|Home | About | Journals | Submit | Contact Us | Français|
Prostate cancer progression to the androgen-independent (AI) state involves acquisition of pathways that allow tumor growth under low-androgen conditions. We hypothesized that expression of molecular chaperones that modulate androgen binding to AR might be altered in prostate cancer and contribute to progression to the AI state. Here, we report that the Hsp90 cochaperone FKBP51 is upregulated in LAPC-4 AI tumors grown in castrated mice and describe a molecular mechanism by which FKBP51 regulates AR activity. Using recombinant proteins, we show that FKBP51 stimulates recruitment of the cochaperone p23 to the ATP-bound form of Hsp90, forming an FKBP51-Hsp90-p23 superchaperone complex. In cells, FKBP51 expression promotes superchaperone complex association with AR and increases the number of AR molecules that undergo androgen binding. FKBP51 stimulates androgen-dependent transcription and cell growth, and FKBP51 is part of a positive feedback loop that is regulated by AR and androgen. Finally, depleting FKBP51 levels by short hairpin RNA reduces the transcript levels of genes regulated by AR and androgen. Because the superchaperone complex plays a critical role in determining the ligand-binding competence and transcription function of AR, it provides an attractive target for inhibiting AR activity in prostate cancer cells.
Defining the etiology of prostate cancer remains an area of active investigation, but there is a growing focus on the role of the androgen receptor (AR) in the context of “androgen-independent” (AI) signaling during disease progression (5). Like other members of the nuclear receptor superfamily, AR is a transcription factor that regulates the expression of scores of genes important for cell growth and development. Nuclear receptors have a conserved domain organization that includes an N-terminal AF-1 domain, a central DNA-binding domain, a hinge region, and a C-terminal ligand-binding domain (LBD). AR activity is normally regulated by androgen binding to the LBD, an event that initiates changes in AR conformation, subcellular localization, and interactions with cofactors that facilitate transcription from target genes (31). AI prostate cancers can grow in a low-androgen environment, which occurs in the setting of androgen ablation therapy in patients and surgical castration in animal models.
Ligand binding to nuclear receptors, including AR, is regulated by molecular chaperones. The central player in this reaction is Hsp90, an ATP-utilizing chaperone that interacts transiently with LBDs to stabilize a conformation that is appropriate for ligand binding (24). Hsp90 operates as a molecular chaperone for numerous client proteins by cycling through conformational changes coupled to ATP binding and hydrolysis (19). Additionally, accessory proteins termed cochaperones facilitate or stabilize changes in Hsp90 conformation and ATPase activity. One well-studied cochaperone originally identified in progesterone receptor (PR) complexes is p23 (15). Through selective binding to the ATP form of Hsp90, p23 stabilizes the “closed” state and promotes Hsp90 interactions with client proteins (14, 33). p23 function is critical for steroid hormone receptors, as it has been shown to act as a limiting component for assembly of the multicomponent Hsp90 “superchaperone” complex that stabilizes the LBD in a conformation that is competent for hormone binding (15). p23 function has also been linked to later steps of nuclear receptor function, including transcription complex disassembly (8).
Less well understood are the cochaperones that belong to the immunophilin class of proteins, FK506-binding proteins FKBP51 and FKBP52, and cyclosporine A-binding protein Cyp40. These cochaperones have N-terminal domains with peptidyl-prolyl isomerase activity and C-terminal domains that contain three tetratricopeptide repeats (TPR) that mediate binding to Hsp90 (26). The functions of FKBP51 and FKBP52, which are 70% identical at the protein level, have been studied mostly in the context of the PR and the glucocorticoid receptor (GR). FKBP51 negatively regulates GR and PR activity by reducing hormone binding affinity (6, 9). In contrast, FKBP52 enhances GR, PR, and AR responsiveness to cognate hormone (28). FKBP52 knockout mice have developmental defects in reproductive tissues (in males) consistent with reduced AR signaling and a failure of embryo implantation (in females) consistent with reduced PR signaling (4, 37). These observations, together with the apparent absence of a phenotype in mice lacking FKBP51, led to the conclusion that FKBP51 does not play a significant role in AR signaling (37).
