He thong quan ly diem sinh vien

To mimic the clinical situation and test the role of AR in progression, we cultured androgen-dependent LNCaP 104-S prostate tumor cells in the presence of the antiandrogen Casodex bicalu

Role of Androgen Receptor in the Progression

of Human ProstateTumor Cells to Androgen

Independence and Insensitivity John M Kokontis, Stephen Hsu, Chih-pin Chuu, Mai Dang, Junichi Fukuchi,

Richard A Hiipakka, and Shutsung Liao*

Ben May Institute for Cancer Research and the Department of Biochemistryand Molecular Biology,

The University of Chicago,Chicago, Illinois

BACKGROUND Various studies have implicated the androgen receptor (AR) in the

pro-gression of androgen-dependent human prostate cancer cells to androgen-independent and

androgen-insensitive phenotypes, but the exact role of AR in progression is unclear

METHODS To mimic the clinical situation and test the role of AR in progression, we cultured

androgen-dependent LNCaP 104-S prostate tumor cells in the presence of the antiandrogen

Casodex (bicalutamide) to derive resistant (CDXR) clones In a second step, we cultured CDXR

cells in the presence of the androgen R1881, which generated androgen- and

Casodex-insensitive (IS) cells These cells were then characterized with regard to AR function and the

effect of ectopic AR expression or AR knockdown on androgen sensitivity

RESULTS CDXR cells showed increased AR expression and transcriptional activity CDXR

cell proliferation was unaffected by Casodex but was repressed by androgen in vitro and in vivo

IS cells, on the other hand, had greatly reduced AR expression and activity compared to CDXR

cells Knockdown of AR expression in CDXR cells produced cells that were insensitive to

androgen Conversely, re-expression of AR in IS cells regenerated cells that were repressed by

androgen Knockdown of AR expression in 104-S cells produced cells that remained stimulated

by androgen, while overexpression of AR in 104-S cells generated an androgen-repressed

phenotype but did not confer androgen-independent growth

CONCLUSIONS Increased AR expression determines whether prostate cancer cells are

repressed by androgen, but is not required for androgen independence These results may have

implications for anti-AR therapy for prostate cancer Prostate 65: 287–298, 2005.

# 2005 Wiley-Liss, Inc.

KEY WORDS: prostate cancer; progression; androgen; androgen receptor; LNCaP

INTRODUCTION Progression of prostate tumors that are dependent

upon androgen for survival and growth to tumors that

are androgen-independent and -insensitive remains

incompletely understood at the molecular level

Anti-androgen or Anti-androgen deprivation therapy of prostate

cancer, pioneered by Charles Huggins [1], is initially

effective in repressing prostate tumor growth

How-ever, endocrine therapy rarely succeeds in killing all

tumor cells and progression of androgen-dependent

tumor cells to androgen-independent cells occurs

often Several mechanisms have been identified that

may participate in the transition of prostate tumor cells

Abbreviations: CS-FBS, charcoal-stripped fetal bovine serum; AR, androgen receptor; 5a-DHT, 5a-dihydrotestosterone; R1881, 17b-hydroxy-17a-methylestra-4,9,11-trien-3-one; IPTG, isopropyl thioga-lactoside; PSA, prostate specific antigen; CDXR, Casodex-resistant;

IS, androgen-insensitive.

This study is dedicated to the memory of Dr Charles B Huggins Grant sponsor: NIH (to SL); Grant numbers: CA58073, AT00850; Grant sponsor: Robert Earp Trust.

*Correspondence to: Shutsung Liao, The Ben May Institute for Cancer Research, The University of Chicago, Box MC6027, 5841 S Maryland Ave., Chicago, Illinois 60637.

E-mail: sliao@huggins.bsd.uchicago.edu Received 8 November 2004; Accepted 21 April 2005 DOI 10.1002/pros.20285

Published online 13 July 2005 in Wiley InterScience (www.interscience.wiley.com).

ß 2005 Wiley-Liss, Inc

to clinical hormone-independence These include

androgen receptor (AR) gene amplification, AR

muta-tion, and bypass of androgenic activation of AR or of

AR signaling itself for cell survival and proliferation

(for reviews, see refs [2–4]) We reported previously

that clonally derived androgen-dependent LNCaP

104-S cells, after long-term androgen deprivation in vitro,

can give rise to 104-R1 and 104-R2 cells, the growth of

which is repressed by androgen [5–8] We

demon-strated that these hormone independent cells exhibited

increased AR expression and transcriptional activity

and their proliferation was repressed by low

concen-trations of androgen The 5a-reductase inhibitor

finas-teride and the antiandrogen Casodex (bicalutamide)

blocked the repressive effects of testosterone

Andro-gen-repressed LNCaP 104-R1 cells can revert to an

androgen-stimulated phenotype when treated with

androgen in vitro [7] and a similar phenomenon is

observed in LNCaP 104-R1 cells grown as tumors in

castrated athymic mice [8]

