary with pCDNA3.1 mechanism of repression is conserved across multiple AR targets. for a total of 5.5 ␮g. Transfected LNCaP cells were stimulated as indicated for Furthermore, NH2-terminal phospho-mutants of the AR retained cy- a period of 72 h. ␤-Galactosidase and luciferase activity were measured as clin D1 sensitivity. In addition, we show that cyclin D1 corepressor reported previously (19). activity regulates AR mutants and polymorphisms associated with Immunoblots. Cells from reporter assays in which H2B-GFP was substi- prostate cancer susceptibility and with the transition to androgen tuted for CMV-␤-galactosidase were pelleted after treatment and lysed in independence. This function is conserved among tumor-derived AR radioimmunoprecipitation assay buffer [150 nM NaCl, 1.0% NP40, 0.5% alleles activated by nonandrogen steroids, indicating that cyclin D1 deoxycholate, 0.1% SDS, and 50 nM Tris (pH 8.0)] solution containing 1 nM ␮ ␮ function is retained with alternate ligands. Lastly, we demonstrate that phenylmethyl sulfonyl fluoride, 10 g/ml 1,10-phenanthroline, 10 g/ml apro- ␮ cyclin D1 is capable of inhibiting wild-type AR induced through tinin, 10 g/ml leupeptin, 10 nM sodium fluoride, 1 mM sodium vanadate, and 60 mM ␤-glycerophosphate. After centrifugation, clarified lysates were sub- cytokine and nonconventional ligand pathways. Taken together, these jected to SDS-PAGE and transferred to Immobilon (Millipore Corp., Bedford, data represent the first in-depth analysis of an AR corepressor, to date, MA). Immunoblots for AR phosphorylation site mutants were then cut in half and demonstrate the potential of cyclin D1 action in the treatment of and blotted separately using antibodies generated against the AR (N-20) and both androgen-dependent and -independent tumors. GFP (Santa Cruz Biotechnology, Santa Cruz, CA). Immunoblots for polymorphic, GFP-tagged ARs were probed with GFP antibody to detect both proteins. MATERIALS AND METHODS Antimouse and antirabbit horseradish peroxidase-conjugated secondary antibodies (Pierce, Rockford, IL) along with enhanced chemiluminescence en- Cell Culture and Treatment. CV1, MCF-7, LNCaP, and PC-3 cells were hancer (Perkin-Elmer Life Sciences) were used to visualize proteins. Endogenous Protein Quantification. GST-AR1-173 and GST-cyclin D1 obtained from American Type Culture Collection and maintained in 5% CO2 incubators. The 22Rv1 cell line was the gift of Dr. James W. Jacobberger (Case were purified from Escherichia coli using standard techniques. The concen- Western Reserve University, Cleveland, OH; Ref. 28). CV1, PC-3, LNCaP, tration of GST fusion proteins was determined by SDS-PAGE electrophoresis and MCF-7 cells were cultured as described previously (29, 30). 22Rv1 cells using Coomassie staining and BSA (Sigma) as a standard. Known concentra- were maintained in RPMI supplemented with 5% FBS, 2 mML-glutamine, and tions of GST-AR and GST-cyclin D1 were then used to quantify AR and cyclin 100-units/ml penicillin/streptomycin. For steroid-free conditions, cells were D1 molecules in LNCaP cells derived from samples transfected as described seeded in phenol red-free media containing charcoal dextran-treated FBS (5% for transcriptional reporter assays. Total protein in LNCaP lysate was meas-

for 22Rv1 and LNCaP, 10% for others; HyClone Laboratories). ured using the Bio-Rad Dc protein assay kit as described by the manufacturer Plasmids. The H2B-GFP, pSG5-AR, PSA61-LUC, CMV-␤-galactosidase, (Bio-Rad Laboratories, Hercules, CA). Purified GST proteins were separated pSG5-AR-T877A, RSV-cyclin D1, RSV-cyclin D1-KE, and pGEX-3Xcyclin by 12% PAGE at known concentrations indicated, alongside LNCaP lysates,

D1 constructs have been described previously (19). The pSV-AR902, pSV- and were subsequently immunoblotted with antibody directed against the NH2 AR715, pSV-AR721, pSV-AR874, pSV-ART877S, and pSV-AR890 mamma- terminus of the AR (N-20; Santa Cruz Biotechnology) and the COOH terminus 4904

36 SPECIFICITY OF CYCLIN D1 ACTION

of cyclin D1 (Ab-3; Neomarkers, Fremont, CA). Signal was quantified using cyclin D1 alone, or cyclin D1 and wild-type AR at a ratio of 3:1. After Metamorph software (Universal Imaging Corporation, West Chester, PA). transfection, LNCaP cells were treated with either 0.1% ethanol Reverse Transcription-PCR. LNCaP cells were transfected in 6-cm (vehicle) or 1 nM DHT for 72 h. Cells were then harvested, lysed, and ␮ dishes containing complete serum using Lipofectin with 1.0 g of pBABE- monitored for both luciferase and ␤-galactosidase activity. Relative ␮ ␮ ␮ Puro, 1.0 g of H2B-GFP, and etiher 6.0 g of RSV-Cyclin D1, 6.0 gof luciferase activity is shown with error bars representing the SD. AR AR⌬46-408 (dnAR), or 6.0 ␮g of CMV-NeoBam (empty vector). Twenty-four activity in the presence of ethanol was set to 1. Addition of 1 nM DHT h after transfection, the media was supplemented with 0.25 ␮g/ml puromycin (Sigma) for rapid selection. After selection, mRNA was harvested using Trizol significantly induced activity of endogenous AR 3.8-fold, whereas reagent as described by the manufacturer (Invitrogen). Random hexamer addition of ectopic wild-type AR increased this activity to 11.3-fold, primers and Superscript II reverse transcriptase (Invitrogen) were used to as expected (Fig. 1A, top). Cotransfection of cyclin D1, however, prepare cDNA from 2.0 ␮g of RNA. To detect and quantify PSA and GAPDH reduced activity to near basal levels in both cases (0.67-fold for levels, radioactive PCR was performed with primers as described previously endogenous and 3.7-fold activation for ectopic). Inhibition of AR (29). cDNA (2 ␮l) was subjected to 27 cycles (94°C30s,51°C30s,and72°C transactivation was not caused by general transcriptional inhibition ␣ 32 ␮ for 30 s) of PCR in the presence of [ - P]dCTP (5.8 Ci/reaction). Resulting because basal AR activity was not affected by cyclin D1 cotransfec- products were separated on a 6% nondenaturing PAGE and visualized via tion. Thus, cyclin D1 is an effective inhibitor of endogenous and phosphoimaging (Molecular Dynamics, Sunnyvale, CA). Quantification of ectopic AR activity as determined by reporter assay. band density was performed using Image Quant software (Molecular Dynam- ics) and PSA expression normalized to GAPDH. To determine the level of cyclin D1 required to inhibit AR, molar concentrations of AR and cyclin D1 were examined after transfection. For these experiments, GST-tagged cyclin D1 and AR NH terminus RESULTS 2 (amino acids 1–173) were isolated and quantified using standard Cyclin D1 Inhibits AR Activity in Androgen-dependent Pros- techniques. Known concentrations of GST fusion proteins (Fig. 1A, tatic Adenocarcinoma Cells. We have previously demonstrated that Lanes 1–4) were then subjected to SDS-PAGE alongside transfected cyclin D1 is a potent AR repressor, inhibiting transactivation of the LNCaP cell lysates (Fig. 1A, Lanes 5–8). Using antibodies with PSA promoter in both C33A and CV1 cell lines, and leading to cell defined epitopes, proteins were detected by immunoblot and quanti- cycle attenuation in androgen-dependent prostatic adenocarcinoma fied using densitometry. Calculation of protein content based on GST cells (LNCaP; Refs. 18, 19). In addition, we demonstrated that AR- standards revealed that after transfection, LNCaP cells express T877A, a common tumor-derived allele present in LNCaP cells, is ϳ0.030 nmol cyclin D1/␮g total protein and 0.033 nmol AR/␮g total effectively repressed by cyclin D1 in the context of CV1 cells (19). To protein (Fig. 1A, bottom). This translates to a 0.9:1 molar ratio determine whether AR repression by cyclin D1 is effective in the achieved in vivo and suggests a stoichometric interaction between context of an androgen-dependent prostate cancer cell, reporter assays corepressor and receptor. were performed on ectopic and endogenous AR in LNCaP cells. Lastly, we examined the ability of the AR to activate endogenous Previous to this study, a titration curve was performed in CV1 cells to target gene expression in the presence of cyclin D1. LNCaP cells were assess the ability of cyclin D1 to repress AR activity on the PSA61- transfected with pBABE-Puro and either empty vector (CMV-Neo- LUC reporter (18). The minimal DNA ratio of cyclin D1 to AR Bam), cyclin D1, or dominant negative AR (AR ⌬46-408) in the required for full inhibition of receptor activity (reduction to basal presence of complete serum. Transfected cells were subject to rapid levels) was determined to be 3:1. On the basis of these prior data, selection for 72 h in puromycin, at which time RNA was harvested LNCaP cells were transfected with the ARR2-LUC (probasin) re- and converted to cDNA. Radioactive PCR was then performed to porter construct, CMV-␤-galactosidase (internal transfection control), detect and accurately quantify PSA levels normalized to GAPDH. As

Fig. 1. Cyclin D1 inhibits endogenous and ectopic AR activity in LNCaP cells. A, top panel, LNCaP cells were transfected in the absence of steroid with the probasin (ARR2) reporter and expression plasmids indicated. After transfection, cells were stimulated as indicated for 72 h. After lysis, luciferase activity was measured and normalized to ␤-galactosidase. Vehicle-treated AR activity was set to 1. Data shown represent at least three independent experiments. Error bars depict the SD. Lower panel, purified GST-AR1-173 (Lanes 1–4, top panel) and GST-cyclin D1 (Lanes 1–4, bottom panel) proteins were subjected to SDS- PAGE at the concentrations indicated alongside whole cell lysates (␮g as indicated) from LNCaP cells transfected with AR and cyclin D1 at a 1:3 ratio. Using the GST standards, AR and cyclin D1 proteins levels were quantified by densitometry as stated in “Materials and Methods.” B, LNCaP cells were transfected in the presence of complete serum as described in “Materials and Methods.” After rapid selection with puromycin, RNA was harvested and converted to cDNA using standard techniques. Amplification of PSA and GAPDH was performed in the presence of [␣-32P]dCTP. PCR products were subject to nondenaturing PAGE and quantified using a phosphoimager. PSA expression was normalized to GAPDH (internal control) and relative expression is shown (bottom panel).

4905

37 SPECIFICITY OF CYCLIN D1 ACTION seen in Fig. 1B, PSA mRNA levels were reduced 43.5% in the for both luciferase and ␤-galactosidase activity. Relative luciferase presence of dominant negative AR when compared with vector only. activity is shown with error bars representing the SD. As expected, Addition of cyclin D1 similarly reduced PSA activity, resulting in a addition of 0.1 nM DHT induced AR activity in all cell types examined 36.0% decrease in mRNA production compared with control. Taken (Fig. 2, A–D, left bars). Induction of AR in response to DHT was most together, these data conclusively demonstrate that cyclin D1 is an pronounced in MCF7 cells (ϳ14.7 fold), with TSUPr1, PC3, and effective inhibitor of AR activity in androgen-dependent prostatic 22Rv1 induction remaining slightly lower at ϳ9.6-, ϳ6.9-, ϳ10.0- adenocarcinoma cells. fold, respectively. However, addition of cyclin D1 or cyclin D1-KE Cyclin D1 Repression of AR Activity Is Independent of Cell reduced AR transactivation capacity to basal levels in all cell types. Type and Promoter. It was recently reported that another AR core- Basal activity (in the presence of ethanol vehicle) remained un- pressor, DAX-1, functions in a cell type-specific manner (37). To changed as reported previously (18, 19). The G1 cyclin, cyclin E, was further examine the specificity of cyclin D1 corepressor activity in also examined as a previous report demonstrated that under specified context of cancer, four distinct tumor cell lines were used: PC-3; conditions this protein serves as an AR coactivator (42). In our 22Rv1; MCF-7; and TSUPr1. 22Rv1 cells retain important character- experiments, AR activity was only slightly enhanced in TSUPr1, istics of clinical androgen-independent prostate cancer as they main- MCF-7, and 22Rv1 cells. These data together demonstrate that unlike tain both AR expression and activity but have bypassed the require- DAX-1, cyclin D1 is a potent AR repressor in multiple cell types, ment for androgen (28). PC-3 cells are derived from prostatic suggesting a highly conserved mechanism of repression. adenocarcinoma but have lost AR and PSA expression (38). TSUPr1 In addition to cell type specificity of comodulators, promoter- cells were originally believed to be prostatic in origin but were later specific effects have been noted for many AR coactivators and re- shown to be identical to T24 bladder carcinoma cells (39). As a result, pressors. It is hypothesized that the AR binds differentially to AREs many early studies characterizing AR comodulators were performed within promoters of target genes and that this may contribute to in this cell type. MCF-7 cells were also examined because AR is promoter and cell type-specific responses to androgens (43). Specif- suspected to play a role in breast cancer, yet their endogenous AR ically, both ARIP3 and PIAS1 are AR coactivators known to enhance appears nonfunctional (40, 41). transcription from minimal AREs, yet ARIP3 (but not PIAS1) re- To test the specificity of cyclin D1 in these four cell types, all cells presses the probasin promoter (44). Herein, we determine that cyclin were transfected with plasmids encoding wild-type human AR and D1 potently represses AR activity in the context of androgen-depen- either cyclin D1, cyclin D1-KE (a mutant defective in CDK4 binding dent prostatic adenocarcinomas, monitoring endogenous PSA pro- 25), cyclin E, or vector control (pCDNA3) at a 1:3 ratio. AR activity moter activity (Fig. 1, A and B). To further examine the promoter was measured using the PSA61-LUC reporter construct, which con- specificity of AR inhibition by cyclin D1, we used two well-charac- tains 6.1-kb of the PSA promoter fused to luciferase. After transfec- terized and widely used androgen responsive promoters, MMTV and tion, cells were stimulated with either 0.1 nM DHT or 0.1% ethanol probasin. CV1 cells were transfected with expression plasmid encod- vehicle for 22–24 h. Cells were then harvested, lysed, and monitored ing wild-type AR, MMTV, or probasin luciferase-based reporters and

Fig. 2. Cyclin D1 is refractory to cellular context. A–D, MCF-7, TSUPr1, PC-3, and 22Rv1 cells were seeded in steroid-free media and transfected as described in “Materials and Methods” with the PSA61-LUC reporter and expression plasmids as indicated. Transfected cells were washed and left to recover for 4–6 h before treatment as indicated for 22–24 h. After stimulation, cells were harvested and luciferase activity measured. Data were normalized to ␤-galactosidase activity and vehicle-treated AR activity set to 1. Data shown represent at least three independent experiments with error bars representing the SD. 4906

38 SPECIFICITY OF CYCLIN D1 ACTION

Fig. 3. Cyclin D1 demonstrates repressor activity across multiple AR target promoters. A, CV1 cells were transfected with MMTV or ARR2 (probasin) reporter constructs as described in Fig. 2 in the presence or absence of cyclin D1 or cyclin D1-KE at a 3:1 ratio with the AR. After stimulation with ligand as indicated, reporter assays were performed as described in “Materials and Methods.” B, CV1 cells were transfected as in Fig. 2 with ER␣, pBS3XERE-LUC, CMV- ␤-galactosidase, and vector or cyclin D1 at a 3:1 ratio. Cells were washed and stimulated as indicated for 24 h. Lysates were obtained and data collected and reported as in Fig. 2.

either cyclin D1, cyclin D1-KE, or vector control (at a 3:1 ratio with using the PSA reporter as described for Fig. 2. Relative luciferase the AR). Cells were stimulated, harvested, and monitored for ␤- activity is reported with error bars representing SDs. In the presence galactosidase and luciferase activity as previously described in Fig. 2. of 0.1 nM DHT, activity of S81,94A and S650A was increased ϳ7.5- Relative luciferase activity is reported with error bars representing and ϳ18.0-fold, respectively (Fig. 4, A and B). Addition of cyclin D1 SDs. Both the MMTV and probasin promoters were strongly activated or cyclin D1-KE at a 3:1 ratio with AR phosphorylation site mutants by 0.1 nM DHT (ϳ107.5- and ϳ252.8-fold, respectively) when com- completely abolished transactivation, returning relative luciferase ac- pared with vehicle control (Fig. 3A). AR activity on each promoter tivity to basal levels. Phosphorylation of the AR was originally was reduced significantly with the addition of cyclin D1 or cyclin hypothesized to regulate its turnover in vivo, but recent evidence D1-KE. Fold repression by cyclin D1 on the MMTV (14.5 fold) and suggests it may contribute to the modulation of AR stability and probasin (8.9 fold) reporters was similar to our previous findings nuclear export, thus altering its activity (45). To verify expression of using the PSA promoter (10–12 fold repression; Refs. 18, 19). To AR phosphorylation site mutants in the presence of ectopic cyclin D1, further demonstrate the specificity of cyclin D1 corepressor activity experiments were conducted in parallel wherein plasmid-encoding for the AR, CV1 cells were transfected as in Fig. 3A with plasmids H2B-GFP (histone H2B tagged with GFP) was substituted for CMV- encoding ER␣, pBS3XERE-LUC (an estrogen-responsive reporter), ␤-galactosidase. Cell lysates were prepared and subjected to SDS- CMV-␤-galactosidase, and cyclin D1 or vector. Transfected cells PAGE followed by immunoblotting for both the AR and GFP. Ex- were treated with 10 nM 17␤-estradiol, a natural ligand for ER␣,or pression of AR phosphorylation mutants was not altered by vehicle (Fig. 3B). Consistent with previous studies, at a 3:1 ratio with coexpression of cyclin D1 (Fig. 4C), indicating that the repression the receptor, cyclin D1 served as a coactivator for ER␣ activity noted was not because of an increase in protein turnover or decrease enhancing relative luciferase activity from 6- to 15-fold (20–22). in receptor production. These findings are consistent with our previ- Thus, cyclin D1 fails to repress promoters driven by ER␣ activity but ously published data showing that cyclin D1 does not affect stability maintains its corepressor function on several androgen-responsive of the wild-type AR (19). Taken together, these data demonstrate that targets. Taken together, these data validate the ability of cyclin D1 the phosphorylation state of these critical residues in the NH2 termi- corepressor activity to span multiple androgen-responsive promoters nus and DNA binding domain of the AR does not affect cyclin D1 and distinguish it from other AR comodulators known to function in corepressor activity. a promoter-specific fashion. Moreover, these data demonstrate an AR Polymorphisms and Tumor-derived AR Alleles Respond to equal efficacy with regard to fold repression. Cyclin D1 Inhibition. We previously mapped cyclin D1 binding to

Phosphorylation of Major AR Residues Residing within the the NH2 terminus of the AR (19). Within this region of the AR exist

NH2 Terminus and DNA Binding Domain Does Not Regulate two polymorphic repeat domains, the polyglutamine and polyglycine Cyclin D1 Corepression Activity. We previously demonstrated that repeats. Although the length of the glycine repeat appears to have little cyclin D1 directly binds to the NH2 terminus of the AR to inhibit prognostic value, the number of NH2-terminal glutamine repeats has transactivation (19). In a cell type-independent fashion, it has also suggested clinical implications (46–48). Expanded repeats (Ͼ22) are been noted that the NH2 terminus and hinge region of the AR have a found in patients with androgen-insensitivity syndrome, whereas in- high frequency of in vivo phosphorylation at three specific sites, dividuals with contracted repeats (Ͻ22) harbor a putative increased serines 81, 94 and 650 (34, 45). Because phosphorylation of the AR risk for prostate cancer development (46, 48). Current evidence sug- is hypothesized to play a role in receptor activation and prostate gests that the contracted repeats result in a more active receptor (49, cancer progression, we examined the significance of these sites in 50). To examine the ability of cyclin D1 to repress such polymorphic cyclin D1-mediated AR repression (34, 45). Using previously de- alleles, we used expression constructs encoding GFP-tagged polymor- scribed phosphorylation site mutants of the AR [S81,94A (double phic ARs with 0 and 48 glutamine repeats (Q0 and Q48, respectively; mutant) and S650A 34], reporter assays were preformed in CV1 cells Ref. 33). Transfections were performed in CV1 cells as described in 4907

39 SPECIFICITY OF CYCLIN D1 ACTION

ligand binding pocket mutant alleles were examined, all of which have been reported in recurrent prostatic adenocarcinoma (17, 51). The PSA61-LUC reporter was used as readout for AR activity and normalized as in previous experiments to ␤-galactosidase. In the presence of ligand (0.1 nM DHT), all mutants demonstrated a significant increase in transactivation of the PSA promoter (Fig. 6, A–F). Addition of cyclin D1 at a 3:1 ratio with the AR-reduced activity to basal levels, consistent with our observations using the wild-type receptor (Figs. 1–3). These data clearly demonstrate that cyclin D1 retains its AR corepressor activity even in the presence of clinically relevant ligand binding mutations known to play a role in prostate cancer progression. Cyclin D1 Is Dominant to Alternatively Activated AR. Confor- mation of nuclear receptors is known to vary in the presence of differential ligands (52). The ligand molecular structure appears to dictate the position of helix 12 within the ligand binding domain of steroid hormone receptors (52). In this fashion, it is hypothesized that differential AR activity is produced in response to individual ligands. Specifically, it has been demonstrated that the potency and specificity of wild-type AR transactivation differs when activated with testosterone versus DHT (53). We previously showed that cyclin D1 is able to inhibit both methyltrienolone (R1881, a synthetic DHT analogue) and DHT-induced AR activity (18, 19). To test the efficacy of cyclin D1 on testosterone-induced AR transactivation, CV1 cells were transfected as in Fig. 2 and treated with either 0.1 nM testosterone or 0.1%

Fig. 4. NH2-terminal phosphorylation sites (serines 81, 94, and 650) are not required for cyclin D1 action. A and B, CV1 cells were transfected as described in Fig. 2 with expression vector encoding phosphorylation site mutants of the AR in the presence or absence of cyclin D1 or cyclin D1-KE at a 1:3 ratio. Reporter assays were conducted as described in Fig. 2. C, parallel experiments were performed wherein H2B-GFP was substituted for CMV-␤-galactosidase. Lysates were harvested and subjected to SDS- PAGE followed by immunoblotting as designated.

Fig. 2, using the PSA61-LUC reporter gene as readout for AR activity. Relative luciferase activity is shown (Fig. 5, A and B). As previously reported, ARs with an expanded number of glutamine repeats (Q48) displayed lower activity in the presence of 0.1 nM DHT than the contracted ARQ0 (compare Fig. 5A with Fig. 5B: ϳ10.6-fold in comparison with ϳ5.0-fold; Ref. 50). Addition of cyclin D1 at a 3:1 ratio with the AR did not affect basal transactivation in any of the polymorphisms and DHT-induced AR transactivation was completely inhibited. In parallel experiments wherein H2B-GFP expression plasmid was substituted for that encoding CMV-␤-galactosidase, immunoblotting revealed that AR expression levels remained unchanged (Fig. 5C). Taken together, these data suggest that cyclin D1 does not require a modal number of NH2-terminal polyglutamine repeats to inhibit AR transactivation (Fig. 5, A and B) and that contracted alleles with enhanced activity are susceptible to cyclin D1 repressor function. In addition to polymorphisms in the glutamine repeat domain of the AR, mutations within the ligand binding pocket of its COOH terminus have been demonstrated to play a role in prostate cancer development and progression (17). Several mutations within the ligand binding domain are selected for during prostate cancer therapy and result in promiscuous ligand binding and AR activation (17, 31). Given the Fig. 5. Contracted and expanded AR polyglutamine polymorphisms are sensitive to cyclin D1. A and B, CV1 cells were seeded and transfected for reporter assays as in Fig. frequency of AR mutation in late-stage prostate cancer, determining 2 with constructs encoding GFP-tagged glutamine repeat AR polymorphisms (Q0 and the efficacy of cyclin D1 on these tumor-derived alleles is of obvious Q48) and in the presence or absence of cyclin D1 or cyclin D1-KE at a 1:3 ratio. C, lysates from parallel experiments to those described in A and B wherein H2B-GFP was substi- importance. To achieve this goal, CV1 cells were transfected as in Fig. tuted for CMV-␤-galactosidase were subject to SDS-PAGE and immunoblotted for GFP 2 with constructs encoding tumor-derived AR alleles (31, 51). Six AR (bottom panel) and GFP-AR (top panel). 4908

40 SPECIFICITY OF CYCLIN D1 ACTION

Fig. 6. Cyclin D1 abrogates the function of clinically relevant prostate cancer-derived AR alleles. A–F, CV1 cells were transfected for reporter assays as described in Fig. 2 with the indicated clinically relevant point mutations of the AR and in the presence or absence of cyclin D1 (at a 1:3 ratio).

ethanol vehicle. Stimulation with 0.1 nM testosterone resulted in an reported, potentiated DHT-mediated AR transactivation, increasing ϳ5-fold induction of luciferase activity over cells treated with vehicle PSA61-LUC activity from 10.5- to 16.3-fold (P Ͻ 0.01; Refs. 15, 60). alone (Fig. 7A). CV1 cells lack the enzyme to reduce testosterone to Cotransfection of cyclin D1 at a 3:1 ratio with the AR, however, DHT and, therefore, PSA61-LUC induction observed was not because completely inhibited the combined effect of these two ligands. These of metabolic conversion. In addition, decreased AR activity in re- data imply that cyclin D1 remains functional, even in the presence of sponse to testosterone versus DHT was expected because testosterone nonligand activators known to potentiate the transition from androgen stimulation of the MMTV promoter is reduced in comparison with dependence to independence in late-stage prostate cancers. DHT (54). Addition of cyclin D1 or cyclin D1-KE inhibited testos- Lastly, nonconventional AR ligands are known to play a role in terone-induced AR transactivation to basal levels, indicating that prostate cancer progression by acting to stimulate ligand binding cyclin D1 corepressor activity is still functional when alternate AR pocket mutants of the AR in the absence of androgen. As such, natural ligands are used. mutations are found in 5–37% of recurrent prostate cancers, and The transition of prostate cancer from androgen dependence to regulation of inappropriate (in the absence of androgen) AR activity independence is hypothesized to involve adaptation of the receptor by cyclin D1 was examined (2). CV1 cells were transfected as in Fig. such that cytokines, low-level androgens, and growth factors can 2 with plasmid encoding AR-T877A, a common ligand binding pocket stimulate AR activity (14, 55). This theory is supported by data mutant of the AR known to be responsive to both progesterone and demonstrating that IL-6, EGF, IGF-I, and KGF activate the AR and 17␤-estradiol. As expected, 17␤-estradiol- and progesterone-stimu- can enhance transactivation driven by low-level androgen (15, 16, 56). lated activity of AR-T877A ϳ2.6- and 17.4-fold, respectively, in Indeed, high levels of growth factors and cytokines are found proxi- comparison to vehicle alone (Fig. 7C). However, addition of ectopic mal to late-stage prostate tumors (57–59). The precise mechanism by cyclin D1 at a 3:1 ratio with AR-T887A resulted in abrogation of AR which the AR is activated through these alternate pathways has yet to activity in the presence of both steroids. These data additionally be uncovered. To determine whether cyclin D1 can inhibit AR activity demonstrate the potency of cyclin D1 corepressor activity and suggest induced by these alternate mechanisms significant to prostate cancer that novel therapeutics modeled after such repression would maintain progression, reporter assays were performed as described in CV1 cells their efficacy in androgen-independent prostate cancers. transfected as described in Fig. 2 but after transfection cells were placed in 0.1% CDT serum to lower the amount of endogenous DISCUSSION growth factors and cytokines available. In our system, treatment with EGF (50 ng/ml) and IGF (50 ng/ml) produced no significant induction Although many coactivators of the AR have been identified, far of the PSA reporter and failed to synergize with low-level androgen fewer corepressors have been established and well characterized. (0.1 nM DHT; data not shown). This result is similar to that recently Because the regulation of AR activity is essential to the current published by Ueda et al. (60). In the presence of 50 ng/ml IL-6, a treatment of prostate cancers as well as the future development of slight but not significant increase (ϳ1.5-fold over basal; P Ͼ 0.05) in novel therapeutics, in-depth characterization of relevant AR corepres- PSA61-LUC transactivation was observed, which was completely sors has obvious clinical relevance. In this study, we demonstrate that inhibited by cotransfection of cyclin D1 (Fig. 7B). IL-6, as previously cyclin D1 is a potent AR corepressor, capable of inhibiting receptor 4909