AR overexpression is a signature of AI disease, and forcing AR expression in model systems is sufficient to generate AI growth and responsiveness to antiandrogens (3). Assuming that AR activity is a key determinant of disease progression, it stands to reason that tumor cells might employ additional mechanisms to increase AR activity. Based on this rationale, we examined whether chaperones known to regulate the ligand-binding cycle of AR are altered during prostate cancer progression and promote AI or androgen-hypersensitive growth. Here we show the Hsp90 cochaperone FKBP51 is overexpressed in a xenograft model of AI prostate cancer and describe a molecular mechanism by which FKBP51 promotes AR signaling in prostate cancer cells. Published microarray data have shown that FKBP51 message levels are higher in metastatic tumors than in tumors confined to the prostate (7). Thus, FKBP51 might be exploited as a drug target for inhibiting AR function in certain advanced prostate cancers.
LAPC-4 tumor cells (3 × 105) were combined 1:1 with Matrigel and injected subcutaneously into the flanks of 6-week-old intact or castrated scid mice as described previously (3). Xenograft tumors were harvested upon reaching ~1 cm3. For androgen-dependent (AD) and AI tumors, this required approximately 8 and 18 weeks of growth in scid mice, respectively. Tumors were flash frozen, ground under liquid N2 using a mortar and pestle, resuspended in 5 volumes of lysis buffer (50 mM Tris-HCl, pH 8.0, containing 50 mM NaCl,1 mM EDTA, 1 mM EGTA, 250 mM sucrose, 1 mM dithiothreitol [DTT], and 1 μg/ml each aprotinin, leupeptin, and pepstatin), disrupted with a Polytron, and clarified by centrifugation (30,000 × g, 30 min). Tumor extracts (5 mg/ml) were diluted with polyacrylamide gel sample buffer and analyzed by immunoblotting. Prostate cancer cell lines (PC-3, AR negative; LNCaP and CWR22Rv1, AR positive) were obtained from the American Type Culture Collection and grown in RPMI 1640 medium containing 5% fetal bovine serum (FBS). Twenty-four hours before ligand addition, cells were transferred to phenol red-free medium containing charcoal- and dextran-treated FBS.
The following antibodies were used for immunoblotting at a dilution of 1:1,000 unless noted otherwise: Hsp90 (H90-10) (2), Hsp70 (BB70) (32), p23 (JJ3) (10), FKBP51 (Hi51B), FKBP52 (Hi52A) (18), PP2Ac (ID6; Upstate), AR (AR21, 1:5,000) (35), and PSA (AR014-5RE, 1:100; Biogenix). FKBP51 (clone 18, 1:20; BD Transduction) was used for immunocytochemistry.
Xenografts were fixed in zinc formalin, dehydrated, and embedded in paraffin. Sections were cut, mounted on glass slides, rehydrated in citrate buffer, and microwaved for antigen recovery. FKBP51 monoclonal antibody clone 18 was used along with biotinylated secondary antibody, avidin-peroxidase, and diaminobenzidine as the chromogen.
LNCaP cells stably expressing Flag-tagged human FKBP51 were prepared using lentivirus generated by standard methods (pLenti4/TO/V5-DEST, psPAX, and pMD2G plasmids) in 293T cells. Cell culture medium from 293T cells containing the lentivirus was used to infect LNCaP cells, with subsequent selection in the presence of Zeocin (Invitrogen). The LacZ control line was generated by similar methods with a plasmid that encodes β-galactosidase (pLenti4/TO/V5-GW/lacZ). FKBP51 knockdown lines of LNCaP and CWR22Rv1 cells were prepared using pLKO.1-puro plasmids (Sigma), and green fluorescent protein (GFP) short hairpin RNA (shRNA) was used as the control.