In this report, we sought to clarify the role of

AR in the progression of LNCaP 104 tumor cells

from hormone-dependence to hormone-independence

Recent reports have demonstrated sustained and

heightened AR expression, function and sensitivity to

androgen in hormone-independent or recurrent

pros-tate cancer cells [9–16] Therefore, it was of interest to

establish whether continued and elevated AR

ex-pression was necessary for the hormone-independent

growth we observed in the LNCaP 104 progression

model For this purpose we derived new

androgen-independent and -repressed clonal sublines from

androgen-dependent 104-S cells by selecting for

growth in the presence of Casodex We extended the

progression model by re-exposing Casodex-resistant

(CDXR) cells to androgen to generate cells (IS) that were

completely androgen and antiandrogen-insensitive

We then studied the effect of enforced reduction or

re-expression of AR in these cells and in progenitor

104-S cells Our results indicate that elevated AR expression

mediates androgenic repression of cell growth but is

not responsible, at least by itself, for

androgen-independent growth

MATERIALS AND METHODS

Materials The LNCaP 104-S, 104-R1, and 104-R2 sublines

and the AN-21 anti-AR polyclonal antibody were

des-cribed previously [5,7] A polyclonal anti-prostate

specific antigen (PSA) antibody was from DAKO

(Glostrup, Denmark), a monoclonal anti-p27Kip1

anti-body was from Transduction Laboratories/BD

Bio-sciences (Lexington, KY), a monoclonal anti-actin

antibody was from Chemicon (Temecula, CA) R1881 (17b-hydroxy-17a-methylestra-4,9,11-trien-3-one) was from Perkin Elmer/NEN Life Science Products (Boston, MA) Casodex1

(ICI 176,334; (2RS)-40 -cyano-3-(4-fluorophenylsulfonyl)-2-hydroxy-2-methyl-30

-(triflu-oromethyl)-propionanilide; bicalutamide) was from AstraZeneca Pharmaceuticals (Wilmington, DE)

Cell Culture and Generation of Casodex-Resistant and Androgen-Insensitive Clones InVitro LNCaP 104-S, 104-R1, and 104-R2 cells were pas-saged and maintained as described previously [7] For the selection of Casodex-resistant (CDXR) cells,

2
106LNCaP 104-S cells were plated in DMEM sup-plemented with 10% charcoal-stripped fetal bovine serum (CS-FBS) and 5 mM Casodex, and were grown for

3 weeks Six colonies were amplified into clonal sub-lines called CDXR1 through CDXR6, which were used

in subsequent characterization of proliferative behav-ior and AR expression For the generation of androgen-insensitive (IS) cells, the CDXR1, CDXR2, and CDXR3 sublines were cultured in DMEM supplemented with 10% CS-FBS and 10 nM R1881 for a period of 3 weeks before replating

Cell Cycle Distribution Analysis Cell cycle distribution analysis on LNCaP S, 104-R1, CDXR, or IS cells was performed as described pre-viously [7] One day after plating, cells were treated with R1881 or Casodex and grown for an additional

4 days with a change of medium on day 2 After over-night fixation at 208C in 70% ethanol/30% phosphate-buffer saline (PBS), cells were treated with 0.1 mg/ml RNAse A in PBS for 30 min at 378C After pelleting, cells were stained with 50 mg/ml propidium iodide in PBS Cell cycle profiles and distributions were

determin-ed by flow cytometric analysis of 2
104 cells using the CellQuest program (v 3.1f) on a BD Facscan flow cytometer (BD Biosciences, San Jose, CA) Cell cycle distribution analysis was performed using ModFit LT software (Verity Software House, Topsham, ME)

Fluorometric Cell Growth Assay For measurement of cell growth, the DNA fluoro-metric method of Rago et al [17] was used Cells (5
103/well) were plated in 96-well plates The next day cells were treated, 6 wells per treatment, with R1881 at 0.1 or 10 nM, Casodex at 5 mM, tetracyline at

