Distinct E€Ects of PIAS Proteins on Androgen-Mediated Gene Activation

Distinct eects of PIAS proteins on androgen-mediated gene activation in prostate cancer cells

Mitchell Gross1, Bin Liu1, Jiann-an Tan3, Frank S French3, Michael Carey2 and Ke Shuai*,1,2

1Division of Hematology-Oncology, Department of Medicine, University of California, Los Angeles, California, CA 90095, USA; 2Department of Biological Chemistry, University of California, Los Angeles, California, CA 90095, USA; 3The Laboratories for Reproductive Biology, Department of Pediatrics, University of North Carolina School of Medicine, Chapel Hill, North Carolina, NC 27599-7500, USA

Androgen signaling in¯uences the development and The androgen receptor (AR) is a member of the growth of prostate carcinoma. The transcriptional nuclear receptor (NR) superfamily (Bentel and Tilley, activity of androgen receptor (AR) is regulated by 1996). NRs have conserved domain structures (Freed- positive or negative transcriptional cofactors. We report man, 1999). At the N-terminus is the ligand- here that PIAS1, PIAS3, and PIASy of the protein independent transcriptional activation domain (AF- inhibitor of activated STAT (PIAS) family, which are 1). The central domain contains two zinc ®nger expressed in human prostate, display distinct eects on structures that are involved in DNA binding (DBD). AR-mediated gene activation in prostate cancer cells. The C-terminal region is the ligand binding domain While PIAS1 and PIAS3 enhance the transcriptional (LBD). In addition, an essential ligand-dependent activity of AR, PIASy acts as a potent inhibitor of AR transactivation domain (AF-2) is localized in the C- in prostate cancer cells. The eects of PIAS proteins on terminus of the LBD. The binding of androgen to AR AR are competitive. PIASy binds to AR but does not triggers the contact of AR to its speci®c androgen aect the DNA binding activity of AR. An NH2- response elements (AREs) in the promoters of terminal LXXLL signature motif of PIASy, although not androgen-responsive genes and the interactions of required for PIASy ± AR interaction, is essential for the AR with other transcription cofactors. These interac- transrepression activity of PIASy. Our results identify tions allow the activation or repression of genes that PIASy as a transcriptional corepressor of AR and regulate development, dierentiation and maintenance suggest that dierent PIAS proteins have distinct eects of male reproductive functions (Trapman and Cleut- on AR signaling in prostate cancer cells. Oncogene jens, 1997). (2001) 20, 3880 ± 3887. The activity of NR is regulated by transcriptional coactivators and corepressors (Freedman, 1999; Hor- Keywords: prostate cancer; androgen signaling; protein witz et al., 1996; Torchia et al., 1998). These inhibitor of activated STAT coregulator complexes modulate NR-mediated gene activation by interaction with general transcription factors or by remodeling chromatin. Many coregula- Introduction tors of AR have been described, most of them are suggested to function as coactivators of AR (a detailed Prostate cancer is the most frequently diagnosed cancer description of these factors can be found at http:// in American men (Parker et al., 1997). Androgen plays www.mcgill.ca/androgendb/). Many nuclear coactiva- an important role in the development and growth of tors contain one or multiple copies of the a-helical prostate carcinoma (Kokontis and Liao, 1999). Andro- LXXLL signature motif (where L is Leucine and X is gen ablation is an eective treatment for advanced any amino acid residue). Biochemical studies have prostate cancer. Unfortunately, after a period of demonstrated that the LXXLL signature motif med- remission, a relapse occurs due to the acquisition of iates ligand-dependent coactivator-NR interactions or androgen-independent tumor growth. A better under- coactivator ± coactivator interactions (Heery et al., standing of the regulation of androgen signaling in 1997; Torchia et al., 1997). human prostate cancer cells will enhance our ability to In studies aimed at the understanding of STAT design a rational therapeutic strategy for prostate (Signal Transducer and Activator of Transcription) cancer treatment. signaling (Darnell, 1997), our laboratory has uncovered a novel protein family termed PIAS (Protein Inhibitor of Activated STAT), of which four members have been identi®ed ± PIAS1, PIAS3, PIASx (consisting of two *Correspondence: K Shuai, Division of Hematology-Oncology, 11- splicing variants: PIASxa and PIASxb), and PIASy 934 Factor Bldg, 10833 LeConte Avenue, Los Angeles, California, (Shuai, 1999; 2000). PIAS1 and PIAS3 interact with CA 9005-1678, USA Received 10 January 2001; revised 20 March 2001; accepted 26 Stat1 and Stat3, respectively. PIAS can block the DNA March 2001 binding activity of STAT and can inhibit STAT- PIAS proteins and androgen-mediated gene-activation M Gross et al 3881 mediated gene activation (Chung et al., 1997; Liu et al., 1998). Recently, two independent yeast two-hybrid screens using the DNA binding domain of AR as the bait have identi®ed ARIP3 (Androgen Receptor Interacting Protein 3), a rat homologue of human PIASxa,as well as PIAS1 as AR-interacting proteins (Moilanen et al., 1999; Tan et al., 2000). Subsequent studies suggest that ARIP3/PIASxa and PIAS1 can function as coactivators of AR. We have carried out a systematic analysis on the roles of various PIAS members of AR signaling in prostate cancer cells. We found that PIAS1, PIAS3 and PIASy are expressed in human prostate. PIASy, unlike other PIAS proteins, is a potent inhibitor of AR-mediated gene activation in prostate cancer cells. PIASy interacts with AR but does not aect the DNA binding activity of AR. Mutational analysis identi®ed an essential function of an LXXLL signature motif at the N-terminus of PIASy for the transrepression activity of PIASy. Our results suggest that PIASy is a transcriptional corepressor of AR. Thus, androgen signaling can be positively or negatively regulated by PIAS proteins in prostate cancer cells.

