Dynamic SUMOylation Is Linked to the Activity Cycles of Androgen Receptor in the Cell Nucleus

Abstract

Despite of the progress in the molecular etiology of prostate cancer, the androgen receptor (AR) remains the major druggable target for the advanced disease. In addition to hormonal ligands, AR activity is regulated by posttranslational modifications. Here, we show that androgen induces SUMO-2 and SUMO-3 (SUMO-2/3) modification (SUMOylation) of the endogenous AR in prostate cancer cells, which is also reflected in the chromatin-bound receptor. Although only a small percentage of AR is SUMOylated at the steady state, AR SUMOylation sites have an impact on the receptor's stability, intranuclear mobility, and chromatin interactions and on expression of its target genes. Interestingly, short-term proteotoxic and cell stress, such as hyperthermia, that detaches the AR from the chromatin triggers accumulation of the SUMO-2/3-modified AR pool which concentrates into the nuclear matrix compartment. Alleviation of the stress allows rapid reversal of the SUMO-2/3 modifications and the AR to return to the chromatin. In sum, these results suggest that the androgen-induced SUMOylation is linked to the activity cycles of the holo-AR in the nucleus and chromatin binding, whereas the stress-induced SUMO-2/3 modifications sustain the solubility of the AR and protect it from proteotoxic insults in the nucleus.

INTRODUCTION

Covalent conjugation of proteins by small ubiquitin-related modifiers (SUMOs), SUMOylation, has emerged as a significant regulatory mechanism, especially in nuclear signaling, transport, transcription, and DNA replication/repair (5, 52). The modification pathway has also been implicated in human diseases, including cancer (1, 40). Humans express three ∼100-amino-acid-long SUMO proteins, SUMO-1, -2, and -3, that can form isopeptide linkages with specific lysine residues of their target proteins. SUMO-2 and SUMO-3 (here called SUMO-2/3) are practically identical, whereas SUMO-1 is only ∼50% identical with SUMO-2/3. SUMO-2 and -3 can form polymeric chains through an internal lysine residue, whereas SUMO-1 is not thought to be modified to form a polymeric chain, but when linked to the end of a poly-SUMO-2/3 chain, it may terminate the chain growth (47). Moreover, conjugation of SUMO-2/3, but not that of SUMO-1, has been reported to be altered in response to cell stress (39). Different SUMO paralogs may thus have (at least partially) distinct regulatory roles.

The SUMOylation pathway requires E1, E2, and E3 activities that are distinct from those in ubiquitylation, and the two modifications have different molecular consequences. The SUMOs are activated by the SAE1 and -2 dimer (E1) and conjugated by UBC9 (E2). PIAS1, -2, -3, and -4 form a major family of SUMO E3 ligases (35). SUMO modifications are thought to be highly dynamic and have been shown to be reversed (deSUMOylated) by a family of SUMO-specific proteases (SENP1, -2, -3, -5, -6, and -7) (15).

An increasing number of proteins, especially transcription factors, have been identified as putative SUMO targets (7, 48). However, most of the previous studies addressing SUMOylation of individual mammalian transcription factors, such as steroid receptors, have been performed by transiently overexpressing the putative target proteins and SUMOs in cells and by using ectopic reporter genes, which limits interpretation of the biological relevance of these studies. Moreover, there is scarce information concerning the signals regulating the SUMO modifications of endogenous transcription factors.

The androgen receptor (AR) acts as a hormone-controlled transcription factor that conveys the messages of both natural and synthetic androgens at the level of genes and gene programs (10). Defective AR signaling leads to a wide array of androgen insensitivity disorders, and deregulated AR function, in particular, overexpression of AR, is critical for the growth and progression of prostate cancer (17). We have previously shown that the AR, when ectopically coexpressed with SUMO-1, is SUMOylated at two conserved lysine residues and that these reversible modifications attenuate the transcriptional potency of the receptor on promoters containing tandem AR-binding sites (13, 32). UBC9, PIAS proteins, and SENP1 act as coregulators for AR, indicating the regulatory relevance of SUMOylation to AR function (13, 18, 32). In this work, we have studied the role of dynamic SUMOylation in the stability, chromatin binding kinetics, and movement of the AR in the nucleus. We also show that the endogenous holo-AR in VCaP (vertebral cancer of prostate) cells is modified by SUMO-2/3 and that the modification level is markedly and rapidly increased in response to cell stress, such as hyperthermia. These stress conditions also detach the AR from its chromatin binding sites, targeting it to the nuclear matrix fraction and resulting in downregulation of AR target gene expression.