2 mg/ml, or isopropyl-b-D-thiogalactopyranosideside (IPTG) at 4 mg/ml and grown for 5 additional days with a medium change on day 4 after plating After

removal of growth medium, 100 ml of water was

added and the plates were frozen at 908C until assay

Cells were thawed and 100 ml of a solution containing

20 mg/ml Hoechst 33258, 2M NaCl, 10 mM Tris HCl

pH 7.4, and 1 mM EDTA was added After thorough

mixing, plates were read in a Wallac 1420 fluorometric

plate reader using a 355/460 nm excitation/emission

filter

Growth of CDXR3 and IS3 Tumors

in Athymic Mice Six to eight-week old castrated male athymic BALB/

c mice (NCI-Frederick) were injected subcutaneously

with 1
106CDXR3 or IS3 cells suspended in 0.5 ml

of Matrigel (BD Bioscience) Four weeks after

in-jection, half of tumor-bearing mice were implanted

with pellets containing 20 mg testosterone propionate

Pellets were replaced with fresh pellets after 30 days

Tumor growth was determined weekly by caliper

measurement Tumor volume was calculated by the

formula (length
width
depth)/2

Real-Time PCR Total RNA was isolated using Trizol reagent

(Invitrogen Life Technologies, Carlsbad, CA)

Contam-inating DNA was removed using DNase I (DNA-free,

Ambion, Austin, TX) cDNA was synthesized from 2 mg

total RNA using an Omniscript RT synthesis kit

(Qiagen, Valencia, CA) For real-time PCR, a

Quanti-Tect Probe PCR kit (Qiagen) and a Prism 7700 cycler

(Applied Biosystems, Foster City, CA) were used in a

dual-labeled probe protocol The following primer

and probe sequences, selected using the Primer

Ex-press program (Applied Biosystems), were used for AR

real-time PCR: forward primer, 50

-CGCCCCTGATCT-GGTTTTC; reverse primer, 50

-TTCGGACACACTG-GCTGTACA; FAM-labeled probe, 6FAM-50

-TGAGTA-CCGCATGCACAAGCTCCG-30-TAMRA The primer

and FAM-labeled probe sequences used for PSA

real-time PCR were described by Gelmini et al [18] A

Ribosomal RNA Control kit (Applied Biosystems) was

used to normalize transcript levels among samples

RNA Interference for AR Knockdown

Complementary 64-base oligonucleotides were

syn-thesized (Integrated DNA Technologies, Inc., Coralville,

IA) containing a 19-base C-terminal AR sequence (50

-GCACTGCTACTCTTCAGCA) in inverted repeat

ori-entation [19] After annealing, the 64-mer duplex was

inserted into BglII/HindIII-digested pH1RP RNAi

expression vector [20] After transfection,

G418-resis-tant 104-S and CDXR3 colonies were expanded and

screened for reduced AR expression by Western blot analysis For tetracycline-induced knockdown of AR,

a 19 bp TETO2 operator sequence was introduced immediately 30of the TATA box of the H1 promoter by PCR amplification [21] The 64-base AR RNAi construct was then inserted into the BglII/HindIII-digested pH1RP-TETO2 vector Prior to transfection with the pH1RP-TETO2-AR vector, 104-S and CDXR3 cells were stably transfected with the pcDNA6/TR vector (Invitrogen) for TET repressor expression Blasticidin S-resistant colonies were screened for TET repres-sor expression using an anti-TET represrepres-sor antibody (MoBiTec; Goettingen, Germany)

AROverexpression Overexpression of AR was achieved by infecting IS cells with the pLNCX2 retrovirus (BD-Clontech, Palo Alto, CA) carrying a 3 kb BglII human AR cDNA frag-ment derived from the pSG5-hAR vector [22] Re-trovirus was generated using the Phoenix-ampho packaging cell line (G Nolan, Stanford University) IPTG-induced AR expression in 104-S cells was carried out using the 30

SS (Stratagene) and pLNXRO2 [23] vectors using methods described previously [7]

RESULTS Progression of LNCaP 104 -S Cells

to CDXR Cells During a 3-week exposure of androgen-dependent 104-S cells to 5 mM Casodex in medium supplemented with CS-FBS, most cells stopped proliferating and detached from the dishes Parallel cultures of these cells, fixed and stained with 4,6-diamidino-2-pheny-lindole (DAPI), did not show typical apoptotic nuclear condensation and fragmentation Discrete colonies soon appeared at low frequency Direct counting of colony number and Poisson estimation of colony fre-quency by tallying growth verses non-growth in plated aliquots of known cell number showed that CDXR cells were present in the 104-S cell population at a frequency

of about 1 in 1.4
105cells at passage 8 Six colonies chosen at random were amplified into sublines, de-signated CDXR1 through CDXR6 A diagram showing the lineages of cells we have derived from the LNCaP 104-S clone is shown in Figure 1a Proliferation of CDXR cells, as measured by the percentage of cells in S phase, was independent of and repressed by androgen

as well as insensitive to Casodex, similar to 104-R1 cells (Fig 1b) CDXR cells treated with androgen accumu-lated in the G1 phase of the cell cycle (data not shown) The percentage of sub-G1 apoptotic cells under all conditions was negligible Although Casodex did not

inhibit growth of CDXR cells, Casodex blocked growth

repression by androgen (Fig 1b)