41 SPECIFICITY OF CYCLIN D1 ACTION

of biological outcomes. Cell type specificity has also been recognized to modulate PSA promoter activity, wherein PC-3 cells have reduced transactivation in comparison with MCF-7 cells (62). The AR corepressor, DAX-1, also functions in a cell type-dependent manner as its activity diminishes in HeLa cells (50% inhibition) in comparison to that noted in the COS-7 (80% inhibition) cell type (37). It is important to note that specificity of DAX-1 action is observed at even higher repressor to receptor ratios than used in this study (10:1 versus 3:1). These findings lead to the hypothesis that cyclin D1 corepressor activity could also be regulated in a cell type specific fashion. We show that cyclin D1 inhibits ligand-dependent activity of both ectopic and endogenous AR in LNCaP cells (Fig. 1, A and B). Strikingly, this repression event occurs at a 0.91:1.00 molar ratio, indicating that even low levels of cyclin D1 are efficient at tempering AR activity (Fig. 1A). This result is in keeping with the model that androgen-dependent induction of cyclin D1 in LNCaP cells likely serves to regulate the rate of future cell cycle progression. To assess the cell type specificity of cyclin D1, we examined AR transactivation of the PSA promoter in two additional cell types derived from androgen-independent cancers (PC-3 and 22Rv1) as well as those that were initially thought to be derived from a prostatic adenocarcinoma (TSUPr1) and thus used previously in the characterization of other AR comodulators. In addition, we examined the effect of cyclin D1 on AR activity in breast carcinoma because AR activity in this cell type is thought to contribute to tumor regression upon administration of medroxyprogesterone acetate as therapy (41). In all cell types, cyclin D1 maintained its corepressor activity, reducing AR transactivation to basal levels in 22Rv1, TSUPr1, PC-3, and MCF-7 cells (Fig. 2, A–D). Overall, cyclin D1 inhibition of the AR appears to be conserved in multiple cell types supporting its efficacy as an AR inhibitor. In addition to cell type specificity, a number of AR comodulators also demonstrate promoter specificity. Both ARIP3 and PIAS1 are AR Fig. 7. Cyclin D1 prevents AR transcription stimulated by natural, nonconventional, coactivators known to enhance transcription from minimal AREs, yet and nonligand (IL-6) activators. A, CV1 cells were transfected as described in Fig. 2 with the plasmids indicated. After 22–24 h treatment as indicated, cell lysates obtained were ARIP3 (but not PIAS1) represses the probasin promoter (44). N-C monitored for luciferase and ␤-galactosidase activity. B, CV1 cells transfected as de- interaction of the AR is essential for both PSA and probasin promoter scribed in Fig. 2 were washed and then placed in 0.1% CDT serum. As described in regulation but is not required for activation of MMTV and sex-limited “Materials and Methods,” cells were treated with either 0.1 nM DHT, 50 ng/ml IL-6, both, or vehicle (0.1% ethanol and 10 mM acetic acid). After stimulation, cells were harvested protein (61). Binding of the AR to AREs on target promoters is and monitored for luciferase activity and normalized to ␤-galactosidase. C, CV1 cells sequence specific as the response element sequence dictates receptor were transfected as in A and stimulated with either 10 nM progesterone, 10 nM 17␤- estradiol, or 0.1% ethanol vehicle for 22–24 h. After stimulation, cells were harvested and binding affinity (43). Thus, examination of cyclin D1 corepressor AR activity depicted as in A. activity on multiple gene promoters was essential to determine the specificity of its action in vivo. We previously demonstrated cyclin D1 inhibition of AR activity on the PSA promoter in the context of CV1 transactivation independent of cell type or AR target promoter ana- cells and data shown herein examine its repressor activity on the lyzed. Cyclin D1 also maintains its corepressor activity on key phos- MMTV and probasin promoters (18, 19). Cyclin D1 effectively in- phorylation-site mutants, polymorphisms, and tumor-derived AR hibited MMTV and probasin transactivation, consistent with our pre- alleles. Both ligand (testosterone, DHT, 17␤-estradiol, and progester- vious result using the PSA promoter (10–12-fold repression, Fig. 3A; one) and nonligand (IL-6) activated AR complexes are repressed by Ref. 19). In addition, we demonstrate that cyclin D1 inhibition of cyclin D1 action. Taken together, our data suggest that cyclin D1 androgen-responsive promoters is not because of general transcrip- functions not only in androgen-dependent prostate cancers but also in tional inhibition as it enhances ER␣ transactivation of target genes as the milieu of AR mutations, polymorphisms, and nontraditional acti- reported previously (Fig. 3B; Refs. 20–22). Finally, we show that vators predisposing to the development of androgen-independent transactivation of endogenous PSA in the presence of steroid is cancers. similarly reduced by ectopic cyclin D1 or dominant negative AR Cyclin D1 Repression Is neither Promoter nor Cell Type Spe- expression (Fig. 1B). These data indicate that cyclin D1 corepressor cific. We previously demonstrated that cyclin D1 fully inhibits AR activity targets a wide array of characterized AR promoters. transactivation of the PSA61-LUC reporter in CV1 and C33A cells The Inhibitory Action of Cyclin D1 Is Independent of NH2- (18, 19). Recent studies suggest that AR-mediated gene transcription Terminal AR Phosphorylation. The AR is a phosphoprotein with is influenced based upon the cell type and promoter examined (37, 43, modification hypothesized to originate from upstream signal trans- 44, 61). Certainly, such differences are biologically essential because duction cascades (34, 45, 60). A recent study demonstrated that in vivo the response of AR-containing tissues to androgens is predicted to phosphorylation of the AR at identified sites (other than at serine 308) vary dependent on cellular context. Within each cell type, expression has seemingly no effect upon its ability to transactivate target gene levels of transcription factors and AR accessory molecules are hy- promoters (45). Instead, it is hypothesized that the stabilization and/or pothesized to regulate receptor transactivation, leading to a diverse set localization of the AR may be regulated through phosphorylation, yet 4910

42 SPECIFICITY OF CYCLIN D1 ACTION no study, to date, has determined the exact mechanism by which this fact, mutation of the AR has been reported to occur in between 5 and regulation may occur (45). With the finding that cytokines and growth 37% of prostate cancer patients, with the higher rates being docu- factors can induce AR phosphorylation and are up-regulated in the mented in patients treated with multiple antiandrogen therapies (re- vicinity of androgen-independent prostate cancer tumors, elucidation viewed in Ref. 65). Mutation of the AR ligand binding domain yields of the role of phosphorylation in AR signaling has become desirable receptors responsive to 17␤-estradiol (T877A/S and H874Y), proges- (57–59). In vivo studies have indicated that at least two major phos- terone (T877A/S, H874Y, and V715M), adrenal androgens (V715M, phorylation sites (serines 81 and 94) exist in the NH2-terminal region H874Y, and T877S/A), cortisol (H874Y and T877A) and even antian- of the AR and one in the hinge region (serine 650; Refs. 34, 45). As drogens such as hydroxyflutamide (T877A/S, V715M, and H874Y), cyclin D1 binding requires the AR NH2 terminus and phosphorylation which is used in the treatment of advanced prostate cancers (17, 65). of the receptor may play a role in prostate cancer transition toward AR mutations at codons 890 (D to N) and 902 (Q to R) have also been androgen independence, we examined the ability of cyclin D1 to identified in flutamide-treated patients (31, 51). Hydroxyflutamide inhibit PSA transactivation by phosphorylation site mutants of the AR failed to stimulate transactivation of the 890 and 902 mutants, but the (19). As previously noted, AR activity was not significantly modu- efficacy of other alternative ligands has yet to be examined (51). lated by phosphorylation site receptor mutants (Fig. 4, A and B). Because ligand binding domain mutations are frequent in androgen- Addition of cyclin D1 or cyclin D1-KE, however, diminished trans- independent prostate cancers, we examined the efficacy of cyclin D1 activation of phosphorylation site AR mutants to basal levels (Fig. 4, inhibition of clinically relevant alleles. In all reporter assays wherein A and B). In addition, cyclin D1 corepressor activity was not caused AR mutants examined (V715M, A721T, H874Y, T877S, T877A, simply by the down-regulation of AR protein levels, as immunoblot- D890N, and Q902R) cyclin D1 served as a potent inhibitor of AR ting revealed in Fig. 4C. These data demonstrate efficacy for cyclin activity, reducing PSA or probasin transactivation to basal levels D1 irrespective of AR phosphorylation at serines 81, 94, and 650. (Figs. 1A and 6, A–F). Moreover, activation of AR-T877A by 17␤- Cyclin D1 Inhibits Transactivation of AR Polymorphisms and estradiol or progesterone was ablated by cyclin D1 (Fig. 7C). These Ligand Binding Pocket Mutants Implicated in Prostate Cancer results together suggest that cyclin D1 will repress the activity of Development and Androgen Independence. Two polymorphic re- mutant ARs frequently arising during the transition of prostate cancer gions exist within the AR NH2 terminus, the polyglutamine and tumors to androgen independence. polyglycine tracts (reviewed in Ref. 7). Polymorphisms within the Cyclin D1 Represses AR Activity Regardless of Ligand Activa- polyglycine repeat appear to be clinically insignificant, whereas ex- tor. It is known that AR response to individual ligands results in pansion and contraction of the polyglutamine repeat is reported to differential gene expression. Previous studies demonstrated that tes- result in significant biological outcomes (46–48, 63, 64). Polyglu- tosterone bound AR exhibits differential regulation of AREs (53, 54). tamine expansion (40–62 repeats) is associated with Kennedy’s dis- Specifically, DHT was more potent in stimulating MMTV-driven ease, dentatorubral and pollidoluysain atrophy, and spinocerebeller promoters, whereas testosterone bound AR showed increased activity ataxia (63, 64). Reduced androgen sensitivity because of repeat exon a multimerized ARE site (53). In addition, binding of individual pansion is thought to result in the neurodegeneration in these diseases. ligands to the AR is known to trigger specific receptor conformations. Contraction of the polyglutamine repeat (Ͻ10 glutamines), however, For example, DHT binding to the AR causes helix 12 closure over the is associated with increased AR activity, leading to a higher propen- AR ligand binding pocket, whereas binding of the DHT analogue, sity to develop both benign prostatic hyperplasia and prostate cancer R1881, results in a bipartite helix 12 conformation (66). Because the (47, 48). It has been reported that men harboring ARs with glutamine AR is seemingly activated through multiple mechanisms within the repeats fewer than 19 amino acids have a 52-fold greater risk of prostate cancer patient and the type of activator can modulate AR developing prostate cancer when compared with those with Ն25 conformation and activity, we examined the ability of cyclin D1 to glutamine residues (46). Shortened repeats were also shown to cor- inhibit AR activity in the presence of testosterone. Our data indicate relate with prostate cancer metastasis and higher mortality (48). Two that cyclin D1 maintains repression of the AR even in the presence of hypotheses have been generated to explain the differential activity of testosterone (Fig. 7A). In addition, we previously demonstrated that glutamine repeat polymorphisms: (a) expansion of the polyglutamine AR activation by R1881 and DHT is inhibited by ectopic cyclin D1 repeat is known to result in the formation of AR aggregates, thus expression (18). Taken together, these data verify the efficacy of inhibiting its ability to transactivate target promoters; and (b) ARs cyclin D1-mediated inhibition of the AR in the presence of multiple with expanded polyglutamine repeats have reduced binding capacity natural ligand activators. for coactivators and may instead cause corepressor association (33). As discussed above, androgen ablation therapy is thought to select Evidence exists to support both of these hypotheses that appear to be for cancer cells with the ability to grow independently of androgen, inclusive, both contributing to the patient phenotypes observed. As we yet analysis of these tumors reveals that the AR remains expressed previously mapped the binding of cyclin D1 to the AR NH2 terminus and active (14). Multiple mechanisms have been proposed for over- (containing the polyglutamine repeat domain; Ref. 19), examination coming androgen dependence, including mutation of the AR ligand of its ability to serve as a corepressor on both expanded (Q48) and binding domain, amplification of the receptor gene, and activation by contracted (Q0) repeats was necessary to assess the efficacy of cyclin nonligands (14, 55). Growth factors and cytokines are up-regulated in D1 as a corepressor. Our data indicate that cyclin D1 maintains its the milieu of androgen-independent tumors (57, 59). This observation inhibitory activity on ARs of varying glutamine repeat length (Fig. 5, led to the hypothesis that these nonligands stimulate AR activity A and B). Using Western blot analysis we also demonstrate that cyclin possibly through triggering of AR phosphorylation downstream of D1 does not alter AR protein levels to down-regulate AR transacti- specific signal transduction pathways. Previous studies have demon- vation (Fig. 5C). These findings classify cyclin D1 as a potent AR strated AR activation by EGF, IGF-I, KGF, and IL-6 (15, 16). As corepressor, capable of inhibiting receptor activity irrespective of inhibition of androgen-independent AR activity is a major target for glutamine repeat polymorphisms. the development of novel prostate cancer therapies, we examined the Androgen ablation therapy, although initially effective, appears to ability of cyclin D1 to inhibit EGF-, IGF-I-, and IL-6-induced AR select for a population of tumor cells adept to growing in an androgen- transactivation. In our hands, EGF and IGF-I failed to stimulate independent fashion. One method by which prostate tumors may AR-mediated transactivation of the PSA promoter (data not shown). subvert androgen dependence is through mutation of the AR itself. In IL-6, however, slightly but not significantly enhanced AR activity 4911

43 SPECIFICITY OF CYCLIN D1 ACTION alone (P Ͼ 0.05) and potentiated low-level DHT activity (P Ͻ 0.01; 20. Neuman, E., Ladha, M. H., Lin, N., Upton, T. M., Miller, S. J., DiRenzo, J., Pestell, Fig. 7B). AR activation by both IL-6 alone and IL-6 plus DHT was R. G., Hinds, P. W., Dowdy, S. F., Brown, M., and Ewen, M. E. Cyclin D1 stimulation of estrogen receptor transcriptional activity independent of cdk4. Mol. completely inhibited by coexpression of cyclin D1. This finding Cell. Biol., 17: 5338–5347, 1997. indicates that cyclin D1 inhibition is dominant to nonligand activators 21. Zwijsen, R. M., Buckle, R. S., Hijmans, E. M., Loomans, C. J., and Bernards, R. Ligand-independent recruitment of steroid receptor coactivators to estrogen receptor and additionally supports its efficacy in inhibiting androgen-indepen- by cyclin D1. Genes Dev., 12: 3488–3498, 1998. dent AR activity. 22. McMahon, C., Suthiphongchai, T., DiRenzo, J., and Ewen, M. E. P/CAF associates In summary, cyclin D1 is a uniquely potent corepressor of the AR with cyclin D1 and potentiates its activation of the estrogen receptor. Proc. Natl. Acad. Sci. USA, 96: 5382–5387, 1999. with broad specificity for ablation of ligand-dependent transactiva- 23. Coqueret, O. Linking cyclins to transcriptional control. Gene (Amst.), 299: 35–55, tion. Cyclin D1 maintains its corepressor activity independently of 2002. cell type, promoter, and agonist examined. Its ability to inhibit clin- 24. Fribourg, A. F., Knudsen, K. E., Strobeck, M. W., Lindhorst, C. M., and Knudsen, E. S. Differential requirements for ras and the retinoblastoma tumor suppressor ically relevant polymorphisms and mutations suggests that analogues protein in the androgen dependence of prostatic adenocarcinoma cells. Cell Growth of cyclin D1 action may be useful in the treatment of initial (androgen Differ., 11: 361–372, 2000. dependent) and recurrent (androgen independent) prostate cancers. 25. Hinds, P. W., Dowdy, S. F., Eaton, E. N., Arnold, A., and Weinberg, R. A. Function of a human cyclin gene as an oncogene. Proc. Natl. Acad. Sci. USA, 91: 709–713, Taken together, our data provide the impetus to examine the in vivo 1994. effects of cyclin D1 on both androgen dependent and independent 26. Reutens, A. T., Fu, M., Wang, C., Albanese, C., McPhaul, M. J., Sun, Z., Balk, S. P., tumors. Janne, O. A., Palvimo, J. J., and Pestell, R. G. Cyclin D1 binds the androgen receptor and regulates hormone-dependent signaling in a p300/CBP-associated factor (P/ CAF)-dependent manner. Mol. Endocrinol., 15: 797–811, 2001. ACKNOWLEDGMENTS 27. Martinez, E. D., and Danielsen, M. Loss of androgen receptor transcriptional activity at the G1-S transition. J. Biol. Chem., 277: 29719–29729, 2002. 28. Sramkoski, R. M., Pretlow, T. G., II, Giaconia, J. M., Pretlow, T. P., Schwartz, S., Sy, We thank Drs. Steven Balk, Sohaib Khan, Erik Knudsen, and both Knudsen M. S., Marengo, S. R., Rhim, J. S., Zhang, D., and Jacobberger, J. W. A new human laboratories for their critical reading of the manuscript. We also thank Dr. prostate carcinoma cell line, 22Rv1. In Vitro Cell Dev. Biol. Anim., 35: 403–409, Randal Morris for his advice concerning statistical analyses. 1999. 29. Wetherill, Y. B., Petre, C. E., Monk, K. R., Puga, A., and Knudsen, K. E. The xenoestrogen bisphenol A induces inappropriate androgen receptor activation and REFERENCES mitogenesis in prostatic adenocarcinoma cells. Mol. Cancer Ther., 1: 515–524, 2002. 30. Sadar, M. D. Androgen-independent induction of prostate-specific antigen gene 1. Denmeade, S. R., and Isaacs, J. T. A history of prostate cancer treatment. Nat. Rev. expression via cross-talk between the androgen receptor and protein kinase A signal Cancer, 2: 389–396, 2002. transduction pathways. J. Biol. Chem., 274: 7777–7783, 1999. 2. Feldman, B. J., and Feldman, D. The development of androgen-independent prostate 31. Fenton, M. A., Shuster, T. D., Fertig, A. M., Taplin, M. E., Kolvenbag, G., Bubley, cancer. Nat. Rev. Cancer, 1: 34–45, 2001. G. J., and Balk, S. P. Functional characterization of mutant androgen receptors from 3. Taplin, M. E., and Ho, S. M. Clinical review 134: the endocrinology of prostate androgen-independent prostate cancer. Clin. Cancer Res., 3: 1383–1388, 1997. cancer. J. Clin. Endocrinol. Metab., 86: 3467–3477, 2001. 32. Palvimo, J. J., Kallio, P. J., Ikonen, T., Mehto, M., and Janne, O. A. Dominant 4. Denmeade, S. R., Lin, X. S., and Isaacs, J. T. Role of programmed (apoptotic) cell negative regulation of trans-activation by the rat androgen receptor: roles of the death during the progression and therapy for prostate cancer. Prostate, 28: 251–265, N-terminal domain and heterodimer formation. Mol. Endocrinol., 7: 1399–1407, 1996. 1993. 5. Brinkmann, A. O., Blok, L. J., de Ruiter, P. E., Doesburg, P., Steketee, K., Berrevoets, 33. Stenoien, D. L., Cummings, C. J., Adams, H. P., Mancini, M. G., Patel, K., C. A., and Trapman, J. Mechanisms of androgen receptor activation and function. DeMartino, G. N., Marcelli, M., Weigel, N. L., and Mancini, M. A. Polyglutamine- J. Steroid Biochem. Mol. Biol., 69: 307–313, 1999. expanded androgen receptors form aggregates that sequester heat shock proteins, 6. Chang, C. S., Kokontis, J., and Liao, S. T. Molecular cloning of human and rat proteasome components and SRC-1, and are suppressed by the HDJ-2 chaperone. complementary DNA encoding androgen receptors. Science (Wash. DC), 240: 324– Hum. Mol. Genet., 8: 731–741, 1999. 326, 1988. 34. Zhou, Z. X., Kemppainen, J. A., and Wilson, E. M. Identification of three proline- 7. Jenster, G., van der Korput, J. A., Trapman, J., and Brinkmann, A. O. Functional directed phosphorylation sites in the human androgen receptor. Mol. Endocrinol., 9: domains of the human androgen receptor. J. Steroid Biochem. Mol. Biol., 41: 605–615, 1995. 671–675, 1992. 35. Marshall, T. W., Link, K. A., Petre-Draviam, C. E., and Knudsen, K. E. Differential 8. He, B., Kemppainen, J. A., Voegel, J. J., Gronemeyer, H., and Wilson, E. M. requirement of SWI/SNF for androgen receptor activity. J. Biol. Chem., in press, Activation function 2 in the human androgen receptor ligand binding domain medi- 2003. ates interdomain communication with the NH -terminal domain. J. Biol. Chem., 274: 2 36. Chen, C., and Okayama, H. High-efficiency transformation of mammalian cells by 37219–37225, 1999. plasmid DNA. Mol. Cell. Biol., 7: 2745–2752, 1987. 9. Marivoet, S., Van Dijck, P., Verhoeven, G., and Heyns, W. Interaction of the 90-kDa 37. Holter, E., Kotaja, N., Makela, S., Strauss, L., Kietz, S., Janne, O. A., Gustafsson, heat shock protein with native and in vitro translated androgen receptor and receptor J. A., Palvimo, J. J., and Treuter, E. Inhibition of androgen receptor (AR) function by fragments. Mol. Cell. Endocrinol., 88: 165–174, 1992. 10. Roy, A. K., Tyagi, R. K., Song, C. S., Lavrovsky, Y., Ahn, S. C., Oh, T. S., and the reproductive orphan nuclear receptor DAX-1. Mol. Endocrinol., 16: 515–528, Chatterjee, B. Androgen receptor: structural domains and functional dynamics after 2002. ligand-receptor interaction. Ann. N. Y. Acad. Sci., 949: 44–57, 2001. 38. Kaighn, M. E., Narayan, K. S., Ohnuki, Y., Lechner, J. F., and Jones, L. W. 11. Trapman, J., and Brinkmann, A. O. The androgen receptor in prostate cancer. Pathol. Establishment and characterization of a human prostatic carcinoma cell line (PC-3). Res. Pract., 192: 752–760, 1996. Investig. Urol. (Berl.), 17: 16–23, 1979. 12. Nash, A. F., and Melezinek, I. The role of prostate specific antigen measurement in 39. van Bokhoven, A., Varella-Garcia, M., Korch, C., and Miller, G. J. TSU-Pr1 and the detection and management of prostate cancer. Endocr. Relat. Cancer, 7: 37–51, JCA-1 cells are derivatives of T24 bladder carcinoma cells and are not of prostatic 2000. origin. Cancer Res., 61: 6340–6344, 2001. 13. Riegman, P. H., Vlietstra, R. J., van der Korput, J. A., Brinkmann, A. O., and 40. Ando, S., De Amicis, F., Rago, V., Carpino, A., Maggiolini, M., Panno, M., and Trapman, J. The promoter of the prostate-specific antigen gene contains a functional Lanzino, M. Breast cancer: from estrogen to androgen receptor. Mol. Cell. Endocri- androgen responsive element. Mol. Endocrinol., 5: 1921–1930, 1991. nol., 193: 121–128, 2002. 14. Culig, Z., Klocker, H., Bartsch, G., and Hobisch, A. Androgen receptors in prostate 41. Bentel, J. M., Birrell, S. N., Pickering, M. A., Holds, D. J., Horsfall, D. J., and Tilley, cancer. Endocr. Relat. Cancer, 9: 155–170, 2002. W. D. Androgen receptor agonist activity of the synthetic progestin, medroxyproges- 15. Hobisch, A., Eder, I. E., Putz, T., Horninger, W., Bartsch, G., Klocker, H., and Culig, terone acetate, in human breast cancer cells. Mol. Cell. Endocrinol., 154: 11–20, Z. Interleukin-6 regulates prostate-specific protein expression in prostate carcinoma 1999. cells by activation of the androgen receptor. Cancer Res., 58: 4640–4645, 1998. 42. Yamamoto, A., Hashimoto, Y., Kohri, K., Ogata, E., Kato, S., Ikeda, K., and 16. Culig, Z., Hobisch, A., Cronauer, M. V., Radmayr, C., Trapman, J., Hittmair, A., Nakanishi, M. Cyclin E as a coactivator of the androgen receptor. J. Cell Biol., 150: Bartsch, G., and Klocker, H. Androgen receptor activation in prostatic tumor cell lines 873–880, 2000. by insulin-like growth factor I, keratinocyte growth factor, and epidermal growth 43. Claessens, F., Verrijdt, G., Schoenmakers, E., Haelens, A., Peeters, B., Verhoeven, factor. Cancer Res., 54: 5474–5478, 1994. G., and Rombauts, W. Selective DNA binding by the androgen receptor as a 17. Balk, S. P. Androgen receptor as a target in androgen-independent prostate cancer. mechanism for hormone-specific gene regulation. J. Steroid Biochem. Mol. Biol., 76: Urology, 60: 132–138; discussion 138–139, 2002. 23–30, 2001. 18. Knudsen, K. E., Cavenee, W. K., and Arden, K. C. D-Type cyclins complex with the 44. Kotaja, N., Aittomaki, S., Silvennoinen, O., Palvimo, J. J., and Janne, O. A. ARIP3 androgen receptor and inhibit its transcriptional transactivation ability. Cancer Res., (androgen receptor-interacting protein 3) and other PIAS (protein inhibitor of acti- 59: 2297–2301, 1999. vated STAT) proteins differ in their ability to modulate steroid receptor-dependent 19. Petre, C. E., Wetherill, Y. B., Danielsen, M., and Knudsen, K. E. Cyclin D1: transcriptional activation. Mol. Endocrinol., 14: 1986–2000, 2000. mechanism and consequence of androgen receptor co-repressor activity. J. Biol. 45. Gioeli, D., Ficarro, S. B., Kwiek, J. J., Aaronson, D., Hancock, M., Catling, A. D., Chem., 277: 2207–2215, 2002. White, F. M., Christian, R. E., Settlage, R. E., Shabanowitz, J., Hunt, D. F., and 4912

44 SPECIFICITY OF CYCLIN D1 ACTION

Weber, M. J. Androgen receptor phosphorylation. Regulation and identification of the 56. Ueda, T., Bruchovsky, N., and Sadar, M. D. Activation of the androgen receptor phosphorylation sites. J. Biol. Chem., 277: 29304–29314, 2002. N-terminal domain by interleukin-6 via MAPK and STAT3 signal transduction 46. Giovannucci, E., Stampfer, M. J., Krithivas, K., Brown, M., Dahl, D., Brufsky, A., pathways. J. Biol. Chem., 277: 7076–7085, 2002. Talcott, J., Hennekens, C. H., and Kantoff, P. W. The CAG repeat within the 57. Giri, D., Ozen, M., and Ittmann, M. Interleukin-6 is an autocrine growth factor in androgen receptor gene and its relationship to prostate cancer. Proc. Natl. Acad. Sci. human prostate cancer. Am. J. Pathol., 159: 2159–2165, 2001. USA, 94: 3320–3323, 1997. 58. Russell, P. J., Bennett, S., and Stricker, P. Growth factor involvement in progression 47. Giovannucci, E., Platz, E. A., Stampfer, M. J., Chan, A., Krithivas, K., Kawachi, I., of prostate cancer. Clin. Chem., 44: 705–723, 1998. Willett, W. C., and Kantoff, P. W. The CAG repeat within the androgen receptor gene 59. Drachenberg, D. E., Elgamal, A. A., Rowbotham, R., Peterson, M., and Murphy, G. P. and benign prostatic hyperplasia. Urology, 53: 121–125, 1999. Circulating levels of interleukin-6 in patients with hormone refractory prostate cancer. 48. Hakimi, J. M., Schoenberg, M. P., Rondinelli, R. H., Piantadosi, S., and Barrack, E. R. Prostate, 41: 127–133, 1999. Androgen receptor variants with short glutamine or glycine repeats may identify 60. Ueda, T., Mawji, N. R., Bruchovsky, N., and Sadar, M. D. Ligand-independent unique subpopulations of men with prostate cancer. Clin. Cancer Res., 3: 1599–1608, activation of the androgen receptor by interleukin-6 and the role of steroid receptor 1997. coactivator-1 in prostate cancer cells. J. Biol. Chem., 277: 38087–38094, 2002. 49. Chamberlain, N. L., Driver, E. D., and Miesfeld, R. L. The length and location of 61. He, B., Lee, L. W., Minges, J. T., and Wilson, E. M. Dependence of selective gene CAG trinucleotide repeats in the androgen receptor N-terminal domain affect transactivation on the androgen receptor NH - and COOH-terminal interaction. J. Biol. activation function. Nucleic Acids Res., 22: 3181–3186, 1994. 2 Chem., 277: 25631–25639, 2002. 50. Beilin, J., Ball, E. M., Favaloro, J. M., and Zajac, J. D. Effect of the androgen receptor 62. Brookes, D. E., Zandvliet, D., Watt, F., Russell, P. J., and Molloy, P. L. Relative CAG repeat polymorphism on transcriptional activity: specificity in prostate and activity and specificity of promoters from prostate-expressed genes. Prostate, 35: non-prostate cell lines. J. Mol. Endocrinol., 25: 85–96, 2000. 18–26, 1998. 51. Taplin, M. E., Bubley, G. J., Ko, Y. J., Small, E. J., Upton, M., Rajeshkumar, B., and Balk, S. P. Selection for androgen receptor mutations in prostate cancers treated with 63. La Spada, A. R., Wilson, E. M., Lubahn, D. B., Harding, A. E., and Fischbeck, K. H. androgen antagonist. Cancer Res., 59: 2511–2515, 1999. Androgen receptor gene mutations in X-linked spinal and bulbar muscular atrophy. 52. Nichols, M., Rientjes, J. M., and Stewart, A. F. Different positioning of the ligand- Nature (Lond.), 352: 77–79, 1991. binding domain helix 12 and the F domain of the estrogen receptor accounts for 64. Orr, H. T., Chung, M. Y., Banfi, S., Kwiatkowski, T. J., Jr., Servadio, A., Beaudet, functional differences between agonists and antagonists. EMBO J., 17: 765–773, A. L., McCall, A. E., Duvick, L. A., Ranum, L. P., and Zoghbi, H. Y. Expansion of 1998. an unstable trinucleotide CAG repeat in spinocerebellar ataxia type 1. Nat. Genet., 4: 53. Hsiao, P. W., Thin, T. H., Lin, D. L., and Chang, C. Differential regulation of 221–226, 1993. testosterone versus 5␣-dihydrotestosterone by selective androgen response elements. 65. Buchanan, G., Greenberg, N. M., Scher, H. I., Harris, J. M., Marshall, V. R., and Mol. Cell. Biochem., 206: 169–175, 2000. Tilley, W. D. Collocation of androgen receptor gene mutations in prostate cancer. 54. Deslypere, J. P., Young, M., Wilson, J. D., and McPhaul, M. J. Testosterone and Clin. Cancer Res., 7: 1273–1281, 2001. 5␣-dihydrotestosterone interact differently with the androgen receptor to enhance 66. Matias, P. M., Donner, P., Coelho, R., Thomaz, M., Peixoto, C., Macedo, S., Otto, N., transcription of the MMTV-CAT reporter gene. Mol. Cell. Endocrinol., 88: 15–22, Joschko, S., Scholz, P., Wegg, A., Basler, S., Schafer, M., Egner, U., and Carrondo, 1992. M. A. Structural evidence for ligand specificity in the binding domain of the human 55. Smith, P. C., Hobisch, A., Lin, D. L., Culig, Z., and Keller, E. T. Interleukin-6 and androgen receptor. Implications for pathogenic gene mutations. J. Biol. Chem., 275: prostate cancer progression. Cytokine Growth Factor Rev., 12: 33–40, 2001. 26164–26171, 2000.