AR transcription was measured using the dual-luciferase reporter system and plotted as firefly/Renilla as described previously (35). The firefly luciferase reporter genes used to measure AR activity were based on the promoters for prostate-specific antigen (PSA), probasin, mouse mammary tumor virus (MMTV), and synthetic androgen response elements (AREs). Reactions were prepared in triplicate, the results were analyzed using the Student t test, and the data presented are representative of at least three experiments. Transcript levels of endogenous AR-regulated genes (KLK2, PSA, S100P, TMPRSS2, and FKBP51) were measured by real-time PCR. Reactions were prepared in duplicate, the results were analyzed by using the Student t test, and the data are representative of at least two experiments. In brief, cells were treated with ligands for 24 h and RNA was prepared by using an RNeasy minikit (Qiagen) including a DNase treatment step. Following quantitation with Ribogreen (Molecular Probes), 1 μg of RNA was reverse transcribed using iScript and amplified using the IQ SYBR green PCR master mix in a MyiQ instrument (all from Bio-Rad). The relative mRNA level was normalized to the β-glucuronidase (GUS) mRNA level in the same sample by the 2−ΔΔCT method. The forward and reverse primers used for real-time PCR were as follows: KLK2, 5′-CACAGCTGCCCATTGCCTAAAGAA-3′ and 5′-GGCCTGTGTCTTCAGGCTCAAA-3′; PSA, 5′-TGGTGCATTACCGGAAAGTGGATCA-3′ and 5′-GCTTGAGTCTTGGCCTGGTCATTTC-3′; S100P, 5′-ATGACGGAACTAGAGACAGCC-3′ and 5′-AGGAAGCCTGGTAGCTCCTT-3′; TMPRSS2, 5′-GGACAGTGTGCACCTCAAAGAC-3′ and 5′-TCCCACGAGGAAGGTCCC-3′; FKBP51, 5′-AGGAGGGAAGAGTCCCAGTG-3′ and 5′-TGGGAAGCTACTGGTTTTGC-3′. Note that FKBP51 is encoded by the FKBP5 gene.
Cell lines were seeded into 96-well dishes (3 × 103 cells/well) and grown for 24 h. Medium containing ligands or drugs was changed on days 1, 3, and 5. On the day of measurement, cells were incubated in medium containing 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium (MTT) at 125 μg/ml for 4 h. Reduced MTT was measured using A595. Samples were performed in sextuplet and plotted as the mean ± the standard deviation.
AR and chaperone complexes were isolated by immunoprecipitation (IP). PC-3 cells were cotransfected with expression plasmids including pcDNA3-Flag-AR with either PCI-FKBP51 or control plasmids. Twenty-four hours posttransfection, cells were treated with 1 nM R1881 for the indicated times and lysed in 500 μl IP buffer (20 mM Tris-HCl, pH 7.5, containing 50 mM NaCl, 20 mM Na2MoO4, 0.5% NP-40, 1 mM EDTA, 1 mM EGTA, 2 mM DTT, protease inhibitors, and a phosphatase inhibitor cocktail [Sigma]). The lysate was sonicated and cleared by centrifugation. Samples were immunoprecipitated for 4 h at 4°C using anti-Flag M2 agarose beads (Sigma). Complexes were washed three times with IP buffer, solubilized in polyacrylamide gel sample buffer, and analyzed by electrophoresis and immunoblotting.
Chaperone complexes were assembled using purified recombinant proteins FKBP51, Hsp90, and p23 expressed in bacteria. Proteins were combined (1 μM each), incubated for 1 h at 30°C, clarified at >100,000 × g, and chromatographed on a Superose 12 or Superdex 75 gel filtration column (40 mM HEPES-KOH, pH 7.4, containing 50 mM KCl and 2 mM MgCl2) at a flow rate of 0.5 ml/min. Column fractions (0.5 ml each) were analyzed by immunoblotting using peroxidase-labeled secondary antibodies and enhanced chemiluminescence. p23 quantitation was performed using Alexa Fluor 680-labeled secondary antibody and infrared imaging (Li-Cor Biosciences). Levels of p23 bound to hsp90 were determined by the levels of p23 in fractions 1 to 6 versus the total p23 levels in fractions 1 to 12.
A whole-cell binding assay with [3H]R1881 was performed with transfected PC-3 cells. pcDNA3-Flag-AR was cotransfected with PCI-FKBP51 or PCI-Neo into PC-3 cells. Cells were treated with the indicated concentrations of [3H]R1881 for 1 h and washed with phosphate-buffered saline, and the total bound [3H]R1881 was extracted in cold ethanol and detected in a scintillation counter. Nonspecific binding was measured at each [3H]R1881 concentration by the addition of 1 μM unlabeled R1881 and then subtracted from the total binding. Nonlinear regression (OriginPro 7.5) was used to determine the Bmax and Kd using the following equation: Specific binding = Bmax [L]/Kd + [L], where [L] is the concentration of [3H]R1881.