Acquisition of Complete Androgen

Insensitivity by CDXR Clones

CDXR1, CDXR2, and CDXR3 clones were cultured

in 10 nM R1881 for several weeks to select for cells that

could proliferate in the presence of androgen CDXR

cells initially underwent G1arrest after the addition of

10 nM R1881 Starting at day 6, however, the majority of

cells underwent apoptosis as determined by DAPI

staining and detached from the plate over the next 6–

8 days A similar result was reported by Joly-Pharaboz

et al [24] using androgen-independent LNCaP cells

that they derived By day 14, actively growing colonies

had formed that were then pooled separately for each

CDXR subline We found that cell proliferation in the

three resultant sublines, called IS1, IS2, and IS3, was

unaffected by R1881, while the three CDXR progenitor

sublines were repressed by androgen (Fig 2a) Growth

of IS cells, like that of CDXR cells, was also unaffected

by Casodex When total cell accumulation was

measur-ed by fluorometric assay of DNA content after 5 days

of growth, similar results were obtained (Fig 2b) IS3 colonies arose from CDXR3 populations (passage 7) at a frequency of approximately 1 in 1.9
103cells plated,

a frequency over 70-fold higher than the frequency of CDXR cells in a 104-S population at a similar passage number

AR Expression in CDXR and IS Cells

AR protein expression in the six CDXR sublines grown in the presence and absence of R1881 was compared to AR expression in S, R1, and 104-R2 cells by Western blot analysis (Fig 3a) All six CDXR sublines expressed AR protein at a level greater than that of 104-S cells, as also observed in 104-R1 and 104-R2 cells [5,7] Western blot analysis of AR protein expres-sion in IS cells showed that AR levels were reduced considerably in IS1, IS2, and IS3 cells compared with CDXR1, CDXR2, and CDXR3 cells (Fig 3b) Expression

of AR mRNA was measured in 104-S, 104-R1, the CDXR

Fig 1 a: Derivation of androgen-independent and -insensitive

sublines from LNCaP 104 -S cells after androgen (A) deprivation

(top) and combined androgen deprivation/Casodex treatment

fol-lowedby androgenre-exposure (bottom) b: Percentage of LNCaP

104 -S, 104 -R1 or CDXR cells in S phase determined by flow

cyto-metry of propidium iodide-stained cells Cells were incubated in

medium containing ethanol vehicle (control), 0.1 nM R1881 and/or

5 mM Casodex (CDX) for 96 hr prior to trypsinization and fixation

Values represent the mean  standard error derived from three to

fiveindependentexperiments AsterisksindicateP value<0.01(*)or

<0.05 (**) in comparisonwith untreated cells

Fig 2 a: Percentage of CDXR and IS cells in S phase determined

by flowcytometryofpropidiumiodide-stainedcells.Cellswereincu-bated in medium containing ethanol vehicle (control), 0.1nM R1881,

10 nM R1881or 5 mM Casodex for 96 hr prior to trypsinization and fixation.Values represent the mean  standard error derived from three to five independent experiments b: Fluorometric assay of LNCaP104 -S,104 -R1,CDXR3,andIS3cellgrowth.Cells(5
103well) were plated in 96 -well plates in 150 ml DMEM/CS-FBS using 6 wells per treatment.The next day150 ml of medium containing hormone was added to give a final concentration of 0.1nM R1881,10 nM R1881,

or 5 mM Casodex, and cells were grown for 5 days before assay (see

‘‘Materials and Methods’’) Values represent the mean  standard error.AsterisksindicateP value<0.01(*)or<0.05(**)incomparison withuntreated cells

clones, and the derived IS sublines by real-time PCR.

While AR mRNA level in 104-R1 cells was about

fivefold higher than in 104-S cells, AR mRNA levels in

104-R2 and in the CDXR sublines were at most only

1.5 times higher than in 104-S cells (Fig 3c) The higher levels of AR protein in 104-R2 and CDXR cells compared with 104-S cells (Fig 3a) suggest that AR protein stability may be greater in these cells than in

Fig 3 Androgenic regulation of AR, PSA, and p27Kip1expression a: AR protein expression in LNCaP sublines after 96 hr of incubation in medium containing ethanol vehicle () or 0.1nM R1881 (þ) b: AR protein expression in LNCaP sublines after 48 hr of incubation in medium containing ethanol vehicle () or1nM R1881 (þ) c: AR mRNA expression in LNCaP sublines after 48 hr of incubation in medium containing ethanolvehicle (control) or10 nMR1881determinedbyreal-time PCR.Normalizedvalues are expressedas foldincrease over104 -Scontrolcells d: PSA mRNA expressionin theindicated sublines after 48 hr of incubationin medium containing ethanolvehicle (control) or10 nMR1881deter-mined by real-time PCR.Normalized values are expressed as fold increase over104 -S cells e: PSA protein expression in LNCaP sublines after