4913

Chapter IV:

Cyclin D1 binding to the androgen receptor NH2-terminal domain inhibits AF2 association and reveals dual roles for AR co-repression.

Burd CJ1, Petre-Draviam CE1, Moghadam H1, Wilson EM3, and Knudsen KE1,2.

1 Department of Cell Biology and 2 Center for Environmental Genetics, University of Cincinnati College of Medicine, Cincinnati, Ohio 45267-0521 3 Laboratories of Reproductive Biology, and the Department of Pediatrics and Department of Biochemistry and Biophysics, University of North Carolina, Chapel Hill, North Carolina 27599-7500

46 Abstract

The androgen receptor (AR) is a member of the nuclear receptor superfamily whose activity is

critical for the development and progression of prostate cancer. Ablation of AR activity is the

goal of prostate cancer therapy; therefore, complete knowledge of AR regulation is essential for

the design of novel, effective therapeutics. AR co-activators that are de-regulated in cancer (e.g.

TIF2) act through recruitment of histone acetyl transferase action, and/or through enhancing

required amino-carboxy terminal AR interactions. In contrast, we and others have previously

demonstrated that cyclin D1 is a potent co-repressor of the AR. Although cyclin D1 is suspected

to recruit histone deacetylases to the AR complex, suppression of histone deacetylase activity

failed to completely reverse cyclin D1 inhibition. Thus, a secondary means of repression must

exist. Here, we probed this additional mechanism of cyclin D1 action. We show that cyclin D1

predominantly binds the amino terminal domain (NTD) of the AR, dependent on the AR

23FxxLF27 motif. We show that cyclin D1 abrogates the ability of the FxxLF motif to interact with the AR carboxy terminus. Secondary NTD sites capable of fostering interaction with the C- terminus were refractory to cyclin D1 action, indicating that the ability of cyclin D1 to modulate

AR amino-carboxy terminal interactions is specific to 23FxxLF27. Although deletion of the N- terminal cyclin D1 binding site severely compromised AR activity (due to loss of FxxLF/C- terminal interaction), we show that this allele unmasks an unexpected role for cyclin D1 in regulation through the AR C-terminus. In summary, our data reveal a novel second mechanism of cyclin D1 action, through modulation of the FxxLF domain and concomitant regulation of

NTD/AF2 interactions. These observations underscore the complexity of cyclin D1 as a co- repressor and highlight the capacity of cyclin D1 to counterbalance AR co-activator function.

47 Introduction

The androgen receptor (AR) is a member of the steroid receptor subclass of the nuclear

receptor superfamily, whose activity is required for prostate development, growth, and function

(1, 2). Treatment for advanced prostate cancer relies almost entirely on elimination of AR

activity, as achieved through either direct AR antagonists or inhibition of AR ligand synthesis

(3). In the majority of cases, androgen ablation therapy results in remission that invariably

relapses. The great majority of relapsed tumors demonstrate restored AR activity, indicating that reconstitution of this pathway is essential for tumor progression (4). Thus, understanding mechanisms that regulate AR function is of critical importance.

Prior to activation by ligand, the AR resides diffusely throughout the cell and is held inactive by heat shock protein complexes (5). In the prostate, the main ligand for the AR is dihydrotestosterone (DHT), which is converted from testosterone through the action of 5alpha- reductase (6). The AR then rapidly translocates to the nucleus, where it binds as an anti-parallel homodimer to DNA at androgen response elements (AREs) found in target gene promoters (7-9).

The ligand binding domain of the AR (found in the C-terminus, amino acids 647 to 919) is

highly conserved among the other steroid receptors and contains the AF2 transcriptional

transactivation domain (10, 11). While AF2 is a potent modulator of transactivation potential and co-activator binding in related nuclear receptors, its action in the AR is relatively weak (10,

12, 13). Fusion of this region to the receptor DNA binding domain fails to demonstrate transactivation potential upon ligand binding (14). By contrast, deletion of AF2 results in a truncated protein whose activity in reporter assays is similar to that observed with the full-length receptor, suggesting that the majority of AR activity resides in the N-terminal region (15, 16).

In the N-terminal domain (NTD), the primary ligand-dependent transactivation domain, AF1, lies between amino acids 142-337, and is thought to harbor the most potent transactivation

48 function of the receptor (17, 18). A second N-terminal transactivation function, AF-5 or Tau5,

lies downstream between residues 360-528 and is not dependent on ligand binding (15, 17).

Rather, the C-terminus of the receptor is hypothesized to modulate AF-5 function (17), and AF-5

is responsive to the effects of the Rho pathway (19). While there is no current evidence that AF-

5 is reactive to co-modulatory proteins, a large body of evidence has demonstrated that AF1 and

AF2 are strongly influenced by association with co-activators and co-repressors (8, 20, 21).

The AR recruits a series of co-activator proteins to promote and enhance transcription of

target promoters, such as the p160 class of co-activators (e.g. SRC-1, TIF2/GRIP1 and

AIB1/SRC3/ACTR/RAC3) (14, 22). Selected members of this highly homologous protein

family contain some intrinsic histone acetyltransferase (HAT) activity, but can also recruit

P/CAF and CBP/p300 to the AR complex, thus increasing the local action of histone acetylation

(8, 23). Recruitment of HATs to the AR complex assists in the formation of active transcriptional complexes by acetylating histones (thereby fostering increased accessibility to the

RNA polymerase II complex), and also by facilitating recruitment of chromatin remodeling complexes (e.g. SWI/SNF) (24, 25). Association of p160 proteins with nuclear receptor AF2

domains occurs through interaction of a ligand-dependent hydrophobic cleft formed by helices 3,

4, 5, and 12 of AF2 upon ligand binding (26-29). Three highly conserved alpha-helical LxxLL motifs (NR boxes) present in the p160 proteins are capable of binding to the hydrophobic groove

(30, 31). For the AR, association of p160 co-activators with the AF2 domain is required for

measurable AF2 activity (32). Strikingly, it has also been shown that SRC1 and TIF2 can

associate with the AR NTD independent of the co-activator NR boxes, and that binding of SRC1

occurs predominantly in this region (14, 22). Deletion of the SRC1 LxxLL motifs fails to

compromise co-activator function whereas ablation of NTD binding capability eliminates SRC1

enhancement of AR function (14, 22). From these and other studies it has been hypothesized

49 that p160 co-activators can act on both AF1 and AF2 to form a ternary complex between the

NTD and AF2 hydrophobic cleft and/or that co-activator function is influenced by interaction between these two domains of the AR (33).

Association of the AR NTD with AF2 has been well-documented and strongly influences

AR function (14, 32, 33). The NTD of the AR contains at least three distinct regions proposed to generate amphipathic alpha helices that can interact with the AF2 hydrophobic groove in response to ligand binding: 23FxxLF27, 179LKDIL183 and 432WxxLF436 (14, 34). In both yeast and mammalian systems, it has been demonstrated that association of the AR NTD with AF2 can occur directly, and is required for full AR activation (13, 33, 35, 36). Association of the NTD with AF2 slows the dissociation rate of bound ligand, thus promoting stabilization of the active receptor complex (34). Mutations that disrupt NTD/AF2 interaction in cultured cells retain compromised, yet detectable AR activity. Several mutations that disrupt NTD/AF2 interaction

(but do not affect ligand binding) have also been identified in patients with androgen insensitivity syndrome, thus demonstrating the biological impact of this interaction in vivo (36).

Therefore, the AF1 transactivation domain is regulated by ligand binding, co-activator association, and interaction with the AF2 domain.

We and others have previously shown that cyclin D1 binds directly to the AR and is a potent repressor of AR function (37, 38). Although first identified based on its function to remove RB mediated cell cycle arrest through its expression at the G1 phase of the cell cycle, it has become increasingly apparent that cyclin D1 plays additional roles as a transcriptional modulatory protein (39, 40). Cyclin D1 is known to act as both a co-activator (e.g. for the estrogen receptor) and a co-repressor of multiple transcription factors (e.g. AR, STAT3, v-Myb, and the thyroid hormone receptor), independent of its role in the cell cycle (41-44). We have previously demonstrated that cyclin D1 directly binds the AR NTD and blocks AR action on

50 multiple target promoters (43, 45). In addition, it can repress the endogenous expression of

prostate specific antigen (PSA) in the LNCaP prostate cancer cell line at stochiometric AR to

cyclin D1 expression levels (45). The ability of cyclin D1 to repress AR mediated transcription does not appear to be cell type specific, as the inhibitory effect of overexpression is maintained in all cell lines tested (45). Additionally, AR activity is cell cycle regulated, as AR activity is lowest at the peak of cyclin D1 expression (G1-S transition) (46).

Although the effects of cyclin D1 on AR activity are well-characterized, the regulatory

target by which it mediates this repression is less fully understood. An AR allele that harbors

constitutive, ligand independent AF5 activity and maintains cyclin D1 binding is refractory to

cyclin D1 repression, suggesting that cyclin D1 manifests its repressor function through the

ligand dependent transactivation domains (AF1 or AF2) (43). Cyclin D1-mediated repression is

not competed by HAT co-activators (including SRC1 and P/CAF) or ARA70, a co-activator

capable of binding both the NTD and AF2 (43). However, we have previously shown that cyclin

D1 co-repressor activity could be partially reversed through the action of trichostatin A, thus

implicating recruitment of histone deacetylase (HDAC) activity as at least one potential

component of cyclin D1 action (43). Subsequently it was shown that cyclin D1 can bind to

HDAC3, further suggesting that its action may be manifested in part through this mechanism

(42). Here, we identify a second mechanism of cyclin D1 action, mediated through modulation

of NTD/AF2 interactions. We show that cyclin D1 binds the first N-terminal alpha helix and

strongly inhibits ligand-induced association of the AR NTD with AF2. Cyclin D1 binding to the

amino terminus requires an intact 23FxxLF27 motif, and functional studies reveal that cyclin D1 impinges specifically on the association of this NTD motif with the hydrophobic cleft. In addition, we uncover the ability of cyclin D1 to bind to the LBD via the cyclin D1 LxxLL motif only in the absence of the AR FxxLF. The consequence of this binding results in further AF1

51 repression, but renders alternate effects on an independent AF2 domain. These studies

underscore the complexity of cyclin D1 action as a co-repressor and demonstrate that cyclin D1 utilizes at least two distinct mechanisms to repress AR activity.

Materials and Methods

Cell Culture and treatment- CV1 and LNCaP cells were obtained from ATCC and cultured as previously described (43) . Charcoal dextran treated serum (CDT) was obtained from Atlanta

Biologicals.

Plasmids- The plasmids ARR2-LUC, pRC/CMV-cyclin D1, pRC/CMV-cyclin D1-LALA,

CMV-β--galactosidase, pCR3.1 hSRC-1A, pGEX 3X-cyclin D1, pCMV-hAR-LFAA, pCMV- hAR, PGEX-KG, and WTAR-pGEM have been previously described (34, 43, 45). The Gal4-

LUC reporter, Gal4-ARLBD, Gal4-ΑΡ∆628−646LBD, and VP16-AR-TAD constructs were gifts of Dr. E. Yong (47, 48). pCRAR1-238 and pCR1-238LFAA were generated by PCR amplification of AR amino acids 1-238 using pCMV-hAR and pCMV-hAR-LFAA as templates, respectively. The following primer pairs were used for amplification:

5’GAGCAAGAGAAGGGGAGCC3’ (sense) and 5’TCACCACTCCTTGGCGTTGTC3’

(antisense). The resulting fragment was inserted into pCR2.1 as directed by the manufacturer

(Invitrogen). pCRAR 34-238 was cloned similarly by PCR from pCMV-hAR template using the primer 5’CGCGAAGTGATGCAGAACCCG3’ in exchange of the sense strand primer. All constructs generated were verified by sequencing and shown to be free of error. To generate the full length allele, pcDNA-AR∆2−34 was cloned by excising the XmaI/BamHI fragment of pAR0 and inserting it into the XmaI/BamHI sites of pCR-AR34-238. The fragment the full-length alleles was subsequently removed from the pCR vector and inserted into pcDNA3.1(-)

52 (Invitrogen) using a NotI /BamHI digest. VP16-AR∆2−34TAD was generated by cleaving

pcDNA-AR∆2−34 with XhoI and HindIII and inserting these fragments into the SalI and HindIII sites of pVP-16 (Clontech).

Immunoprecipitation- LNCaP cells were cultured in IMEM containing 5% charcoal dextran treated serum for 48 h. Cells were then switched IMEM containing 5% heat inactivated fetal bovine serum for 16h, at which time cells were harvested via trypsinisation. Cell pellets were lysed and sonicated in NETN (20mM Tris-Cl, pH 8.0, 100mM NaCl, 1mM EDTA, pH 8.0, 0.5%

NP-40) containing proteinase inhibitors (1mM PMSF, 10µM benzamidine-HCl, 50 µM 1,10 phenanthroline-HCl, 15µM aprotinin, 20µM leupeptin, 15µM pepstatin) followed by centrifugation. Supernatants were immunoprecipitated with 1ug of antibody directed against AR

(Santa Cruz, N-20), cyclin D1 (Santa Cruz, HC-295), or DBF4 (Santa Cruz, H-300) and pulled

down with protein A-sepherose beads (Amersham Biosciences, CL-4B). Bead complexes were

washed 4 times with NETN. Samples were loaded on 12 % SDS-PAGE, transferred to a PVDF

membrane, and immunoblotted with the designated antibodies.

In Vitro Binding Assay- AR and CD44 constructs were in vitro transcribed/translated using the

Promega T7 rabbit reticulate lysate kit as per manufacturer’s instructions in the presence of [35S] methionine (Easy Tag Express [35S] Protein Labelling Mix, Perkin Elmer). Recombinant protein

was then incubated with GST alone or GST-cyclin D1 immobilized on glutathione-agarose beads

(Sigma) in NETN plus protease inhibitors for 3 hours at 4°C with rotation. Preparation of GST

and GST-cyclin D1 was performed as previously described (43). Total volume for each reaction

was 500 uL. Following incubation, samples were washed 5 times with 750 uL NETN. Samples

(bound and 5% input) were then denatured in SDS-PAGE loading buffer and run on a 12% SDS-

polyacrylamide gel. Signal was enhanced by incubation with Fluoro-hance (Research Products

53 International) as indicated by the manufacturer and results were visualized and quantified on a

Storm 860 phosphoimager (Molecular Dynamics).

Mammalian Two Hybrid and Reporter Assays- For all reporter assays, 6 well dishes of CV1 cells

were cultured for 24 hours in phenol-red free DMEM supplemented with 10% CDT. Cells were

then transfected using the BES/Calcium phosphate method with a total of 4ug DNA per well

(43). For mammalian two hybrid assays, 0.5 ug of Gal4-LUC reporter, VP16-ARTAD, and

Gal4-ARLBD were used in the presence of 1.5 ug of either empty vector or co-regulator (cyclin

D1, p53, TIF2). In reporter assays, 1.0 ug of ARR2-LUC, 0.5 ug of AR, and 1.25 ug of either

cyclin D1 or empty vector were used. All transfections also contained 0.25 ug of β-galactosidase

reporter and empty vector to achieve 4 ug of total DNA. Post-transfection, cells were washed

with PBS, media was replaced, and cells were treated with indicated concentration of DHT or

ethanol vehicle (not to exceed 0.1%) for 24 h. Following stimulation, cells were harvested by

trypsinisation and monitored for luciferase and β--galactosidase activity using the Promega

Luciferase kit and Tropix Galactar-Star systems, respectively, as per manufacturer’s instructions.

Transfections were performed at least twice in triplicate. Averages and standard deviations are

shown.

Results

Cyclin D1 inhibits AR transcriptional activity- We and others have previously shown that cyclin

D1 overexpression represses AR transcriptional activity on multiple androgen responsive

promoters (37, 38). The dose dependent effect of cyclin D1 on the AR was verified using

ARR2-LUC reporter, which contains the probasin promoter repeated in tandem. CV1 cells were

utilized, as these spontaneously immortalized epithelial cells lack endogenous AR (yet support

ectopic AR activity) and are unaffected by culture in steroid free conditions (charcoal dextran

54 treated serum; CDT) or in the presence of androgen (data not shown). CV1 cells were

transfected with AR, ARR2-LUC, β−galactosidase, and either cyclin D1 or empty vector in the

absence of androgens. Post transfection, cells were treated with 0.1 nM DHT or ethanol vehicle

for 24h. Luciferase activity was corrected for transfection efficiency against β-galactosidase

activity and AR activity in the presence of ligand set to ‘100’. Consistent with previous reports

(37, 38), cyclin D1-mediated repression is dose dependent (Fig. 1A).

Cyclin D1 interacts with the AR in the prostatic adenocarcinoma cells- Interactions between

cyclin D1 and the androgen receptor have been previously shown both in vitro and in vivo (38,

43). Moreover, interaction of endogenous proteins has been shown in liver extracts (37, 38). To examine interaction in prostatic cells, we utilized the AR-positive LNCaP cell line. Alhough AR positive prostatic adenocarcinoma cells express low levels of cyclin D1 (49), we have previously shown that androgen stimulation induces cyclin D1 enrichment (37, 38). Therefore, LNCaP cells were initially cultured in media devoid of androgens to induce cell cycle arrest. Cells were then stimulated with media containing 5% heat inactivated fetal bovine serum for 16h, at which time cells were harvested and subjected to immunoprecipitation as indicated. As shown in Figure 1B,

AR effectively co-immunoprecipitates with cyclin D1 (lane 2). It should be noted that AR

positive prostatic adenocarcinoma cell lines express low levels of cyclin D1 (49). Therefore, it is

not surprising that only a small fraction of the AR pool is associated with cyclin D1.

Cyclin D1 binds the AR N-terminal alpha helix - We have previously demonstrated that cyclin

D1 binds preferentially to the first 502 amino acids of AR (43). In addition, cyclin D1 action is

directed at only the ligand dependent transactivation functions within this region, as we have

previously demonstrated that AF5 (ligand independent) activity is refractory to cyclin D1 (43).

Therefore, we sought to delineate the effector point of cyclin D1 action on the NTD. To do so,

55 amino terminal AR truncations (AR1-238 and AR∆46-408) were generated for in vitro binding

assays (Fig. 2A). Both wild-type and mutant AR constructs were in vitro translated/transcribed

in the presence of [35S]-methionine and incubated with GST-cyclin D1 immobilized on

glutathione-agarose beads. To control for non-specific binding, all AR alleles were also

incubated with GST alone immobilized on glutathione-agarose. Input (5% of reaction) and

bound fractions were resolved by SDS-PAGE and visualized on a phosphoimager (Fig. 2B).

Inputs confirmed that these constructs generate stable proteins (lanes 1-3), although the AR1-238 construct also results in a truncated protein (determined to be a degradation production through immunoprecipitation studies, data not shown). As can be observed in the right panel, wild-type

AR bound to GST-cyclin D1 above the GST control in both the presence and absence of ligand

(compare lanes 4-6), consistent with previous reports (43). Both AR∆46-408 and AR1-238

retained specific binding to GST-cyclin D1 above the GST control (Fig. 2B, compare lanes 7,8 and 9,10). These results implicate the first 46 amino acids of AR as a major binding site for cyclin D1.

Subsequently, sub-deletions of the 1-46 amino acid region were generated, both in the context of the N-terminus (AR 34-238) and full-length receptor (AR∆2−34) (Figure 3A). These constructs were engineered to include an initiating methionine and encode predicted proteins detectable after in vitro transcription/translation in the presence of [35S]-methionine (Fig. 3B,

lanes 1,2,7,8). The radiolabeled recombinant proteins were incubated with GST-cyclin D1 or

GST alone, and binding analyses performed as described above (Fig. 2B). Recombinant CD44,

labeled via in vitro transcription/translation, was also included as a negative control, and failed to

bind GST-cyclin D1 (data not shown) consistent with previous reports (43). As expected, AR 1-

238 demonstrated marked binding capacity (Fig. 3B, compare lanes 3,4). In contrast, deletion of

the first 34 amino acids completely abrogated binding (compare lanes 1, 4 with lanes 2, 6).

56 Interestingly, in the context of the full length receptor, the allele with a deletion of the first 34

amino acids (AR∆2−34) retained some minimal binding (Fig. 3B, lane 12). These data suggest

that an alternate binding site lies outside the first 238 amino acid region. However, binding of

this truncation to GST-cyclin D1 was significantly lower (51%) than wtAR (compare lanes 7,10

with lanes 8,12). These data indicate that a primary binding site of cyclin D1 to the AR requires the extreme N-terminus of the nuclear receptor, which is predicted to encode a long alpha helix that regulates interaction with the AR C-terminus.

It has been shown that the first 34 amino acids of the AR are critical for its transcriptional transactivation function (22, 34). To determine the impact of deletion of the amino terminal alpha helix, reporter assays were performed using the ARR2-LUC reporter (Figure 3C). Briefly,

CV1 cells cultured in CDT for 24 hours were co-transfected with AR constructs (wild-type or

AR∆2−34), the ARR2-LUC reporter, and CMV-β-galactosidase (as an internal control for transfection efficiency). Post-transfection, cells were stimulated with either 1 nM dihydrotestosterone (DHT) or ethanol vehicle for 24 hours, at which time cells were harvested, lysed and analyzed for luciferase and β-galactosidase activity. This concentration of ligand was essential to obtain measurable activity of the truncated AR allele. For comparison, basal activity

(wtAR treated with vehicle) was set to “1” and relative luciferase activity is shown. As expected, wtAR demonstrates a high level of transactivation potential (56.3 fold induction over basal activity), whereas AR∆2−34 is significantly compromised (23.2 percent of wtAR activity), similar to that observed with wtAR and cyclin D1 (Fig. 1).

Cyclin D1 inhibits NTD/AF2 interactions- Encompassed within the first 34 amino acids is the

23FxxLF27 motif, which binds with high affinity to the AF2 domain; this interaction has been shown to stabilize the receptor/ligand interaction and is required for full AR activity (34). Two

57 lower affinity NTD motifs capable of binding AF2 reside outside the cyclin D1 binding region,

at residues 179-183 and 432-436 (14, 34). The ability of the NTD to bind AF2 in the presence of

ligand can be monitored using a well-defined mammalian two hybrid assay (47), as depicted in

Figure 4A. This system is capable of specifically monitoring the interaction of the amino and

carboxy termini of the AR (Fig. 4B). Briefly, CV1 cells were transfected under steroid depleted

conditions with expression constructs encoding the Gal4-LUC reporter, Gal4-ARLBD (the

carboxyl terminus of AR, amino acids 614-919, fused to the Gal4 DNA binding domain) and/or

VP16-ARTAD (the transcriptional activation domain of VP16 fused with AR amino acids 1-

565). CMV-β-galactosidase was included in all transfections as an internal control for transfection efficiency. Post-transfection, cells were stimulated with 0.1 nM DHT or ethanol vehicle for 24h prior to harvest and analyzed for luciferase activity. VP16-ARTAD activity induced by interaction with Gal4-ARLBD in the presence of ligand was set to “100”. Either construct alone in the presence of ligand or together in the absence of ligand failed to induce luciferase activity (Figure 4B). As expected, transfection of both constructs in the presence of ligand fostered NTD/AF2 interaction, thus stimulating a 99-fold activity from the Gal4-LUC reporter. In addition we examined mammalian two hybrid activity after additional cotransfection of expression plasmid encoding TIF2 (known to bind both NTD and AF2), p53 (a known repressor of the NTD/AF2 interaction), cyclin D1, or empty vector control (Figure 4C)

(14, 50, 51). As expected, p53 significantly reduced NTD/AF2 interactions (95.6% repression)

(51), whereas overexpression of TIF2 increased Gal4 promoter activity (14, 50). Cyclin D1 significantly inhibited NTD/AF2 interactions, reducing activity to 41% of vector transfected cells in the presence of ligand. These data suggest that cyclin D1 binding to the amino terminus reduces NH2/COOH interactions required for full transcriptional AR activity. However, we have previously shown that the repressing effects of cyclin D1 in reporter assays can be partially

58 rescued by inhibiting histone deacetylases (HDAC) with an optimum concentration of 50 nM trichostatin A (TSA) (43). To ensure that the inhibitory effect seen on the Gal4 reporter by cyclin D1 in this system is not due to HDAC recruitment and consequential reduced luciferase gene expression, the experiment shown in Figure 4C was repeated in the presence of TSA

(Figure 4D). As shown, inhibition of HDAC had no significant effect on the ability of cyclin D1 to inhibit the mammalian two hybrid system (37% of vector control compared to 41% in Figure

4C). Thus, cyclin D1 is an effective inhibitor of NTD/AF2 association.

Cyclin D1 requires the FxxLF motif for binding and modulation of NTD/AF2 interaction- Since cyclin D1 binds the first 34 amino acids of the AR and is an effector of NTD/AF2 interactions, we hypothesized that cyclin D1 may exert its inhibitor action through the 23FxxLF27 motif.

Previously it has been shown that mutation of leucine-26 and phenylalanine-27 to alanine abolishes AF2 interaction (34). To determine the impact of this mutation on cyclin D1 binding and function, an allele of AR was generated wherein the leu-26 and phe-27 residues at positions

26 and 27 were mutated to alanine in the context of the NTD (AR1-238LFAA). After generation

and labeling of AR 1-238 or AR 1-238LFAA with [35S] methionine via in vitro transcription/translation (Figure 5, left panel), binding assays were performed as described in

Figure 2B with either GST alone or GST-cyclin D1. Neither AR 1-238 nor AR 1-238LFAA bound to GST alone (lanes 3,5). Interaction of the AR constructs with GST-cyclin D1 was quantitated and AR 1-238 binding set to “100”. As shown, AR 1-238LFAA had a significantly reduced ability to bind GST-cyclin D1 compared to AR 1-238 (34% of AR 1-238) (Fig. 5, right panel). Combined these data suggest that cyclin D1 binds to the first 34 amino acids of AR and that a functional FxxLF motif within this region is required for full association. Through this interaction cyclin D1 modulates NTD/AF2 interaction.

59 To test this hypothesis, VP16-ARTAD was mutated to delete the first 34 amino acids

(resulting in VP16-AR∆2−34TAD) (Figure 6A). Constructs were examined in the mammalian two hybrid assay as described in Figure 4. Although compromised for NTD/AF2 interaction, the

VP16-AR∆2−34TAD proteins retained some association with the Gal4-LBD (approximately

20% activation of intact VP16-TAD, data not shown). This retention of interaction has precedent, since two additional NTD motifs exist which are capable of binding to AF2:

433WxxLF437 and 179LKDIL183. For accurate comparison of cyclin D1 action, activity of the

VP16-AR∆2−34TAD and Gal4-ARLBD interaction was set to “100” in the presence of ligand

(Fig. 6A). Interaction of VP16-AR∆2−34TAD with Gal4-ARLBD was refractory to inhibition by cyclin D1 (right panel); in these experiments, cyclin D1 actually increased NTD/AF2 interactions. The underlying cause of this unexpected result is addressed in Figure 7.