MATERIALS AND METHODS

Cell culture.

VCaP cells and COS-1 cells were from ATCC and maintained as described in references 13 and 24. Stably AR-expressing isogenic Flp-In-293 (HEK293) (Invitrogen) cell lines were maintained in Dulbecco's modified Eagle's medium (DMEM) (Gibco-Invitrogen) supplemented with 10% (vol/vol) fetal bovine serum (FBS), 25 U/ml penicillin and 25 μg/ml streptomycin, and 100 μg/ml hygromycin-B.

Plasmid constructions.

For generation of pcDNA5/FRT-hAR, human AR (hAR) cDNA was first transferred into pNTAP-B (Stratagene) with the aid of PCR and BamHI and XhoI restriction sites in the primers and pSG5-hAR as a template. The NTAP-tagged hAR cDNA was subsequently cloned by PCR to pcDNA5/FRT as an NheI/XhoI fragment. pcDNA5/FRT-hAR-K386,520R was generated using a QuikChange II XL site-directed mutagenesis kit (Stratagene) and the pNTAP-B-hAR as a template. pcDNA5/FRT-EGFP-hAR and pcDNA5/FRT-EGFP-hAR-K386,520R were created by transferring EGFP-hAR/-K386,520R as a NheI-BamHI fragment (32).

Generation of isogenic HEK293 cells.

Stably AR-expressing HEK293 (Flp-In-293; Invitrogen) cell lines were generated according to the manufacturer's instructions with minor modifications. Briefly, Flp-In-293 cells were cotransfected with a 9:1 ratio of pOG44:pcDNA5/FRT-hAR or pcDNA5/FRT-hAR-K386,520R using Lipofectamine 2000 (Invitrogen). At 48 h after transfection, the cells were split to 1:2 and hygromycin-B (100 μg/ml) (Invitrogen) was added. Ten days after transfection, 15 hygromycin-resistant foci were picked with pipette tips and transferred to wells in a 24-well plate, grown for 7 days, and transferred to 12 wells to further growth and analyses. These wells were split to four portions for Zeocin (Invitrogen) sensitivity tests, β-galactoside assays, Western blot analyses, and further growth.

Preparation of nuclear matrix.

Nuclear matrices were prepared as previously described (9) with minor modifications. Cells were initially processed on 6 wells and, after the addition of ammonium sulfate, transferred into microcentrifuge tubes. Vanadyl RNase complex and phenylmethylsulfonyl fluoride (PMSF) in cytoskeleton and digestion buffers were replaced with RNase inhibitor (20 U/ml) and 1:100-diluted protease inhibitor cocktail (PIC) (catalog no. P8340; Sigma-Aldrich), respectively. Digestion buffer contained 100 U/ml DNase I (Fermentas). Nuclear matrices were further extracted with 2 M NaCl, and the resultant pellets were solubilized with SDS-PAGE sample buffer.

Immunoprecipitation and immunoblotting.

VCaP cells were harvested in phosphate-buffered saline (PBS) containing 10 mM N-ethylmaleimide (NEM [a SENP inhibitor]), and the cell lysates were prepared, immunoprecipitated, and analyzed by immunoblotting as described in references 36 and 38. Whole-cell lysates were obtained by lysing the cells in SDS-PAGE sample buffer containing 10 mM NEM and 1:200-diluted protease inhibitor cocktail (PIC) (catalog no. P8340; Sigma-Aldrich). For immunoprecipitations, the harvested cells were first incubated in lysis/immunoprecipitation buffer (50 mM Tris-HCl [pH 8.0], 140 mM NaCl, 1 mM EDTA, 1% Triton X-100, 10% glycerol, 10 mM sodium phosphate [pH 7.0], 50 mM NaF, 10 mM NEM, and 1:200-diluted PIC) for 20 min on ice and processed as described in reference 36. For immunoprecipitation under more stringent conditions, samples were lysed in lysis buffer supplemented with 2% (wt/vol) SDS, the lysates were heated at 95°C for 10 min, briefly sonicated, and centrifuged, and the supernatants were transferred to fresh tubes and diluted 1:10 with the lysis buffer before immunoprecipitation. For protein half-life measurements and quantification, goat anti-mouse IgG DyLight 680 fluorescent dye-conjugated secondary antibody was used and membranes were analyzed with a Li-COR Odyssey infrared imaging system (LI-COR Inc.) according to the manufacturer's instructions. The following primary antibodies were used: rabbit antiserum against AR (14), mouse monoclonal anti-AR (sc-7305), antitubulin (sc-5286), anti-glyceraldehyde-3-phosphate dehydrogenase (anti-GAPDH) (sc-25778), anti-lamin B1 (sc-6216), anti-SUMO-2/3 (catalog no. M114-3; MBL), anti-SUMO-1 (catalog no. 33-2400; Zymed), antiubiquitin (sc-8017), anti-PIAS1 (sc-8152), anti-RNA polymerase II N-20 (sc-899), and anti-c-Myc (9E10) (sc-40).

