Vol. 2, 277-285, February 1996 Clinical Cancer Research 277

Advances in Brief

Mutations in the Androgen Receptor Gene Are Associated with Progression of Human Prostate Cancer to Androgen Independence I

Wayne D. Tilley, 2 Grant Buchanan, Theresa E. Hickey, and Jacqueline M. Bentel Cancer Cell Biology Laboratory, Department of Surgery, School of Medicine, Flinders University of South Australia, Bedford Park, SA 5042, Australia

Abstract Progression to androgen-independent growth of human

prostate cancers may be mediated by alterations in the structure and/or expression of the androgen receptor (AR) gene. To date, mutations in the AR gene have largely been identified in hormone refractory tumors. In this study, sin- gle-strand conformational polymorphism analysis and DNA sequencing of the entire AR gene coding region was per- formed on 25 primary prostate tumors sampled prior to initiation of hormonal (i.e., androgen ablation) therapy. Base changes leading to amino acid substitutions in the AR were identified in 11 (44%) tumors. The presence of AR amino acid substitutions was associated with decreased im- munohistochemical staining for AR in tumor cells and the rapid failure of subsequent hormonal therapies. Single- strand conformational polymorphism analysis of exons 2, 3, and 8 of the X-linked hypoxanthine guanine phosphoribosyl transferase (HPRT) gene in the same samples revealed no bandshifts, suggesting that the high frequency of AR gene mutations detected was not a consequence of generalized genetic instability. These data indicate that AR gene muta- tions occur commonly in advanced prostate cancers prior to endocrine treatment of disease and may contribute to al- tered androgen responsiveness of the tumors.

Introduct ion Prostate cancer is the most frequently diagnosed invasive

cancer and the second leading cause of cancer deaths in men in Western countries (1). The predominant form of systemic treat- ment for patients with metastatic disease is hormonal or andro- gen ablative therapy (i.e., orchidectomy, luteinizing hormone- releasing hormone analogues, and AR 3 antagonists; Ref. 2).

Received 9/22/95; revised 12/18/95; accepted 12/21/95. 1This work was supported by grants from the National Health and Medical Research Council of Australia, Anti-Cancer Foundation of the Universities of South Australia, and Clive and Vera Ramaciotti Foun- dations. 2 To whom requests for reprints should be addressed. Phone: 61-8-204- 5511; Fax: 61-8-374-0832; E-mail: suwdt@cc.flinders.edu.au. 3 The abbreviations used are: AR, androgen receptor; SSCP, single- strand conformational polymorphism; BPH, benign prostatic hyperpla- sia; PSA, prostate-specific antigen; HPRT, hypoxanthine guanine phos- phoribosyl transferase; MIOD, mean integrated optical density; AIS, androgen insensitivity syndrome.

Although an initial response to hormonal therapy is observed in 70-80% of patients with advanced disease, most tumors progress rapidly to androgen-independent growth and only 10- 20% of patients are alive 5 years following diagnosis (3). At the present time, the molecular changes in tumor cells that lead to resistance to endocrine treatments and independence of andro- gens for growth are poorly understood.

The effects of androgens on the development of the normal prostate gland and the growth of prostate tumors are mediated by the AR, which is a member of a superfamily of ligand- activated nuclear transcription factors (4, 5). Proposed mecha- nisms of progression to androgen-independent growth include loss of AR expression (6, 7), amplification of the AR gene (8), and structural alterations in the receptor protein (9). Other mechanisms that do not involve the AR pathway [e.g., activation of androgen-independent growth factor pathways (10)] have also been proposed.

Immunohistochemical studies have demonstrated expres- sion of AR in primary, advanced, and hormone refractory pros- tate cancers, suggesting that development of androgen-indepen- dence is unlikely to be a consequence of loss of AR expression (11). Indeed, AR gene amplification, which possibly facilitates disease progression by enhancing the growth-promoting effects of androgens at low cellular concentrations, has been demon- strated in a subset of hormone refractory prostate cancers (8). The often rapid progression of advanced prostate cancers from androgen-dependent to androgen-independent growth has led to investigation of structural and/or functional alterations in the AR that could account for the development of resistance to hormonal therapies and disease progression.

Mutations in the AR gene have been identified in a stage B (12) and a small number of advanced (i.e., stage D2) prostate cancers (13-15). In addition, a recent report has demonstrated that 5 of 10 hormone refractory prostate cancer metastases contained mutations in the ligand-binding domain of the AR (16). Where functional analyses of mutated ARs have been performed, broadening of ligand specificity was demonstrated, suggesting that hormone refractory disease may in part be caused by activation of these receptors in the androgen-depleted environment (9, 16). Although previous studies have demon- strated AR gene mutations in hormone refractory disease, their occurrence prior to initiation of androgen ablation treatment has only rarely been reported (12, 17). This finding, while incon- sistent with the frequently short-lived responses of prostate tumors to hormonal therapies, may in part be attributed to the examination of small segments of the AR (e.g., the ligand- binding domain) rather than the complete AR coding region.

In the present study, the frequency of mutations in the entire AR coding region in advanced primary prostate cancers was examined using the PCR and SSCP analysis. The presence of base substitutions was confirmed by sequencing of cloned

278 AR Gene Mutations in Prostate Cancer

PCR-amplified fragments exhibiting SSCP bands of reproduc- ibly altered mobility. Expression of AR protein was evaluated using quantitative immunohistochemistry and video image anal- ysis. Our results indicate that mutations in the AR gene are present in advanced prostate cancers prior to initiation of treat- ment and are associated with reduced AR imnmnostaining in tumor cells and the rapid failure of subsequent androgen abla- tive therapy.