48 hr of incubation in medium containing ethanol vehicle () or 1 nM R1881 (þ) Cells were lysed in Laemmli gel loading buffer without bro-mophenol blue dye and aliquots (20 mg protein) were separated on10% SDS ^PAGE gels.Upper panel shows equal loading of identical lysate preparations stained with anti-actin antibody f:PSAprotein expressionin LNCaP104 -R1,CDXR1,CDXR2, and CDXR3 cells after 48 hr incuba-tioninmediumcontainingethanolvehicle (),1nMR1881or 5 mMCasodex g: p27Kip1protein expressionin LNCaP sublines after 72 hr ofgrowth

in medium containing ethanolvehicle (), or R1881 (þ) at 0.1nM (104 -S cells) or10 nMR1881 (104 -R1through IS3 cells)

104-S cells Expression of AR mRNA in the three IS

sublines was greatly decreased compared to the CDXR

cell progenitors

PSA Expression in CDXR and IS Cells

Transcriptional activity of AR in CDXR and IS cells

was assessed by examining androgen induction of PSA

mRNA levels by real-time PCR PSA mRNA levels after

48 hr of treatment with androgen were three to fourfold

higher in the CDXR sublines than in 104-S cells, while

PSA mRNA levels were five and twofold higher in

104-R1 and 104-R2 cells, respectively, than in 104-S cells

(Fig 3d) PSA mRNA levels were low in the three IS

sublines The basal level of PSA mRNA in 104-R1 cells

was significantly higher than in 104-S cells and could be

reduced by treatment with Casodex (not shown) This

may indicate that AR in 104-R1 cells can scavenge low

levels of endogenous androgen, but this ability is

pro-bably not important in driving cell proliferation since

Casodex had no effect on 104-R1 cell cycle progression

(Figs 1b,2b) Induction of PSA protein by androgen in

the IS sublines was undetectable (Fig 3e) Casodex had

no effect by itself on PSA expression in CDXR cells,

indicating that Casodex does not act as an AR agonist in

these cells (Fig 3f)

p27Kip1Expression in CDXR and IS Cells

Previously we showed that androgenic control of

p27Kip1expression at a post-transcriptional level was

partly responsible for androgen-induced G1arrest in

104-R1 and 104-R2 cells [7] We examined p27Kip1

pro-tein expression in IS cells to determine whether

in-duction of p27Kip1is lost in these cells after androgen

treatment Expression of p27Kip1 in androgen-treated

and untreated IS cells was indeed lower than that in

104-R1, 104-R2, and CDXR cells (Fig 3g) This result

suggests that loss of androgenic control of p27Kip1

ex-pression may enable IS cells to grow rapidly in the

presence of androgen

Sensitivity of CDXR3 and IS3 Tumor

Growth toTestosterone

CDXR3 and IS3 tumors were grown in the flanks

of male castrated athymic mice to test the effect of

androgen on tumor growth in vivo Four weeks after

injection of cells, half of the tumor-bearing mice were

implanted subcutaneously with pellets containing

20 mg testosterone propionate Tumor size was

monitored over the next 30 days for all tumors and up

to 150 days for the CDXR3 tumors in mice with

testo-sterone implants Testotesto-sterone repressed CDXR3

tumor growth initially but had no effect on IS3 tumor

growth (Fig 4) After about 60 days post-implantation,

one group of CDXR3 tumors (N ¼ 3) began to re-grow and reached pre-implantation volume by 150 days Another group (N ¼ 5), composed of tumors of gen-erally smaller size at the time of pellet implantation, regressed completely Removal of testosterone pellets from the mice with regenerated tumors followed by injection of Casodex did not affect tumor volume, indicating that the tumors were hormone-insensitive (not shown) These results corroborate observations with these sublines grown in vitro (Figs 1 and 2) Ex-pression of AR was significantly reduced in CDXR3 tumors that had re-grown after implantation of testo-sterone pellets (Fig 4, inset), consistent with the

lower-ed expression of AR in IS cells

Knockdown of AR Expression in CDXR3 Cells Abolishes Sensitivity to Androgen Using vector-derived interfering RNA [19], we de-termined whether CDXR3 cells would tolerate knock-down of AR expression CDXR3 cells were transfected with a pH1RP vector [20] containing a 19-base C-terminal AR sequence or an empty pH1RP vector as control After colony selection and screening, three clones (6,15,17) showing the least expression of AR were used for further analysis (Fig 5a) Induction of PSA and p27Kip1 by androgen was also lost in these clones Flow cytometric analysis revealed that the clones had lost sensitivity to R1881 while CDXR3 cells