Results

PIAS1, PIAS3 and PIASy are expressed in human prostate To study the role of PIAS proteins in prostate cancer, we ®rst performed Northern blot analysis to determine which PIAS family members are expressed in human prostate. It has been shown previously that PIAS3 is ubiquitously expressed in a variety of human tissues including prostate (Chung et al., 1997). A human tissue blot was probed for the expression of PIAS1, PIASx and PIASy. Two messages with the sizes of 2.3 and Figure 1 Expression of PIAS in human tissues and cell lines. A 3.5 kb, probably resulting from dierent RNA proces- human tissue blot (Clontech) was hybridized with radiolabeled sing, were detected in various human tissues including human PIASy cDNA (a), human PIAS1 cDNA (b), or a 200 bp prostate when the blot was probed with PIASy cDNA sequence of the 3'-end of the human PIASxa cDNA (c), following (Figure 1a). Similarly, two major forms of PIAS1 manufacturer's instructions. (d) Immunoblot analysis of LNCaP cytoplasmic (Cyt) and nuclear (Nuc) extracts with PIASy, PIAS1 messages with the sizes of 2.6 and 7.4 kb were present and PIAS3 antibodies as indicated in prostate and other tissues (Figure 1b). However, PIASxa is mainly expressed in human testis (Figure 1c), consistent with the result obtained when ARIP3, a rat homologue of PIASxa, was used as the probe be detected in the nuclear fractions (Figure 1d). The (Moilanen et al., 1999) (expression of PIASxa was nuclear locatization of PIAS1 is consistent with the detectable in other tissues when the blot was over- pattern seen in the germinal epithelium of human testis exposed (data not shown)). A similar expression by immunohistochemistry using the same PIAS1 anti- pattern of PIASxb was also detected (data not shown). body (Tan et al., 2000). We conclude that PIAS1, PIAS3, and PIASy are expressed in human prostate tissue. Distinct effects of PIAS proteins on AR-mediated gene We next wanted to con®rm that PIAS proteins are activation in prostate cancer cells indeed expressed in prostate epithelial cells. LNCaP human prostate cancer cells were subjected to We examined the eects of PIAS1, PIAS3 and PIASy biochemical fractionation, and expression of PIASy, proteins on AR-mediated gene activation in the PIAS3, and PIAS1 was examined by Western blotting LNCaP cell line. LNCaP cells express a mutated AR using speci®c anti-PIAS antibodies (Chung et al., 1997; but androgen response appears to be normal in these Liu et al., 1998, 2001). PIASy, PIAS1 and PIAS3 can cells (McDonald et al., 2000). Expression vectors