Materials and Methods Prostate Tissues and Cell Lines. Twenty-five primary

prostate tumors and 12 BPH specimens were obtained from patients undergoing transurethral resections of the prostate per- formed for acute urinary obstruction. The presence of carcinoma or BPH in the tissue blocks was confirmed by pathological assessment on hematoxylin and eosin-stained paraffin sections (Department of Pathology, Flinders Medical Center, Bedford Park, Australia). All tumor specimens were obtained from pa- tients prior to initiation of hormonal therapy. Tumor stage was determined according to the Modified Whitmore Jewett System (18). The cohort consisted of 9 stage C and 16 stage D~ tumors•

Following surgery and commencement of hormonal ther- apy (orchidectomy and/or androgen receptor antagonists), an initial decrease in serum PSA levels was observed in all patients. Serum PSA measurements were determined by the Department of Clinical Biochemistry at Flinders Medical Center using a solid-phase, two-site immunoenzymatic assay (Tandem-E PSA; Hybritech, Inc., San Diego, CA) as described previously (19). The normal range of serum PSA levels in healthy men is 0 - 4 ng/ml for this assay. In this study, serum PSA levels were categorized into three groups: category 1, levels <4 ng/ml; category 2, levels between 4 and 10 ng/ml; and category 3, levels >10 ng/ml. Patients were followed clinically with serum PSA measurements and bone and computerized tomographic scans as required. Clinical follow-up and survival data of 3-7 years were available for 20 of the patients. Five patients were lost to follow-up. Tumor response was assessed by standard clinical and biochemical features (i.e., sustained reduction in serum PSA measurements, improvement in bone and comput- erized tomographic scans, and survival).

Outcome groups were defined at 2 years following initia- tion of hormonal therapy as follows: (a) response--consisted of those patients with stable disease (i.e., no clinical evidence of disease progression, with PSA levels remaining in the original posthormonal therapy category), or improving disease status (i.e., clinical evidence of decreasing tumor mass and a fall in PSA category) or (b) failure--consisted of those patients with progressive disease (i.e., increasing tumor mass with or without a change in PSA category) and/or those who died from prostatic carcinoma within the first 2 years following initiation of hor- monal therapy.

The human prostate cancer cell lines PC-3 and LNCaP were obtained from the American Type Culture Collection (Rockville, MD) and maintained in culture as previously de- scribed (6).

Amplification of Genomic DNA. Genomic DNA was isolated from frozen or paraffin-embedded prostate specimens, peripheral blood lymphocytes from normal volunteers, and the

_J

v

Q

0

w

v

~ | O---- ~ e

. ,& / - - - - -.

O q 3 ~ 0 0

~ ~ 0 - ~ = ~ o .

~.~.-, ~ "-~ = e , ~ o *~

""o ""= ~ . o ~0

, 4

< ~

~ = ~

Clinical Cancer Research 279

Table 1 PCR primers used for amplification of the coding region of the human AR and HPRT genes

Name Exon Oligonucleotide sequence (5'-3')

AR 1-1s a 1-1 as ~ 1-2s a l-2as a 1-3s a 1-3as a 1-4s ~ 1-4as a 1-5s a l-5as ~ l-6s a 1-6as ~ 1-7s a 1-7as a 1-8s ~ 1-8as 2-1s 2-1as 3-1s 3-1as 4-1s 4-1as ~ 4-2s a 4-2as 5s 5as 6s 6as 7s 7as 8s 8as a

HPRT h2s h2as h3s h3as h8s h8as

1 CGCCTGTTGAACTCTTCTGAG 1 CTGTGAAGGTTGCTGTTCCTCA 1 AAGCCCATCGTAGAGGCCCCA 1 TGTCCTTGGAGGAAGTGGGAG 1 AGGAAGCAGTATCCGAAGGCA 1 GCAGCCTAGGCTCTCGCCTTCT 1 CACTGAAGATACTGCTGAGTA 1 GGAAAGTTGTAGTAGTCGCGACTCTG 1 GCACTGGACGAGGCAGCTGCGTACCAG 1 TCGCCAGGTCCCCATAGCGGCACT 1 CGCATCAAGCTGGAGAACCCGCTG 1 CCACCACACGGTCCATACAACT 1 ACAGCCGAAGAAGGCCAGTTGT 1 CAGGTGCGGTGAAGTCGCTTTCCT 1 AGGAAAGCGACTTCACCGCACCTG 1 AGAACACAGAGTGACTCTGCC 2 TGCAGGTTAATGCTGAAGACC 2 GTTATTTGATAGGGCCTTGCC 3 TGGTGCCATACTCTGTCCACT 3 TATCTGGTCTAAAGAGAGACT 4 TTGACCACTGATGATAAATTC 4 ACACACTACACCTGGCTCAAT 4 CAGTGTCACACATTGAAGGCT 4 AATATGATCCCCCTTATCTCA 5 CCGTCAGTACCCAGACTGACC 5 CCAGGTCTGGCCAAGCTGCTG 6 TGGGCTTATTGGTAAACTTCC 6 CAAAAGTGGTCCTCTCTGAAT 7 TCTAATGCTCCTTCGTGGGCA 7 CTCTATCAGGCTGTTCTCCCT 8 GAGGCCACCTCCTTGTCAACC 8 CAGGCAGAAGACATCTGAAAG

ATCCTGTAATGCTCTCATTGA AACAGCTGCTGATGTTTGAAA GAAACTTTCTATTAAATTCCTG AATATAAGAAAACCTACTGTTG GAGAAAACAATTCTCTTTCCT TATTTGTAGAGAGGCACATTT

a Primer located in indicated exon. Nondesignated primers are located in the intronic sequence flanking the indicated exon.

LNCaP and PC-3 prostate cancer cell lines according to pub- lished procedures (6, 20). In the case of tumors, DNA was isolated from a thick (20-1xm) section, and the presence of tumor in the majority of the section used for DNA extraction was confirmed by hematoxylin and eosin staining of an adjacent 5-txm section.