Fig 4 CDXR3 and IS3 tumor growth in castrated male athymic mice after subcutaneous implantation (þT) of a testosterone pro-pionate (20 mg) pellet.Control animals received no pellet.‘‘G’’ indi-catesre-growthgroup;‘‘R’’ indicatesregressedgroup.Thenumberof tumors per group ranged from 3 to 8.Values represent the mean  standarderror.Insetis Westernblot showing ARprotein expression in104 -S tumor cells (lane1),CDXR3 tumor cells (lane 2) and in two CDXR3tumors thatre-grew followingimplantationof testosterone pellets (lanes 3 and 4) Upper band represents a non-specifically-stainedprotein

transfected with empty vector were repressed by R1881

(Fig 5b) Similar results were obtained when total cell

accumulation was measured by fluorometric assay

after five days of growth in the absence or presence of

R1881 (Fig 5c) Transfection and selection of colonies

was repeated in the presence of DHT to determine

whether an androgen-dependent phenotype could

be generated in this context However, clones with

lowered AR expression exhibited only an

androgen-insensitive phenotype (not shown) With constitutive

knockdown of AR, the possibility exists that colonies

underwent unknown changes that conferred growth in the absence of AR expression To assess the effect of conditional knockdown of AR, which lessens this pos-sibility, the pH1RP-TETO2-AR vector was

transfect-ed into CDXR3 cells expressing the TET repressor Tetracycline-induced knockdown of AR expression in CDXR3 cells (Fig 5d) resulted in no disruption of cell cycle and blocked androgen-induced arrest (Fig 5e), similar to constitutive knockdown of AR We con-clude that elevated AR expression is responsible for androgenic repression of growth and that continuous

Fig 5 KnockdownofARexpressioninCDXR3cellsbyinterferingRNA.a:Westernblots showingexpressionofAR,PSA,p27Kip1,andactinin CDXR3 cells transfected with empty pH1RP vector andin three CDXR3 cell clones transfected with pH1RP-AR knockdownvector.Cells were grown for 4 days in the presence of10 nMR1881 (þ) or ethanol vehicle ().Cells were lysed in Laemmligel loading buffer withoutbromophenol blue dye andaliquots(40 mgprotein) were separatedon10% or12% (p27kip1) SDS ^PAGE gels b:Percentage ofCDXR3 cellsin Sphase determined

by flow cytometry of propidium iodide-stained cells.CDXR3 cells transfected with empty pH1RP vector or the pH1RP-AR vector grown for

4 daysin thepresence of10 nMR1881orethanolvehicle (control) prior to trypsinization andfixation.Valuesrepresent themean  standarderror

of four independent experiments Asterisk indicates P value <0.01 in comparison with untreated cells c: Fluorometric cell growth assay of CDXR3 pH1RP-AR clones after 5 days of growth in the presence of10 nM R1881or ethanol vehicle (control) Asterisk indicates P value <0.01in comparison with untreated cells d: Western blot showing tetracycline (Tet) induction of AR knockdown in CDXR3 pH1RP-TetO2-AR cells (clones1and 2).Cells were grown for 4 days in the presence or absence of 2 mg/ml tetracycline (Tet) e: Percentage of CDXR3 pH1RP-TetO2-AR cells (clones1and 2) in S phase determined by flow cytometry of propidium iodide-stained cells.Cells were grown for 4 days in the presence of

10 nMR1881 (R) and/or 2 mg/ml tetracycline (Tet) or ethanolvehicle (Con) prior to trypsinization and fixation

elevated AR expression in CDXR cells is not required

for androgen-independent cell growth

Overexpression of AR in IS3 Cells Regenerates

the Androgen-Repressed Phenotype

We infected IS3 cells with a pLNCX2 retroviral

vector carrying the wild-type hAR cDNA to determine

whether an androgen-repressed phenotype could be

regenerated After selection and screening, four clones

that exhibited the highest level of AR expression

(clones 1, 2, 4, and 6; Fig 6a) were used to assess the

effect of androgen on growth Androgenic induction

of PSA and p27Kip1 was restored in these cells Flow

cytometric cell cycle analysis showed that proliferation

of all four clones was inhibited by R1881, while control

IS3 cells remained insensitive to R1881 (Fig 6b)

Fur-thermore, the level of repression correlated with the

level of AR expression among the clones Fluorometric

assay of cell growth showed similar results (Fig 6c)