Oncogene PIAS proteins and androgen-mediated gene-activation M Gross et al 3882 encoding Flag-tagged PIAS1, PIAS3 or PIASy were PIASy does not influence AR DNA bindings cotransfected together with a wild-type AR expression vector and a luciferase reporter (PSA-E4 Luc) with PIAS1 and PIAS3 can block the DNA binding activity the androgen-responsive promoter of PSA (Prostate of STATs in vitro (Chung et al., 1997; Liu et al., 1998). Speci®c Antigen) (Figure 2). Transfected cells were We wished to determine if PIASy aects the DNA untreated or treated with R1881 (methyltrienolone, a binding activity of AR by electrophoretic mobility shift synthetic androgen). We found that in the absence of analysis (EMSA). An anity puri®ed His-tagged DNA PIAS proteins, R1881 induced a 10 ± 15-fold activation binding domain of AR (His-DBD) was incubated with of the PSA reporter. In the presence of an increasing 32P-labeled ARE oligonucleotide in the presence or amount of PIAS1, the R1881-induced luciferase absence of GST or GST-PIASy fusion proteins. The activity was dramatically enhanced, up to 70-fold when DNA binding ability of AR was not aected by either 2 mg of PIAS1 was employed (Figure 2a). PIAS3 GST or GST-PIASy at various concentrations used showed similar eects as PIAS1 (Figure 2b). Most (Figure 5a). The lack of a PIASy-AR-DBD supershift interestingly, unlike other PIAS proteins, PIASy caused band was probably due to the instability of such a a dose-dependent inhibition of R1881-induced lucifer- complex under the conditions used in these assays. The ase activity. In fact, when 1 mg of PIASy was used, the His-tagged AR-DBD showed no binding to a mutant AR-mediated gene activation was completely inhibited ARE oligonucleotide, con®rming the binding speci®city (Figure 2c). PIAS proteins did not show signi®cant of His-tagged AR-DBD (Figure 5b). We conclude that eects on the basal level of reporter transcription in the PIASy does not aect the DNA binding activity of absence of R1881 treatment (data not shown). These AR. results suggest that PIASy, in contrast to other PIAS proteins, is a potent inhibitor of AR signaling. Thus, The PIASy LXXLL motif is required for repression dierent members of the PIAS family have distinct eects on AR-mediated gene activation in prostate A putative a-helical LXXLL signature motif is present cancer cells. at the NH2-terminal of PIASy (Figure 6a). The LXXLL motif has been shown to mediate NR- coactivator or coactivator ± coactivator interactions. Competitive effects of PIASy and PIAS1 on AR-mediated We next tested if the PIASy LXXLL motif is involved gene activation in PIASy-AR interaction by in vitro GST pull-down To test the hypothesis that PIASy and other PIAS analysis. A mutant PIASy (PIASy-AA) with mutations proteins can in¯uence androgen response in a in the LXXLL motif (LXXLL to LXXAA) was competitive fashion, LNCaP cells were transiently generated. Both 35S-labeled wild-type PIASy and tranfected with the PSA luciferase reporter together mutant PIASy-AA proteins bound to GST-AR-DBD with AR, PIAS1, and PIASy. The dramatic activation but not GST beads (Figure 6b, the upper panel). of AR-mediated transcription in the presence of Similar amounts of GST and GST-AR-DBD were PIAS1 was abolished in the presence of PIASy (Figure present in each sample (Figure 6b, the lower panel). 3). These results imply that the relative expression Thus, the LXXLL signature motif is not involved in level of PIASy and other PIAS proteins in prostate PIASy-AR interaction. cancer cells can in¯uence the overall androgen To test if the LXXLL motif is required for the response. transrepression activity of PIASy, LNCaP cells were transiently transfected with the PSA luciferase reporter together with wild-type PIASy or PIASy-AA mutant. The NH2-terminal portion of PIASy interacts with the While wild-type PIASy eectively inhibited the R1881- DBD domain of AR induced luciferase activity, the PIASy-AA mutant It has been previously shown that ARIP3/PIASxa and protein completely lost its ability to inhibit AR- PIAS1 can interact with the DBD of AR (Moilanen et mediated gene activation (Figure 7a). To further al., 1999; Tan et al., 2000). We next tested if PIASy con®rm these results, similar luciferase assays were can also interact with the DBD of AR by in vitro performed in 293T cells which are known to allow for binding assays. Glutathione-S-transferase (GST) or ecient transfection. Again, wild-type PIASy, but not GST fused to AR-DBD (GST-AR-DBD) proteins PIASy-AA mutant, was able to inhibit the R1881- were bound to glutathione-agarose beads and mixed induced luciferase activity in a dose dependent manner with 35S-labeled PIASy. 35S-labeled luciferase protein (Figure 7b, upper panel). Interestingly, PIASy-AA was used as a control. A speci®c interaction between mutant protein enhanced the reporter activity when AR-DBD and PIASy was observed (Figure 4a). When expressed at higher levels. This dominant negative a portion of PIASy containing the NH2-terminal 406 eect of PIASy-AA, which is able to interact with AR amino acid residues (PIASy-N) was tested, the AR- but lacks the transrepression activity, may result from PIASy interaction remained intact (Figure 4a). Similar the competition of PIASy-AA with the endogenous amounts of GST and GST-AR-DBD were present on PIASy for binding to AR. Protein blot analysis beads (Figure 4b). Thus, the NH2-terminal region of indicated that wild-type PIASy and mutant PIASy- PIASy can directly interact with the DBD domain of AA were expressed at comparable levels (Figure 7b, AR. lower panel). These results demonstrate that the