Individual exons or overlapping fragments (exons 1 and 4) of the AR gene were amplified from genomic DNA by PCR using oligonucleotide primers (Fig. 1 and Table 1) based on published sequences (21-23). Eight sets of primers were used for amplification of exon 1 to ensure that the size of the PCR products was within an acceptable range for maintaining the sensitivity of the SSCP analysis (24). Aliquots of 100-250 ng DNA were amplified by 35 cycles of PCR in 25-1xl reactions containing 50 IXM of each of the four deoxyribonucleotide triphosphates, 1.0 p, Ci [e~-32p]dCTP, 1.5 IXM MgC12, 50 mM KC1, 10 mM Tris-HC1 (pH 8.3; at room temperature), 0.45% (v/v) Triton-X 100, 200 p,g/ml gelatin, 0.25 IXM of the appro-

priate sense and antisense oligonucleotide primer pairs, and 0.2 units BresaTaq DNA polymerase (Bresatec, Adelaide SA, Aus- tralia). After an initial denaturation step of 3 min at 95°C, the cycle parameters were as follows: 45 s at 94°C for denaturation, 1 rain at 55-65°C (temperature optimized for primer sequence) for primer annealing, and 1 min at 72°C for primer extension. The CAG trinucleotide repeat region in exon 1 was amplified using conditions as previously reported (23), and the glycine homopolymeric region was amplified in the presence of 5% DMSO (Sigma, St. Louis, MO) and a 3:1 ratio of 7-deaza- 2'GTP (Boehringer Mannheim Australia, Castle Hill, Australia). The denaturation and annealing temperatures for the modified PCR reactions were 96°C and 60°C, respectively. Exons 2, 3, and 8 of the X-linked H P R T gene were also amplified from genomic DNA extracted from each of the prostate samples. The primers were located in the intronic sequence flanking individ- ual exons and designed on the basis of the published H P R T gene sequence (25). The standard amplification conditions used for the AR gene were also used for the H P R T gene. The size and integrity of all PCR products were confirmed on 2% agarose gels.

SSCP A n a l y s i s . SSCP analysis was performed using a modification of the procedure of Orita et al. (26). Radiolabeled PCR products were diluted 1:10 in loading buffer that contained 95% formamide (v/v), 50 rnM EDTA, 20 rnM NaOH, and 0.05% of each of xylene cyanol and bromophenol blue. Samples were denatured at 100°C for 10 rain, snap frozen on dry ice, and thawed on wet ice before electrophoresis in nondenaturing 5.4% polyacrylamide gels. Parameters for SSCP analysis were opti- mized according to Hayashi (24) for each amplified fragment of genomic DNA to ensure maximum sensitivity for detection of mutations (i.e., gels contained 0-10% glycerol and were run at 10°C, 22°C, or 30°C). A nondenatured aliquot of the amplifi- cation reaction was also electrophoresed with the denatured sample to determine the mobility of the double-stranded DNA. Following electrophoresis, gels were transferred onto Whatman filter paper, dried under vacuum, and imaged by Phosphor Imager and Image Quant software (Molecular Dynamics, Sunnyvale, CA). Mobility shifts identified by SSCP analysis were confirmed using an independent PCR amplification reac- tion.

DNA Sequencing, Regions of the AR gene exhibiting a reproducible mobility shift on SSCP analysis were amplified in independent PCR reactions and subcloned using the TA Cloning kit (Invitrogen, San Diego, CA). The sense and antisense strands of 3-10 resultant clones were sequenced with a fmol Sequenc- ing kit (Promega, Madison, WI) and [~/-32P]dATP labeled M-13 forward or reverse primers according to the manufacturers' protocols. The products of the sequencing reactions were elec- trophoresed on denaturing polyacrylamide gels, dried, and vi- sualized with a Phosphor Imager. Alterations in the AR gene sequence were confirmed on two independent PCR-amplified DNA fragments and from sequence derived from both sense and antisense DNA strands. Repetition of all PCR-SSCP analyses and confirmation of the DNA sequencing results on multiple clones indicates that mutations detected did not occur as a result of misincorporation of bases by Taq DNA polymerase.

A R I m m u n o h i s t o c h e m i s t r y . Five-p~m frozen and paraf- fin sections of the prostate samples were stained with polyclonal

280 AR Gene Mutations in Prostate Cancer

antisera to either the amino (U402) or carboxyl (R489) termini

of the human AR using an avidin-biotin immunoperoxidase method as previously described (19, 27, 28). AR immunostain- ing in paraffin sections was enhanced using an antigen unmask- ing procedure as previously described (29). Paraffin sections were mounted onto Histogrip-coated microscope slides (Zymed Laboratories, Inc., San Francisco, CA) and microwaved in 0.01 M citrate buffer (U402: pH 6.0, 8 min; R489: pH 6.5, 18 min) before immunohistochemical staining. Immunoreactive prod- ucts were visualized with the chromogen 3'3'-diaminobenzidine tetrahydrochloride, and the area and intensity of staining were

measured by computer-assisted color video image analysis us- ing the Video Pro 32 system (Leading Edge Pty Ltd., Adelaide SA, Australia) according to established procedures (19, 28). The areas of unstained (A1) and positively stained nuclei (A2) and the intensity of AR immunostaining (i.e., the integrated optical density) were measured for at least 15 fields in both the glan- dular regions of BPH tissues and in sections containing predom- inantly carcinoma in the case of prostate cancer specimens. From these measurements, the level of AR staining (i.e., MIOD [MIOD = integrated optical density/(A1 + A2)]) was calcu- lated.

Statistical Analysis. All data were expressed as mean + SE. Immunostaining parameters and clinical outcome of patients were analyzed using a paired t test. The relationship between the presence of amino acid substitutions in the AR and response to hormonal therapy was analyzed using a ×2 test. Significance was

established at P < 0.05.

Results PCR-SSCP Analysis of the AR Gene in Prostate Cancer

Specimens. SSCP analysis of PCR products encompassing

the entire coding region of the AR gene resulted in the identi- fication of reproducible band shifts corresponding to variant DNA conformers in 13 of 25 primary human prostate tumor samples. No mobility shifts were identified using DNA derived from peripheral blood lymphocytes of six normal individuals (data not shown). The human prostate cancer cell line LNCaP, which is known to have an A to G base transition at codon 868 in exon 8 of the AR gene (30), consistently exhibited a band of altered mobility following PCR-SSCP analysis and was used as a control for this exon (Fig. 2a). None of the AR gene exon 8 DNA fragments amplified from the 25 prostate tumor samples exhibited a mobility shift similar to that seen for the LNCaP cell line. Typical examples of the SSCP shifts for exons 1, 5, and 8 of the AR gene amplified from the prostate tumors are shown in Fig. 2, b-d. At least one normal DNA conformer was observed with variant DNA conformers in the SSCP analysis (Fig. 2, b-d), suggesting the presence of both wild-type and mutant AR gene sequences within the prostate tumor specimens.