Changes in the number of AR molecules that undergo androgen binding in response to FKBP51 expression were also assessed by using the AR-PP2A interaction, which is strictly dependent on the agonist conformation of AR (35). PC-3 cells were cotransfected with expression plasmids including pcDNA3-Flag-AR and simian virus 40 (SV40) small t antigen with either PCI-FKBP51 or a control plasmid. Twenty-four hours posttransfection, cells were pretreated with 10 μM FK506 for 1 h (where indicated) and then treated with 1 nM R1881 for 1 h. The cells were lysed in 500 μl IP buffer containing 20 mM Tris-HCl (pH 7.5), 100 mM NaCl, 0.5% NP-40, 1 mM EDTA, 1 mM EGTA, 2 mM DTT, and protease inhibitors. The lysate was then sonicated and cleared by centrifugation. AR complexes were immunoprecipitated and washed as described above. Levels of transfected AR and endogenous PP2Ac in the complex were quantified by immunoblotting using the Odyssey infrared imaging system (LI-COR Biosciences, Lincoln, NE).
We reasoned that chaperones involved in regulating the ligand-binding cycle of AR might be altered during prostate cancer progression and promote AI or androgen-hypersensitive growth. We tested this hypothesis by comparing expression levels of chaperones in LAPC-4 xenografts propagated in scid mice (12). LAPC-4 AD and AI cells were grown in intact and castrated mice, respectively. By immunoblotting the tumor extracts, the levels of Hsp90, Hsp70, FKBP52, Cyp40, and p23 were found to be comparable in LAPC-4 AD and AI samples (Fig. (Fig.1A1A and data not shown). In contrast, FKBP51 protein was elevated about threefold in the LAPC-4 AI samples relative to that in AD samples (Fig. (Fig.1A).1A). Immunocytochemistry with a monoclonal antibody to FKBP51 confirmed that FKBP51 overexpression occurs in the LAPC-4 AI tumors (Fig. (Fig.1B).1B). Real-time PCR was used to show that there is a 7- to 10-fold increase in FKBP51 mRNA (Fig. (Fig.1C).1C). PSA transcript levels, which normally are tightly regulated by AR and androgen, were slightly higher in two of three AI xenografts (Fig. (Fig.1C).1C). Another androgen-regulated gene, S100P, was upregulated 6- to 15-fold in the AI xenografts (Fig. (Fig.1C).1C). We interpreted these data as provisional evidence for a mechanism that promotes AR activity by compensating for the reduced-androgen environment of a castrated host. FKBP51 levels are also elevated in the AI xenograft generated from the AD xenograft CWR22 (7). Thus, FKBP51 is overexpressed in the two most commonly used xenograft models of AI prostate cancer.
FKBP51 is an androgen-regulated gene in prostate cancer cells (13). FKBP51 overexpression in the AI xenografts LAPC-4 and CWR22Rv1 could therefore be a result of enhanced AR signaling characteristic of the AI state. We considered the alternative hypothesis, namely, that FKBP51 overexpression enhances AR function and thereby facilitates the AI phenotype. To address this hypothesis, we tested whether FKBP51 can modulate AR activity. We assayed AR-dependent transcription in the absence or presence of cotransfected FKBP51 in the prostate cancer cell line PC-3, which is negative for both AR and FKBP51. Reintroduction of AR into PC-3 cells activates the endogenous FKBP51 gene, an observation that is consistent with AR regulation of FKBP51 expression (11). FKBP51 expression resulted in a strong enhancement of androgen-stimulated transcription from a reporter gene containing the 6-kb PSA promoter, also consistent with a previous report (7). There was no obvious effect on AR protein levels (Fig. (Fig.2A,2A, inset), a good indication that the FKBP51 effect on androgen-stimulated transcription is not simply a chaperone effect on AR protein stabilization. Additional evidence that FKBP51 regulates AR activity was obtained by showing that it enhances androgen-stimulated transcription from other promoters known to be regulated by AR. FKBP51 enhanced androgen-stimulated transcription measured on AREs from the probasin (22) promoter (AAR3tk-Luc; Fig. Fig.2B)2B) and the MMTV promoter (Fig. (Fig.2C)2C) and on synthetic AREs (Fig. (Fig.2E).2E). FKBP51 clearly has opposite effects on AR and GR, as FKBP51 inhibited dexamethasone (Dex)-stimulated GR transcription from the MMTV promoter (Fig. (Fig.2D)2D) (6). FKBP51 enhancement of AR activity was inhibited by treating cells with FK506, a ligand that binds with submicromolar affinity to FKBP51 (Fig. (Fig.2F)2F) (36). FKBP52 was also capable of enhancing AR activity in our assays, though unlike that of FKBP51, its effect was not inhibited significantly by FK506. This result is consistent with studies from Smith and coworkers who showed that FKBP52 regulation of steroid hormone receptor activity is independent of its enzyme activity (27).