A second binding site for cyclin D1 has been suggested between amino acids 633 and

666 of the hinge region within AR, although in our previous experiments NTD binding is the primary site for D1 association (38). Interestingly, an NTD/AF2 inhibitory region within the

hinge region has been identified (48). We therefore utilized a previously described deletion of

this inhibitory region (aa 628-646) fused to the Gal4 DNA (47) binding domain in the

mammalian two hybrid assay to measure any affect cyclin D1 hinge binding may have on

NTD/AF2 interactions (Fig. 6B). As shown, expression of cyclin D1 reduced luciferase activity

of Gal4-AR∆628-646LBD (right panel) comparable that observed with the full length Gal4-

ARLBD (44% compared to 41%) (Fig. 4B). Together, these data indicate that the cyclin D1

utilizes the high affinity NTD binding site within the first 34 amino acids to regulate FxxLF/AF2

interaction in the presence of ligand.

Cyclin D1 activates AR AF2 function in the absence of the FxxLF motif. To elucidate whether

the enhanced activity in the mammalian two hybrid system in figure 6A is the result of AF2

60 activation by cyclin D1 similar to that seen in ER (41), the assay was repeated in the absence of

the VP16-NTAD construct. In this assay, the ability of the AR AF2 transactivation domain was

monitored via the Gal4-LUC reporter. The GAL4-ARLBD expression plasmid was transfected

into CV-1 cells as performed previously with cyclin D1, cyclin D1-LALA (containing two

leucine to alanine mutations within the cyclin D1 254LxxLL258 motif), or vector control.

Consistent with the literature, ligand was unable to initiate a significant increase of transcription above basal AF2 activity (Fig 7A, vector), which was set to ‘1’. Strikingly, cyclin D1 initiated an induction of AF2 activity (Fig. 7A). However, mutation of the LxxLL (LALA) ablated the enhancement of AF2 activity. It should be noted that we have previously shown that the LALA allele of cyclin D1 represses wtAR activity with equal effectiveness of the wild type cyclin D1 allele, suggesting that this effect is only applicable in the absence of a functional FxxLF motif within AR (43) and/or that the activation of AF2 is insignificant relative to the loss of AF1 activity.

Although we have shown that loss of residues 2-34 ablates cyclin D1 binding to the AR

NTD, some minimal binding was retained in the context of full length receptor (Fig. 3B).

Therefore, the effect of cyclin D1 on the AR∆2-34 allele was examined on the ARR2-Luc reporter. This allele possesses only a fraction of the activity of the wild type allele (Fig. 3C). As shown in Fig. 7B, wild type cyclin D1 maintains some capability of repression on the AR∆2-34 allele which is alleviated by mutation of the cyclin D1 LxxLL motif, suggesting that recruitment to the hydrophobic cleft within the carboxy terminus may further repress overall activity

although enhancing AF2 function. Consistent with this idea, repression was alleviated through

mutation of the LxxLL motif.

61 Discussion

We have previously demonstrated that cyclin D1 binds the AR NTD and is a potent

repressor of ligand-dependent AR activity (43). Our previous studies demonstrated that cyclin

D1 likely employs HDACs as one component of its repressor action, but that additional

mechanisms must exist (43). Here, we identify a second mechanism, and demonstrate that cyclin

D1 modulates association of the NTD domain with the AR AF2 region. We demonstrate that

while multiple cyclin D1 binding sites are present within the AR, a predominant binding site for

cyclin D1 resides in the most amino terminal alpha helix (residues 1-34) (Figures 2, 3). Using a

well-defined mammalian two-hybrid assay we show that binding of cyclin D1 to this region

results in a dramatic reduction in association between the NTD and AF2 (Figure 4). This

inhibition was not reversed by addition of TSA, eliminating the possibility that cyclin D1

recruited HDACs to the Gal4-luciferase reporter. We demonstrate that the AF2 contact site

within this region (23FxxLF27) is required for full cyclin D1 binding, as mutation to a nonfunctional motif (23FxxAA27) significantly reduces cyclin D1 association with the NTD (Figure

5). Alternate motifs capable of association with AF2 (i.e. 179LKDIL183 and 432WHTLF436) were not inhibited by cyclin D1 function, indicating that cyclin D1 exclusively utilizes the FxxLF motif to regulate NTD/AF2 interaction (Figure 6). Furthermore, loss of the NTD binding site allows for recruitment of cyclin D1 to the hydrophobic cleft within the LBD via the cyclin D1

LxxLL motif. Recruitment to AF2 confers differing co-regulatory properties to the respective activation functions of AR. While overall activity is repressed, specific AF2 function is enhanced. Both of these functions of cyclin D1 are dependent upon the LxxLL motif within cyclin D1 and are consequential only in the absence of the AR FxxLF motif. Thus, these studies have identified a second mechanism of cyclin D1 co-repressor activity, through modulation of the FxxLF motif and NTD/AF2 interactions.

62 Cyclin D1 high affinity binding site lies in the amino terminal alpha helix and requires the intact

23FxxLF27 motif. We had previously narrowed the predominant site of cyclin D1 binding within

amino acids 1-502 of the AR, which encompassed both the AF1 and AF-5 transactivation

domains. Functional studies revealed that AF-5 function is resistant to cyclin D1 action,

indicating that cyclin D1 repressor function is restricted to the ligand dependent transactivation

domains (43). Here, deletion mapping revealed that the binding site in the NTD lies within

amino acids 1-34. This region of the AR is highly conserved throughout evolution, and is

predicted to encode a long alpha helix containing two putative protein interaction motifs

(23FxxLF27 and 30VxxVI34) (34, 52). This stretch of 34 amino acids is highly influential in the

regulation of both co-activator and AF2 binding, and is therefore pivotal for full AR function

(22, 53). Mutation of 23FxxLF27 to 23FxxAA27 resulted in significantly decreased cyclin D1 binding to the truncated NTD (amino acids 1-238), suggesting that the structure of this motif is essential for cyclin D1 association. However, the alpha helix mutant AR∆2-34 did retain residual cyclin D1 binding using in vitro assays (Figure 3). Thus, an alternate binding site likely exists outside amino acids 1-238. In functional studies, cyclin D1 action on AR alleles that lack the NTD binding site (and FxxLF) requires an intact LxxLL motif within the cyclin D1, suggesting that secondary binding may be to the hydrophobic cleft vacated by the loss of FxxLF.

In addition, a binding site for AR has been suggested in the hinge region of AR (residues 633-

668) (38), although this was not observed in our previous interaction studies (43).

Cyclin D1 binding regulates FxxLF/AF2 interactions. We demonstrate that cyclin D1 inhibits interaction of the NTD with the AF2 hydrophobic cleft, as mediated specifically through the

23FxxLF27 motif (Figures 4-6). Although three NTD interaction motifs have been described for

association with AF2 (23FQNLF27, 179LKDIL183 and 432WHTLF436), 23FxxLF27 is considered to be the predominant interaction site and has the highest affinity for AF2 (14, 34). WxxLF is less

63 highly conserved among species, and is not required in the context of the full-length receptor for

NTD/AF2 binding. The function of 179LKDIL183 has not been extensively investigated, although mutation of this site impairs NTD/AF2 interaction and reduces overall AR transactivation potential (14). We demonstrated that deletion of the FxxLF motif reduced but did not ablate

NTD/AF2 interaction, consistent with the ability of these lower affinity sites to mediate AF2 interactions. Interestingly, interaction of these weaker motifs with AF2 was not inhibited by cyclin D1 (Fig.6A).

Consequence of cyclin D1 recruitment to the AR LBD. The ability of cyclin D1 to enhance activity of AR carboxy terminal constructs raises interesting questions as to the functional significance of each of these interactions. Cyclin D1 is still capable of conferring an overall inhibitory affect on AR∆2-34 activity through carboxy terminal binding dependent upon its

LxxLL motif (Fig. 7B). However, in the absence of the AR C-terminus, LBD recruitment of cyclin D1 activates AF2 function similar to that seen in the ER, presumably through SRC1 and/or P/CAF recruitment to this transactivation domain (Fig. 7A) (41, 54). This suggests that recruitment to the LBD, which promotes AF2 activation, still results in repression of AF1 activity. The significance of this finding as it relates to an FxxLF intact AR is inconclusive. We have previously shown that the LALA allele of cyclin D1 retains equivalent inhibitory properties as wild type cyclin D1 (43). In addition, we have generated an allele of cyclin D1 which fails to bind AR, inhibit NTD/AF2 interaction, and inhibit AR transcriptional activity (unpublished data). This allele retains its LxxLL motif, yet possesses no inhibitory or activation ability with regard to the AR, suggesting that recruitment to the hydrophobic cleft of AR may only be relevant in the absence of the FxxLF motif. Although, it is possible that activation of AF2 by this allele is insignificant in the context of full AR activity.

64 In summary, these data demonstrate that cyclin D1 binds to the NTD of the AR, and that

binding to this site abrogates NTD/AF2 interaction and confers transcriptional repression to the

full length receptor independent of its LxxLL motif. In the absence of the NTD binding site, a

secondary binding site of cyclin D1 exists, at which cyclin D1 confers transcriptional repression,

but dependent on LxxLL. The ability of cyclin D1 to activate AF2 occurs only in the complete

absence of the NTD, and thus holds little biological relevance

Consequence of cyclin D1-regulated NTD/AF2 interaction. The importance of FxxLF

interactions with AF2 in the context of AR activity and prostatic epithelial proliferation are becoming increasingly apparent. It has been recently demonstrated that overexpressing peptide representing the first 34 amino acids of AR can inhibit androgen dependent prostate growth, presumably through competition for the hydrophobic cleft and inhibition of NTD/AF2

interactions (55). In addition, abrogation of this association can result in androgen insensitivity

syndrome, thus underscoring the importance of this interaction for overall AR function in vivo

(36, 56). With regard to mechanism, it is clear that FxxLF/AF2 interaction slows ligand

dissociation and therefore stabilizes the active AR complex (34). The AF2 hydrophobic cleft

harbors a five-fold higher affinity for FxxLF than p160 NR boxes, and it has been shown that

FxxLF can exclude co-activator binding to AF2 (53). This event precludes AF2 activity (for

which co-activators are essential) and thus maintains AF1 as the predominant transactivation

domain. If true, cyclin D1 action on the FxxLF motif would act predominantly through

abrogation of a stable AR/ligand complex and thus reduce overall AF1 function.

In this same model, p160 overexpression (such as has been observed in advanced prostate

cancer) effectively competes for the AF2 binding site and induces AF2 activity (57, 58).

Although, both the FxxLF and WxxLF motif act to repress p160 recruitment to AF2 through

occupancy of the binding pocket, the overall action on AR activity of these two motifs is quite

65 different. While the FxxLF motif confers activity to AF1 and modulates strong AR transactivation potential, the WxxLF motif lacks the ability to activate strong transactivation potential (34). In fact, this motif acts mainly to repress AF2 activity (57). It is surprising that cyclin D1, through activation of AF2, acts to reverse the negative regulatory domain of WxxLF.

However, the activation of AF2 is most likely inconsequential when compared to the loss of AF1 activity.

Alternate models for p160 function (although not mutually exclusive) suggest that

NTD/AF2 interaction influences the sequence of co-activator binding, provides a novel platform for co-activator association, and/or is facilitated by the ability of selected p160 co-activators to simultaneously bind the NTD and LBD. These hypotheses are generally based on the observation that p160s can bind the AR NTD and LBD, and that overexpression of TIF2 or

SRC1A increased activity in the mammalian two-hybrid assay (14, 22, 33). Under these models, the ability of p160 co-activators to promote AR activity through two distinct mechanisms

(promotion of NTD/AF2 association and acetylation of histones) would be effectively foiled by the dual repressor actions of cyclin D1 (abrogation of NTD/AF2 association and recruitment of

HDACs). The potency of these dual mechanisms likely underscores our previous observations that cyclin D1 repressor function cannot be rescued by ectopic expression of co-activators

(SRC1, P/CAF, CBP or ARA70) (43), and that cyclin D1 is an effective repressor of AR function at stoichiometric levels with the receptor (45). Dual mechanisms have also been identified for the Dax1 orphan receptor and SMRT, which both act as AR co-repressors. These repressors have been recently reported to block AR NTD/AF2 interaction, and can also recruit other co-repressors (HDACs, NCoR and Alien) (59-61).

In summary, we reveal a second, novel function of cyclin D1 co-repressor action in

modulating FxxLF/AF2 interaction in the AR. This function of cyclin D1 is manifested through

66 its ability to bind the first NTD alpha helix and acts largely on the AF1 ligand dependent transactivation function. These studies culminate in the hypothesis that cyclin D1 utilizes at least two distinct mechanisms to exert its repressor function, thus negating the opposing dual mechanisms utilized by co-activators to hyperstimulate the AR. It is our belief that in-depth knowledge of receptor-specific co-repressor function will lead to the design of novel therapeutic strategies for AR-dependent cancer.

Acknowledgements

We thank Erin Williams, Kevin Link, Dr. Erik Knudsen, Dr. E.L. Yong, and Dr. Sohaib Khan for technical assistance and critical reading of this manuscript. Plasmids were kindly provided by Drs. J. Wang, A. Brinkman, R. Bernards, M. Danielson, M. Roussel, K. Fukasawa, E.L.

Yong, B. O’Malley, and R. Brackenbury. This work was supported by National Institute of

Health grant R01-CA-099996 to K.K. C.B. and C.P-D. are supported through University Of

Cincinnati Distinguished Graduate Assistantships and the Albert Ryan Foundation. C.B. is also supported by N.I.H. training grant # HD07200-15.

References

1. Trapman, J. and Brinkmann, A. O. The androgen receptor in prostate cancer. Pathol Res Pract, 192: 752-760, 1996. 2. Isaacs, J. T. Antagonistic effect of androgen on prostatic cell death. Prostate, 5: 545-557, 1984. 3. Leewansangtong, S. and Soontrapa, S. Hormonal ablation therapy for metastatic prostatic carcinoma: a review. J Med Assoc Thai, 82: 192-205, 1999. 4. Feldman, B. J. and Feldman, D. The development of androgen-independent prostate cancer. Nat Rev Cancer, 1: 34-45, 2001. 5. Pratt, W. B. and Toft, D. O. Steroid receptor interactions with heat shock protein and immunophilin chaperones. Endocr Rev, 18: 306-360, 1997. 6. Chang, C. S., Kokontis, J., and Liao, S. T. Molecular cloning of human and rat complementary DNA encoding androgen receptors. Science, 240: 324-326, 1988. 7. Zhou, Z. X., Sar, M., Simental, J. A., Lane, M. V., and Wilson, E. M. A ligand-dependent bipartite nuclear targeting signal in the human androgen receptor. Requirement for the DNA-binding domain and modulation by NH2-terminal and carboxyl-terminal sequences. J Biol Chem, 269: 13115-13123, 1994. 8. Jenster, G. Coactivators and corepressors as mediators of nuclear receptor function: an update. Mol Cell Endocrinol, 143: 1-7, 1998. 9. Langley, E., Zhou, Z. X., and Wilson, E. M. Evidence for an anti-parallel orientation of the ligand-activated human androgen receptor dimer. J Biol Chem, 270: 29983-29990, 1995. 10. Brinkmann, A. O., Blok, L. J., de Ruiter, P. E., Doesburg, P., Steketee, K., Berrevoets, C. A., and Trapman, J. Mechanisms of androgen receptor activation and function. J Steroid Biochem Mol Biol, 69: 307-313, 1999. 11. Jenster, G., van der Korput, J. A., Trapman, J., and Brinkmann, A. O. Functional domains of the human androgen receptor. J Steroid Biochem Mol Biol, 41: 671-675, 1992. 12. Hong, H., Kohli, K., Trivedi, A., Johnson, D. L., and Stallcup, M. R. GRIP1, a novel mouse protein that serves as a transcriptional coactivator in yeast for the hormone binding domains of steroid receptors. Proc Natl Acad Sci U S A, 93: 4948-4952, 1996. 13. Berrevoets, C. A., Doesburg, P., Steketee, K., Trapman, J., and Brinkmann, A. O. Functional interactions of the AF-2 activation domain core region of the human androgen receptor with the amino-terminal domain and with the transcriptional coactivator TIF2 (transcriptional intermediary factor2). Mol Endocrinol, 12: 1172-1183, 1998. 14. Alen, P., Claessens, F., Verhoeven, G., Rombauts, W., and Peeters, B. The androgen receptor amino-terminal domain plays a key role in p160 coactivator-stimulated gene transcription. Mol Cell Biol, 19: 6085-6097, 1999. 15. Jenster, G., van der Korput, H. A., van Vroonhoven, C., van der Kwast, T. H., Trapman, J., and Brinkmann, A. O. Domains of the human androgen receptor involved in steroid binding, transcriptional activation, and subcellular localization. Mol Endocrinol, 5: 1396- 1404, 1991. 16. Rundlett, S. E., Wu, X. P., and Miesfeld, R. L. Functional characterizations of the androgen receptor confirm that the molecular basis of androgen action is transcriptional regulation. Mol Endocrinol, 4: 708-714, 1990.

68 17. Jenster, G., van der Korput, H. A., Trapman, J., and Brinkmann, A. O. Identification of two transcription activation units in the N-terminal domain of the human androgen receptor. J Biol Chem, 270: 7341-7346, 1995. 18. Simental, J. A., Sar, M., Lane, M. V., French, F. S., and Wilson, E. M. Transcriptional activation and nuclear targeting signals of the human androgen receptor. J Biol Chem, 266: 510-518, 1991. 19. Metzger, E., Muller, J. M., Ferrari, S., Buettner, R., and Schule, R. A novel inducible transactivation domain in the androgen receptor: implications for PRK in prostate cancer. Embo J, 22: 270-280, 2003. 20. Collingwood, T. N., Urnov, F. D., and Wolffe, A. P. Nuclear receptors: coactivators, corepressors and chromatin remodeling in the control of transcription. J Mol Endocrinol, 23: 255-275, 1999. 21. Bevan, C. and Parker, M. The role of coactivators in steroid hormone action. Exp Cell Res, 253: 349-356, 1999. 22. Bevan, C. L., Hoare, S., Claessens, F., Heery, D. M., and Parker, M. G. The AF1 and AF2 domains of the androgen receptor interact with distinct regions of SRC1. Mol Cell Biol, 19: 8383-8392, 1999. 23. Chen, H., Lin, R. J., Schiltz, R. L., Chakravarti, D., Nash, A., Nagy, L., Privalsky, M. L., Nakatani, Y., and Evans, R. M. Nuclear receptor coactivator ACTR is a novel histone acetyltransferase and forms a multimeric activation complex with P/CAF and CBP/p300. Cell, 90: 569-580, 1997. 24. Louie, M. C., Yang, H. Q., Ma, A. H., Xu, W., Zou, J. X., Kung, H. J., and Chen, H. W. Androgen-induced recruitment of RNA polymerase II to a nuclear receptor-p160 coactivator complex. Proc Natl Acad Sci U S A, 100: 2226-2230, 2003. 25. DiRenzo, J., Shang, Y., Phelan, M., Sif, S., Myers, M., Kingston, R., and Brown, M. BRG-1 is recruited to estrogen-responsive promoters and cooperates with factors involved in histone acetylation. Mol Cell Biol, 20: 7541-7549, 2000. 26. Shibata, H., Spencer, T. E., Onate, S. A., Jenster, G., Tsai, S. Y., Tsai, M. J., and O'Malley, B. W. Role of co-activators and co-repressors in the mechanism of steroid/thyroid receptor action. Recent Prog Horm Res, 52: 141-164; discussion 164-145, 1997. 27. Glass, C. K. and Rosenfeld, M. G. The coregulator exchange in transcriptional functions of nuclear receptors. Genes Dev, 14: 121-141, 2000. 28. Bledsoe, R. K., Montana, V. G., Stanley, T. B., Delves, C. J., Apolito, C. J., McKee, D. D., Consler, T. G., Parks, D. J., Stewart, E. L., Willson, T. M., Lambert, M. H., Moore, J. T., Pearce, K. H., and Xu, H. E. Crystal structure of the glucocorticoid receptor ligand binding domain reveals a novel mode of receptor dimerization and coactivator recognition. Cell, 110: 93-105, 2002. 29. Darimont, B. D., Wagner, R. L., Apriletti, J. W., Stallcup, M. R., Kushner, P. J., Baxter, J. D., Fletterick, R. J., and Yamamoto, K. R. Structure and specificity of nuclear receptor- coactivator interactions. Genes Dev, 12: 3343-3356, 1998. 30. Voegel, J. J., Heine, M. J., Tini, M., Vivat, V., Chambon, P., and Gronemeyer, H. The coactivator TIF2 contains three nuclear receptor-binding motifs and mediates transactivation through CBP binding-dependent and -independent pathways. Embo J, 17: 507-519, 1998. 31. Heery, D. M., Kalkhoven, E., Hoare, S., and Parker, M. G. A signature motif in transcriptional co-activators mediates binding to nuclear receptors. Nature, 387: 733-736, 1997.

69 32. He, B., Kemppainen, J. A., Voegel, J. J., Gronemeyer, H., and Wilson, E. M. Activation function 2 in the human androgen receptor ligand binding domain mediates interdomain communication with the NH(2)-terminal domain. J Biol Chem, 274: 37219-37225, 1999. 33. Ikonen, T., Palvimo, J. J., and Janne, O. A. Interaction between the amino- and carboxyl- terminal regions of the rat androgen receptor modulates transcriptional activity and is influenced by nuclear receptor coactivators. J Biol Chem, 272: 29821-29828, 1997. 34. He, B., Kemppainen, J. A., and Wilson, E. M. FXXLF and WXXLF sequences mediate the NH2-terminal interaction with the ligand binding domain of the androgen receptor. J Biol Chem, 275: 22986-22994, 2000. 35. Doesburg, P., Kuil, C. W., Berrevoets, C. A., Steketee, K., Faber, P. W., Mulder, E., Brinkmann, A. O., and Trapman, J. Functional in vivo interaction between the amino- terminal, transactivation domain and the ligand binding domain of the androgen receptor. Biochemistry, 36: 1052-1064, 1997. 36. Langley, E., Kemppainen, J. A., and Wilson, E. M. Intermolecular NH2-/carboxyl- terminal interactions in androgen receptor dimerization revealed by mutations that cause androgen insensitivity. J Biol Chem, 273: 92-101, 1998. 37. Knudsen, K. E., Cavenee, W. K., and Arden, K. C. D-type cyclins complex with the androgen receptor and inhibit its transcriptional transactivation ability. Cancer Res, 59: 2297-2301, 1999. 38. Reutens, A. T., Fu, M., Wang, C., Albanese, C., McPhaul, M. J., Sun, Z., Balk, S. P., Janne, O. A., Palvimo, J. J., and Pestell, R. G. Cyclin D1 binds the androgen receptor and regulates hormone-dependent signaling in a p300/CBP-associated factor (P/CAF)- dependent manner. Mol Endocrinol, 15: 797-811, 2001. 39. Ewen, M. E. and Lamb, J. The activities of cyclin D1 that drive tumorigenesis. Trends Mol Med, 10: 158-162, 2004. 40. Coqueret, O. Linking cyclins to transcriptional control. Gene, 299: 35-55, 2002. 41. Zwijsen, R. M., Buckle, R. S., Hijmans, E. M., Loomans, C. J., and Bernards, R. Ligand- independent recruitment of steroid receptor coactivators to estrogen receptor by cyclin D1. Genes Dev, 12: 3488-3498, 1998. 42. Lin, H. M., Zhao, L., and Cheng, S. Y. Cyclin D1 Is a Ligand-independent Co-repressor for Thyroid Hormone Receptors. J Biol Chem, 277: 28733-28741, 2002. 43. Petre, C. E., Wetherill, Y. B., Danielsen, M., and Knudsen, K. E. Cyclin D1: mechanism and consequence of androgen receptor co-repressor activity. J Biol Chem, 277: 2207- 2215, 2002. 44. Bienvenu, F., Gascan, H., and Coqueret, O. Cyclin D1 represses STAT3 activation through a Cdk4-independent mechanism. J Biol Chem, 276: 16840-16847, 2001. 45. Petre-Draviam, C. E., Cook, S. L., Burd, C. J., Marshall, T. W., Wetherill, Y. B., and Knudsen, K. E. Specificity of cyclin D1 for androgen receptor regulation. Cancer Res, 63: 4903-4913, 2003. 46. Martinez, E. D. and Danielsen, M. Loss of androgen receptor transcriptional activity at the G(1)/S transition. J Biol Chem, 277: 29719-29729, 2002. 47. Lim, J., Ghadessy, F. J., Abdullah, A. A., Pinsky, L., Trifiro, M., and Yong, E. L. Human androgen receptor mutation disrupts ternary interactions between ligand, receptor domains, and the coactivator TIF2 (transcription intermediary factor 2). Mol Endocrinol, 14: 1187-1197, 2000. 48. Wang, Q., Lu, J., and Yong, E. L. Ligand- and coactivator-mediated transactivation function (AF2) of the androgen receptor ligand-binding domain is inhibited by the cognate hinge region. J Biol Chem, 276: 7493-7499, 2001.

70 49. Linja, M. J., Porkka, K. P., Kang, Z., Savinainen, K. J., Janne, O. A., Tammela, T. L., Vessella, R. L., Palvimo, J. J., and Visakorpi, T. Expression of androgen receptor coregulators in prostate cancer. Clin Cancer Res, 10: 1032-1040, 2004. 50. He, B., Lee, L. W., Minges, J. T., and Wilson, E. M. Dependence of selective gene activation on the androgen receptor NH2- and COOH-terminal interaction. J Biol Chem, 277: 25631-25639, 2002. 51. Shenk, J. L., Fisher, C. J., Chen, S. Y., Zhou, X. F., Tillman, K., and Shemshedini, L. p53 represses androgen-induced transactivation of prostate-specific antigen by disrupting hAR amino- to carboxyl-terminal interaction. J Biol Chem, 276: 38472-38479, 2001. 52. Steketee, K., Berrevoets, C. A., Dubbink, H. J., Doesburg, P., Hersmus, R., Brinkmann, A. O., and Trapman, J. Amino acids 3-13 and amino acids in and flanking the 23FxxLF27 motif modulate the interaction between the N-terminal and ligand-binding domain of the androgen receptor. Eur J Biochem, 269: 5780-5791, 2002. 53. He, B. and Wilson, E. M. Electrostatic modulation in steroid receptor recruitment of LXXLL and FXXLF motifs. Mol Cell Biol, 23: 2135-2150, 2003. 54. McMahon, C., Suthiphongchai, T., DiRenzo, J., and Ewen, M. E. P/CAF associates with cyclin D1 and potentiates its activation of the estrogen receptor. Proc Natl Acad Sci U S A, 96: 5382-5387, 1999. 55. Minamiguchi, K., Kawada, M., Ohba, S., Takamoto, K., and Ishizuka, M. Ectopic expression of the amino-terminal peptide of androgen receptor leads to androgen receptor dysfunction and inhibition of androgen receptor-mediated prostate cancer growth. Mol Cell Endocrinol, 214: 175-187, 2004. 56. Thompson, J., Saatcioglu, F., Janne, O. A., and Palvimo, J. J. Disrupted amino- and carboxyl-terminal interactions of the androgen receptor are linked to androgen insensitivity. Mol Endocrinol, 15: 923-935, 2001. 57. He, B., Bowen, N. T., Minges, J. T., and Wilson, E. M. Androgen-induced NH2- and COOH-terminal Interaction Inhibits p160 coactivator recruitment by activation function 2. J Biol Chem, 276: 42293-42301, 2001. 58. Gregory, C. W., He, B., Johnson, R. T., Ford, O. H., Mohler, J. L., French, F. S., and Wilson, E. M. A mechanism for androgen receptor-mediated prostate cancer recurrence after androgen deprivation therapy. Cancer Res, 61: 4315-4319, 2001. 59. Liao, G., Chen, L. Y., Zhang, A., Godavarthy, A., Xia, F., Ghosh, J. C., Li, H., and Chen, J. D. Regulation of androgen receptor activity by the nuclear receptor corepressor SMRT. J Biol Chem, 278: 5052-5061, 2003. 60. Agoulnik, I. U., Krause, W. C., Bingman, W. E., 3rd, Rahman, H. T., Amrikachi, M., Ayala, G. E., and Weigel, N. L. Repressors of androgen and progesterone receptor action. J Biol Chem, 278: 31136-31148, 2003. 61. Altincicek, B., Tenbaum, S. P., Dressel, U., Thormeyer, D., Renkawitz, R., and Baniahmad, A. Interaction of the corepressor Alien with DAX-1 is abrogated by mutations of DAX-1 involved in adrenal hypoplasia congenita. J Biol Chem, 275: 7662- 7667, 2000.

Figure 1. Cyclin D1 binds to and inhibits the AR in a dose dependent manner. A. CV1 cells were transfected with ARR2-LUC, CMV-β-galactosidase, AR, and either cyclin D1 or vector. Cells were treated with 0.1 nM DHT or 0.1% ethanol vehicle for 24h prior to harvest and

analysis. Luciferase activity was corrected for transfection efficiency against β-galactosidase activity and activation in the presence of ligand of vector control set to ‘100’. Data represents at least two experiments each with a minimum of three independent samples B. LNCaP cells were arrested in G0 through culture in steroid depleted serum. Cells were stimulated with heat inactivated fetal bovine serum for 16h and harvested. Cells were lysed and immunoprecipitated with antibodies specific for AR, cyclin D1, and DBF4 (non-specific control). Immunocomplexes were subjected to SDS-PAGE and immunoblotted for AR.