Oncogene PIAS proteins and androgen-mediated gene-activation M Gross et al 3883 LXXLL motif of PIASy, although not required for PIASy-AR interaction, is essential for PIASy to repress AR-mediated gene activation.

Discussion

To understand the role of PIAS proteins in androgen signaling in prostate cancer cells, we systematically examined and compared the eects of PIAS1, PIAS3 and PIASy, which are expressed in human prostate, on the transcriptional activity of AR in prostate cancer cells. Our studies revealed two signi®cant ®ndings: (1) PIAS proteins display distinct eects on AR-mediated gene activation in prostate cancer cells. PIAS1 and PIAS3 enhance while PIASy represses the transcriptional activity of AR. These results support a hypothesis that the relative level of PIAS expression in prostate cancer cells may contribute partly to the overall androgen response; (2) PIASy acts as a transcriptional corepressor of AR. The N-terminal LXXLL signature motif of PIASy is essential for the transrepression activity of PIASy. To the best of our knowledge, PIAS represents the ®rst known example of a family of related proteins that can both positively and negatively regulate AR- mediated transcriptional activity. PIAS proteins contain several conserved domains, including a zinc binding motif and an acidic region. In addition, all

Figure 2 Eects of PIAS proteins on AR-mediated gene activation in a human prostate carcinoma cell line. LNCaP cells were transiently transfected with empty expression plasmid, exogenous wild-type human AR expression plasmid, and increasing doses (0.1, 1.0 and 2.0 mg) of PIAS expression plasmids as indicated. PSA-E4 Luc and pRL-TK were included as reporter and control vectors, respectively. Cells were grown in phenol red- free RPMI with charcoal-stripped fetal bovine serum and Figure 3 Competitive eects of PIASy and PIAS1 on AR- harvested 24 h after stimulation. Fold induction was calculated mediated gene activation. LNCaP cells were transiently trans- from paired samples grown in the presence or absence of 10 nM fected with 1 mg PIAS1 and PIASy expression vectors as R1881. Total DNA content was kept constant in all wells. Results indicated. Exogenous hAR was expressed in all wells. PSA-E4 are displayed as the average of at least three independent Luc reporter and pRL-TK control plasmids were included as in experiments+s.e. Figure 2

Oncogene PIAS proteins and androgen-mediated gene-activation M Gross et al 3884

Figure 4 The PIASy interacts with AR-DBD. (a) Glutathione-agarose beads coated with approximately 100 mg GST or GST- ARDBD were incubated with 35S-labeled PIASy or a non-speci®c control protein (luciferase). The samples were washed and analysed by SDS ± PAGE and autoradiography. Input represents 10% of the added protein. (b) Coomassie blue staining of GST and GST-ARDBD used in (a)

Figure 6 The LXXLL signature motif is not required for PIASy-AR interaction. (a) Sequence of the LXXLL signature motif of PIASy. (b) GST pull-down experiments were performed as described in Figure 4 except that the LXXLL-AA mutant PIASy was also included in the assays as indicated