To address the contribution of FKBP51 to endogenous gene expression, we used lentivirus to generate LNCaP cell lines that stably express Flag-FKBP51 with LacZ as a control. Flag-FKBP51 enhanced AD expression of the endogenous PSA gene measured by real-time PCR (Fig. (Fig.3A).3A). Since AR and androgens are critical regulators of prostate cell growth, we examined whether FKBP51 expression promotes AD cell growth. We assayed growth of the LNCaP-FKBP51 and LNCaP-lacZ lines in the presence of androgen over 6 days (Fig. (Fig.3B).3B). In the absence of androgen, relatively little growth of LNCaP-lacZ and LNCaP-FKBP51 cells was observed. Both lines underwent growth in response to a low concentration (50 pM) of the synthetic androgen R1881; however, FKBP51 expression provided an advantage for cell growth in this assay (P < 0.001; Fig. Fig.3B3B).
Because the FKBP51 knockout mouse does not display overt defects in androgen signaling (4, 37), the contribution of FKBP51 to AR function has been unclear. We considered the possibility that FKBP51 is dispensable in normal cells but that FKBP51 might be important for AR activity in prostate cancer cells. We tested whether androgen signaling through AR is dependent on endogenous FKBP51 by a knockdown approach, using lentivirus to transduce LNCaP and CWR22Rv1 cells with shRNA targeted to FKBP51. Lentivirus containing shRNA targeted to GFP was used as a control. Real-time PCR was used to measure androgen induction of AR-regulated genes in control and knockdown cell lines. In cells containing the FKBP51 shRNA, FKBP51 transcript and protein levels after androgen treatment (24 h) were reduced significantly (Fig. 3C and E). Reducing FKBP51 levels led to substantial reductions in the transcript levels for the KLK2, PSA, and S100P genes. There was no significant effect on the AR-regulated TMPRSS2 gene (Fig. (Fig.3C).3C). Transcript levels for the KLK2, PSA, and S100P genes were reduced in CWR22Rv1 cells that contained the FKBP51 shRNA (Fig. (Fig.3D),3D), an AI derivative of CWR22 that overexpresses FKBP51 (17). These data provide clear evidence that FKBP51 contributes to the regulation of AR activity in prostate cancer cells. This is despite the fact that FKBP51 is not required for AR function in the development of the male mouse reproductive system, which instead relies on the closely related cochaperone FKBP52 (4, 37).
Overexpression of AR increases transcription and growth in response to the antiandrogen Casodex (Cas) (3). Reasoning that FKBP51 regulates some aspect of the AR ligand-binding cycle, we analyzed whether its overexpression alters AR-dependent transcription in response to several AR-binding ligands. FKBP51 stimulated AR transcriptional activity in the presence of Cas, cyproterone acetate, and hydroxyflutamide, while little enhancement was observed in the presence of flutamide (Fig. (Fig.4A).4A). FKBP51 also stimulated AR activity in the presence of the adrenal androgen androstenedione and high concentrations of estrogen, but not of Dex (Fig. (Fig.4A).4A). These results prompted us to explore whether FKBP51 affects the efficacy of the antiandrogen Cas, which is used as a competitive inhibitor of adrenal androgens in patients undergoing androgen ablation therapy. We tested this by measuring AR transcription activity in the presence of synthetic androgen (R1881) and a range of Cas concentrations. While FKBP51 did not change the apparent Ki for Cas, it did render AR resistant to maximal inhibition by Cas (Fig. (Fig.4B).4B). In fact, the level of AR activity measured in the presence of FKBP51 overexpression and a saturating concentration of Cas (10 to 100 μM) was comparable to the level of AR activity measured simply in the presence of R1881 (gray shading). These data show that FKBP51 overexpression promotes AR activity in response to a variety of ligands, and it has the potential to influence the efficacy of antiandrogens used in prostate cancer treatment.