Overexpression of AR in IS cells, therefore, regenerated

an androgen-repressed phenotype

Knockdown of AR Expression and Induction

of AROverexpression in104 -S Cells Previous reports have shown that reduction of AR signaling, either through antisense oligonucleotides [25–27] or by decoy ARE-containing oligonucleotide [28], suppresses growth and induces apoptosis, respec-tively, in androgen-dependent LNCaP cells To deter-mine the effect of reduced AR expression on 104-S cell growth, we used the pH1RP-TETO2-AR vector to con-ditionally knockdown AR expression Interestingly, tetracycline-induced knockdown of AR (Fig 7a) re-sulted in no change in responsiveness to androgen measured by either flow cytometric cell cycle analysis (Fig 7b) or by fluorometric cell growth assay (Fig 7c) However, it should be noted that AR was not com-pletely eliminated by tetracycline treatment Complete blockade of AR expression would be expected to arrest cells similarly to Casodex

We then attempted to generate an androgen-independent phenotype in 104-S cells by overex-pressing AR However, infection of 104-S cells with

Fig 6 Ectopic overexpression of ARin IS3 cells a:Westernblots showingexpression of AR,PSA, p27Kip1, and actinincontrol IS3 cells and IS3 cellsinfectedwithpLNCX2-ARretrovirus.Cellsweregrownfor3daysin thepresence of10nMR1881(þ)orethanolvehicle().b:Percentage of IS3 cells in S phase determined by flow cytometry of propidium iodide-stained cells IS3 cells and IS3 cell clones infected with pLNCX2-AR re-trovirus grown for 4 days in the presence of 10 nM R1881 or ethanol vehicle (Control) prior to trypsinization and fixation.Values represent the mean  standard error of four independent experiments Asterisk indicates P value <0.01 in comparison with untreated (Control) cells c:Fluorometric cell growth assay of IS3 clones overexpressing AR after 5 days of growth in the presence of10 nMR1881or ethanol vehicle (Control) Asterisk indicates P values <0.01in comparison with untreated cells

retroviral vectors encoding AR followed by selection in

the presence or absence of androgen resulted in only a

few antibiotic-resistant clones that exhibited no

in-creased expression of AR We therefore used an

IPTG-inducible system to conditionally overexpress AR One

clone that exhibited substantial IPTG-induced AR ex-pression, clone 22 (Fig 7d), was grown in the presence

or absence of R1881 and IPTG Induction of AR ex-pression by IPTG did not confer hormone-independent growth measured by either flow cytometry (Fig 7e) or

Fig 7 Conditional knockdown and overexpression of AR in104 -S cells a: AR protein expression in two clones of TETrepressor-expressing

104 -S cells transfectedwith thepH1RP-TETO2-AR knockdownvector after 4 days ofgrowthinpresence or absence of 2 mg/ml tetracyline (Tet) b:Percentage of104 -Scellsin Sphase determinedby flowcytometryofpropidiumiodide-stainedcells.104 -Scellclones transfectedwithpH1RP-TETO2-ARvector grown for 4 daysin thepresence of 0.1nMR1881 (R) and/or 2 mg/ml tetracyline (Tet) or ethanolvehicle (Con) prior to trypsini-zation and fixation.Values represent the mean  standard error of three to five independent experiments Asterisk indicates P value <0.01in comparison with LNXRO2 AR-22 cells treated with R1881 alone c: Fluorometric cell growth assay of104 -S clones transfected with pH1RP-TETO2-AR vector.Cells (5
103per well) were grown for 5 days in the presence of 0.1nM R1881 (R) and/or 2 mg/ml tetracyline (Tet) or ethanol vehicle (Con) prior to assay d: ARprotein expressionincontrol104 -S cells transfectedwith the 30

SSvector (expressing thelac repressor) alone

or also infected with the lac repressor-controlled, AR expressing LNXRO2-AR retrovirus Cells were grown for 4 days in the presence or absence of 4 mMIPTG e: Percentage of104 -S cells in S phase determinedby flowcytometry of propidiumiodide-stained cells.104 -S cells trans-fected with the 30

SSvector alone and also infected with the LNXRO2-AR retrovirus were grown for 4 days in the presence of 0.1nM R1881 (R) and/or4mMIPTGorethanolvehicle(Con)prior to trypsinizationandfixation.Valuesrepresentthemean  standarderrorof threeindependent experiments Asterisk indicates P value <0.01in comparison with LNXRO2 AR-22 cells treated with R1881alone f: Fluorometric cell growth assayof104 -Scells transfectedwith the30SSvector aloneandalsoinfectedwith theLNXRO2-ARretrovirus.Cells(5
103per well)weregrown for 5 daysin thepresence of0.1nMR1881 (R) and/or 4 mMIPTGorethanolvehicle (Con) prior to assay Asteriskindicates P value <0.01incompar-ison with cells treated with R1881alone

fluorometric cell growth assay (Fig 7f) Quite the

oppo-site, overexpression of AR reduced R1881-induced cell

proliferation in these cells and generated an

androgen-repressed phenotype IPTG androgen-repressed cell growth

slightly in parental 104-S 30

SS cells (Fig 7f) This may have been due to a slight induction of AR expression by