PIAS proteins, except PIASy, contain a serine/ threonine region at the COOH terminus (Shuai, 2000). The domain(s) responsible for the observed distinct eects of PIAS proteins on AR signaling is not known. Future studies on the molecular basis of the PIAS speci®city on androgen signaling should enhance our understanding of the regulation of androgen- mediated gene activation in prostate cancer cells. Figure 5 PIASy does not inhibit AR-DBD DNA-binding Several lines of evidence presented in this report activity. (a) Anity puri®ed AR-DBD (100 ng) was incubated suggest that PIASy may function as a transcriptional with 40 fmole ARE-consensus oligonucleotide alone or with corepressor of AR. First, PIASy directly interacts with puri®ed GST or GST-PIASy at 20, 200, 400 ng. Arrow indicates the AR-DBD/ARE speci®c complex. (b) AR-DBD incubated as the DBD domain of AR. Second, PIASy represses the in (a) with mutant ARE oligonucleotide transcriptional activity of AR without aecting the

Oncogene PIAS proteins and androgen-mediated gene-activation M Gross et al 3885 DNA binding activity of AR. Third, an NH2-terminal LXXLL signature motif is required for the transrepression activity of PIASy, but is not involved in PIASy-AR interaction. The LXXLL signature motif was originally identi®ed in several NR coactivators (Heery et al., 1997; Torchia et al., 1997). The LXXLL motif can mediate NR-coactivator or coactivator ± coactivator interactions (Darimont et al., 1998; Glass and Rosenfeld, 2000; McInerney et al., 1998; Neaear et al., 1999; Rachez et al., 1999). Recently, highly related sequences are also found to be present in NR corepressor (Hu and Lazar, 1999; Nagy et al., 1999; Perissi et al., 1999). It is possible that the LXXLL motif of PIASy is involved in interacting with other corepressors which are required for PIASy to repress transcription. Consistent with this hypothesis, the PIASy-AA mutant displays a dominant negative eect on AR-mediated gene activation (Figure 7). Further studies are required to understand the exact role of this LXXLL motif in PIASy-mediated repression on AR. PIAS proteins are known to be negative regulators of STATs. The discovery of the involvement of PIAS in AR signaling suggests that PIAS proteins may have a broad role in transcriptional regulation. In STAT signaling, PIAS1 and PIAS3 can inhibit the DNA binding activity of Stat1 and Stat3, respectively. In this paper, we showed that PIASy can inhibit AR-mediated gene activation without blocking the DNA binding activity of AR, suggesting PIASy is a corepressor of AR. Consistently, our most recent studies indicate that PIASy can also function as a corepressor of Stat1 (Liu et al., 2001). Thus, PIAS proteins can modulate transcription through distinct mechanisms. PIAS proteins interact with STATs only in cells stimulated with cytokines (Chung et al., 1997; Liu et al., 1998, 2001). The cytokine-dependent PIAS ± STAT interaction is due to the ability of PIAS to interact with the dimeric form, but not the monomeric form of STAT (Liao et al., 2000). Whether the PIAS ± AR interaction in prostate cancer cells requires androgen stimulation is not known. However, at least in yeast cells, the PIAS1 ± AR interaction is dependent on ligand stimulation (Tan et al., 2000). The association of PIAS with a transcription factor only upon ligand stimulation can allow the eective participation of PIAS in transcriptional regulation in response to a variety of cellular signals.