To gain insight into how FKBP51 regulates AR, we analyzed the chaperone composition of AR complexes from PC-3 cells cotransfected with FKBP51. Flag-AR complexes were isolated by IP under nondenaturing conditions, and molybdate was included to preserve chaperone interactions. Chaperones that coprecipitated with Flag-AR included Hsp90, Hsp70, FKBP51, and p23 (Fig. (Fig.5A).5A). FKBP51 induced a striking increase in the level of chaperones in AR complexes, particularly p23 and Hsp90 (Fig. (Fig.5A).5A). We viewed this effect as potentially significant because the cochaperone p23 binds directly to the ATP form of Hsp90, and this stabilizes Hsp90-client protein interactions (1, 14, 15). In this context, p23 is known to mark steroid receptors in the apo state that are poised to bind hormone (15). These data imply that in prostate cancer cells, FKBP51 levels may be limiting for the formation of p23-Hsp90-client protein complexes.
Our data suggested that FKBP51 might control AR activity by regulating chaperone complex assembly. Such a pathway could help define the number of AR molecules that attain the conformation required for the LBD to bind androgen. To more clearly define the contribution of FKBP51 in this pathway, we used recombinant proteins to test whether FKBP51 regulates chaperone complex assembly in a purified system. Hsp90, p23, and FKBP51 (1 μM each) were incubated alone and in various combinations. The samples were then subjected to gel filtration chromatography, and the eluted fractions were examined by immunoblotting. In this assay, protein assembly into the chaperone complex results in coelution with Hsp90. We found that p23 interacted weakly with Hsp90 (Fig. (Fig.5B,5B, lanes 3 and 4, upper panels), but the interaction was enhanced markedly by the addition of FKBP51 and resulted in the formation an FKBP51-p23-Hsp90 ternary complex (Fig. (Fig.5B,5B, lanes 3 and 4, lower panels).
Using the gel filtration assay, we examined the nucleotide requirements and effects of inhibitors on FKBP51-p23-Hsp90 ternary complex formation in vitro. Nucleotides and inhibitors were added to the assembly reaction mixtures prior to chromatography. The amount of p23 coeluting with Hsp90 was quantified and is shown as bar graphs. Assembly was supported by ATP or AMP-PNP, indicating a requirement for ATP binding to Hsp90 but not hydrolysis. Moreover, ternary complex assembly could be inhibited by geldanamycin (GA) and FK506, drugs that bind to Hsp90 and FKBP51, respectively (Fig. (Fig.5C).5C). We corroborated the results of our chaperone assembly assays (with recombinant proteins) by isolating chaperone complexes from cultured cells. Following a brief (1-h) treatment of cells with GA and FK506, FKBP51 complexes were isolated by IP and the chaperone content was analyzed by immunoblotting. Treatment with GA essentially abolished p23 recruitment into the FKBP51-Hsp90 complex, though it did not appear to affect Hsp90 binding to FKBP51 (Fig. (Fig.5D,5D, lane 3). FK506 reduced the levels of both Hsp90 and p23 associated with FKBP51 (Fig. (Fig.5D,5D, lane 4). Our data suggest that Hsp90 mediates the interaction between FKBP51 and p23. We tested this by using recombinant proteins and gel filtration chromatography. When Hsp90 is omitted from the binding reaction mixture, FKBP51 and p23 do not form a complex (Fig. (Fig.5E).5E). These data suggest that FKBP51 modulates p23 binding to Hsp90 through its contact with Hsp90 and not by direct contact with p23.
Since Hsp90 binds the LBD of steroid hormone receptors (24) and FKBP51 promotes Hsp90 ternary complex assembly, we anticipated that FKBP51 would regulate the LBD of AR. We tested this by deleting the LBD of AR and analyzing the chaperone content of AR isolated by IP. Neither FKBP51 nor p23 was detected in AR complexes when the LBD was deleted from AR (Fig. (Fig.6).6). Our observations, together with the established role of Hsp90 in the steroid hormone-binding cycle, led us to hypothesize that FKBP51 promotes a step in the AR ligand-binding cycle. To formally address whether FKBP51 affects androgen binding to AR, we transfected AR into AR-negative PC-3 cells without and with FKBP51 and performed whole-cell hormone-binding assays. Remarkably, FKBP51 expression increased the number of androgen-binding sites by 38% without changing AR levels (Fig. (Fig.7A).7A). Moreover, FKBP51 increased androgen binding in the whole-cell assay without increasing the affinity of AR for androgen (Fig. (Fig.7A7A).