IPTG in these cells (Fig 7d) These observations explain

why 104-S cells did not tolerate constitutive

over-expression of AR: the cells cannot proliferate in either

the presence or absence of androgen Based on these

results, overexpression of AR by itself cannot confer

androgen-independent growth Increased AR

expres-sion may be locked in as a compensatory response to

androgen deprivation or may be linked indirectly to the

changes that are responsible for independent growth

Collectively, our results show that the level of AR

expression determines whether prostate cancer cells

are repressed by androgen, but other additional factors

or events are required for cells to become

androgen-independent

DISCUSSION

We have developed a progression model wherein

androgen-dependent prostate tumor cells progress to

androgen-independent and then to

androgen-insensi-tive stages in two discrete steps In the first step,

androgen-independent but androgen-repressed CDXR

cells arise at low frequency when androgen-dependent

cells are cultured in the presence of antiandrogen

CDXR cells are typified by increased expression and

activity of AR In the second step, progression to

com-plete androgen insensitivity occurs at higher frequency

when androgen is added to eliminate

androgen-repressed cells The genetic or epigenetic changes in

CDXR cells responsible for androgen-independent

growth have not been determined in the present study

It is clear from our results, however, that events other

than increased AR expression must occur in LNCaP

104-S cells during the transition to androgen

indepen-dence Conditional overexpression of AR in 104-S cells

generated an androgen-repressed phenotype but did

not confer androgen-independent growth Mutation of

the LNCaP AR gene to forms that utilize Casodex as an

agonist was proposed as a mechanism responsible for

Casodex withdrawal syndrome [29] However, the

ob-servation that androgen but not Casodex induces PSA

expression in the 104-R1 and CDXR sublines (Fig 3f)

indicates that AR mutation is not the mechanism

res-ponsible for progression to Casodex resistance We did

not detect the mutations in the ligand-binding domain

of AR in 104-R1/R2 cells or CDXR cells that Hara et al

[29] detected in their Casodex-resistant LNCaP cells

Nor did we detect the mutations in the AR N-terminal

domain reported by Han et al [30,31] that confer

ligand-independent transcriptional activity Our re-sults also differ from those of Culig et al [32] and Chen

et al [16] who reported agonistic effects of Casodex in androgen-independent LNCaP cells Chen et al also reported that LNCaP cells readily overexpress AR con-stitutively after retroviral infection, which our clonal LNCaP 104-S cells did not tolerate In the study by Chen

et al LNCaP cells overexpressing AR apparently pro-liferated by scavenging low levels of available andro-gen CDXR as well as 104-R1 and 104-R2 cells appear not to do this, however, because the growth of these cells is insensitive to the antagonist Casodex The ob-servation that CDXR cells can easily tolerate loss of AR indicates that continuous elevated AR expression is not required in these cells However, we cannot rule out the possibility that these cells require ligand-independent signaling from AR expressed at very low levels Our results contrast with those of Zegarra-Moro et al [33] who reported that antibody- or ribozyme-mediated blockade of AR activity in hormone-refractive LNCaP cells repressed cell growth The mechanism that LNCaP 104-S cells utilize to progress to androgen-independency therefore appears to be different than the mechanisms reported by others LNCaP and pros-tate tumor cells in general most likely have multiple mechanisms available to achieve growth in a low-androgen environment

Knockdown of AR in CDXR3 cells resulted in cells that were insensitive to androgen and conversely, re-expression of AR in IS3 cells regenerated cells that were repressed by androgen These results indicate that the level of AR expression is the sole determinant of whether cells are repressed by androgen Evidence for the androgen-repressed phenotype outside of the LNCaP cell line is accumulating [34,35] The ARCaP cell line is another prostate tumor cell line that is repressed by androgen [36] Hara et al [14] recently noted suppression of MDA PCa 2b-hr prostate cancer cell growth by high dose androgen after long term androgen deprivation Overexpression of AR in PC-3 cells has been shown to confer growth repression by androgen [37–39] Bruchovsky et al [40] noted pos-sible androgenic repression of tumor growth in an interim analysis of the Canadian Prospective Trial

of intermittent androgen deprivation therapy Andro-gen has also been reported to repress cell growth during androgen-induced differentiation of immorta-lized human and rodent prostate epithelial cells in vitro [41,42] Bruchovsky et al [43] described the repression

by androgen of rat ventral prostate epithelial cell proliferation in the hormonal regulation of prostate homeostasis The intermediate androgen-independent but androgen-repressed stage of LNCaP 104 tumor cell progression may therefore have a real but under-recognized counterpart in clinical prostate tumor