Figure 2. Cyclin D1 interacts with the amino terminus of the AR. A. Diagram of AR deletion constructs used in GST-cyclin-D1 binding assays. Previously described functional domains of the AR are designated and amino acid residues shown. B. AR alleles were in vitro translated with [35S] methionine and incubated with GST-cyclin D1 or GST immobilized on

glutathione agarose. Bound proteins were subjected to 12% SDS-PAGE and visualized on a

phospoimager (right panel). Inputs are also shown (left panel).

Figure 3. Cyclin D1 binds preferentially to the first 34 amino acids of the AR.

A. Diagram of AR truncations generated for binding assays containing different amino terminal

deletions of either the full length AR or AR 1-238 used in Fig. 1. B. AR truncations were in vitro

translated with [35S] methionine and incubated with GST-cyclin D1 or GST immobilized on

glutathione agarose beads. Input and bound proteins show the relevance of the first 34 amino

72 acids in the context of the amino terminus (left panel) and full length AR (right panel). C. CV1

cells were transfected with ARR2-LUC, CMV-β-galactosidase, and expression plasmids

encoding either wtAR or AR∆2−34. Cells were treated with 1 nM DHT or 0.1% ethanol vehicle for 24 h prior to harvest. Luciferase activity was corrected for transfection efficiency against β- galactosidase activity and numbers represent fold activation over basal wtAR activity (ethanol treated). Data represents at least two experiments each performed with triplicate independent samples.

Figure 4. Cyclin D1 inhibits the NTD/AF2 interaction in the mammalian two-hybrid assay.

A. Diagram illustrating mammalian two-hybrid constructs. The AR amino terminal

transactivation domain (amino acids 1-565) was fused to the VP16 activation domain and the AR

carboxy terminal ligand binding domain (amino acids 628-919) was fused to the Gal4 DNA

binding domain. B. CV1 cells were transfected with Gal4-luciferase reporter and CMV-

βgalactosidase along with GAL4-ARDBD and/or VP16-ARTAD. Cells were treated with either

0.1nM DHT or ethanol vehicle for 24h prior to harvest. Activation of both constructs in the

presence of ligand was set to “100”, and relative luciferase activities are shown. C. CV1 cells were transfected as in A with either empty vector, cyclin D1, p53, and TIF2 at a 3:1 ratio of coregulator to AR constructs. Relative luciferase activity was determined as previously described and activity in the absence of overexpressed co-regulators was set to “100”. All mammalian two hybrid data is the result of a minimum of two experiments, each performed with triplicate independent samples. D. CV1 cells were transfected as in B, except all cultures were treated with

50 nM TSA in addition to either DHT or ethanol vehicle 24 h prior to harvest. Relative luciferase activity was determined as previously described and activity in the absence of overexpressed co-regulators was set to “100”.

Figure 5. The FxxLF motif is required for full cyclin D1 binding. AR 1-238 and AR 1-

238LFAA were in vitro translated with [35S] methionine and bound to GST-cyclin D1 or GST

beads as in Figure 2. Bound proteins were visualized on a phosphoimager (middle panel) and

quantitated as a percentage of input (left panel). Relative binding of the constructs is depicted

setting AR 1-238 to “100” (right panel). Binding is the average of two independent experiments.

Figure 6. Cyclin D1 regulated NTD/AF2 interaction through the FxxLF motif. A. An

amino terminal truncation of the mammalian two hybrid construct was generated to measure the

effect of this binding region on NTD/AF2 interactions. CV1 cells were transfected as in Figure

4B with the VP16-AR∆2-34NTD to monitor the effects of cyclin D1 on FxxLF independent

NTD/AF2 interactions. Post transfection, cells were treated with DHT, and activity was set to

“100”. Cyclin D1 or vector was co-expressed at a ratio of 3:1 (cyclin D1:AR). Data represents at

least two experiments each performed with triplicate independent samples. B. AR hinge region

deletions of the Gal4-ARLBD were utilized to identify the effects of secondary cyclin D1

binding on NTD/AF2 interactions in the context of the hinge inhibitory domain. CV1 cells were transfected as in Figure 4B with 1.0 ug of Gal4-luciferase reporter and 0.5 ug of β-galactosidase,

VP-16ARTAD, and Gal4-AR∆628-646LBD in the presence or absence of 1.5 ug of cyclin D1.

Figure 7. Cyclin D1 is recruited to AF2 through its LxxLL domain to selectively activate

AF2 function. A. CV1 cells were transfected with Gal4-ARLBD, Gal4-Luciferase, CMV-

βgalactosidase, and either vector, cyclin D1, or cyclin D1-LALA. Cells were treated with 0.1

nM DHT or ethanol vehicle for 24 h prior to harvest. Data represents at least two experiments

each performed with triplicate independent samples. Activity in the absence of cyclin D1 was set

74 to “1”. Relative luciferase activity is shown. B. CV1 cells were transfected with AR∆2-34,

ARR2-LUC, CMV-βglactosidase, and either vector, cyclin D1, or cyclin D1-LALA. Cells were treated with 1 nM DHT or ethanol vehicle for 24h prior to harvest. Activity of vector transfected in the presence of ligand is set to “100”.

75 Figure 2 NTAD(1-555) DBD(556-623) LBD/AF2 A 1 AF1(142-337) Hinge(624-665) (666-919) 919 wtAR 1238 AR1-238

1 46 408 919 ∆46-408

A ∆

4 R

6 w 1 - - 4 t 2 A 0 3 B R 8 ∆46-408 AR1-238 8 wtAR kDa 201 120 100

29.5

1 2 3 4 5 6 7 8 9 10 - - + - - - - DHT INPUT + - - + - + - GST - + + - + - + GST-Cyclin D1

76 Figure 3A,B 1 238 A AR1-238

34 238 AR34-238

1 919 wtAR

34 919 AR∆2−34

A A A A R R A A R R 3 3 R R

4 1 4 w 1 w ∆ ∆ - - - - 2 2 t 2 2 2 2 t A A - - 3 3 3 3 3 3 R 8 8 8 R 8 4 4 B kDa kDa 201 100 120 100 56 38 56

29.5 38

1 2 3 4 5 6 7 8 9 10 11 12 Input + - + - GST Input + - + - GST -+ -+GST-Cyclin D1 -+ -+ GST-Cyclin D1

77 Figure 3C

C 70

60 ETOH

50 1 nM DHT

Relative Luciferase Activity 0 hAR AR∆2−34

78 Figure 4 A B 120 ETOH 100 0.1nM DHT 1 565 80 VP16 VP16-ARTAD 60

628 919 40 Gal4 Gal4-ARLBD 20

Relative Luciferase Activity 0 Gal4-ARLBD VP16-ARTAD Gal4-ARLBD +VP16-ARTAD

C 350 D 140 ETOH ETOH 300 50 nM TSA 0.1nM DHT 120 0.1nM DHT

250 100

200 80

150 60

100 40 50 20 Relative Luciferase Activity 0 Relative Luciferase Activity 0 vector cyclin D1 p53 TIF2 vector cyclin D1 p53

79 Figure 5 AR1-238LFAA AR1-238 0

80 60 40 20

120 100 Relative Binding Relative 80

AA LF 38 1-2 AR AA LF 38 1-2 AR 38 1-2 AR 4 5 6 8

3 -+ - - + - + GST D1 GST-cyclin + 1-2 AR

AA LF 38 1-2 AR INPUT 38 1-2 1 2 3 AR Figure 6 A 500 450 ETOH 400 0.1 nM DHT 34 565 VP16 VP16-AR∆2−34TAD 350 300 628 919 250 Gal4 Gal4-ARLBD 200 150 100

Relative Luciferase Activity 50 0 Vector Cyclin D1

B 120 ETOH 100 0.1 nM DHT 1 565 VP16 VP16-ARTAD 80 646 919 60 Gal4 Gal4-AR∆628-646LBD 40

Relative Luciferase Activity 0 Vector Cyclin D1

81 A B 34 Gal4 628 919 919 Gal4-ARLBD AR ∆ 2 − 34

Relative Luciferase Activitiy Relative Luciferase Activity 100 120 140 20 40 60 80 82 0 0 1 2 3 4 5 6 etrCci 1Cyclin D1-LALA Cyclin D1 Vector etrCci 1Cyclin D1-LALA Cyclin D1 Vector . MDHT 0.1 nM ETOH MDHT 1 nM ETOH Figure 7

Chapter V:

The Central Domain of Cyclin D1 Elicits Nuclear Receptor Co-repressor Activity

Christin E. Petre-Draviam1, Erin B.Williams1, Craig J. Burd1, Andrew Gladden3, Hamed Moghadam1, Jaroslaw Meller2, J. Alan Diehl3, and Karen E. Knudsen1.

1 Department of Cell Biology, 2 Department of Biomedical Engineering, Children's Hospital Research Foundation, University of Cincinnati College of Medicine, Cincinnati, Ohio 45267-0521, and the 3 Abramson Family Cancer Research Center, University of Pennsylvania Cancer Center, Philadelphia, PA 19104.

83 SUMMARY

Cyclin D1 functions in a cell cycle independent fashion to regulate the activity of

a growing number of transcription factors. Nuclear receptors represent the most

prominent group of cyclin D1-regulated transcription factors, yet the region(s) of cyclin

D1 responsible for such co-modulatory activity remain poorly defined. We previously

demonstrated that cyclin D1 is a potent inhibitor of androgen receptor (AR)

transactivation in the context of numerous cellular backgrounds and androgen responsive

promoters. Herein, we provide evidence that the region of cyclin D1 responsible for

mediating AR binding and repression is located outside the N-terminal cyclin box and

within a central, exclusively α-helical domain. Deletion of this domain results in

compromised AR binding and co-repressor activity, but does not perturb cell cycle

progression. We show that the central domain of cyclin D1 binds HDAC3 and is

sufficient for both AR interaction and repression. Strikingly, over expression of this domain attenuates cell cycle progression in prostatic adenocarcinoma cells. Furthermore,

cyclin D1 co-activator and co-repressor activities can be distinguished, as the central

domain is required for AR and TRβ1 co-repression, but dispensable for co-activation of

ERα. Taken together, these data identify a minimal co-repressor domain responsible for cyclin D1 regulation of the AR. In addition, our data suggest that the co-activator and co- repressor functions of cyclin D1 are distinct and could be targeted in a specific manner to develop novel therapeutics for a multitude of nuclear receptor regulated disease states.

84 INTRODUCTION

Through distinct mechanisms, cyclin D1 regulates a variety of transcriptional programs. Cyclin D1 was first identified based upon its critical role in promoting cell

cycle progression (1-3). In this role, cyclin D1 regulates the transcriptional repressor

function of the retinoblastoma tumor suppressor protein, RB (4). Prior to mitogenic

stimulation, RB forms repressor complexes on target gene promoters required for S-phase

progression (e.g. cyclin A, MCMs) (Reviewed in: 5). To induce proliferation, mitogens

stimulate the expression and stabilization of cyclin D1. Accumulated cyclin D1 binds

and activates cyclin dependent kinase 4/6 (CDK4/6) directing it to phosphorylate RB, and

leading to the subsequent derepression of downstream target gene promoters (6). The

ability of cyclin D1 to disrupt RB repressor complexes is required for cell cycle

progression into S-phase, thus highlighting the importance of cyclin D1 action in cellular

proliferation (7). However, it is increasingly apparent that cyclin D1 regulates a

multitude of additional transcription factors independent of CDK4 and outside of its role

in the cell cycle (e.g. v-Myb, DMP1, Sp-1, Myo-D, and a host of nuclear receptors)

(Reviewed in: 8). Interestingly, these cell cycle independent roles of cyclin D1 appear to

play a significant role in oncogenesis (9). Still, the mechanism by which cyclin D1 exerts

control over the action of these transcription factors remains poorly defined.

Nuclear receptors are currently the most widely defined family of cyclin D1-

regulated factors. In fact, cyclin D1 has been linked to the regulation of estrogen receptor

alpha (ERα), thyroid hormone receptor β (TRβ1), PPARγ, and the androgen receptor

(AR) (10-17). Although the response of these receptors to cyclin D1 action is varied, they

85 are each highly regulated by association with ligand. Conventionally, ligand association

causes nuclear receptor release from inhibitory heat shock proteins, allowing the

subsequent formation of active dimer complexes (18). Such activated nuclear receptor

complexes bind target DNA elements with high affinity and facilitate the recruitment of

co-activators required for target gene transactivation (reviewed in: 19). Interestingly,

cyclin D1 is known to activate ERα in a ligand-independent fashion through recruitment of co-activators (SRC-1 or P/CAF) to the receptor complex (10-12,20). Through this mechanism, cyclin D1 is also perceived to enhance ligand-dependent ERα transactivation

(10). Cyclin D1 co-activator activity is manifested through its C-terminal LxxLL motif, a nuclear receptor interaction domain also common in many co-activators (including SRC-

1) (11). In contrast to its function as an ERα co-activator, cyclin D1 acts as a potent co- repressor of TRβ1 and PPARγ (13,14). We initially demonstrated that cyclin D1 binds directly to the AR, and specifically inhibits its ligand-dependent transactivation functions

(15,16). We showed that this function of cyclin D1 is independent of both CDK4 and RB association (15). Moreover, the C-terminal cyclin D1 LxxLL motif was shown to be dispensable for AR modulation (16). Thus, the mechanism utilized by cyclin D1 to mediate AR co-repression is distinct from its ERα co-activator function. In addition, we demonstrated a marked biological consequence of cyclin D1 AR co-repressor activity, as its ectopic expression attenuated androgen-dependent cellular proliferation in prostate cancer cells (16). Indeed, cyclin D1 expression appears to be important in prostate cancer expression, as a recent report demonstrates a “cyclin switch” occurs during tumor progression wherein cyclin D1 levels decline and other G1 cyclins (e.g. cyclin E) markedly increase (21). Since permanent ablation of AR activity is the goal of prostate

86 cancer therapy, there is a strong impetus to delineate the mechanisms by which cyclin D1 exerts its repressor activity on the AR.

The cyclin D1 protein itself has yet to be crystallized, but its structure can be inferred from that of highly homologous viral (V, M, and K) and mammalian (A and H) cyclins (22-25). A conserved cyclin box domain consisting of five tightly packed alpha helicies lies within the cyclin D1 N-terminus (26). It is through interaction of this domain with cyclin dependent kinase (CDK) 4/6 that the complex becomes competent for

CDK-activating kinase (CAK) phosphorylation and subsequent activation (27). In addition, sequences within the cyclin D1 C-terminal PEST domain are known to regulate cyclin D1 localization and degradation (28,29). Although numerous studies have linked these functional domains to the cell cycle related functions of cyclin D1, few have examined the role of such regions in transcriptional regulation.

Herein, we identify the region of cyclin D1 responsible for both AR binding and repression and determine its role in regulating other nuclear receptors. We have localized the cyclin D1 co-repressor domain to residues 142-253. Deletion of this domain resulted in compromised AR binding and co-repressor activity, but did not perturb proper cell cycle transitions. This region efficiently bound HDAC3 and was sufficient for AR interaction and co-repression. Remarkably, ectopic expression of the cyclin D1 central domain sufficiently hindered androgen dependent growth in prostatic adenocarcinoma cells. In addition, we assessed the requirement of cyclin D1 co-repressor domain for the regulation of other nuclear receptors (ERα and TRβ1). We demonstrate that the cyclin

D1 nuclear receptor co-repressor function can be dissected from its co-activator activity, indicating that nuclear receptor modulation is manifested through distinct cyclin D1

87 motifs. Taken together, these data reveal numerous insights into the mechanism by

which cyclin D1 regulates nuclear receptor function.

EXPERIMENTAL PROCEDURES

Cell Culture and Treatment- CV1 and COS-7 cells were obtained from ATCC and

cultured in DMEM supplemented with 10% heat-inactivated fetal bovine serum (∆FBS;

Atlanta Biologicals, Norcross, GA), 100 units/ml penicillin/streptomycin (Mediatech,

Herndon, PA), and 2 mM L-glutamine. LNCaP cells from ATCC were maintained in

IMEM supplemented with 5% ∆FBS, penicillin/streptomycin, and L-glutamine. For

steroid-free conditions used in reporter assays, 10% charcoal-dextran treated FBS (CDT;

Hyclone Laboratories, Logan UT) was utilized.

Transfection and Transcriptional Reporter Assays- CV1 or COS-7 cells were seeded in

the absence of steroid hormones and transfected with 4µg of DNA using the BES/calcium phosphate transfection protocol as previously described (30). Following overnight transfection, cells were washed and allowed to recover for 4-6 hours prior to stimulation as indicated with 0.1nM or 1nM dihydrotestosterone (24-48 h, as indicated, Sigma), 10 nM 17-β-estradiol (48 h, Sigma), 10 nM T3 (24 h; Sigma), or 0.1% ethanol vehicle.

Cells were then harvested, lysed and monitored for luciferase activity using the Promega luciferase assay kit (Promega, Madison, WI). β-galactosidase activity was measured as an internal control for transfection efficiency using Galacto-Star reagent (Tropix,

Bedford, MA). Appropriate p-values were obtained using ANOVA followed by a

Newman-Kuels Multiple Comparison post-test.

88 To detect protein expression levels of cyclin D1 in reporter assays, CV1 cells were

transfected in 6 cm dishes and green fluorescent protein tagged histone 2B (H2Β-GFP) substituted for CMV-β-galactosidase. Following treatment, cells were harvested and subject to SDS-PAGE on a 10% polyacrylamide gel. The gel was then transferred to

Immobilon (Millipore Corp., Bedford, MA) and immunoblotted for GFP (Santa Cruz

Biotechnology, Santa Cruz, CA; B-2), AR (Santa Cruz Biotechnology; N-20), and cyclin

D1 (Neomarkers, Freemont, CA; Ab-3). Proteins were detected via autoradiography using either goat-anti-mouse (GFP) or goat-anti-rabbit horseradish peroxidase conjugated antibodies (Pierce, Rockford, IL) and enhanced chemiluminescence enhancer

(PerkinElmer Life Sciences, Boston, MA).

Plasmids- The pSG5-AR, H2Β-GFP, CMV-β-galactosidase, PSA-61-LUC, pBS3XERE-LUC, p-CMV5-hERα, pGEX 3X cyclinD1-GST and CMV-CD44 constructs were previously described (16,17). GST-AR1-660 contains the first 660 amino acids of the wild type AR fused to GST and was generously provided by E. Wilson (University of

North Carolina at Chapel Hill, Chapel Hill, NC). The 8Palaa-LUC thyroid hormone responsive luciferase reporter and receptor expression plasmid, pCDNA3.1-TRβ1, were generously provided by Dr. R. Koenig (University of Michigan, Ann Arbor, MI). pCDNA3-HDAC3-Flag was kindly supplied by Dr. E. Seto (University of South Florida,

Tampa, Florida). Plasmid constructs encoding cyclin D1 variants (wild type, ∆1-99, ∆61, and ∆XMN) for baculoviral expression were previously described and kindly provided by

Dr. C. Sherr (St. Jude Children’s Research Hospital, Memphis, TN; (31)). Plasmid encoding cyclin D1-∆XMN (pBJ5-cyclin D1∆XMN) for mammalian expression was previously described (31). The pFlex-Flag-cyclin D1∆174-253 construct was made by

89 removing amino acids 174 to 253 of pFlex-Flag-cyclin D1 (28) by QuikChange PCR

(Stratagene). To generate pCDNA3.1-cyclin D1 wild type, ∆XMN, and ∆174-253 constructs for mammalian expression, the cyclin D1 insert was first removed from the parental vector (pRSV-cyclin D1 (11), pBJ5-cyclin D1∆XMN (31), and pFlex-

FlagCycD1∆174-253, respectively) via BamHI digestion. The inserts were then ligated into the BamHI site of pCDNA3.1 (Invitrogen) and screened for orientation. Additional mammalian expression constructs encoding cyclin D1∆152-174 and cyclin D1∆142-152 were made using pCDNA3.1-cyclin D1 (wild type) as a template for the QuickChange

Site-Directed Mutagenesis Kit as described by the manufacturer (Stratagene, La Jolla,

CA). GST-cyclin D1-central domain (GST-CD) was generated by PCR amplification of amino acids 141-250 of wild type cyclin D1 using primers flanked by BamHI recognition sites. The generated PCR product was then inserted into the BamHI site in pGEX-KG

(32). To generate GFP-tagged cyclin D1 central domain (pEGFP-CD), the PCR product utilized to make GST-cyclin D1-XMN was inserted into the BamHI site of pEGFP-C1

(Clontech, Palo Alto, CA). All cyclin D1 mutants generated were verified by sequencing at the University of Cincinnati DNA core.

In vitro Binding Assays- For assays shown in Figure 1B, cyclin D1 wild type and mutant proteins were generated using the baculovirus purification system as previously described (33,34) and lysed in NETN (20 mM Tris (pH 8.0), 100 mM NaCl, 1 mM

EDTA (pH 8.0), and 0.5% Nonidet P-40) solution containing 1 mM phenylmethylsulfonyl fluoride, 10 µg/ml 1,10-phenanthroline, 10 µg/ml aprotinin, 10

µg/ml leupeptin, 10 mM sodium fluoride, 1 mM sodium vanadate, and 60 mM β- glycerophosphate (PPI). Wild type, ∆1-99, and ∆61 cyclin D1 proteins were

90 subsequently immunoprecipitated on protein A-sepharose beads (Amersham) using 6µL

cyclin D1-Ab3 (Oncogene) and 1uL rabbit anti-mouse (ICN) anti-sera. For cyclin

D1∆XMN, 2µL of D1-17-13G (33) along with 1µL rabbit anti-mouse were utilized. 35S- labeled AR and CD44 proteins were made using the TnT-coupled Reticulocyte Lysate system (Promega) and then added to the immunoprecipitated reactions as indicated. In vitro transcription/translation of AR and CD44 proteins was carried out in the presence of

NEG-772 Easytag Express Labeling Mix (PerkinElmer Life Sciences). Following incubation for 3 h at 4°C with rotation, the reactions were washed 5 times in NETN+PPI.

The samples were then subjected to SDS-PAGE in duplicate. One 10% acrylamide gel was incubated in Fluoro-Hance solution (Research Products International Corp., Mount

Prospect, IL) as specified by the manufacturer, dried, and proteins detected via

autoradiography. The duplicate gel was transferred to Immobilon (Millipore Corp.,

Bedford, MA) and immunoblotted for cyclin D1 using antibody directed against full

length cyclin D1 (Neomarkers, Freemont, CA; Ab-3, Lot #010P010). Horseradish

peroxidase-conjugated protein-A secondary antibody (Zymed) and enhanced

chemiluminescence enhancer (PerkinElmer Life Sciences) were used to visualize

proteins.

For GST-AR binding assays shown in Figure 3B, GST-AR1-660 and GST were generated and purified from bacteria using standard procedures. Lysates from COS-7 cells transfected with 16 µg of the indicated cyclin D1 expression plasmids were incubated with equal amounts of GST-AR or GST-conjugated agarose for 4 h at 4°C.

Following binding, the beads were washed five times in NETN+PPI and prepared for

SDS-PAGE. Transferred western blots were probed for both cyclin D1 and AR. GST in

91 vitro binding assays shown in Figure 4A, were preformed as previously described (16).

Bound AR was detected and visualized on a Phospho Imager.

Co-immunoprecipitation Assays- For co-immunoprecipitations, COS-7 cells were seeded in 10 cm dishes and transfected the following day using the BES/calcium phosphate protocol (30). Cyclin D1 and pSG5-AR or GFP-XMN and HDAC3 constructs were introduced at a 1:1 ratio with DNA concentrations totaling 16 µg. Post-transfection, cells were washed and allowed to recover for 24 h prior to harvest and lysed in 1 mL

NETN +PPI. Lysates were sonicated and clarified by centrifugation before being divided equally amongst control (DBF4; Santa Cruz biotechnology), AR (N-20; Santa Cruz), and cyclin D1 (Figure 1C; Neomarkers) or control (DBF4), GFP (Santa Cruz) and HDAC3

(Santa Cruz) immunoprecipitation reactions. Bound fractions were subjected to SDS-

PAGE followed by western blotting. Protein-A horseradish peroxidase conjugated secondary antibody was utilized for protein visualization (Zymed, San Francisco, CA).

Flow Cytometry- COS-7 cells were transfected with 16 µg of plasmid encoding cyclin

D1 wild type or ∆XMN and 1.0 µg H2B-GFP. 24 h post-transfection cells were harvested and fixed in 80% ice-cold ethanol. Following fixation, cells were stained with propidium iodide (PI; 0.2 µg/µL) and subject to flow cytometry to detect PI intensity in both the GFP positive and negative (mock) populations. Histograms represent approximately 10,000 cells.

BrdU Incorporation- LNCaP cells were seeded on poly-L-lysine coated coverslips in 6- well dishes and transfected the following day using Lipofectin reagent (Invitrogen) and

2.0 µg of either GFP or GFP-central domain (GFP-CD). Following transfection, the media was replaced and cells allowed to recover for 32 h. Transfected cells were then

92 pulsed overnight with bromodeoxyuridine (BrdU; Amersham Biosciences), fixed and

stained as previously described (16). Results represent duplicate experiments performed

in triplicate wherein at least 100 cells were counted per coverslip.

Structural Modeling- Two-dimensional models as depicted in Figure 4A and C were

generated using the POLYVIEW server (35). The secondary structure of cyclin D1 is

based upon sequence to structure alignment of residues 22 to 261 of cyclin D1 with the

sequence of the highly homologous cyclin K, as defined in (22). Solvent accessibilities

were assigned to this region (without side chain geometry optimization) based upon the

cyclin K-CDK6-INK4C crystal structure (PDB code 1G3N) (36). For the remaining N-

and C-terminal cyclin D1 fragments, as well as the AR N-terminus (a.a. 1-50) depicted in

4C, the secondary structures and solvent accessibilities were predicted by using the

SABLE server (37). Modeling of homologous cyclin K structures as shown in Figure 4B

was performed using the RASMOL program (38) and depicted using the POLYVIEW

server.

RESULTS

Residues 142-253 are required for cyclin D1-androgen receptor interactions –

We previously demonstrated that cyclin D1 interacts directly with the AR N-terminus to

potently inhibit transactivation of androgen responsive genes (15-17). This co-repressor

activity of cyclin D1 is independent of RB and CDK association and does not require the

LxxLL domain previously demonstrated to be essential for ERα co-activator activity

(11). In order to determine the region(s) of cyclin D1 responsible for AR interaction,

93 specific cyclin D1 deletion mutants (Figure 1A) were generated in a baculovirus

expression system and immunoprecipitated on protein-A beads. CD44 (negative control)

and AR protein were then generated by in vitro transcription/translation in the presence of 35S-methionine and added to the bead-conjugated cyclin D1, as indicated. Binding reactions were washed and subjected to SDS-PAGE along with input proteins. Input and bound radiolabeled proteins were detected via autoradiography. As anticipated, CD44 failed to interact with any of the cyclin D1 derivative proteins (Figure 1B; lanes 1, 3, 5 and 7). In contrast, AR effectively bound wild type cyclin D1 as well as the ∆1-99 mutant (see lanes 2, and 4). Binding of the ∆61 mutant was slightly reduced, but maintained (see lane 6). Cyclin D1-∆XMN, however, failed to associate with the AR in vitro, indicating that the AR binding domain may lie within this region (lane 8).

Expression and efficient immunoprecipitation of the cyclin D1 constructs was verified in parallel by SDS-PAGE followed by immunoblot analysis. (Fig. 1B, bottom panel). Both wild type and mutant cyclin D1 proteins were effectively expressed and immunoprecipitated. Thus, failure of cyclin D1-∆XMN to bind AR suggests that the deleted central region (amino acids 142-253) is required for the in vitro association of these two proteins.

To further examine the requirement of this region for AR binding, we analyzed cyclin D1-AR interactions through co-immunoprecipitation experiments. Here, COS-7 cells were transfected with AR and either cyclin D1 or cyclin D1-∆XMN at a 1:1 ratio.

Transfected cells were then harvested, lysed, and incubated with anti-sera generated against AR, DBF4 (negative control), or cyclin D1. Complexes were then subjected to

SDS-PAGE followed by immunoblotting as indicated (Figure 1C). As expected, wild

94 type cyclin D1 efficiently associated with AR (lane 1), and AR co-immunoprecipitated

with cyclin D1 (lane 3). In contrast, cyclin D1-∆XMN failed to co-immunoprecipitate

with the AR in both the forward and reverse experiment (Fig. 1C, lanes 4, 6). Together,

these data clearly demonstrate that direct binding of cyclin D1 to the AR requires amino

acids 142-253.