Figure 7 The LXXLL motif is required for the transrepression Materials and methods activity of PIASy activity in AR signaling. (a) LNCaP cells were transfected with empty expression vector and wild-type PIASy or Cells PIASy-AA expression plasmids along with PSA-E4 Luc reporter and human AR expression plasmids as indicated. Media was 293T and LNCaP cells were maintained in DMEM and changed to phenol red-free RPMI supplemented with charcoal- phenol red-free RPMI media, respectively. All media stripped fetal bovine serum with or without 10 nM R1881 for contained 10% fetal bovine serum (Life Technologies, Inc.), 24 h. Luciferase activity is expressed relative to Renilla luciferase 100 u/ml penicillin and 100 mg/ml streptomycin. control. (b) 293T cells were transfected with empty vector, and increasing amounts of PIASy, and PIASy-AA expression plasmids. Human AR and b-galactosidase expression plasmids Immunoblotting were also included. Results are expressed relative to b- galactosidase control. The same cell extracts from (b) were Rabbit polyclonal antibodies were raised against the carboxy- analysed by Western blot with anti-Flag (bottom panel) terminal regions of PIAS1, PIAS3, and PIASy as previously

Oncogene PIAS proteins and androgen-mediated gene-activation M Gross et al 3886 described (Chung et al., 1997; Liu et al., 2001; Tan et al., immunoblotting using the M2-FLAG monoclonal antibody 2000). LNCaP cells were separated by biochemical fractiona- (Sigma). tion into cytoplasmic and nuclear extracts as described (Shuai et al., 1992). Extracts were standardized for protein Electrophoretic mobility shift assay (EMSA) concentration, resolved on a 7.5% polyacrylamide-sodium dodecyl sulfate gel (SDS ± PAGE), and analysed by standard EMSA analysis was performed as described (Khan et al., immunoblotting procedures. 1993). Brie¯y, anity puri®ed AR-DBD produced in Escherichia coli (Huang et al., 1999) was incubated with or without GST-fusion protein for 5 min at room temperature Plasmids in gel shift buer. Approximately, 40 fmol oligonucleotide FLAG-PIASy was constructed by subcloning the full-length probe containing a consensus AR DNA binding site or human PIASy cDNA into the BglII ± SalI sites of pCMV- mutant AR DNA binding site (Huang et al., 1999) was added FLAG vector. FLAG-PIASy-AA was produced by PCR- and incubated for an additional 20 min at room temperature. based site-directed mutagenesis using primers to produce L23 The complexes were resolved on a 4.5% non-denaturing to A and L24 to A missense mutations and subcloned into polyacrylamide gel. Gels were dried and analysed by pCMV-FLAG. GST-PIASy, GST-PIASy-N (1 ± 406), and phosphoimaging (Molecular Dynamics, Inc.). GST-PIASy AA were produced by subcloning PIASy and PIASy AA cDNAs into pGEX4T-1 (Amersham Pharmacia GST pull-down assays