DNA Sequencing of the AR Gene. DNA sequencing (Fig. 3) of cloned PCR products derived from independent amplification reactions using DNA isolated from specimens yielding variant DNA conformers on SSCP analysis resulted in the identification of base changes in the AR gene, which were confined to 13 of the 25 prostate tumor samples (Fig. 1 and Table 2). AR gene sequences were confirmed in three or more clones sequenced in both sense and antisense orientations. At

b

Fig. 2 PCR-SSCP analysis of the AR gene in human prostate tumors. Location of the AR gene on the X chromosome results in detection of two bands in normal samples (sense and antisense), indicating a single copy of the gene. Closed arrowheads, variant DNA conformers of altered mobility; open arrowhead, absence of a normal DNA conformer. The mobility of the double-stranded (ds) DNA is indicated, a, PCR- SSCP analysis of exon 8. PCR fragments amplified for exon 8 using control DNA from normal human blood and four human prostate tumor specimens exhibited identical mobilities on nondenaturing gel electro- phoresis. In contrast, the same fragment amplified from the human prostate cancer cell line LNCaP (which is known to contain a point mutation in exon 8 at amino acid 868) exhibited altered mobility, b, PCR-SSCP analysis of the AR gene in tumor p383 showing bands of altered mobility for exon 5. Sequence analysis revealed a C-T base change in the intronic sequence six bases upstream of the intron 4/exon 5 splice junction, c, PCR-SSCP analysis of exon 1 of the AR gene showing the absence of a normal band and the presence of a variant DNA conformer. Sequence analysis identified a single T-C base change in codon 54 resulting in a Leu-Ser amino acid substitution, d, PCR- SSCP analysis of exon 8 showing bands of altered mobility in prostate tumor p424. Sequence analysis showed this to be associated with a A-C base substitution in codon 882 resulting in conservation of the serine amino acid residue.

least one wild-type AR gene sequence was identified in each sample. Eleven (44%) of the tumors contained missense muta- tions; 4 of these tumors contained base changes, resulting in more than one amino acid substitution in the AR, and a nonsense mutation was identified in exon 6 of tumor p346 (in addition to a missense mutation in exon 1; Table 2). The only sequence

Clinical Cancer Research 281

NORMAL / i l l A C G T Cys

Gin

o

-COOH

-NH 2

rT~ MUTANT . c

GIn/A\ : : : : ~ ~ c

Asn780 r LA _,.~.::~ : :~::i

c Tyr[~ / ~ ~

b NORMAL l i ] -COOH Gin Gin [

A C G T / c ~ ~ :~,: / T'I ~ ~ J C l Leu

T" Trp/"4

\G ] G,,

~ T ] Phe

Leu[ STOP794 [

Gly [G / I ,,~

P"eEij Fig. 3 Sequence analysis of the AR gene in two prostate tumor sam- ples. a, DNA sequence of the AR gene in tumor p413 showing a G-A base substitution resulting in substitution of an asparagine for a serine amino acid residue in codon 780 in exon 6. b, sequence analysis from exon 6 of tumor p346 showing a G-A transition resulting in the introduction of a premature stop codon in place of Trp794.

alterations in the AR gene identified in two tumors were a silent base change in exon 1 (p396) and base substitutions in three separate introns within close proximity to the intron/exon splice junctions (p383). Eleven silent base changes in the AR coding region and seven base substitutions in the intronic sequence were also identified in the subgroup of tumors with missense mutations (data not shown). No base changes in the AR gene were indicated in the remaining 12 tumors by SSCP analysis.

One tumor (p408) contained a normal AR gene sequence with the exception of a low number (n = 13) of CAG trinucle- otide repeats in the glutamine homopolymeric region of exon 1. Analysis of this region of the AR gene in tumor p408 using a sample containing >50% nonmalignant cells (lymphocytes, smooth muscle cells) demonstrated the presence of a single PCR product, suggesting that the germline DNA of this patient also contained 13 CAG repeats. With the exception of tumor sample p259 with 30 CAG repeats in exon 1 of the AR gene, the remaining tumor specimens contained 19-26 CAG repeats.

PCR-SSCP Analysis of the A R Gene in BPI-I Specimens. PCR-SSCP analysis of exons 1-8 of the AR gene in 12 BPH specimens identified band shifts in two samples (data not shown). In a single patient (p617), an A-C base transition in codon 628 that resulted in the substitution of a threonine residue for a lysine, an A-G base transition in codon 715 that resulted

Table 2 Summary of mutations identified in the AR gene in 25 human prostate tumors

Tumour Exon Mutation Consequence

Missense p205 1 A T G - A C G Met265a-Thr

2 C T C - C C C Leu572-P:o p245 1 CCA-TCA Pro26s_C e~ p259 1 T T G - T C G Leusa-Ser

8 TTT-CTT Phe889-Leu p332 1 C T G - C A G Leu57-Gln p346 1 C A G - C A T Gln111-His p347 1 C A G - C G G Gln64-Arg

7 CTT-CCT Leu828-Pro p367 1 G A T - G G T Asps26-Gly p395 4 A T C - A C C Ile670-Thr p405 1 AAA-AGA Lys 179-Arg p413 6 A G C - A A C Ser78o-Asn p424 4 C A G - C G G Gln668-Arg

6 T C T - C C T Ser789-Pro Nonsense

p346 6 T G G - T G A TrP794-sto p

a Amino acid positions according to Tilley et aL (5).

in the substitution of a glutamine residue for a lysine, and a silent mutation in codon 699 were identified in exon 4 of the AR

gene. A single silent base change in codon 203 was found in exon 1 of sample p635. SSCP analysis of the AR gene coding region detected only bands of similar mobility to control DNA in the remaining 10 BPH samples (data not shown).