To establish that the increase in androgen binding in the whole-cell assay was due to androgen binding directly to AR (as opposed to a cellular factor whose expression was induced by AR), we used a quantitative coimmunoprecipitation assay developed in our laboratory (35) that discriminates between androgen-free and androgen-bound forms of AR. In the presence of SV40 small t antigen, protein phosphatase 2A (PP2A) undergoes highly selective loading onto androgen-bound AR (35). Using the PP2A-AR interaction assay, we found that FKBP51 expression approximately doubled the ratio of PP2A to AR in an immunoprecipitate from cells treated with androgen. Treating cells with FK506 blocked the stimulatory effect of FKBP51 in this assay, but it did not reduce the basal level of androgen binding to AR (Fig. (Fig.7B).7B). This result suggests that, even in the absence of FKBP51 activity, a pool of AR molecules acquire competence for androgen binding. While neither FKBP51 nor FKBP52 may be required for the androgen-binding cycle of AR, the expression of these cochaperones is critical for physiological levels of AR signaling during development and in prostate cancer. In summary, our data show that FKBP51 promotes the assembly of a superchaperone-AR complex marked by p23, and this leads to a substantial increase in the number of androgen-bound AR molecules in the cell. FKBP51 overexpression therefore provides a pathway for increasing AR signaling, which is known to promote AI-type growth in human prostate cancer models (3).
We used mutant forms of FKBP51 to determine whether prolyl isomerase activity and Hsp90 binding is important for FKBP51 stimulation of AR activity on the PSA promoter. The M1 mutant form characterized by Smith and coworkers (2) has two amino acid changes in the FK1 domain that reduce prolyl isomerase activity. Mutations in the TPR domain that disrupt binding to Hsp90, described originally by Chinkers and associates in PP5 (29), were also constructed (M2). FKBP51 enhancement of AR activity measured on the PSA promoter was impaired by the M1 and M2 mutations, which, in combination (M1/M2), gave a larger reduction in FKBP51 activity toward AR (Fig. (Fig.8B).8B). The prolyl isomerase mutation (M1) reduced the ability of FKBP51 to promote p23 assembly into the Hsp90 chaperone complex without affecting FKBP51 binding to Hsp90. Mutating the TPR domain (M2) resulted in a loss of FKBP51 binding to Hsp90, which in turn reduced p23 assembly into the chaperone complex (Fig. (Fig.8C).8C). Finally, the M1, M2, and M1/M2 mutant forms have reduced activity in the PP2A-AR interaction assay, which is a readout of the number of AR molecules that are bound with androgen (Fig. (Fig.8C).8C). Our data indicate that physical contact with Hsp90 (via the TPR domain) and the prolyl isomerase activity (in the FK1 domain) are important for chaperone complex assembly and enhancement of AR activity.
Inhibitors of Hsp90 are promising therapeutics for cancer because Hsp90 regulates client proteins critical for signal transduction and cell growth (20). The Hsp90 inhibitor 17-AAG is a GA derivative with reduced hepatotoxicity in stage II clinical trials for cancers, including prostate cancer (34). 17-AAG and the immunophilin ligand FK506 each can inhibit prostate cancer cell growth in culture (21, 30). Given our finding that in vitro assembly of the Hsp90-p23-FKBP51 superchaperone complex was prevented by inhibitors of Hsp90 and FKBP51 (Fig. (Fig.5C),5C), we tested whether chaperone inhibitors were effective in blocking prostate cancer cell growth in response to androgen. 17-AAG reduced the androgen-stimulated growth of FKBP51-overexpressing and control cells to similar extents, showing that Hsp90 inhibition is sufficient to overcome the growth benefit provided by FKBP51 (Fig. 9A and B). Treating cells with a combination of 17-AAG (5 nM) and FK506 (1 μM) resulted in an additive effect on growth inhibition in both the control and FKBP51-overexpressing cells (Fig. 9A and B). Other groups have shown that high concentrations of 17-AAG inhibit AR by affecting protein stability (30); however, AR levels did not change significantly when cells were maintained in low concentrations of 17-AAG (10 nM) and FK506 (10 μM) for 4 days. CDK4, an established client of Hsp90, also showed minimal change in response to 17-AAG and FK506 treatment (Fig. (Fig.9C).9C). Our data suggest that a drug combination that targets the Hsp90 superchaperone complex can effectively inhibit androgen-stimulated growth of prostate cancer cells without causing major changes in CDK4 and Hsp70 levels.