Biotech, Inc., Piscataway, NJ, USA). The coding regions of all vectors were con®rmed by sequencing. PSA-E4 Luc The E. coli strain BL21 (DE3) was transformed with containing a 496-kp region from the PSA enhancer upstream pGEX4T-1 (Clontech) or pGST-AR-DBD. Overnight cul- of the adenovirus E4 core promoter and luciferase gene and tures were diluted into 0.5 liters of LB medium and grown to pET11dHis6ARDBD which codes for a His-tagged protein OD600 of 0.8 ± 1.0. Cultures were induced with 1 mM encompassing AR amino acids 544 ± 634 have been previously isopropyl-l-thio-b-D-galactopyranoside at 378C for 2 h and reported (Huang et al., 1999). The plasmid, pGST-AR-DBD, GST or GST-AR-DBD puri®ed with glutathione-agarose as which codes for glutathione-S-transferase fused to AR previously described (Chung et al., 1997). PIASy and PIASy- residues 544 ± 634 has also been previously described (Tan AA coding regions were cloned into pcDNA3.1 (Invitrogen, et al., 2000). Inc.) and 35S-labeled protein synthesized using TNT Quick Coupled Transcriptional/Translation kit (Promega Corp.). Equal amounts of agarose-immobilized GST or GST-AR- Transfections and reporter assays DBD were incubated with 20 mcl of protein mixture in PBS 293T were transfected via the calcium phosphate method (Liu (pH 7.5) containing 1 mg/ml BSA, 0.02% NP-40, 1 mg/ml et al., 1998). Brie¯y, cells were plated at 56105 cells per well aprotonin, 1 mg/ml leupeptin, 1 mM DTT and 1 mM PMSF into a 6-well plate and transfected with 1 ± 3 mg total DNA. essentially as previously described (Tan et al., 2000). Beads Empty vector DNA was included to equalize amount of total were thoroughly washed and bound proteins were removed DNA transfected in each experiment. After transfection, cells and analysed on a 10% SDS ± PAGE gel which was dried were grown in appropriate media supplemented with 10% and developed by ¯uorography. An aliquot of bound charcoal-dextran stripped fetal bovine serum (Omega Scien- material was analysed by Coomassie staining of a 15% ti®c). Cells were stimulated with methyltrienolone (R1881; SDS ± PAGE gel to con®rm equal protein loading. Dupon-NEN) for 18 ± 24 h before harvest. A b-galactosidase reporter vector was included in each well to adjust for transfection eciency. LNCaP cells were transfected with TFX-50 (Promega Corp., Madison, WI, USA) as previously described (Abreu-Martin et al., 1999). The Renilla luciferase Acknowledgments reporter vector, pRL-TK, was included in all wells to control We thank C Sawyers for discussion and reading the for transfection eciency. Fire¯y and Renilla luciferase manuscript; X Rao and M Sohn for technical assistance. activities were measured (Promega Corp). A portion of each B Liu is a Cancer Research Institute fellow. Supported by lysate was analysed on a 7.5% SDS ± PAGE gel followed by a grant from the NIH (AI43438).