PCR-SSCP Analysis of the H P R T Gene in Prostate Cancer and BPH Specimens. The 25 prostate cancer and 12 BPH specimens examined using SSCP analysis exhibited bands with identical mobilities to control DNAs for exons 2, 3, and 8 of the X-linked HPRT gene (data not shown). DNA derived from the human prostate cancer cell line PC-3 exhibited a PCR-SSCP band shift in exon 8 of the HPRT gene, while LNCaP DNA showed the wild-type pattern for all three exons analyzed (data not shown).

Immunocytochemical Analysis of AR Protein in Pros- tate Cancer and BPH Specimens. The level of AR immu- nostaining (MIOD) in the 25 prostate cancers determined using an anti-NH 2 terminal antibody (U402) ranged between 0 and 57 pixel density units, with a mean (___SE) of 28 (___4). In contrast, the MIOD of AR staining in the BPH specimens ranged between 12 and 50 pixel density units, with a mean value of 39 (---3), which was significantly greater than the MIOD of AR staining in the tumors (P < 0.05). AR immunostaining (MIOD) of prostate cancer specimens identified as having an amino acid substitution in the receptor (21 ___ 6) was found to be signifi- cantly lower than that in prostate cancers without amino acid substitutions (33 +-- 6; P = 0.03).

Clinical follow-up data were available for 20 of the 25 patient samples used in these studies. Of the 20 patients, 12 responded to hormonal therapy and 8 progressed rapidly to hormone refractory disease. Immunohistochemical analysis of prostate cancer cells in these specimens demonstrated that the mean AR level (_SEM) in responders to hormonal manipula- tion [32 (___ 6) pixel density units] was not statistically different from AR staining in cancers that progressed on hormonal ther- apy in <2 years (21 -L-_ 7 pixel density units), although this result

282 AR Gene Mutations in Prostate Cancer

Fig. 4 AR immunostaining of prostate tumor p346. Top panel, paraffin section showing positive nuclear staining in the majority of tumor cells with the anti-NH 2 terminal AR antibody U402. A few stromal cells are also positively stained for AR. Bottom panel, absence of AR staining in the same tumor sample using the anti-COOH terminal receptor antibody R489. Inset, AR immunoreactivity in the glandular epithelial cells of a BPH specimen (positive control) stained with antibody R489 at the same time as tumor p346. ×250.

approached significance (P = 0.051; data not shown). Six of the eight tumors that progressed rapidly following initiation of hormonal therapy contained amino acid substitutions in the AR gene. In contrast, only 3 of 12 tumors that responded to andro- gen ablation therapy :>2 years contained mutations in the AR gene which resulted in amino acid substitutions in the receptor. [An additional tumor in this latter group, p346, contained a premature termination codon in codon 794 of the ligand-binding domain which was likely to have resulted in a nonfunctional AR (see below).] The potential clinical importance of amino acid substitutions in the AR in untreated primary tumors was dem-

onstrated by a significant correlation between their presence and

the rapid failure of subsequent hormonal therapies (X 2 ---~ 4.85;

P < 0.015). Although receptor protein was detected immunohisto-

chemically in the majority of prostate cancer specimens con- taining mutations in the AR gene, in most samples it was not possible to determine the proportion of tumor cells expressing mutant AR. In a single tumor (p346), a G-A transition in codon 794 in exon 6 of the AR resulted in the introduction of a premature termination codon (Fig. 1 and Table 2). Expression of a truncated AR protein in this tumor was suggested by the positive immunoreactivity of the tumor cells when stained with an antiserum (U402) directed against the NH 2 terminus of the AR

Clinical Cancer Research 283

and the absence of staining of tumor cells with antiserum (R489) directed against the COOH terminus of the receptor (Fig. 4).

Discussion

Mutations in the AR gene that alter the amino acid se- quence of the receptor were identified in a high proportion (44%) of 25 primary prostate tumors sampled prior to the commencement of hormonal therapy. In contrast to mutations previously identified in the AIS, which are germline, those identified in the present study in prostate cancers appear to be somatic mutations, since both wild-type and mutant DNA se- quences were detected. Base changes resulting in amino acid substitutions were found in exons 1, 2, 4, and 6 - 8 of the AR gene. Amino acid alterations in the ligand-binding domain of the AR have been described previously in hormone refractory pros- tate cancers (9, 16). With the exception of the study by Schoen- berg et al. (31), where the polymorphic CAG repeat region of the AR was analyzed, and a single report of a mutation at codon 340 of the AR in human prostate cancer (17), previous studies have not analyzed exon 1 of the AR which encompasses 58% of the coding region of the gene. In the present study, 50% of amino acid substitutions identified in the AR were within exon 1. In view of a recent report demonstrating that residues 1-485 of exon 1 are required for full wild-type AR activity (32), it is likely that at least seven of the eight exon 1 amino acid substitutions located within this region of the exon will have significant effects on androgen or antiandrogen action in the prostate tumors.

The relatively high frequency of mutations detected in advanced prostate cancers in this study suggests that AR gene mutations may be involved in disease progression. The recent identification of AR gene mutations in 50% of hormone refrac- tory prostate cancers also supports this association (16). Ad- vanced malignancy, including prostatic cancer, is often associ- ated with tumor cell aneuploidy. Previous studies have documented the increased frequency of mutations in exons 3 and 8 of the X-linked HPRT gene in spontaneous and chemi- cally induced tumors (33). The location of mutations in the HPRT gene is associated with sequences susceptible to mutation (i.e., "hot spots" in exons 3 and 8), rather than specific selec- tion for genotype or phenotype (33). However, the absence of mutations in exons 2, 3, and 8 of the HPRT gene in the same DNA samples used for the analysis of the AR gene suggests that the relatively high frequency of AR gene mutations detected in this study is not due to generalized genetic instability in prostate tumors or instability of the X chromosome. This finding and the association between the presence of mutations in the AR gene in the primary tumor and the rapid failure of subsequent hormonal therapy suggest that specific mechanisms for mutation of the AR gene are involved in the progression of human prostate cancers.