Though androgen ablation remains the cornerstone of prostate cancer treatment, prostate cancer patients develop resistance to this therapy. Prostate cancer progresses from localized AD tumors to metastatic AI disease with an attendant poor prognosis. Defining the mechanisms that drive AI disease is therefore one of the major goals of prostate cancer research. We took a highly focused approach that involved testing whether the expression of protein chaperones with known links to steroid hormone binding are altered in AI prostate cancer. Utilizing the LAPC-4 xenograft system developed by Sawyers and colleagues (3), we found that the cochaperone FKBP51 is overexpressed in AI xenografts propagated in castrated mice. FKBP51 is a peptidyl-prolyl isomerase that binds the immunosuppressants FK506 and rapamycin. FKBP51 clearly antagonizes GR and PR function by decreasing binding affinity for hormone, but its specific mode of action on Hsp90 and steroid hormone receptors has been enigmatic. In mouse prostate and in human prostate cancer cells, FKBP51 expression is tightly regulated by androgen. In the CWR22 xenograft, FKBP51 expression is strongly reduced upon host castration but FKBP51 levels in the recurrent tumor recover to a level that exceeds those in the primary tumor (16). We found the same trend in LAPC-4 xenografts, with significantly higher FKBP51 expression levels at both the protein and mRNA levels in the AI LAPC-4 tumor grown in castrated mice compared with AD tumors grown in intact mice. These results obtained with CWR22 and LAPC-4 xenografts could reflect FKBP51 expression changes during human prostate cancer progression. In a sampling of 52 local and 9 metastatic prostate cancers, Febbo and coworkers (7) reported that FKBP51 displays a metastatic-to-local tumor expression ratio of 2.9.
Because FKBP51 transcription is regulated by AR, a key question that arose is whether the elevation of FKBP51 transcript and protein levels in the AI state reflects enhanced AR activity or whether the FKBP51 protein promotes AR activity. We found that FKBP51 promotes AR-dependent transcription using both reporter- and endogenous-gene-based assays, consistent with published data (7). Moreover, FKBP51 enhanced LNCaP cell growth in response to a low androgen concentration (50 pM). The growth benefit of FKBP51 in LNCaP cells was inhibited by 17-AAG, which links FKBP51 to Hsp90 function. We obtained direct evidence that FKBP51 induces assembly of the Hsp90 superchaperone complex and that this regulates the LBD of AR (Fig. (Fig.9D).9D). This involves FKBP51 contact with Hsp90 and recruitment of p23, a cochaperone known to stabilize Hsp90 binding to client proteins (14). Crystallographic analysis revealed that p23 recognizes the ATP-bound, closed conformation of Hsp90; a major contact surface for p23 is formed by “lid” closure of Hsp90 (1). Thus, FKBP51 may promote ATP binding or lid closure or stabilize another mobile element in the Hsp90 structure such as the catalytic loop that is contacted by p23 (1).
Hsp90 inhibitors are promising as therapeutics because Hsp90 regulates client proteins, including AR, that are critical for signal transduction and cell growth in cancer (20). High-dose 17-AAG, however, causes skeletal complications and increases Hsp70 levels, the latter of which may promote tumor resistance to therapy (23, 25). The use of Hsp90 inhibitors in combination with drugs that target cochaperones that function in the same multisubunit complex is a potential strategy for inhibiting clients such as AR that are highly dependent on the superchaperone complex.
By promoting Hsp90 superchaperone complex assembly, FKBP51 has a quantitative effect on androgen signaling by increasing the number of AR molecules that are subsequently loaded with hormone (Fig. (Fig.9D).9D). Because androgen-bound AR regulates FKBP51 expression, this creates a feedforward mechanism that could amplify AR signaling under the low-hormone conditions that occur during androgen ablation.
We thank David Smith for providing FKBP51 reagents, Charles Sawyers for LAPC-4 cells, Len Neckers for Hsp90 constructs, Hongwu Xin and Adam Spencer for technical assistance, and members of the Paschal lab for helpful discussions.
This work was supported by an IDEA award from the DOD (PC 080439) and by a Program Project Grant from the NCI (PO1-CA104106).
Published ahead of print on 4 January 2010.