Compromised co-repressor function by the Cyclin D1-∆XMN Mutant - Since

cyclin D1-∆XMN is severely compromised for AR binding, we hypothesized that its AR co-repressor activity would be similarly compromised. To assess the consequences of the

142-253 deletion, we employed mammalian expression constructs encoding wild type or cyclin D1-∆XMN in reporter assays using the PSA-61-LUC reporter construct (6.1kb of the prostate specific antigen (PSA) promoter linked to luciferase (39)) to monitor AR activity. CV1 cells, which express no endogenous AR (data not shown), were transfected

with reporter (0.75 µg), human AR (pSG5-AR; 0.5 µg), CMV-β-galactosidase (internal

control for transfection efficiency; 0.5 µg) and cyclin D1 (wild type or mutant; 1.5 µg) or

empty vector (pCDNA3). Following transfection, cells were stimulated for 24 h with 0.1

nM DHT or 0.1% ethanol vehicle as indicated. Luciferase activity was monitored and

normalized to β-galactosidase. Relative luciferase activity is shown and represents at least three independent experiments performed in triplicate. As expected, cyclin D1 effectively inhibited AR transactivation reducing its activity by approximately 92% (Fig.

2A). However, cyclin D1-∆XMN failed to significantly inhibit AR co-repressor activity

(~25% repression). To verify that diminished cyclin D1 co-repressor activity was not due to instability of the central domain deleted mutant protein, parallel immunoblots were performed. Lysates from parallel transfections wherein H2B-GFP was substituted for

95 CMV-β-galactosidase were subjected to SDS-PAGE followed by immunoblotting. As

shown in Figure 2B, cyclin D1-∆XMN and cyclin D1 wild type were equally expressed.

In addition, the level of AR remained unchanged in the presence of either construct, verifying that the inhibitory action of cyclin D1 does not involve regulation of AR protein levels (16). Thus, failure of cyclin D1-∆XMN to repress AR transactivation is not due to

protein instability or deregulated AR expression. However, it has been suggested that

AR activity is regulated in a cell cycle specific manner (40). AR transcriptional activity

is greatly reduced at the G1-S transition wherein cyclin D1 levels are highest (40).

Although regions within the central domain were not previously reported to disrupt CDK

binding, the integrity of threonine 156 appears to dictate CAK association and activation of the cyclin D1-CDK complex (41). In order to assess the effect of central domain deletion on cell cycle transitions, the cell cycle profile of cyclin D1-∆XMN expressing cells was examined. COS-7 cells were transfected with H2B-GFP and wild type or cyclin D1-∆XMN. 24 h post-transfection, cells were harvested, fixed, and stained with propidium iodide to determine DNA content. Flow cytometry was performed and both

GFP positive (transfected) and negative (non-transfected) cells from the same plate were analyzed. Histograms represent at least 10,000 (~120,000 for the mock sample) cells analyzed and experiments were performed in duplicate with the standard deviation shown. Transfection with either wild type or cyclin D1-∆XMN did not significantly alter cell cycle profiles relative to mock-transfected cells (Figure 2C). Therefore, our data confirm that deletion of the cyclin D1 central domain does not affect proper cell cycle progression, but compromises its ability to bind the AR and regulate its transcriptional activity.

96 Multiple Sequences Within the Cyclin D1 Central Domain (a.a. 142-153) Mediate

AR Binding and Regulation- The central domain of cyclin D1 encodes a largely understudied region of the protein. Due to its high level of structural similarity with known viral (V, K and M) and mammalian cyclins (A and H), one may use homology modeling to obtain a putative three-dimensional structure of cyclin D1. As noted in other cyclins, the central domain of cyclin D1 contains a linker domain (a.a. 148-156) followed by five sequential alpha helicies (a.a. 157-251) (22-25), see also Figure 4). In order to identify a specific AR-interacting motif within this cyclin D1 domain, we developed numerous internal deletion constructs and tested them for functional competency. To assess the binding capacity of these mutants, GST-AR1-660 was generated and purified on glutathione agarose. Cyclin D1 mutant and wild type proteins were then obtained through transfection of COS-7 cells with pCDNA3.1-based constructs as shown in Figure

3A. 24 h post-transfection, the lysates were harvested and incubated with immobilized

GST-AR1-660. Following binding, the reactions were washed, samples prepared for

SDS-PAGE analysis, and subsequently transferred to PVDF membrane. The resulting membranes were then immunoblotted with antibodies directed against the AR N-terminus and cyclin D1 C-terminus, as these epitopes are present in the truncated AR and internally deleted cyclin D1 proteins. As expected, wild type cyclin D1 bound to the

GST-AR1-660 column, whereas cyclin D1-∆XMN demonstrated compromised binding capacity (Figure 3B, bottom panel; compare lanes 4 and 6). Interestingly, N- or C- terminal deletions within the cyclin D1 central domain (∆142-152, ∆152-174, ∆174-253) retained some binding capacity to the AR N-terminus, thus indicating that multiple

97 contact sites may be present within this region (Figure 3B, bottom panel; compare lanes

8, 10 and 12).

To assess the AR co-repressor function of each subdeletion, reporter assays were performed in CV1 cells transfected with PSA-61-LUC (0.75 µg), AR (0.5 µg), CMV-β- galactosidase (0.5 µg) and either cyclin D1 wild type, internal central domain deletion constructs, or empty pCDNA3 vector (1.5 µg). Following transfection, cells were stimulated for 24 h, harvested, and assayed as described in Figure 2A. The percent inhibition was determined relative to AR activity in the presence of ligand and absence of cyclin D1. Bars represent the average inhibition from at least three independent experiments performed in triplicate. Wild type cyclin D1 effectively reduced AR activity on the PSA promoter to basal levels (~88.9% repression), whereas cyclin D1-∆XMN demonstrated compromised AR co-repressor activity (~12.7% repression) (Figure 3C).

Internal deletion of amino acids 142-152 had no effect upon the ability of cyclin D1 to inhibit AR transactivation, supporting the hypothesis that these residues are dispensable for AR interaction. Deletion of N- or C-terminal regions within the central domain

(∆152-174, ∆174-253) slightly reduced cyclin D1 inhibition (~67.0% and ~60.0%, respectively), supporting a model wherein several binding sites within the central domain

mediate AR binding. However, in contrast to Cyclin D1-∆XMN, severe aberrations in

cell cycle progression were observed upon transfection of the internal central domain

deletion mutants (∆152-253 and ∆174-253, data not shown) therefore reducing the

potential to accurately assess AR co-repressor function. As such, deletion of the entire

central domain (142-254) is optimal for ablation of AR co-repressor function.

98 The Cyclin D1 Central Domain is Sufficient for AR Binding and Inhibition- Our

data demonstrate that deletion of the cyclin D1 central domain yields a protein defective

in its ability to bind and regulate the AR. Although this domain appears to be folding

into a separate structural module (see Figure 4B) it remains possible that elimination of

this region alters the conformation of neighboring domains, thus abrogating cyclin D1-

AR interaction. We previously mapped cyclin D1 binding to the first 34 amino acids of

the AR (42). The predicted two-dimensional structure of this region is depicted in Figure

4C. When examining the cyclin D1-AR interaction, it is also necessary to note that the linker regions, as well as several residues forming the last α-helix within the cyclin box are deleted in the cyclin D1-∆XMN mutant (Figure 1A). The cyclin box motif removed in the deletion mutant (residues LLVNKLK) is replaced by a similar C-terminal motif

(residues LLESSLR). Therefore, this fragment is likely to adopt a similar conformation in the deletion construct, preserving the ability to bind CDKs and linking the cyclin box directly with the C-terminal (glutamic acid rich) signaling domain (see Figure 4A and

4B). The above hypothesis is addressed experimentally in two steps: we first show that the central domain is sufficient for AR binding and inhibition and then we demonstrate that cyclin D1-∆XMN maintains its co-activator activity in the absence of the central domain.

In order to test the sufficiency of the central domain for AR binding, GST-central domain (GST-CD), GST-cyclin D1, and GST alone were generated and purified on glutathione agarose. AR was subject to in vitro transcription/translation in the presence

of 35S-methionine and added to each reaction. Following binding, the columns were washed and both input and bound fractions subject to SDS-PAGE. The resulting gel was

99 dried and bound AR detected on a Phospho Imager. In addition, aliquots of glutathione

bound GST proteins were subjected to SDS-PAGE and Coomassie stained to verify

protein production and purification. The central domain of cyclin D1 bound AR with

efficiency comparable to that observed in the presence of wild type protein (Figure 5;

compare lanes 2 and 3). GST alone, however, failed to bind AR demonstrating the low

background binding observed in this experiment (lane 4). In addition, co-

immunoprecipitation experiments utilizing a GFP-tagged central domain confirmed these

findings in vivo (data not shown). Thus, these data demonstrate sufficiency of the cyclin

D1 central domain for AR binding.

Although we show that the central domain of cyclin D1 is sufficient for AR

binding, it remains possible that co-repressor activity lies outside of this region. In this

case, deletion of the central domain would result in decreased AR co-repressor activity

purely due to the compromised ability of cyclin D1 to bind active AR complexes. Thus,

we assessed the ability of GFP-central domain (GFP-CD) to inhibit AR activity in

reporter assays. First, the expression and integrity of GFP-CD and GFP in COS-7 cells

were verified by immunoblot (Figure 5B, bottom panel). Next, COS-7 cells were

transfected as in Figure 2A with PSA-61-LUC (0.75 µg), pSG5-AR (0.5 µg), CMV-β-

galactosidase (0.5 µg), and GFP-CD or GFP alone at increasing concentrations (0.25, 0.5

and 1.0 µg, respectively). Following 48 h stimulation, cells were harvested and assayed

for luciferase and β-galactosidase activity. Activity of the AR in the presence of ethanol vehicle was set to “1”. Addition of 1 nM DHT in the presence or absence of GFP

(parental vector) stimulated PSA-LUC transactivation ~7 fold as expected (Figure 5B,

upper panel). However, addition of GFP-CD repressed AR activity in a significant, and

100 dose dependent manner (reducing activity from ~7 to ~2 fold), indicating that the central domain contains AR repressor capacity.

To test the impact of GFP-CD on androgen dependent proliferation, LNCaP cells were utilized. These cells are dependent on AR activity for proliferation and we have previously shown that ectopic cyclin D1 attenuates both AR dependent transcription and

AR-dependent proliferation in this model system (16,17). To test the ability of the central domain to modulate androgen dependent growth, LNCaP cells were transfected with GFP or GFP-CD and monitored for S-phase progression via BrdU incorporation assays. Data shown represent two independent experiments performed in triplicate. LNCaP cells transfected with GFP-CD demonstrated a significantly lower proliferation index than those transfected with GFP alone (Figure 5C; ~11.4% vs. ~18.5% BrdU incorporation).

Therefore, over expression of the central domain itself abrogates cell cycle progression in a manner similar to the full-length protein (see: 16), indicating that the cyclin D1 repressor function lies within this region. Taken together, these data demonstrate that the central domain of cyclin D1 is sufficient for both AR binding and inhibition, thus identifying a novel AR co-repressor domain within a largely unstudied region of cyclin

D1.

We have also shown that the effect of cyclin D1 on AR activity can be partially reversed through inhibition of histone deacetylases (HDACs), thus suggesting that cyclin

D1 may recruit HDACs to the AR complex (16). Subsequently, it was shown that cyclin

D1 can bind to HDAC3 and recruit this repressor molecule to the TRβ1 complex (13).

Given our observation that the central domain is sufficient for co-repressor activity, we hypothesized that HDAC3 may bind to this region of cyclin D1. To test this hypothesis,

101 COS-7 cells were transfected with expression plasmids encoding GFP-CD and HDAC3.

Transfected cells were harvested and lysates subjected to co-immunoprecipitation

experiments, as shown (Figure 5D). Immunoprecipitation of GFP-CD complexes

revealed a strong association with HDAC3, which was also noted in the reverse reaction

(compare lanes 1 and 3). Control antibody (DBF4), however, failed to pull down either

HDAC3 or GFP-CD (lane 2). Thus, the cyclin D1-CD binds HDAC3 specifically,

uncovering one potential mechanism by which this domain mediates AR co-repression.

The Cyclin D1 Central Domain Mediates Nuclear Receptor Co-repressor, but Not

Co-activator Activity- Since cyclin D1 has been reported to both positively (e.g. ERα) and negatively (e.g. TRβ1) regulate nuclear receptor transactivation, we examined the role of the central domain in each of these functions (10-12,20). Specifically, regulation of TRβ-1 has been shown to occur through recruitment of HDAC3 to thyroid hormone responsive genes (13). To assess the consequence of central domain deletion upon the repression of other nuclear receptors by cyclin D1, we performed reporter assays in CV-1 cells transfected as in Figure 2A with TRβ1 (0.5 µg) and 8Palaa-LUC (0.5 µg). The

8Palaa-LUC reporter construct contains 2 copies of 8Pal, a palidromic thyroid hormone responsive element (TRE) with two perfect octameric half sites, upstream of luciferase.

Transfected cells were treated for 24 h with either 10 nM thyroid hormone (T3) or 500

µM sodium hydroxide (vehicle). In the absence of cyclin D1 expression, T3 activated

TRβ1 transcription approximately 5.5 fold. At a 1:4 ratio of cyclin D1 to TRβ1, receptor activity was significantly inhibited (Figure 6A; p<0.05). Interestingly, addition of cyclin

D1-∆XMN demonstrated significantly reduced TRβ1 co-repressor activity in comparison to wild type cyclin D1 (p<0.05). Thus, deletion of the central domain abrogates cyclin

102 D1 repression of multiple nuclear receptors including AR and TRβ1. Since the co-

activator function of cyclin D1 has been mapped to an LxxLL motif external to the

central domain deletion, we hypothesized that ERα regulation would be retained in the presence of the cyclin D1-∆XMN mutant. To test this hypothesis, reporter assays were

performed in CV-1 cells transfected with ERα (0.5 µg) and the 3XERE-LUC (0.75 µg)

reporter. 48 h post-stimulation with 10nM estradiol (E2) or 0.1% ethanol vehicle, cells

were harvested and reporter assays performed. ERα activity in the presence of ethanol

vehicle was set to “1”. E2 stimulated ERα activity approximately 4.6 fold over vehicle alone (Figure 6B). This activity was significantly enhanced by expression of both wild type cyclin D1 and cyclin D1-∆XMN (5.8 and 6.7 fold, p<0.05, respectively) demonstrating maintenance of co-activator activity in the absence of the cyclin D1 central domain. Interestingly, ligand-independent activation of ERα was not noted in these assays as previously reported (11). This observation may be cell type or reporter dependent as previous studies were conducted in U2-OS and COS-7 cells, which display increased ligand dependent co-activation by cyclin D1 (approximately 30-40 fold) and with the ERE-TATA-LUC reporter construct. Thus, cyclin D1 possesses two distinct mechanisms of nuclear receptor modulation. The cyclin D1 co-repressor function appears to be conserved within the central domain, whereas co-activator activity lies fully outside of this region. In summary, our data identify a transcriptional co-repressor motif within cyclin D1, capable of binding to both AR and HDAC3, and sufficient to repress

AR transactivation potential. Moreover, we demonstrate that this co-repressor region is distinct from the co-activator function of cyclin D1, revealing dual functions for cyclin

D1 in nuclear receptor modulation.

103

DISCUSSION

Our data examining the cyclin D1 AR co-repressor function identifies a critical and conserved nuclear receptor regulatory domain. We find that the central domain of cyclin D1 (amino acids 142-253) is critical for AR binding and that deletion of these residues abrogates cyclin D1 co-repressor activity. Interestingly, we also demonstrate that the central domain directly interacts with HDAC3 and is sufficient for both AR binding and inhibition of androgen dependent proliferation, implying that this region represents a minimal co-repressor domain. Further studies to generally examine the role of the cyclin D1 central domain in nuclear receptor regulation demonstrates that its co- repressor function is conserved. Specifically, cyclin D1 lacking the AR co-repressor domain also demonstrates compromised ability to inhibit TRβ1 transactivation.

However, nuclear receptor co-activation by cyclin D1 appears to be distinct from its co- repressor function and clearly lies outside of the central domain. Our data show that ERα co-activation occurs in the absence of the central co-repressor domain and support previous findings implicating a C-terminal LxxLL motif in nuclear receptor co-activation by cyclin D1. These data have significant mechanistic and clinical implications, demonstrating that a small peptide encodes a conserved nuclear receptor co-repressor motif distinct from cyclin D1 co-activator function.

The Central Domain of Cyclin D1 is Critical for AR Binding and Regulation-

Although the crystal structure of cyclin D1 has yet to be solved, its marked homology with previously crystallized viral (K, M, A and H) cyclins gives insight into its putative

104 configuration (22-25). Through such homology studies, the cyclin box has been

identified to lie within amino acids 56-152 of the protein, an area containing five distinct alpha helices (24,26,31). Within this N-terminal domain is the CDK4 binding site

(K114), required for kinase activation and subsequent RB phosphorylation (27), see also

Figure 4A and 4B). Immediately following the cyclin box is a short linker peptide and a second set of five sequential helices. It is within this widely uncharacterized domain that we map the interaction of cyclin D1 with the AR. We show that this domain is both required and sufficient for AR binding (Figures 1, 5). In addition, we demonstrate that this domain alone can efficiently bind and regulate AR activity, inhibiting target gene transactivation and the proliferation of androgen dependent prostate cancer cells (Figure

5). We previously demonstrated that the N-terminal AR fragment (a.a. 1-34), whose computationally predicted secondary structures and solvent accessibilities are shown in

Figure 4C, mediates the interaction with cyclin D1 (42). As can be seen from Figure 4C and 4A the solvent exposed surfaces of the AR N-terminal (residues 17-33) and the central domain C-terminal (residues 238-253) helices appear to be complementary in terms of charged residue distribution. Therefore, one plausible mode of interaction between AR and cyclin D1 could involve these two helices packed against each other

(especially in light of the similarity to cyclin D3 discussed below). Testing this hypothesis will be the subject of future work.

Our data also demonstrate a direct interaction of this domain with HDAC3, providing insight into the mechanism by which AR inhibition is achieved. We previously demonstrated that cyclin D3 is also capable of AR co-repression, although the mechanism behind this activity has yet to be defined (15). It is intriguing that the cyclin

105 D1 central domain contains several regions with virtual identity to their cyclin D3 counterparts, including, but not exclusive to, amino acids 142-161, 178-192, and 242-252

(100% homology). Indeed, it is possible that several of these homologous regions participate in AR binding, as internal deletions of the cyclin D1 central domain demonstrates impaired AR binding and co-repressor activity.

Interestingly, deletion of amino acids 142-152, which contain a putative nuclear receptor interaction motif (LLLxxxLxxxL), failed to compromise cyclin D1 co-repressor activity (Figure 3C). Consistent with this finding, mutation of residues within this region

(L142-144A) of cyclin D1 in the presence of the L254, 255A yielded a mutant capable of full AR repression (data not shown). Thus, multiple novel motifs within the cyclin D1 central domain are likely to mediate AR interaction. Significantly, the amino acids of the linker domain are also required for proper phosphorylation of the CDK T-loop by CAK

(41). In fact, mutation within this domain (T156A) compromises cyclin D1 function and also decreases its affinity for p21, a factor thought to be required for nuclear import of the cyclin–CDK complex (41). These previous findings may explain why internal deletions of the cyclin D1 central domain caused defects in cell cycle progression only in the presence of the linker domain (data not shown). Although AR transactivation is cell cycle dependent, this function has yet to be examined for many other nuclear receptors

(40). In a recent report by Padma et al (43), the translation and activity of TRβ is also linked to cell cycle progression. Specifically, TRβ transactivation is lowest at G0 and increases as cell cycle progresses, peaking at G2/M. Our findings suggest that examination of both cell cycle dependent and independent roles of cyclin D1 is critical

106 for proper interpretation of its co-modulatory effects. In addition, it will be of interest to

determine if PPARγ and/or ERα activities are regulated in a cell cycle dependent manner.

Conservation of the cyclin D1 co-repressor domain for nuclear receptor

regulation- Recent evidence has linked cyclin D1 to the regulation of numerous

transcription factors including many from the nuclear receptor super family. Estrogen

receptor alpha (ERα) was among the first transcription factors documented to be cyclin

D1-regulated. Through the recruitment of ERα co-activators (P/CAF or SRC-1), cyclin

D1 enhances receptor transactivation (10,12). Such interaction requires the N-terminal

LxxLL motif of cyclin D1. First identified for their role in co-activator binding to nuclear receptors, LxxLL motifs are known to interact with the hydrophobic groove created by helix 12 closure over the receptor ligand binding pocket (44-46).

Interestingly, cyclin D1-mediated co-activator recruitment has significant clinical

implications as high levels of cyclin D1 expression correlate with tumor relapse and poor

survival rates in breast cancer patients (47,48). In contrast to its role as an ERα co-

activator, cyclin D1 is a potent co-repressor of numerous nuclear receptors including

TRβ1, PPARγ, and the AR (13-17,49). In these cases several mechanisms of inhibition have been uncovered. For the AR, recruitment of HDAC activity and inhibition of N- to

C-terminal interactions results in decreased receptor transactivation (16,42). In the case of TRβ1, cyclin D1 works as a co-repressor through recruitment of HDAC3 to the receptor (13). Here, we demonstrate that the cyclin D1 central domain binds HDAC3 and mediates both AR interaction and co-repression (Fig. 4), suggesting that the recruitment of histone deactylases by cyclin D1 may represent a conserved mechanism of nuclear receptor repression. This hypothesis is strongly supported by the finding that cyclin D1

107 co-repression of TRβ1 is lost upon deletion of the central domain, similar to that observed with the AR (see Fig. 2A and 6A). However, deletion of this domain does not compromise ERα co-activation by cyclin D1, indicating that the intact C-terminal LxxLL motif remains functional (Fig. 6B). Surprisingly, we fail to see co-activation of ERα in the absence of hormone as previously reported (10,12). It remains possible that estrogen independent activation by cyclin D1 is cell type or promoter specific and not observed in

CV1 cells expressing the ERE-LUC reporter due to the lack of an essential activating component. In a manuscript recently published by Wang et al (14), PPARγ regulation by cyclin D1 is also mapped to a region within the central domain (amino acids 143-179).

Our data suggest that the cyclin D1 central domain contains several conserved nuclear receptor interacting motifs, yet these sequences are distinct from those utilized for ERα co-activation. Notably, cyclin D1 binding to the Dmp1 transcription factor was also mapped to the central domain, however the effect of the ∆XMN mutation upon Dmp1 regulation was never determined (31). Thus, it remains possible that the nuclear receptor co-repressor motif of cyclin D1 is also utilized in the regulation of other transcription factors such as STAT3, v-Myb, and Sp1 (50-53). To date, the interaction motif required for regulation of these transcription factors has yet to be identified. Clearly, further investigations are needed to determine the overall significance of the cyclin D1 central domain in transcriptional regulation.

In summary, we have narrowed the cyclin D1 nuclear receptor co-repressor motif to a central region (amino acids 142-152) within the protein. Ablation of the repressor region does not disrupt cell cycle control, thus supporting previous evidence that the cell

cycle function of cyclin D1 can be distinguished from its function in G1-S control. This

108 region is sufficient for AR binding and contains several motifs that may contribute to the interaction and inhibition of nuclear receptor activity. Cyclin D1 co-repressor activity appears distinct from its ability to co-activate other nuclear receptors, which may prove useful in the development of protein mimetics to block nuclear receptor function. In addition, current evidence suggests that the identified co-repressor domain of cyclin D1 extends beyond nuclear receptor control, as the Dmp1 transcription factor also interacts with this motif. Together, the data shown here reveal a novel regulatory region of cyclin

D1 and demonstrate that its role as a co-repressor can be cleanly segregated from its function in cell cycle control or as a nuclear receptor co-activator. Given both the cell cycle dependent and independent roles of cyclin D1 hypothesized to regulate androgen dependent growth, it will be of interest to determine if AR binding to this region is altered during prostatic tumorigenesis. Furthermore, it remains to be determined if this domain harbors additional clues into the development and progression of other disease states involving cyclin D1 regulated pathways.

ACKNOLEDGEMENTS

The authors wish to thank Dr. G. Babcock and S. Schwemberger for the use and aid in acquiring flow cytometry data. We thank Dr. E. Knudsen, and members of the Knudsen laboratories for their technical assistance and critical reading of the manuscript. This work was supported by National Institute of Health grant R01-CA-099996. The

University of Cincinnati Distinguished Graduate Fellowship and the Albert J. Ryan

Foundation support the work of both C.E.P. and C.J.B.

109 REFERENCES

1. Motokura, T., Bloom, T., Kim, H. G., Juppner, H., Ruderman, J. V., Kronenberg, H. M., and Arnold, A. (1991) Nature 350, 512-515 2. Arnold, A., Kim, H. G., Gaz, R. D., Eddy, R. L., Fukushima, Y., Byers, M. G., Shows, T. B., and Kronenberg, H. M. (1989) J Clin Invest 83, 2034-2040 3. Rosenberg, C. L., Kim, H. G., Shows, T. B., Kronenberg, H. M., and Arnold, A. (1991) Oncogene 6, 449-453 4. Kato, J., Matsushime, H., Hiebert, S. W., Ewen, M. E., and Sherr, C. J. (1993) Genes Dev 7, 331-342 5. Dyson, N. (1998) Genes Dev 12, 2245-2262 6. Sherr, C. J. (1994) Cell 79, 551-555 7. Resnitzky, D., and Reed, S. I. (1995) Mol Cell Biol 15, 3463-3469 8. Coqueret, O. (2002) Gene 299, 35-55 9. Lamb, J., Ramaswamy, S., Ford, H. L., Contreras, B., Martinez, R. V., Kittrell, F. S., Zahnow, C. A., Patterson, N., Golub, T. R., and Ewen, M. E. (2003) Cell 114, 323-334 10. Zwijsen, R. M., Wientjens, E., Klompmaker, R., van der Sman, J., Bernards, R., and Michalides, R. J. (1997) Cell 88, 405-415 11. Zwijsen, R. M., Buckle, R. S., Hijmans, E. M., Loomans, C. J., and Bernards, R. (1998) Genes Dev 12, 3488-3498 12. Neuman, E., Ladha, M. H., Lin, N., Upton, T. M., Miller, S. J., DiRenzo, J., Pestell, R. G., Hinds, P. W., Dowdy, S. F., Brown, M., and Ewen, M. E. (1997) Mol Cell Biol 17, 5338-5347 13. Lin, H. M., Zhao, L., and Cheng, S. Y. (2002) J Biol Chem 277, 28733-28741 14. Wang, C., Pattabiraman, N., Zhou, J. N., Fu, M., Sakamaki, T., Albanese, C., Li, Z., Wu, K., Hulit, J., Neumeister, P., Novikoff, P. M., Brownlee, M., Scherer, P. E., Jones, J. G., Whitney, K. D., Donehower, L. A., Harris, E. L., Rohan, T., Johns, D. C., and Pestell, R. G. (2003) Mol Cell Biol 23, 6159-6173 15. Knudsen, K. E., Cavenee, W. K., and Arden, K. C. (1999) Cancer Res 59, 2297- 2301 16. Petre, C. E., Wetherill, Y. B., Danielsen, M., and Knudsen, K. E. (2002) J Biol Chem 277, 2207-2215 17. Petre-Draviam, C. E., Cook, S. L., Burd, C. J., Marshall, T. W., Wetherill, Y. B., and Knudsen, K. E. (2003) Cancer Res 63, 4903-4913 18. Pratt, W. B., and Toft, D. O. (1997) Endocr Rev 18, 306-360 19. Khan, O. Y., and Nawaz, Z. (2003) Curr Opin Drug Discov Devel 6, 692-701 20. McMahon, C., Suthiphongchai, T., DiRenzo, J., and Ewen, M. E. (1999) Proc Natl Acad Sci U S A 96, 5382-5387 21. Maddison, L. A., Huss, W. J., Barrios, R. M., and Greenberg, N. M. (2004) Prostate 58, 335-344 22. Card, G. L., Knowles, P., Laman, H., Jones, N., and McDonald, N. Q. (2000) Embo J 19, 2877-2888 23. Schulze-Gahmen, U., and Kim, S. H. (2002) Nat Struct Biol 9, 177-181

110 24. Kim, K. K., Chamberlin, H. M., Morgan, D. O., and Kim, S. H. (1996) Nat Struct Biol 3, 849-855 25. Brown, N. R., Noble, M. E., Endicott, J. A., Garman, E. F., Wakatsuki, S., Mitchell, E., Rasmussen, B., Hunt, T., and Johnson, L. N. (1995) Structure 3, 1235-1247 26. Jeffrey, P. D., Russo, A. A., Polyak, K., Gibbs, E., Hurwitz, J., Massague, J., and Pavletich, N. P. (1995) Nature 376, 313-320 27. Hinds, P. W., Dowdy, S. F., Eaton, E. N., Arnold, A., and Weinberg, R. A. (1994) Proc Natl Acad Sci U S A 91, 709-713 28. Diehl, J. A., Zindy, F., and Sherr, C. J. (1997) Genes Dev 11, 957-972 29. Diehl, J. A., Cheng, M., Roussel, M. F., and Sherr, C. J. (1998) Genes Dev 12, 3499-3511 30. Chen, C., and Okayama, H. (1987) Mol Cell Biol 7, 2745-2752 31. Inoue, K., and Sherr, C. J. (1998) Mol Cell Biol 18, 1590-1600 32. Guan, K. L., and Dixon, J. E. (1991) Anal Biochem 192, 262-267 33. Gladden, A. B., and Diehl, J. A. (2003) J Biol Chem 278, 9754-9760 34. Summers, M. a. S. G. (1987) Tex. Agric. Exp. St. Bull., 1555 35. Porollo A., A. R., Meller J. (In Review) 36. Jeffrey, P. D., Tong, L., and Pavletich, N. P. (2000) Genes Dev 14, 3115-3125 37. Adamczak R., P. A., and Meller J. (2004) Proteins: Structure, Function and Bioinformatics 38. Sayle, R. A., and Milner-White, E. J. (1995) Trends Biochem Sci 20, 374 39. Cleutjens, K. B., van der Korput, H. A., van Eekelen, C. C., van Rooij, H. C., Faber, P. W., and Trapman, J. (1997) Mol Endocrinol 11, 148-161 40. Martinez, E. D., and Danielsen, M. (2002) J Biol Chem 277, 29719-29729 41. Diehl, J. A., and Sherr, C. J. (1997) Mol Cell Biol 17, 7362-7374 42. Burd CJ, P.-D. C., Moghadam H, Wilson EM, and Knudsen KE. (In Review) 43. Padma M., D. N. I., East-Palmer J., and Yen, P.M. (2004) MBC 44. McInerney, E. M., Rose, D. W., Flynn, S. E., Westin, S., Mullen, T. M., Krones, A., Inostroza, J., Torchia, J., Nolte, R. T., Assa-Munt, N., Milburn, M. V., Glass, C. K., and Rosenfeld, M. G. (1998) Genes Dev 12, 3357-3368 45. Heery, D. M., Kalkhoven, E., Hoare, S., and Parker, M. G. (1997) Nature 387, 733-736 46. Darimont, B. D., Wagner, R. L., Apriletti, J. W., Stallcup, M. R., Kushner, P. J., Baxter, J. D., Fletterick, R. J., and Yamamoto, K. R. (1998) Genes Dev 12, 3343- 3356 47. Adnane, J., Shao, Z., and Robbins, P. D. (1999) Oncogene 18, 239-247 48. Schuuring, E., Verhoeven, E., van Tinteren, H., Peterse, J. L., Nunnink, B., Thunnissen, F. B., Devilee, P., Cornelisse, C. J., van de Vijver, M. J., Mooi, W. J., and et al. (1992) Cancer Res 52, 5229-5234 49. Reutens, A. T., Fu, M., Wang, C., Albanese, C., McPhaul, M. J., Sun, Z., Balk, S. P., Janne, O. A., Palvimo, J. J., and Pestell, R. G. (2001) Mol Endocrinol 15, 797- 811 50. Ganter, B., Fu, S., and Lipsick, J. S. (1998) Embo J 17, 255-268 51. Bienvenu, F., Gascan, H., and Coqueret, O. (2001) J Biol Chem 276, 16840- 16847

111 52. Shao, Z., and Robbins, P. D. (1995) Oncogene 10, 221-228 53. Ratineau, C., Petry, M. W., Mutoh, H., and Leiter, A. B. (2002) J Biol Chem 277, 8847-8853

FIGURE LEGENDS

Figure 1. Cyclin D1-∆XMN fails to bind the AR both in vivo and in vitro. A, Models representing cyclin D1 wild type (WTD1) and mutant proteins utilized for binding analysis. B, Cyclin D1 wild type and mutant proteins were expressed and purified using a baculoviral expression system. The proteins were immunoprecipitated and incubated with radiolabeled AR generated by in vitro transcription/translation. Following incubation, the beads were washed extensively and subjected to SDS-PAGE in duplicate.