Amino acid substitutions in two exons of the AR were detected in four of the tumors in the present study. These findings may be the result of two populations of cells within tumor specimens that each contain a single amino acid substi- tution or a single population of cells with two amino acid substitutions. Where base changes occurred in different PCR- amplified fragments, these two alternatives could not be distin- guished. Multiple amino acid substitutions in the DNA-binding domain of the AR gene have been documented previously in a

patient with AIS (34), in the ligand-binding domain of the AR in an androgen-independent prostate cancer metastasis (16), and in a latent prostatic cancer (35). Similarly, multiple amino acid substitutions have been described in the 5oL-reductase type 2 gene in cases of 5a-reductase deficiency (36) and in the p53 gene in an individual breast tumor (37). In vitro studies of the functional consequences of two amino acid substitutions in codons 595 and 615 of the AR in a patient with incomplete AIS demonstrated that the alteration at codon 595 was able to par- tially restore DNA-binding activity to mutant AR that contained an inactivating amino acid substitution at codon 615 (34). Thus, the functional consequences of more than one alteration in the coding region of the AR in vivo will most likely be determined by the extent of the interaction between the amino acids in- volved with other receptor accessory proteins that are required for ligand-binding and/or transcriptional activity.

Although AR gene mutations have not been described previously in BPH, base changes resulting in two amino acid substitutions in the AR gene in 1 of 12 clinical BPH specimens were detected in the present study. Functional consequences of the amino acid substitutions in the BPH tissues have not yet been determined. Histopathological examination of specimens confirmed the diagnosis of BPH, and the patient has remained asymptomatic >2 years following transurethral prostatic resec- tion. Although it is not possible to determine the cell types or proportion of cells containing AR gene mutations identified in this study, previous studies have detected foci of cancer cells in up to 8% of surgical BPH specimens (38). Therefore, the pres- ence of cancer cells containing AR gene mutations in the tissues provided for SSCP and DNA sequence analysis or the existence of benign hyperplastic prostate (epithelial or stromal) cells con- taining AR gene mutations cannot be eliminated in the present study.

In contrast to MS where mutations in the AR gene usually result in loss of ligand-binding and/or AR function (39, 40), the mutations described thus far in prostate cancers appear to cause a broadening of ligand specificity and/or activation of AR by estrogens, progestins, adrenal androgens, or antiandrogens (e.g., flutamide) in addition to testicular androgens (9, 16). As such, these mutant ARs may contribute to the proliferation of prostate tumors in an androgen-depleted environment (e.g., following orchidectomy) or during antiandrogen therapy. The flutamide withdrawal syndrome (41), in which tumor growth appears to be paradoxically driven by the antiandrogen flutamide, may also be explained by the presence of mutant ARs in prostate tumor cells which are transcriptionally activated rather than inhibited by flutamide similar to the mutant AR expressed in the LNCaP cell line.

The exon 1 polymorphic CAG repeat region in the AR gene of patient p408, a white Caucasian male, contained 13 glutamine residues. In population studies of normal individuals, 11-31 CAG repeats have been reported (42). The most commonly observed CAG repeat frequency in white Caucasian men was 21 (range, 15-31), whereas the repeat number in black American males was significantly lower (i.e., n = 18; range, 11-29). In vitro studies indicate that elimination of the CAG tract of the AR results in increased transcriptional activity of the receptor (43). In contrast, expansion of the CAG repeat region, as ob- served in Kennedy's disease, results in partial loss of AR func-

284 AR Gene Mutations in Prostate Cancer

tion (44). Based on these studies, the increased risk of develop- ing prostate cancer in black American males has been proposed to be related to a reduced frequency of CAG repeats in this population (42). Similarly, it is feasible that the low number of glutamine residues in patient p408 may have contributed to the development of his prostate cancer. A reduction in CAG repeat number in exon 1 of the AR from 24 to 18 has been reported previously in the tumor of a prostate cancer patient with meta- static disease who exhibited a paradoxical agonist response to flutamide treatment (31). In that study, however, the entire AR coding region was not analyzed to eliminate the possibility of an amino acid substitution in the hormone-binding domain, which could account for the agonist action of flutamide.

The increased detection of AR gene mutations in prostate tumors as compared to BPH specimens suggests that tumor cells contain the receptor mutations. However, with the exception of a single tumor (p346), where evidence of a truncated AR protein was detectable in the majority of tumor cells by differential staining with AR antibodies directed against the NH 2 and COOH termini of the receptor, the proportion of cells (malig- nant and/or nonmalignant) expressing mutant ARs could not be deduced using the current methodology. AR protein expression in tumor cells ranged from almost negative to strongly positive. Although no causal relationship has been established between the presence of AR gene mutations identified in the present study and reduced AR expression in the tumors, significantly lower levels of immunoreactive AR were observed in prostate tumors containing amino acid substitutions in the AR. Similar findings have been reported in individuals with AIS where mutations in the AR were associated with decreased expression of AR protein (45). Conversely, reduced cellular AR protein levels have also been documented following ligand-binding and receptor activation (46). Therefore, it is feasible that decreased AR levels in prostate cancer specimens containing mutant re- ceptors may reflect ligand-mediated activation rather than al- tered transcription, translation, or stability of the mutant AR. Nevertheless, the findings of the present study suggest that the androgen responsiveness of prostate tumors may be determined by both the functional consequences of the AR gene mutations and the cellular levels of receptor expression. This conclusion is supported by the association of amino acid substitutions in the AR and reduced receptor immunostaining with the rapid failure of hormonal therapy.

The present study indicates that AR gene mutations exist in a subset of advanced stage primary prostate tumors prior to initiation of hormonal therapy, and that within a cohort of 20 of these patients able to be followed up clinically, the presence of AR gene mutations was associated with failure of subsequent hormonal therapy. These findings support the hypothesis that the often rapid onset of androgen-independent progression of prostate cancer may arise due to selective outgrowth of existing cells with mutant AR rather than the acquisition of new muta- tions following the initiation of treatment. Future quantitation of the proportion of cells expressing mutant ARs within prostatic tumors prior to initiation of therapy and in new metastatic tumors arising following failure of hormonal treatment will confirm whether prostate cancer progression is at least in part due to selective outgrowth of cells expressing AR gene muta- tions.