References

Abreu-Martin MT, Chari A, Palladino AA, Craft NA and Horwitz KB, Jackson TA, Bain DL, Richer JK, Takimoto Sawyers CL. (1999). Mol. Cell. Biol., 19, 5143 ± 5154. GS and Tung L. (1996). Mol. Endocrinol., 10, 1167 ± 1177. Bentel JM and Tilley WD. (1996). J. Endocrinol., 151, 1 ± 11. Hu X and Lazar MA. (1999). Nature, 402, 93 ± 96. Chung CD, Liao J, Liu B, Rao X, Jay P, Berta P and Shuai Huang W, Shostak Y, Tarr P, Sawyers C and Carey M. K. (1997). Science, 278, 1803 ± 1805. (1999). J. Biol. Chem., 274, 25756 ± 25768. Darimont BD, Wagner RL, Apriletti JW, Stallcup MR, Khan KD, Shuai K, Lindwall G, Maher SE, Darnell Jr JE Kushner PJ, Baxter JD, Fletterick RJ and Yamamoto KR. and Bothwell AL. (1993). Proc. Natl. Acad. Sci. USA, 90, (1998). Genes Dev., 12, 3343 ± 3356. 6806 ± 6810. Darnell Jr JE. (1997). Science, 277, 1630 ± 1635. Kokontis JM and Liao S. (1999). Vitam. Horm., 55, 219 ± Freedman LP. (1999). Cell, 97, 5±8. 307. Glass CK and Rosenfeld MG. (2000). Genes Dev., 14, 121 ± Liao J, Fu Y and Shuai K. (2000). Proc. Natl. Acad. Sci. 141. USA, 97, 5267 ± 5272. Heery DM, Kalkhoven E, Hoare S and Parker MG. (1997). Liu B, Gross M, ten Hoeve J and Shuai K. (2001). Proc. Natl. Nature, 387, 733 ± 736. Acad.Sci.USA,98, 3203 ± 3207.

Oncogene PIAS proteins and androgen-mediated gene-activation M Gross et al 3887 LiuB,LiaoJ,RaoX,KushnerSA,ChungCD,ChangDD Rachez C, Lemon DB, Suldan Z, Bromleigh V, Gamble M, and Shuai K. (1998). Proc. Natl. Acad. Sci. USA, 95, Neaear AM, Erdjument-Bromage H, Tempst P and 10626 ± 10631. Freedman LP. (1999). Nature, 398, 824 ± 828. McDonald S, Brive L, Agus DB, Scher HI and Ely KR. Shuai K. (1999). Prog. Biophys. Mol. Biol., 71, 405 ± 422. (2000). Cancer Res., 60, 2317 ± 2322. Shuai K. (2000). Oncogene, 19, 2638 ± 2644. McInerney EM, Rose DW, Flynn SE, Westin S, Mullen TM, Shuai K, Schindler C, Prezioso VR and Darnell Jr JE. (1992). Krones A, Inostroza J, Torchia J, Nolte RT, Assa-Munt Science, 258, 1808 ± 1812. N, Milburn MV, Glass CK and Rosenfeld MG. (1998). TanJ,HallSH,HamilKG,GrossmanG,PetruszP,LiaoJ, Genes Dev., 12, 3357 ± 3368. Shuai K and French FS. (2000). Mol. Endocrinol., 14, 14 ± Moilanen AM, Karvonen U, Poukka H, Yan W, Toppari J, 24. Jeanne OA and Palvimo JJ. (1999). J. Biol. Chem., 274, Torchia J, Glass C and Rosenfeld MG. (1998). Curr. Opin. 3700 ± 3704. Cell. Biol., 10, 373 ± 383. Nagy L, Kao HY, Love JD, Li C, Banayo E, Gooch JT, Torchia J, Rose DW, Inostroza J, Kamei Y, Westin S, Glass Krishna V, Chatterjee K, Evans RM and Schwabe JW. CK and Rosenfeld MG. (1997). Nature, 387, 677 ± 684. (1999). Genes Dev., 13, 3209 ± 3216. Trapman J. and Cleutjens KB. (1997). Semin. Cancer Biol., 8, Neaear AM, Beaurang PA, Zhou S, Abraham S, Solomon W 29 ± 36. and Tjian R. (1999). Nature, 398, 828 ± 832. Parker SL, Tong T, Bolden S and Wingo PA. (1997). CA Cancer J. Clin., 47, 5 ± 27. Perissi V, Staszewski LM, McInerney EM, Kurokawa R, Krones A, Rose DW, Lambert MH, Milburn MV, Glass CK and Rosenfeld MG. (1999). Genes Dev., 13, 3198 ± 3208.

Oncogene