One gel (top) was dried and bound AR detected via autoradiography. The duplicate gel

(bottom) was transferred and immunoblotted for cyclin D1 wild type and mutant proteins.

C, COS-7 cells were transfected with either wild type (left panel) or cyclin D1-∆XMN

(right panel) and AR at a 1:1 ratio as described in Materials and Methods. 24 h post- transfection, cells were harvested and lysates subjected to immunoprecipitation with antibody directed against either cyclin D1, AR, or non-specific (NS) protein. Following binding, reactions were washed and prepared for SDS-PAGE analysis. Transferred gels were cut in half and immunoblotted for AR and cyclin D1.

Figure 2. Cyclin D1-∆XMN demonstrates compromised AR co-repressor activity.

A, CV1 cells were seeded in steroid free conditions and transfected with PSA-LUC

112 reporter, AR, β-galactosidase, and wild type cyclin D1 (WTD1) or cyclin D1-∆XMN

(∆XMN). Post transfection, cells were treated with 0.1 nM DHT or 0.1% ethanol vehicle

for 24 h. Cells were harvested and analyzed for β-galactosidase and luciferase activity.

AR activity in the absence of ligand (ethanol; ETOH) was set to “1” with the average fold induction by DHT presented. Error bars represent the standard deviation. Experiments were performed at least 3 times in triplicate. B, Cells were transfected and treated as in A with H2B-GFP in place of CMV-β-galactosidase. The mock sample was not transfected.

Following treatment, cells were harvested and subjected to SDS-PAGE followed by immunoblotting for AR, GFP and cyclin D1 (wild type and ∆XMN). C, Wild type or

cyclin D1-∆XMN was transfected into COS-7 cells along with H2B-GFP. 24 hours post- transfection, cells were harvested, fixed, and stained with propidium iodide to detect

DNA content. Both GFP positive and GFP negative cells (mock) were analyzed via flow cytometry for changes in cell cycle profile. These experiments were performed in duplicate and the estimated percentage of cells in each phase of the cell cycle determined

using ModFIT software.

Figure 3. The central domain of cyclin D1 is required in its entirety to bind and

regulate AR activity. A, Schematic representations of cyclin D1 deletion mutants

internal to a.a.142-253 utilized in binding and functional assays. B, COS-7 cells were

transfected with wild type or cyclin D1 mutant proteins as described in Materials and

Methods. Post-transfection, cells were harvested and lysates incubated on a column of

GST-AR1-660. Following binding, the reactions were washed extensively and input (I)

and bound (B) fractions subjected to SDS-PAGE analysis. The gels were transferred, cut

113 in half, and immunoblotted for AR and cyclin D1. C, CV1 cells were seeded, transfected,

and treated as in Figure 2A. Samples were harvested and analyzed for β-galactosidase and luciferase activity. Bars represent the average percent inhibition of wild type AR activity. Error bars represent standard deviation. Experiments were performed at least 3 times in triplicate.

Figure 4. Structural analysis of cyclin D1 and the AR N-terminus. A, Shown is a two-dimensional model of cyclin D1 structure as predicted by sequence to structure alignment and described in Materials and Methods. Arrows and braids respectively represent β-strands and α-helices and the relative solvent accessibility of each amino acid residue is depicted by shaded boxes. Black boxes indicate buried residues, whereas white boxes represent fully exposed residues. Amino acids composing the cyclin D1 central domain are highlighted in blue and the residues in contact with Cdk6 in red. B, The overall three-dimensional structure of the cyclin D1 homologue, cyclin K, is shown in complex with CDK6 (PDB code 1G3N) as defined using the RASMOL program (38).

CDK6 is shown in blue, the N-terminal HN1 helix and the cyclin box domain in red, and the cyclin K counterpart to the cyclin D1 central domain (CD) in yellow. Note that cyclin K lacks the C-terminal signaling domain that extends from the last helix of the central domain in cyclin D1 (compare with panel A). C, A two-dimensional model of the

AR N-terminus (a. a. 1-50) generated as described in Materials and Methods. The secondary structures and solvent accessibilities are represented as in A.

114 Figure 5. The central domain is sufficient for AR binding and inhibition. A, GST-

central domain (GST-CD), GST-cyclin D1 (wild type) and GST were generated and

purified as described in Materials and Methods. In vitro translated AR was incubated on the columns and then washed to remove the unbound fraction. Input and bound fractions were subject to SDS-PAGE analysis and visualized on a phosphoimager (top panel). The lower panel represents coomassie staining of beads utilized for binding analysis. B, upper panel, COS-7 cells were transfected, stimulated with 1 nM DHT for 48 h, and harvested for reporter assay as in Figure 2A. Increasing concentrations of GFP and GFP- central domain (GFP-CD) were added to the reactions as indicated (0.25, 0.50, 1.0 µg, respectively). Lysates were analysed for luciferase activity and normalized for transfection efficiency. Data shown represent the average of three independent experiments. lower panel, COS-7 cells transfected with GFP-CD or GFP alone were subject to SDS-PAGE followed by immunoblotting to verify protein expression and integrity. C, LNCaP cells were transfected with GFP or GFP-CD as indicated. 48 h post-transfection, cells were labeled overnight with BrdU, fixed, and stained. Histograms represent data from at least two independent experiments performed in triplicate with error bars indicating the standard deviation. D, COS-7 cells were transfected as in Figure

1C, with plasmids encoding HDAC3 and GFP-CD at a 1:1 DNA ratio. 48 h post- transfection, cells were harvested and subjected to immunoprecipitation with antibodies directed against GFP-CD (GFP), HDAC3, and non-specific protein (NS). Following binding, the reactions were washed extensively and then analysed by SDS-PAGE followed by immunoblotting as indicated.

115 Figure 6. Cyclin D1 nuclear receptor co-repressor activity is distinct from its co-

activator function. A, CV-1 cells were transfected as described in Material and

Methods with 0.5 µg CMV-β-galactosidase, 0.5 µg TRβ1, 0.5 µg Palaa-LUC reporter, and 2.0 µg wild type (WT) or cyclin D1-∆XMN. Post-transfection, cells were stimulated for 24 h with 10nM T3 or 0.1% NaOH (sodium hydroxide (1M); vehicle). Reporter assays were harvested and monitored for β-galactosidase and luciferase activity.

Receptor activity in the presence of vehicle was set to “1”. Bars represent the average fold induction and error bars show the standard deviation. Experiments were performed at least three times in triplicate with asterisks indicating statistical significance (p<0.05).

B, CV-1 cells were transfected as in A with 0.5 µg CMV-β-galactosidase, 0.5 µg ERα,

0.75 µg 3XERE-LUC reporter, and 1.5 µg wild type (WTD1) or cyclin D1-∆XMN.

Following transfection, the cells were washed and stimulated for 48 h with either 10nM

17-β-estradiol (E2) or ethanol (EtOH) vehicle. Cells were harvested, analyzed, and reported as in A. Experiments were performed at least three times in triplicate. Asterisks indicate a statistical significance of p<0.05.

116 A. LxxLL 1 295 WTD1 CYCLIN BOX PEST 107 295 ∆1-99 1 234 ∆61 1 142 254 295 ∆XMN

B. Input WTD1 ∆1-99 ∆61 ∆XMN CD44 AR CD44 AR CD44AR CD44 AR CD44 AR KDa 115 115 -AR 93 93 Bound to Cyclin D1 -CD44

49.8 49.8

35.7 IB:Cyclin D1 29.2

20.9 1234567 8 117 Figure 1A, B C. IP: IP:

R S 1 R S 1 A N D A N D α α α α α α KDa KDa AR - 120 AR - 120 100 100

38.3 WTD1 - ∆XMN - 29.6

123 456

118 Figure 1C A. B.

20 XMN Mock WTD1 ∆ KDa 18 0.1% ETOH 120 AR - 16 0.1 nM DHT 100 14 12 H2B GFP - 38.3 10 38.3 8 WTD1 - 6 4 29.6 Relative LUC Activity 2 ∆XMN - 0 123 Vector WTD1 ∆XMN C. 250

600 Mock 300 WTD1 200 ∆XMN G0/G1: 32.4 ± 2.4% G0/G1: 32.2 ± 0.7% G0/G1: 27.7 ± 0.2% S: 41.7 ± 2.3% S: 46.2 ± 0.3% 150 S: 42.2 ± 8.4% 400 G2/M: 25.9 ± 1.2% 200 G2/M: 24.2 ± 7.3% G2/M: 30.1 ± 8.6% 100 # Cells # Cells # Cells 200 100 50

0 0 0 0 200 400 600 800 1000 0 200 400 600 800 1000 0 200 400 600 800 1000 PI Intensity PI Intensity PI Intensity

119 Figure 2A, B A. B.

1LxxLL 295 WT CYCLIN BOX PEST WT (no AR) WT ∆XMN ∆142-152 ∆152-174 ∆174-253 1 295 IB IB IB IB IB IB ∆XMN KDa 124 GST-AR 1-660 - 1 295 101 ∆142-152 35.7 1 295 Anti- ∆152-174 CycD1 29.2

1 295 12 34 56 78 9101112 ∆174-253

120

100

% Inhibition % 40

0 WT D1 ∆XMN ∆142- ∆152- ∆174- 152 174 253

120 Figure 3A,B, C A. B.

CD HN1 Cdk6

Cyclin Box

Figure 4 121 A. B. 10 0.1% ETOH 9 CD 8 0.1nM DHT INPUT GST-WTGST- D1 GST 7 KDa 6 120 AR - 5

100 4 3 2

55.9 Relative LUC Activity 1 Coomassie 0 38.2 GFP GFP-CD 29.7

1234 55.9 GFP-CD- GFP- 29.7 1 2 C. D. 20 18 IP: 16 3 14 C P A F S D 12 G N H α α α 10 KDa 8 HDAC3 - 55.9 6 4 55.9 GFP-CD - % BrdU% Incorporation 2 0 GFP GFP-CD 1 23 Figure 5 122 A.TRβ1 B. 8 9 ERα * 8 7 0.1% NaOH 0.1% EtOH * 7 10nM E2 6 10nM T3 * 6 5 5 4 4 3 * 3 2 2

1 Relative LUC Activity 1 Relative LUC Activity 0 0 Vector WTD1 ∆XMN Vector WTD1 ∆XMN

123 Figure 6 Chapter VI:

Summary and Conclusions

Characterization of Cyclin D1 Action

Our initial findings suggested that cyclin D1 was a key regulator of

prostatic adenocarcinoma cell growth, functioning to modulate both cell cycle

progression and AR transactivation (1). In this thesis, we extensively examine

the mechanism and specificity of cyclin D1 action. We find in Chapters II and IV

that cyclin D1 directly binds the AR N-terminus to inhibit the activity of both AF-1 and AF-2 ligand dependent transactivation functions. This activity is extremely potent as cyclin D1 maintains its ability to repress AR-mediated transactivation of

the PSA promoter even in the presence of ectopic co-activator expression (e.g.

SRC-1, P/CAF, p300, ARA70) (Chapter II, Fig. 5). Numerous AR co-activators

have been shown to bind the AR N-terminus including, but not exclusive to: SRC-

1, ARA70, CBP, BRCA-1, RB, cyclin E, and TIF2 (2-7). Thus, our findings imply

that co-activator association with the AR N-terminus fails to disrupt cyclin D1

interaction. In addition, our data suggest that cyclin D1 co-repressor activity is

dominant to both type I and type II co-activator function, although the ligand

independent AR transactivation (through AF-5) is not subject to cyclin D1

regulation (Chapter II). Interestingly, several additional co-repressors are known

to bind the AR N-terminus including: silencing mediator of retinoic acid and

thyroid hormone receptors (SMRT), ErbB-3 binding protein 1 (Ebp1), and amino-

terminal enhancer of split (AES) (8-10). Although the mechanism by which these

co-repressors inhibit AR transactivation remains somewhat speculative, it is

124 hypothesized that they serve as competitors for co-activator binding and may, in addition, disrupt N- to C-terminal receptor interactions (11). In fact, Ebp1 interaction and regulation of the AR requires an intact LxxLL motif, similar to that employed by SRC-1 (8). Thus, in many ways the functional characterization of such N-terminal AR co-repressors appears to parallel that which we observe with cyclin D1, however, few studies to date have examined the actions of these proteins in depth.

The ability of cyclin D1 to regulate AR activity is likely to play a significant role in prostate cancer development and progression. The AR is known to mediate growth, differentiation, and secretion within the normal prostate (12). In addition, deregulation of AR activity is implicated in the development of androgen independent prostate cancers (13). To characterize the role of cyclin D1 in androgen dependent growth, we examined its function in the presence of multiple factors thought to be involved in the development and progression of prostate cancer (Chapter III). At the highest risk for developing prostate cancer is the

African American population, who harbor a contracted N-terminal AR polyglutamine repeat (14, 15). It is logical to hypothesize that cyclin D1 regulation of the AR is compromised in this population, as we mapped the cyclin

D1-AR interaction to the N-terminus of the receptor (Chapter II Fig. 3 and

Chapter IV Fig. 2). However, our findings illustrate that cyclin D1 co-repressor function is maintained irrespective of polyglutamine tract length (Chapter III,

Figure 5). In androgen independent prostate cancers, several additional mechanisms of AR deregulation have been proposed including AR gene

125 amplification, co-activator over expression, ligand binding domain mutation, and receptor activation by high level cytokines and growth factors (16). Mutation of the AR ligand-binding pocket is frequently (up to 50%) observed in patients undergoing anti-androgen therapy (17). Such mutation leads to promiscuous and inappropriate AR activation by non-canonical ligands (e.g. progesterone, 17-

βestradiol, corticosteroids) (18). Interestingly, cyclin D1 co-repressor activity is unaffected by clinically relevant AR ligand-binding pocket mutations (Chapter III,

Fig. 6). In addition, activation of the AR T877A allele by progesterone is abrogated by ectopic cyclin D1 expression, demonstrating that cyclin D1 co- repressor activity is maintained in the presence of non-canonical AR ligands

(Chapter III, Fig. 7). Still, few studies have examined the role of N-terminal AR mutations in prostate cancer progression. Thus, in future studies it will be necessary to determine the role, if any, of N-terminal AR mutations in cyclin D1 co-repressor activity.

In addition to receptor mutation, AR gene amplification is observed in up to 30% of androgen independent prostate cancers, and represents one potential mechanism by which cyclin D1 co-repressor activity may be subverted (19, 20).

In addition, a recent study by Chen et al. shows that inappropriate AR activity as achieved through increased receptor expression leads to the development of a hormone refractory phenotype (13). In Chapter III Fig. 1, we illustrate that full cyclin D1 co-repressor activity is achieved at a 1:1 molar ratio with the AR.

Indeed, titration assays indicate that lower levels of cyclin D1 fail to completely inhibit AR transactivation (Introduction, Figure V), suggesting that subversion of

126 the cyclin D1 feedback mechanism may occur with increased receptor expression (1). In addition, AR co-activator over expression is observed frequently in androgen independent cell lines and tumors. However, our data suggest that cyclin D1 co-repressor function is dominant to increased co- activator expression (Chapter I, Fig. 5). Supporting a model wherein AR regulation by cyclin D1 is maintained in spite of co-activator over expression, we show that cyclin D1 is an effective AR co-repressor in the context of several cell types (including androgen independent prostate cancers) with differential co- activator expression patterns (Chapter III, Fig. 1&2). These cell type specificity studies also suggest that the co-repressor function of cyclin D1 requires readily available co-factors, common to all cell types examined. In contrast to cyclin D1, the activity of other known AR co-modulators is highly dependent upon cellular background. For example, the DAX-1 co-repressor function is diminished in

HeLa (cervical carcinoma) cells in comparison with that observed in the COS-7

(green monkey kidney epithelial) cell type (21). Interestingly, AR activation by non-ligand growth factors and cytokines is also reported to involve cell type specific co-factors (22-24). Our initial findings indicate that cyclin D1 co- repressor activity is dominant to IL-6-mediated receptor activation, however we failed to observe non-ligand stimulation of the receptor in the presence of EGF and forskolin as previously published (Chapter III, Fig. 7 and data not shown).

These initial characterization studies reveal a potent cyclin D1 co-repressor function capable of regulating AR action in the context of polymorphisms, gene mutations, and non-conventional ligands/non-ligand activators known to facilitate

127 prostate cancer progression and androgen independence. Still, much of the

biological characterization of cyclin D1 action remains to be addressed in vivo.

Our initial experiments suggest that ectopic cyclin D1 expression would

prevent the growth of androgen dependent tumors in vivo (Chapters I and V). To begin to assess the biological consequence of cyclin D1 action, androgen dependent prostatic adenocarcinoma (LNCaP) cells were employed.

Interestingly, ectopic cyclin D1 or cyclin D1-KE (a mutant defective in CDK 4 binding 25) significantly inhibited LNCaP cell cycle progression, demonstrating that cyclin D1 co-repressor activity is capable of regulating AR-dependent tumor cell growth (Chapter I, Fig. 7). Still, further studies will be required in order to determine the consequence of ectopic cyclin D1 expression in additional androgen dependent and independent cell lines. Furthermore, it would be of interest to examine the functionality of cyclin D1-AR regulatory axis in patients wherein therapeutic treatment may select for tumors with deregulated growth. In addition to these studies, the role of cyclin D1 in prostatic development should be examined. Knockout mice are already available that fail to express cyclin D1, although the prostate of these mice has yet to be examined in detail for abnormalities (26, 27). However, evidence suggests that in other cyclins (cyclin

D2, D3 and cyclin E) may compensate for the loss of cyclin D1 (28-30). In fact, we have previously shown that cyclin D3 also possesses AR co-repressor

activity, although the mechanism and consequence of such action remains

elusive (1). In addition, dogs and humans are the only two species known to naturally acquire prostate cancer (31). Therefore, mouse knockout models may

128 not clearly parallel what is seen in human prostate cancer patients. Clearly, such patient studies will be needed in the future to properly characterize the role of the

cyclin D1-AR axis in prostate cancer development and progression.

Mechanism of Cyclin D1 co-repressor activity

Many of the over 16 AR co-repressor identified to date were first identified

based upon their ability to modulate type II NR function. Type II nuclear

receptors, including TR, retinoic acid receptor (RAR), vitamin D receptor and

some orphan receptors, do not associate with heat shock proteins in the absence

of ligand (32). Instead, these NRs associate with co-repressors to block target

gene activation, until ligand becomes available (33-35). It is through the study of

these interactions that NR co-repressors were first identified. More recently, the

role of these same co-repressors in type I NR (e.g. ER, AR) function has been

examined. HDAC proteins represent a great majority of the NR co-repressors

identified to date and are grouped into 3 classes (types I-III) based upon their

sequence homology (36). The type I HDAC1 has been shown to bind the AR C-

terminus to inhibit PSA transactivation in a histone deacetylase dependent

manner (37). Still, the target of HDAC I activity remains unclear as both

chromatin remodeling (via regulation of nucleosome acetylation) and AR

transactivation (via acetylation of the DNA-binding domain) are dependent upon

acetylation status (37-40). We show that the central domain of cyclin D1 directly

interacts with HDAC3, and suggest that this interaction may be utilized to

regulate NR function (Chapter V, Fig. 5). Indeed, we find that addition of TSA

129 partially reverses cyclin D1 co-repression of the AR (Chapter II, Fig. 6).

Potentially, HDAC3 recruitment by cyclin D1 could antagonize the activity of type

I HAT co-activators, resulting in AR repression. However, it remains is unclear if

cyclin D1 serves to directly recruit histone deacetylase activity to the AR, or if

adaptor proteins such as NCoR and SMRT are required. In addition, cyclin D1

regulation of the Sp-1 transcription factor is known to involve direct binding to

TAFII250, a core component of the transcriptional machinery (41). Cyclin D1

interaction with TAFII250 may prevent the formation of active transcriptional

complexes at AR target genes by competing away required initiation factors (42).

Clearly, further investigation into the physiological interactions required for cyclin

D1 regulation of the AR is needed to identify additional co-factors required for its activity in vivo.

In addition to histone deacetylase activity, numerous identified AR co-

repressors serve to antagonize type II co-activator function through inhibition of

receptor N- to C- terminal interactions, DNA binding, and protein stability. The

AR co-repressor, calreticulin, is known to disrupt receptor DNA binding in the

presence of ligand (43). However, we do not observe such activity associated

with cyclin D1, as the AR effectively binds AREs in the presence of cyclin D1

(data not shown). In addition, the finding that AF-5 activity is not subject to cyclin

D1 regulation illustrates that the receptor is capable of DNA interactions in the

presence of ectopic cyclin D1 (Chapter I, Fig. 4). Thus, we investigated the role

of cyclin D1 in regulating receptor N- to C- terminal interactions. The interaction

of the AR N- and C- termini is perceived to stabilize ligand binding and promote

130 an active receptor conformation (7, 44, 45). Three LxxLL-like motifs, 23FxxLF27,

179IXXIL183 and 432WXXLF436, in the AR N-terminus mediate interaction with the

C-terminus and may be bridged by type II co-activators (45, 46). Binding of the

AR co-repressors DAX-1, p53, and SMRT has been hypothesized to disrupt N- to

C- terminal receptor interactions, causing dramatic decreases in AR

transactivation potential (21, 47, 48).

NCoR/ ? In chapter IV, we show that cyclin D1 1 SMRT? 42 -25 3 HDAC3 Cyclin D1 is a potent repressor of AR N- to C- terminal interactions and that this N C N activity most likely involves

AR AR disruption of the FxxLF function. ARE Figure 1. Cyclin D1 uses two distinct Thus cyclin D1 utilizes two distinct mechanisms to inhibit AR transactivation. mechanisms to antagonize both type

I and type II AR co-activators (Figure 1). In addition to disruption of N- to C- terminal interactions, binding of AR co-repressor to the receptor N-terminus is hypothesized to prevent co-activator association with the AF-1 transactivation function. Such competition has been observed between the AR co-activator

Tip60 and HDAC1 implying that over expression of AR co-activators in advanced prostate cancer may stifle co-repressor function (37). However, our initial studies indicate that cyclin D1 co-repressor activity is dominant to co-activator function

(Chapter II, Fig. 5). Still, titration experiments need to be performed in order to determine the relative strength of cyclin D1 action. Taken together, our mechanistic data illustrate that cyclin D1 is a potent and unique AR co-repressor,

131 serving to target receptor transactivation through at least two distinct mechanisms. These findings suggest that novel therapeutics mimicking the action of cyclin D1 may prove useful in the treatment of both androgen dependent and independent prostate cancers.

Cyclin D1 as a transcriptional regulator

While CDK independent cyclin D1 action is linked to the regulation of numerous transcription factors, the regions responsible for such transcriptional modification are in general poorly defined. Although co-activation of ERα activity by cyclin D1 was originally linked to co-activator recruitment, recent evidence suggests that the effect of cyclin D1 over expression noted in mammary carcinomas is both CDK and ER independent (49-52). Instead, these data implicate the transcription factor, C/EBPβ, as the major target of cyclin D1 co- activator action (49). Still, several major caveats exists in these studies, as genetic analysis of cyclin D1 over expressing cell lines may fail to identify genes

A. Cyclin D1 co-repressor function B. Cyclin D1 co-activator function

NCoR/ ? 1 42 14 SMRT? -25 2-2 3 53 HDAC3 C/EBPβ L L Cyclin D1 x TF TF + + x Cyclin D1 L RE Co-Act

RE RE

Figure 2. Proposed model demonstrating the distinct transcriptional regulator functions of cyclin D1

132 repressed by endogenous cyclin D1 levels. In these cases, the role of C/EBPβ

remains unclear. Since our data support a model wherein cyclin D1 co-repressor

and co-activator functions are distinct and separable, it remains possible that

C/EBPβ involvement is function specific (See Figure 2). Supporting such as

model, cyclin D1 co-repression of PPARγ induced gene transcription was

recently reported to be independent of C/EBPβ function (53). However, we have yet to address the role of C/EBPβ in AR/TRβ1 co-repression and the cyclin D1 domain responsible for C/EBPβ interaction remains elusive.

Although previous reports suggest that both Dmp1 and PPARγ directly bind the cyclin D1 central domain, these studies fail to investigate the mechanistic function of this region (53, 54). Thus, our findings demonstrating

LxxLL 1295

Cyclin D1 CYCLIN BOX PEST

1 142 152 295 ∆142-152 1 174 253 295 ∆174-253 ∆152-253 1 152 253 295

300 120 ∆142-152 150 ∆174-253 ∆152-253 90 G0/G1: 31.1% G0/G1: 14.5% G0/G1: 16.0% 200 S: 43.4% S: 54.29% S: 47.9% G2/M: 25.6 % 100 G2/M: 31.24% G2/M: 36.1% 60 # Cells # Cells # Cells 100 50 30

0 0 0 0 200 400 600 800 1000 0 200 400 600 800 1000 0 200 400 600 800 1000 PI Intensity PI Intensity PI Intensity

Figure 3. Internal deletion of cyclin D1 results in cell cycle abnormalities. COS-7 cells were transfected with the cyclin D1 mutants shown and analyzed as in Chapter V Fig. 2. Cyclin D1 mutants possessing a complete cyclin box (including residues 142-152), but lacking a complete central domain show marked cell cycle perturbation.

133 that HDAC3 is capable of associating with the cyclin D1 central domain are the

first to implicate a mechanism behind its sufficiency (Chapter V Fig. 5).