Acknowledgments We thank Professor Villis Marshall for his continuing support of

this research, and Professors Grant Sutherland and John Funder and Drs. David Horsfall and Rosemary Hall for critical evaluation of the manu- script. We also gratefully acknowledge the excellent technical assistance of Michele Grimbaldeston. AR antibodies were generously donated by Drs. C. M. Wilson, M. J. McPhaul, and J. D. Wilson (Department of Internal Medicine, University of Texas Southwestern Medical Center, Dallas, TX).

References 1. Boring, C. C., Squires, T. S., and Tong, T. Cancer statistics, 1993. CA Cancer J. Clin., 43: 7-26, 1993. 2. Santen, R. J. Clinical review 37: endocrine treatment of prostate cancer. J. Clin. Endocrinol. Metab., 75: 685-689, 1992. 3. Kozlowski, J. M., Ellis, W. J., and Grayhack, J. T. Advanced pros- tatic carcinoma: early v's late endocrine therapy. In: G. L. Andriole and W. J. Catalona (eds.), The Urologic Clinics of North America, pp. 15-24. Philadelphia: W. B. Saunders Co., 1991. 4. O'Malley, B. The steroid receptor superfamily: more excitement predicted for the future. Mol. Endocrinol., 4: 363-369, 1990. 5. Tilley, W. D., Marcelli, M., Wilson, J. D., and McPhaul, M. J. Characterization and expression of a cDNA encoding the human andro- gen receptor. Proc. Natl. Acad. Sci. USA, 86: 327-331, 1989. 6. Tilley, W. D., Wilson, C. M., Marcelli, M., and McPhaul, M. J. Androgen receptor gene expression in human prostate carcinoma cell lines. Cancer Res., 50: 5382-5386, 1990. 7. Quarmby, V. E., Beckman, W. J., Cooke, D. B., Lubahn, D. B., Joseph, D. R., Wilson, E. M., and French, F. S. Expression and local- ization of androgen receptor in the R-3327 Dunning rat prostatic ade- nocarcinoma. Cancer Res., 50: 735-739, 1990. 8. 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. 9. Klocker, H., Culig, Z., Hobisch, A., Cato, A. C. B., and Bartsch, G. Androgen receptor alterations in prostatic carcinoma. Prostate, 25: 266- 273, 1994. 10. Thompson, T. C. Growth factors and oncogenes in prostate cancer. Cancer Cells, 2: 345-354, 1990. 11. van der Kwast, T. H., Schalken, J., de Winter, J. A., van Vroon- hoven, C. C., Mulder, E., Boersma, W., and Trapman, J. Androgen receptors in endocrine-therapy-resistant human prostate cancer. Int. J. Cancer, 48: 189-193, 1991. 12. Newmark, J. R., Hardy, D. O., Tonb, D. C., Carter, B. S., Epstein, J. I., Isaacs, W. B., Brown, T. R., and Barrack, E. R. Androgen receptor gene mutations in human prostate cancer. Proc. Natl. Acad. Sci. USA, 89: 6319-6323, 1992. 13. Culig, Z., Hobisch, A., Cronauer, M. V., Cato, A. C. B., Hittmair, A., Radmayr, C., Eberle, J., Bartsch, G., and Klocker, H. Mutant androgen receptor detected in an advanced-stage prostatic carcinoma is activated by adrenal androgens and progesterone. Mol. Endocrinol., 7: 1541-1550, 1993. 14. Gaddipati, J. P., Mcleod, D. G., Heidenberg, H. B., Sesterhenn, I. A., Finger, M. J., Moul, J. W., and Srivastava, S. Frequent detection of codon 877 mutation in the androgen receptor gene in advanced prostate cancers. Cancer Res., 54: 2861-2864, 1994. 15. Suzuki, H., Sato, N., Watabe, Y., Masai, M., Seino, S., and Shimazaki, J. Androgen receptor gene mutations in human prostate cancer. J. Steroid Biochem. Mol. Biol., 46: 759-765, 1993. 16. 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. 17. Castagnaro, M., Yandell, D., Dockhorn-Dworniczak, Wolfe, H., and Poremba, C. Androgen receptor gene mutations and p53 gene

Clinical Cancer Research 285

analysis in advanced prostate cancer. Verh. Dtsch. Ges. Pathol., 77: 119-123, 1993.

18. Khoury, S. Staging of prostate cancer. In: S. Khoury, C. Chatelain, G. Murphy, and L. Denis (eds.), Urology: Prostate Cancer, pp. 149-153. France: FIIS, 1990. 19. Tilley, W. D., Lim-Tio, S. S., Horsfall, D. J., Aspinall, J. O., Marshall, V. R., and Skinner, J. M. Detection of discrete androgen receptor epitopes in prostate cancer by immunostaining: measurement by color video image analysis. Cancer Res., 54: 4096-4102, 1994.

20. Sambrook, J., Fritsch, E. F., and Maniatis, T. Molecular Cloning: A Laboratory Manual, pp. 9.14-9.23. Plainview, NY: Cold Spring Harbor Laboratory Press, 1989. 21. Marcelli, M., Tilley, W. D., Wilson, C. M., Griffin, J. E., Wilson, J. D., and McPhaul, M. J. Definition of the human androgen receptor gene structure permits the identification of mutations that cause andro- gen resistance: premature termination of the receptor protein at amino acid residue 588 causes complete androgen resistance. Mol. Endocrinol., 4: 1105-1116, 1990. 22. Lubahn, D. B., Brown, T. R., Simental, J. A., Higgs, H. N., Migeon, C. J., Wilson, E. M., and French, F. S. Sequence of the intron/exon junctions of the coding region of the human androgen receptor gene and identification of a point mutation in a family with complete androgen insensitivity. Proc. Natl. Acad. Sci. USA, 86: 9534-9538, 1989.

23. La Spada, A. R., Wilson, E., Lubahn, D., Harding, A., and Fisch- beck, K. H. Androgen receptor gene mutations in X-linked spinal and bulbar muscular atrophy. Nature (Lond.), 352: 77-79, 1991.

24. Hayashi, K. PCR-SSCP: a simple and sensitive method for detec- tion of mutations in the genomic DNA. PCR Methods. Applic., 1: 34-38, 1991. 25. Edwards, A., Voss, H., Rice, P., Civitello, A., Stegemann, J., Schwager, C., Zimmermann, J., Erfle, H., Caskey, C. T., and Ansorge, W. Automated DNA sequencing of the human HPRT locus. Genomics, 6: 593-608, 1990. 26. Orita, M., Suzuki, Y., Sekiya, T., and Hayashi, K. Rapid and sensitive detection of point mutations and DNA polymorphisms using the polymerase chain reaction. Genomics, 5: 874-879, 1989. 27. Husmann, D. A., Wilson, C. M., McPhaul, M. J., Tilley, W. D., and Wilson, J. D. Antipeptide antibodies to two distinct regions of the androgen receptor localize the receptor protein to the nuclei of target cells in the rat and human prostate. Endocrinology, 126: 2359-2368, 1990. 28. Greenberg, N. M., Demayo, F., Finegold, M. J., Medina, D., Tilley, W. D., Aspinall, J. O., Cunha, G. R., Donjacour, A. A., Matusik, R. J., and Rosen, J. M. Prostate cancer in a transgenic mouse. Proc. Natl. Acad. Sci. USA, 92: 3439-3443, 1995. 29. Shi, S. R., Chaiwun, B., Young, L., Cote, R. J., and Taylor, C. R. Antigen retrieval technique utilizing citrate buffer or urea solution for immunohistochemical demonstration of androgen receptor in formalin- fixed paraffin sections. J. Histochem. Cytochem., 41: 1599-1604, 1993. 30. Veldscholte, J. C. R., Kuiper, G. G., Jenster, G., Berrevoets, C., Claassen, E., vanRooij, H. C., Trapman, J., Brinkmann, A. O., and Mulder, E. A mutation in the ligand binding domain of the androgen receptor of human LNCaP cells affects steroid binding characteristics and response to anti-androgens. Biochem. Biophys. Res. Commun. 173: 534-540, 1990. 31. Schoenberg, M. P., Hakimi, J. M., Wang, S. P., Bova, G. S., Epstein, J. I., Fischbeck, K. H., Isaacs, W. B., Walsh, P. C., and Barrack, E. R. Microsatellite mutation (CAG(24-18)) in the androgen receptor gene in human prostate cancer. Biochem. Biophys. Res. Commun., 198: 74-80, 1994.

32. Jenster, G., Vanderkorput, H. A. G. M., Trapman, J., and Brink- mann, 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.

33. Cariello, N. F., and Skopek, T. R. Analysis of mutations occurring at the human HPRT locus. J. Mol. Biol., 231: 41-57, 1993.

34. Zoppi, S., Marcelli, M., Deslypere, J. P., Griffin, J. E., Wilson, J. D., and McPhaul, M. J. Amino acid substitutions in the DNA-binding domain of the human androgen receptor are a frequent cause of receptor- binding positive androgen resistance. Mol. Endocrinol., 6: 409-415, 1992.

35. Takahashi, H., Furusato, M., Allsbrook, W. C., Nishii, H., Wakui, S., Barrett, J. C., and Boyd, J. Prevalence of androgen receptor gene mutations in latent prostatic carcinomas from Japanese men. Cancer Res., 55: 1621-1624, 1995.

36. Thigpen, A. E., Davis, D. L., Milatovich, A., Mendonca, B. B., Imperato, M. J., Griffin, J. E., Francke, U., Wilson, J. D., and Russell, D. W. Molecular genetics of steroid 5 alpha-reductase 2 deficiency. J. Clin. Invest., 90: 799-809, 1992.

37. Mazars, R., Spinardi, L., BenCheikh, M., Simony Lafontaine, J., Jeanteur, P., and Theillet, C. p53 mutations occur in aggressive breast cancer. Cancer Res., 52: 3918-3923, 1992.

38. Mebust, W. K., Holtgrewe, H. L., Cockett, A. T. K., and Peters, P. C. Transurethral prostatectomy: immediate and postoperative com- plications: a cooperative study of 13 participating institutions evaluating 3,885 patients. J. Urol., 141: 243-247, 1989.

39. McPhaul, M. J., Marcelli, M., Zoppi, S., Wilson, C. M., Griffin, J. E., and Wilson, J. D. Mutations in the ligand-binding domain of the androgen receptor gene cluster in two regions of the gene. J. Clin. Invest., 90: 2097-2101, 1992.

40. Sultan, C., Lumbroso, S., Poujol, N., Belon, C., Boudon, C., and Lobaccaro, J. M. Mutations of androgen receptor gene in androgen insensitivity syndromes. J. Steroid Biochem. Mol. Biol., 46: 519-530, 1993.

41. Scher, H. I., and Kelly, W. K. Flutamide withdrawal syndrome-- its impact on clinical trials in hormone-refractory prostate cancer. J. Clin. Oncol., 11: 1566-1572, 1993.

42. Coetzee, G. A., and Ross, R. K. Prostate cancer and the androgen receptor. J. Natl. Cancer Inst. 86: 872-873, 1994.

43. Chamberlain, N. L., Driver, E. D., and Miesfeld, R. L. The length and location of CAG trinucleotide repeats in the androgen receptor N-terminal domain affect transactivation function. Nucleic Acids Res., 22: 3181-3186, 1994.

44. Mhatre, A. N., Trifiro, M. A., Kaufman, M., Kazemiesfarjani, P., Figlewicz, D., Rouleau, G., and Pinsky, L. Reduced transcriptional regulatory competence of the androgen receptor in x-linked spinal and bulbar muscular atrophy. Nat. Genet., 5: 184-188, 1993.

45. Marcelli, M., Tilley, W. D., Zoppi, S., Griffin, J. E., Wilson, J. D., and McPhaul, M. J. Androgen resistance associated with a mutation of the androgen receptor at amino acid 772 (Arg--Cys) results from a combination of decreased messenger ribonucleic acid levels and impair- ment of receptor function. J. Clin. Endocrinol. Metab., 73: 318-325, 1991. 46. Shan, L. X., Rodriguez, M. C., and Janne, O. A. Regulation of androgen receptor protein and mRNA concentrations by androgens in rat ventral prostate and seminal vesicles and in human hepatoma cells. Mol. Endocrinol., 4: 1636-1646, 1990.