Interestingly, TRβ1 co-repression by cyclin D1 requires the central domain and has been previously linked to HDAC3 recruitment (55). Therefore, it remains possible that cyclin D1 co-repressor activity is also conserved with respect to other transcription factors including Dmp1 and PPARγ. Further studies will be needed to assess the functional conservation of central domain co-repressor activity. Unfortunately, we have observed that perturbation of this region affects

CDK-dependent cyclin D1 functions, causing inappropriate cell cycle transitions

(Figure 3). Thus, CDK defective cyclin D1 mutant alleles are required for future analyses of transcriptional regulation.

Summary

In summary, we propose that cyclin D1 co-repressor activity is essential for proper AR regulation. Cyclin D1 is a negative feedback inhibitor of the AR functioning through both CDK-dependent and independent mechanisms (Figure

4). As a regulator of CDK action, cyclin D1 is induced by androgen to drive cell cycle progression. Following induction, increased levels of cyclin D1 may then feedback to inhibit further, inappropriate mitogenesis through CDK independent mechanisms including the disruption of N- to C-terminal receptor interactions and recruitment of histone deacetylase activity. Cyclin D1 co-repressor activity is extremely potent and appears to be mediated through its central domain. The requirement of this central domain for NR co-repression is conserved and may

134 involve histone deacetylase Androgen

N recruitment. However, co-activator NC NC 2 AR AR activity is mediated through a

NCoR/ ? 14 SMRT? 2-2 53 HDAC3 CycD1 distinct region of cyclin D1 and

Cyclin D1 appears to contribute to mammary

1 tumorigenesis through interaction

CycD1 CDK4 + with the C/EBPβ transcription P P RB RB CycA CDK2 P + factor. Together these results CycA Cell Cycle Progression provide the impetus to study the

Figure 4. Summary of cyclin D1 action in role of the cyclin D1 central prostate cancer. We find that cyclin D1 regulates prostate cancer growth through two distinct mechanisms. domain in prostate cancer progression and suggest that protein mimetics may prove useful in the development of novel therapeutics.

References:

1. Knudsen, K. E., Cavenee, W. K., and Arden, K. C. D-type cyclins complex with the androgen receptor and inhibit its transcriptional transactivation ability. Cancer Res, 59: 2297-2301, 1999. 2. Lu, J. and Danielsen, M. Differential regulation of androgen and glucocorticoid receptors by retinoblastoma protein. J Biol Chem, 273: 31528-31533, 1998. 3. Yamamoto, A., Hashimoto, Y., Kohri, K., Ogata, E., Kato, S., Ikeda, K., and Nakanishi, M. Cyclin E as a coactivator of the androgen receptor. J Cell Biol, 150: 873-880, 2000. 4. Park, J. J., Irvine, R. A., Buchanan, G., Koh, S. S., Park, J. M., Tilley, W. D., Stallcup, M. R., Press, M. F., and Coetzee, G. A. Breast cancer susceptibility gene 1 (BRCAI) is a coactivator of the androgen receptor. Cancer Res, 60: 5946-5949, 2000. 5. Zhou, Z. X., He, B., Hall, S. H., Wilson, E. M., and French, F. S. Domain interactions between coregulator ARA(70) and the androgen receptor (AR). Mol Endocrinol, 16: 287-300, 2002.

135 6. Bevan, C. L., Hoare, S., Claessens, F., Heery, D. M., and Parker, M. G. The AF1 and AF2 domains of the androgen receptor interact with distinct regions of SRC1. Mol Cell Biol, 19: 8383-8392, 1999. 7. Ikonen, T., Palvimo, J. J., and Janne, O. A. Interaction between the amino- and carboxyl-terminal regions of the rat androgen receptor modulates transcriptional activity and is influenced by nuclear receptor coactivators. J Biol Chem, 272: 29821-29828, 1997. 8. Zhang, Y., Fondell, J. D., Wang, Q., Xia, X., Cheng, A., Lu, M. L., and Hamburger, A. W. Repression of androgen receptor mediated transcription by the ErbB-3 binding protein, Ebp1. Oncogene, 21: 5609- 5618, 2002. 9. Dotzlaw, H., Moehren, U., Mink, S., Cato, A. C., Iniguez Lluhi, J. A., and Baniahmad, A. The amino terminus of the human AR is target for corepressor action and antihormone agonism. Mol Endocrinol, 16: 661- 673, 2002. 10. Yu, X., Li, P., Roeder, R. G., and Wang, Z. Inhibition of androgen receptor-mediated transcription by amino-terminal enhancer of split. Mol Cell Biol, 21: 4614-4625, 2001. 11. Heinlein, C. A. and Chang, C. Androgen receptor (AR) coregulators: an overview. Endocr Rev, 23: 175-200, 2002. 12. Jenster, G. The role of the androgen receptor in the development and progression of prostate cancer. Semin Oncol, 26: 407-421, 1999. 13. Chen, C. D., Welsbie, D. S., Tran, C., Baek, S. H., Chen, R., Vessella, R., Rosenfeld, M. G., and Sawyers, C. L. Molecular determinants of resistance to antiandrogen therapy. Nat Med, 10: 33-39, 2004. 14. Hakimi, J. M., Schoenberg, M. P., Rondinelli, R. H., Piantadosi, S., and Barrack, E. R. Androgen receptor variants with short glutamine or glycine repeats may identify unique subpopulations of men with prostate cancer. Clin Cancer Res, 3: 1599-1608, 1997. 15. Giovannucci, E., Stampfer, M. J., Krithivas, K., Brown, M., Dahl, D., Brufsky, A., Talcott, J., Hennekens, C. H., and Kantoff, P. W. The CAG repeat within the androgen receptor gene and its relationship to prostate cancer. Proc Natl Acad Sci U S A, 94: 3320-3323, 1997. 16. Culig, Z., Klocker, H., Bartsch, G., and Hobisch, A. Androgen receptors in prostate cancer. Endocr Relat Cancer, 9: 155-170, 2002. 17. Taplin, M. E., Bubley, G. J., Shuster, T. D., Frantz, M. E., Spooner, A. E., Ogata, G. K., Keer, H. N., and Balk, S. P. Mutation of the androgen- receptor gene in metastatic androgen-independent prostate cancer. N Engl J Med, 332: 1393-1398, 1995. 18. Balk, S. P. Androgen receptor as a target in androgen-independent prostate cancer. Urology, 60: 132-138; discussion 138-139, 2002. 19. Koivisto, P., Kononen, J., Palmberg, C., Tammela, T., Hyytinen, E., Isola, J., Trapman, J., Cleutjens, K., Noordzij, A., Visakorpi, T., and Kallioniemi, O. P. Androgen receptor gene amplification: a possible molecular mechanism for androgen deprivation therapy failure in prostate cancer. Cancer Res, 57: 314-319, 1997.

136 20. Visakorpi, T., Hyytinen, E., Koivisto, P., Tanner, M., Keinanen, R., Palmberg, C., Palotie, A., Tammela, T., Isola, J., and Kallioniemi, O. P. In vivo amplification of the androgen receptor gene and progression of human prostate cancer. Nat Genet, 9: 401-406, 1995. 21. Holter, E., Kotaja, N., Makela, S., Strauss, L., Kietz, S., Janne, O. A., Gustafsson, J. A., Palvimo, J. J., and Treuter, E. Inhibition of androgen receptor (AR) function by the reproductive orphan nuclear receptor DAX-1. Mol Endocrinol, 16: 515-528, 2002. 22. Hobisch, A., Eder, I. E., Putz, T., Horninger, W., Bartsch, G., Klocker, H., and Culig, Z. Interleukin-6 regulates prostate-specific protein expression in prostate carcinoma cells by activation of the androgen receptor. Cancer Res, 58: 4640-4645, 1998. 23. Ueda, T., Mawji, N. R., Bruchovsky, N., and Sadar, M. D. Ligand- independent activation of the androgen receptor by interleukin-6 and the role of steroid receptor coactivator-1 in prostate cancer cells. J Biol Chem, 277: 38087-38094, 2002. 24. Lin, D. L., Whitney, M. C., Yao, Z., and Keller, E. T. Interleukin-6 induces androgen responsiveness in prostate cancer cells through up-regulation of androgen receptor expression. Clin Cancer Res, 7: 1773-1781, 2001. 25. Hinds, P. W., Dowdy, S. F., Eaton, E. N., Arnold, A., and Weinberg, R. A. Function of a human cyclin gene as an oncogene. Proc Natl Acad Sci U S A, 91: 709-713, 1994. 26. Fantl, V., Stamp, G., Andrews, A., Rosewell, I., and Dickson, C. Mice lacking cyclin D1 are small and show defects in eye and mammary gland development. Genes Dev, 9: 2364-2372, 1995. 27. Sicinski, P., Donaher, J. L., Parker, S. B., Li, T., Fazeli, A., Gardner, H., Haslam, S. Z., Bronson, R. T., Elledge, S. J., and Weinberg, R. A. Cyclin D1 provides a link between development and oncogenesis in the retina and breast. Cell, 82: 621-630, 1995. 28. Chen, B. and Pollard, J. W. Cyclin D2 compensates for the loss of cyclin D1 in estrogen-induced mouse uterine epithelial cell proliferation. Mol Endocrinol, 17: 1368-1381, 2003. 29. Geng, Y., Whoriskey, W., Park, M. Y., Bronson, R. T., Medema, R. H., Li, T., Weinberg, R. A., and Sicinski, P. Rescue of cyclin D1 deficiency by knockin cyclin E. Cell, 97: 767-777, 1999. 30. Ciemerych, M. A., Kenney, A. M., Sicinska, E., Kalaszczynska, I., Bronson, R. T., Rowitch, D. H., Gardner, H., and Sicinski, P. Development of mice expressing a single D-type cyclin. Genes Dev, 16: 3277-3289, 2002. 31. Waters, D. J., Sakr, W. A., Hayden, D. W., Lang, C. M., McKinney, L., Murphy, G. P., Radinsky, R., Ramoner, R., Richardson, R. C., and Tindall, D. J. Workgroup 4: spontaneous prostate carcinoma in dogs and nonhuman primates. Prostate, 36: 64-67, 1998. 32. Horlein, A. J., Naar, A. M., Heinzel, T., Torchia, J., Gloss, B., Kurokawa, R., Ryan, A., Kamei, Y., Soderstrom, M., Glass, C. K., and et al. Ligand-

137 independent repression by the thyroid hormone receptor mediated by a nuclear receptor co-repressor. Nature, 377: 397-404, 1995. 33. Huang, E. Y., Zhang, J., Miska, E. A., Guenther, M. G., Kouzarides, T., and Lazar, M. A. Nuclear receptor corepressors partner with class II histone deacetylases in a Sin3-independent repression pathway. Genes Dev, 14: 45-54, 2000. 34. Nagy, L., Kao, H. Y., Chakravarti, D., Lin, R. J., Hassig, C. A., Ayer, D. E., Schreiber, S. L., and Evans, R. M. Nuclear receptor repression mediated by a complex containing SMRT, mSin3A, and histone deacetylase. Cell, 89: 373-380, 1997. 35. Chen, J. D. and Evans, R. M. A transcriptional co-repressor that interacts with nuclear hormone receptors. Nature, 377: 454-457, 1995. 36. Gray, S. G. and Ekstrom, T. J. The human histone deacetylase family. Exp Cell Res, 262: 75-83, 2001. 37. Gaughan, L., Logan, I. R., Cook, S., Neal, D. E., and Robson, C. N. Tip60 and histone deacetylase 1 regulate androgen receptor activity through changes to the acetylation status of the receptor. J Biol Chem, 277: 25904-25913, 2002. 38. Collingwood, T. N., Urnov, F. D., and Wolffe, A. P. Nuclear receptors: coactivators, corepressors and chromatin remodeling in the control of transcription. J Mol Endocrinol, 23: 255-275, 1999. 39. Thomas, M., Dadgar, N., Aphale, A., Harrell, J. M., Kunkel, R., Pratt, W. B., and Lieberman, A. P. Androgen receptor acetylation site mutations cause trafficking defects, misfolding, and aggregation similar to expanded glutamine tracts. J Biol Chem, 279: 8389-8395, 2004. 40. Fu, M., Rao, M., Wang, C., Sakamaki, T., Wang, J., Di Vizio, D., Zhang, X., Albanese, C., Balk, S., Chang, C., Fan, S., Rosen, E., Palvimo, J. J., Janne, O. A., Muratoglu, S., Avantaggiati, M. L., and Pestell, R. G. Acetylation of androgen receptor enhances coactivator binding and promotes prostate cancer cell growth. Mol Cell Biol, 23: 8563-8575, 2003. 41. Adnane, J., Shao, Z., and Robbins, P. D. Cyclin D1 associates with the TBP-associated factor TAF(II)250 to regulate Sp1-mediated transcription. Oncogene, 18: 239-247, 1999. 42. Coqueret, O. Linking cyclins to transcriptional control. Gene, 299: 35-55, 2002. 43. Dedhar, S., Rennie, P. S., Shago, M., Hagesteijn, C. Y., Yang, H., Filmus, J., Hawley, R. G., Bruchovsky, N., Cheng, H., Matusik, R. J., and et al. Inhibition of nuclear hormone receptor activity by calreticulin. Nature, 367: 480-483, 1994. 44. He, B., Kemppainen, J. A., Voegel, J. J., Gronemeyer, H., and Wilson, E. M. Activation function 2 in the human androgen receptor ligand binding domain mediates interdomain communication with the NH(2)-terminal domain. J Biol Chem, 274: 37219-37225, 1999. 45. Alen, P., Claessens, F., Verhoeven, G., Rombauts, W., and Peeters, B. The androgen receptor amino-terminal domain plays a key role in p160

138 coactivator-stimulated gene transcription. Mol Cell Biol, 19: 6085-6097, 1999. 46. He, B., Kemppainen, J. A., and Wilson, E. M. FXXLF and WXXLF sequences mediate the NH2-terminal interaction with the ligand binding domain of the androgen receptor. J Biol Chem, 275: 22986-22994, 2000. 47. Liao, G., Chen, L. Y., Zhang, A., Godavarthy, A., Xia, F., Ghosh, J. C., Li, H., and Chen, J. D. Regulation of androgen receptor activity by the nuclear receptor corepressor SMRT. J Biol Chem, 278: 5052-5061, 2003. 48. Shenk, J. L., Fisher, C. J., Chen, S. Y., Zhou, X. F., Tillman, K., and Shemshedini, L. p53 represses androgen-induced transactivation of prostate-specific antigen by disrupting hAR amino- to carboxyl-terminal interaction. J Biol Chem, 276: 38472-38479, 2001. 49. Lamb, J., Ramaswamy, S., Ford, H. L., Contreras, B., Martinez, R. V., Kittrell, F. S., Zahnow, C. A., Patterson, N., Golub, T. R., and Ewen, M. E. A mechanism of cyclin D1 action encoded in the patterns of gene expression in human cancer. Cell, 114: 323-334, 2003. 50. Zwijsen, R. M., Wientjens, E., Klompmaker, R., van der Sman, J., Bernards, R., and Michalides, R. J. CDK-independent activation of estrogen receptor by cyclin D1. Cell, 88: 405-415, 1997. 51. Zwijsen, R. M., Buckle, R. S., Hijmans, E. M., Loomans, C. J., and Bernards, R. Ligand-independent recruitment of steroid receptor coactivators to estrogen receptor by cyclin D1. Genes Dev, 12: 3488- 3498, 1998. 52. Neuman, E., Ladha, M. H., Lin, N., Upton, T. M., Miller, S. J., DiRenzo, J., Pestell, R. G., Hinds, P. W., Dowdy, S. F., Brown, M., and Ewen, M. E. Cyclin D1 stimulation of estrogen receptor transcriptional activity independent of cdk4. Mol Cell Biol, 17: 5338-5347, 1997. 53. Wang, C., Pattabiraman, N., Zhou, J. N., Fu, M., Sakamaki, T., Albanese, C., Li, Z., Wu, K., Hulit, J., Neumeister, P., Novikoff, P. M., Brownlee, M., Scherer, P. E., Jones, J. G., Whitney, K. D., Donehower, L. A., Harris, E. L., Rohan, T., Johns, D. C., and Pestell, R. G. Cyclin D1 repression of peroxisome proliferator-activated receptor gamma expression and transactivation. Mol Cell Biol, 23: 6159-6173, 2003. 54. Inoue, K. and Sherr, C. J. Gene expression and cell cycle arrest mediated by transcription factor DMP1 is antagonized by D-type cyclins through a cyclin-dependent-kinase-independent mechanism. Mol Cell Biol, 18: 1590- 1600, 1998. 55. Lin, H. M., Zhao, L., and Cheng, S. Y. Cyclin D1 Is a Ligand-independent Co-repressor for Thyroid Hormone Receptors. J Biol Chem, 277: 28733- 28741, 2002.

139 Chapter VI:

Ongoing/Future Directions

Cyclin D1 Transcript [b]

Alternative splicing of the cyclin D1 mRNA has been linked to a single nucleotide

polymorphism (A870G) at the exon 4-intron 4 boundary (1). At this junction, individuals

harboring the A allele harbor an increased propensity for improper splicing, resulting in

production of a cyclin D1 mRNA containing intron Cyclin D1-A (294 a.a.) G PEST 4 (1). Translation of such alternative mRNAs cDNA 1 2345

LxxLL (termed transcript [b]) results in the production of a mRNA protein lacking the transcript [a] C-terminal Cyclin D1-B (274 a.a.) A regulatory domain (Figure 1). The C-terminus of cDNA 1 234

mRNA cyclin D1-A protein is involved in protein turnover Figure 1. Cyclin D1 polymorphism at nucleotide 870 causes the production as well as nuclear export (2, 3). Hence, deletion of of two distinct splice variant mRNAs. Cyclin D1-B contains sequences within these motifs in the transcript [b] form of cyclin D1 intron 4 and lacks the PEST and LxxLL motifs present in cyclin D1-A. was presumed to cause inappropriate cell cycle

regulation, possibly influencing the development and progression of numerous cancers.

In fact, some studies have begun to examine the role of this polymorphism in tumor

susceptibility, finding a potential link between A allele frequency and cancer

susceptibility, onset, and prognosis (4-7). Two independent laboratories studying the

Japanese population recently noted that an increased risk of developing both prostate

cancer and BPH is associated with the AA genotype (8, 9). Although these results may be attributed to the maintained cell cycle dependent functions of cyclin D1-B, the role of

140 alternative splicing in CDK-independent cyclin D1 function remains elusive (10).

Interestingly, the C-terminal co-activator (LxxLL) motif of cyclin D1-A is absent in the B isoform, yet the nuclear receptor co-repressor motif remains intact (See Figure 1).

We have already begun to assess the functional capacity of cyclin D1-B in reporter and co-immunoprecipitation assays. Our initial data suggest that the cyclin D1 co-repressor function remains intact with 8 7 0.1 % ETOH regards to AR regulation (see Figure 2). 6 0.1 nM DHT These results are not surprising, as the 5 4 central domain is intact in both cyclin D1 3 2 Relative LUC activity LUC Relative alleles. Still, further studies will be required 1 0 Vector to address the role of cyclin D1 Cyclin Cyclin D1-A D1-B polymorphism in vivo and confirm that CDK Figure 2. Reporter assay to detect cyclin D1-B co-repressor activity. COS- independent cyclin D1-B function fails to 7 cells were transfected via the calcium phosphate/BES protocol with PSA- luciferase reporter, CMV-β-galactosiadse influence prostate cancer susceptibility. In (for normalization of transfection efficiency), AR and/or cyclin D1 variants as order to perform these studies the dual roles indicated. Following washing, cells were treated overnight (18h, as indicated) and of cyclin D1 must first be uncoupled through subsequently harvested and analyzed. At a 3:1 ratio with the AR, both cyclin D1 variants significantly inhibit receptor development of a CDK-binding defective transactivation. Bars represent the average of at least 2 experiments cyclin D1-B allele. Using this strategy the performed in triplicate. Standard deviation is depicted using error bars. role of cyclin D1 polymorphism in additional nuclear receptor related cancers might also be discerned.

Identification of AR-Cyclin D1 complex members

141 Regulation of AR activity is documented to involve hundreds of cellular proteins, yet the transcriptional complex in the presence of cyclin D1 has yet to be elucidated.

Already our data suggest that cyclin D1 functions to facilitate AR interaction with the co- repressor, HDAC3 (Chapter V, Fig. 5). However, whether N-CoR or SMRT are required to bridge such an interaction remains unclear. In addition, it is unknown if co-activator association with the AR is blocked by cyclin D1. Through 2D gel electrophoresis and chromatin immunoprecipitation (ChIP) assays we hope to identify the members of the cyclin D1- AR complex both in vitro and in vivo. We have already begun to assess the components of this complex in vitro, using GST-cyclin D1 or GST-cyclin D1-KE (CDK binding defective (11) as bait for proteins derived from LNCaP cell lysate. The associated complexes are then analyzed by 2D gel electrophoresis followed by tandom mass spectroscopy of individual spots to identified bound proteins. Through these experiments we are likely to uncover components of AR-cyclin D1 complexes as well as other novel targets of cyclin D1 regulation in prostate cancer cells. To verify these findings in vivo, ChIP assays will be utilized to monitor the occupation of endogenous androgen responsive elements within the prostate specific antigen (PSA) promoter by cyclin D1-AR complexes. In addition, these assays may be utilized to monitor chromatin remodeling (through ChIP of acetylated histone) and potentially the activity of bound HDAC3.

Therapeutic Testing/Development

Our data suggest that cyclin D1 mimetics would have the potential to reduce

prostate cancer proliferation in vivo. In order to test this hypothesis, efficient gene

transduction mechanisms will be required. Numerous gene transfer systems have been

142 identified to date, including viral vectors, naked DNA injection, liposomal DNA

encapsulation, and DNA-protein conjugates. When choosing a delivery system, the

target cell type, transgene size, transduction efficiency, immune/inflammatory response

and therapy duration must be considered. DNA sequence encoding the cyclin D1

central domain is relatively small (~300 base pairs), making vector capacity a non-issue.

Although deletion of the cyclin box prevents inappropriate cell cycle stimuli, CDK

independent cyclin D1 functions may remain within the central domain. Thus, it is

important to consider the gene transfer specificity of each system to avoid non-specific cell targeting. In addition, our data indicate that cyclin D1 must be expressed at a 1:1 protein ratio with the AR for complete co-repressor activity (Chapter III, Fig. 1). As AR is frequently over expressed/amplified in prostate cancer, effective mimetics must reach high titers in target cells (12, 13). These considerations rule out inefficient transduction methodologies such as liposomal delivery, which is estimated to require 1,000+ liposomes/per cell (14). Of the current gene therapy systems available, adeno- associated virus (AAV) therapy appears most promising for the development and testing of therapeutic cyclin D1 mimetics. AAV has the capability to infect both dividing and non-dividing cell types, a property required for the efficient transduction of the characteristically slow growing prostate cancer cell phenotype (15). AAV integration is site-specific (localized to chromosome 19q14.3) offsetting the potential safety risks associated with random integration (e.g. in retroviral transduction), and elicits minimal immune/inflammatory response (16). In addition, AAV fails to undergo lytic replication in the absence of helper virus co-infection preventing viral spread to non-target tissues

(17). As the prostate can be directly accessed via bracytherapy, site-specific injection

143 of high titer AAV may be achieved in human patients, should an effective cyclin D1 therapy be developed in mouse models.

To begin to test the efficacy of cyclin D1-derived therapies in prostate cancer, the

GFP-tagged central domain or GFP alone will be cloned into the AAV vector, pAAV2

(18). Once cloned, AAV can be produced at high titer and used to infect LNCaP cells either in cell culture or mouse xenograph tumor models. These androgen dependent prostate cancer cells will then be monitored for proliferation in both cell culture and nude mice models. In addition, RNA from these samples can be utilized to compare gene expression profiles in control versus central domain expressing samples, thus allowing us to identify possible cytotoxic profiles, which may present in future in vivo studies. In particular, the effect of cyclin D1 central domain expression on normal cell growth will be important to address, as AAV infection shows little cell type specificity. Overall, cyclin D1-derived therapeutics are alone unlikely to be successful as androgen independent prostate cancers may already subvert this pathway through AR amplification. Combination therapy, wherein androgen ablation is supplemented by gene therapy, is prevalent among ongoing clinical trails and may represent the most effective use for cyclin D1-derived treatments.

References

1. Betticher, D. C., Thatcher, N., Altermatt, H. J., Hoban, P., Ryder, W. D., and Heighway, J. Alternate splicing produces a novel cyclin D1 transcript. Oncogene, 11: 1005-1011, 1995. 2. Diehl, J. A., Zindy, F., and Sherr, C. J. Inhibition of cyclin D1 phosphorylation on threonine-286 prevents its rapid degradation via the ubiquitin-proteasome pathway. Genes Dev, 11: 957-972, 1997. 3. Diehl, J. A., Cheng, M., Roussel, M. F., and Sherr, C. J. Glycogen synthase kinase-3beta regulates cyclin D1 proteolysis and subcellular localization. Genes Dev, 12: 3499-3511, 1998.

144 4. Zheng, Y., Shen, H., Sturgis, E. M., Wang, L. E., Eicher, S. A., Strom, S. S., Frazier, M. L., Spitz, M. R., and Wei, Q. Cyclin D1 polymorphism and risk for squamous cell carcinoma of the head and neck: a case-control study. Carcinogenesis, 22: 1195-1199, 2001. 5. Wang, L., Habuchi, T., Takahashi, T., Mitsumori, K., Kamoto, T., Kakehi, Y., Kakinuma, H., Sato, K., Nakamura, A., Ogawa, O., and Kato, T. Cyclin D1 gene polymorphism is associated with an increased risk of urinary bladder cancer. Carcinogenesis, 23: 257-264, 2002. 6. Kong, S., Amos, C. I., Luthra, R., Lynch, P. M., Levin, B., and Frazier, M. L. Effects of cyclin D1 polymorphism on age of onset of hereditary nonpolyposis colorectal cancer. Cancer Res, 60: 249-252, 2000. 7. Zhang, Y. J., Chen, S. Y., Chen, C. J., and Santella, R. M. Polymorphisms in cyclin D1 gene and hepatocellular carcinoma. Mol Carcinog, 33: 125-129, 2002. 8. Koike, H., Suzuki, K., Satoh, T., Ohtake, N., Takei, T., Nakata, S., and Yamanaka, H. Cyclin D1 gene polymorphism and familial prostate cancer: the AA genotype of A870G polymorphism is associated with prostate cancer risk in men aged 70 years or older and metastatic stage. Anticancer Res, 23: 4947-4951, 2003. 9. Wang, L., Habuchi, T., Mitsumori, K., Li, Z., Kamoto, T., Kinoshita, H., Tsuchiya, N., Sato, K., Ohyama, C., Nakamura, A., Ogawa, O., and Kato, T. Increased risk of prostate cancer associated with AA genotype of cyclin D1 gene A870G polymorphism. Int J Cancer, 103: 116-120, 2003. 10. Sawa, H., Ohshima, T. A., Ukita, H., Murakami, H., Chiba, Y., Kamada, H., Hara, M., and Saito, I. Alternatively spliced forms of cyclin D1 modulate entry into the cell cycle in an inverse manner. Oncogene, 16: 1701-1712, 1998. 11. Hinds, P. W., Dowdy, S. F., Eaton, E. N., Arnold, A., and Weinberg, R. A. Function of a human cyclin gene as an oncogene. Proc Natl Acad Sci U S A, 91: 709-713, 1994. 12. Koivisto, P., Kononen, J., Palmberg, C., Tammela, T., Hyytinen, E., Isola, J., Trapman, J., Cleutjens, K., Noordzij, A., Visakorpi, T., and Kallioniemi, O. P. Androgen receptor gene amplification: a possible molecular mechanism for androgen deprivation therapy failure in prostate cancer. Cancer Res, 57: 314- 319, 1997. 13. Visakorpi, T., Hyytinen, E., Koivisto, P., Tanner, M., Keinanen, R., Palmberg, C., Palotie, A., Tammela, T., Isola, J., and Kallioniemi, O. P. In vivo amplification of the androgen receptor gene and progression of human prostate cancer. Nat Genet, 9: 401-406, 1995. 14. Yotsuyanagi, T. and Hazemoto, N. [Cationic liposomes in gene delivery]. Nippon Rinsho, 56: 705-712, 1998. 15. Mabjeesh, N. J., Zhong, H., and Simons, J. W. Gene therapy of prostate cancer: current and future directions. Endocr Relat Cancer, 9: 115-139, 2002. 16. Kotin, R. M., Siniscalco, M., Samulski, R. J., Zhu, X. D., Hunter, L., Laughlin, C. A., McLaughlin, S., Muzyczka, N., Rocchi, M., and Berns, K. I. Site-specific integration by adeno-associated virus. Proc Natl Acad Sci U S A, 87: 2211-2215, 1990.

145 17. Richardson, W. D. and Westphal, H. A cascade of adenovirus early functions is required for expression of adeno-associated virus. Cell, 27: 133-141, 1981. 18. Handa, A., Muramatsu, S., Qiu, J., Mizukami, H., and Brown, K. E. Adeno- associated virus (AAV)-3-based vectors transduce haematopoietic cells not susceptible to transduction with AAV-2-based vectors. J Gen Virol, 81: 2077- 2084, 2000.

146