FRAP analysis, immunofluorescence, and confocal imaging.

Stably enhanced green fluorescent protein-AR (EGFP-AR)- or EGFP-ARK386,520R-expressing isogenic Flp-In-293 (HEK293) cells were grown on Ibidi 8-well chambers (Integrated BioDiagnostics) and treated with R1881 or bicalutamide. Fluorescence recovery after photobleaching (FRAP) analyses were performed using a Zeiss LSM 700 confocal microscope with Zen 2009 software (version 5.5). Bleach pulses were performed with maximal laser intensity in regions of interest (ROIs) (2.8 μm by 9.9 μm), and serial images were collected over a 30-s period. Background fluorescence and general bleaching during acquisitions were measured, and the fluorescence intensity in the ROI was normalized to the measured values. For immunofluorescence analysis, VCaP cells seeded onto glass coverslips were fixed with 3% (wt/vol) paraformaldehyde–PBS, permeabilized with 0.1% Triton X-100–PBS, processed with antibodies, and examined using a Zeiss LSM 700 microscope as described in reference 37. The following primary antibodies were used: rabbit antiserum against AR (14), anti-promyelocytic leukemia (anti-PML) (sc-9863), and anti-SUMO-2/3 (M114-3; MBL). The Z-stacked image represents a maximum projection of 19 images obtained every 0.54 μm from a single cell and visualized with the Zen software. The effect of heat stress on AR nuclear distribution was studied in COS-1 cells transiently expressing EGFP-AR. At 24 h after transfection, the cells were treated with 10 nM R1881 for 2 h and exposed to 43°C for 30 min.

RNA interference (RNAi).

VCaP cells were transfected with small interfering RNA (siRNA) oligomers (ON-TARGETplus pool or corresponding control; Dharmacon) for 96 h using Trans-IT-siQUEST reagent (Mirus Bio Corp.) according to the manufacturer's instructions.

ChIP, re-ChIP, and RT-qPCR analyses.

Chromatin immunoprecipitation (ChIP) assays were performed according to the method described in reference 25. Briefly, the cells were treated with or without testosterone and exposed to 43°C as indicated in the figure legends. Immunoprecipitation was performed with 1 μg of the indicated antibody, and immunocomplexes were collected with protein A (AR, SUMO-2/3, IgG) or protein G (PIAS1) beads (Millipore). DNA was purified using a QIAquick PCR purification system (Qiagen). Quantitative PCR (qPCR) analyses were carried out with FastStart SYBR green Master (Roche Diagnostics) and an Mx3000P real-time PCR system (Stratagene). Specific primers for different regions are listed in Table SA1 in the supplemental material. Results were calculated using E^−(ΔCT) × 20, where E (efficiency of target amplification) represents a coefficient of DNA amplification by one PCR cycle for a particular primer pair, C_T represents threshold cycle, and ΔC_T represents C_{T(ChIP-template)} − C_T(Input). Results are presented as fold increases over the values for normal rabbit IgG-precipitated samples. Re-ChIP analyses (in which chromatin was sequentially precipitated with two different antibodies) were performed essentially as described above with the following differences. After the first immunocomplexes were collected, protein A beads were washed and chromatin fragments were released from the beads by incubating with 10 mM dithiothreitol. The second immunoprecipitation and immunocomplex collection were performed as described for normal ChIP. Total RNA was isolated from samples by the use of TRIzol (Invitrogen) according to the manufacturer's instructions, and expression of AR target genes was measured by reverse transcriptase qPCR (RT-qPCR) as described in reference 25. Primers for RT-qPCR are listed in Table SA2 in the supplemental material.

RESULTS

Cell stress induces accumulation of SUMO-2/3-modified AR pool in prostate cancer cells.

In VCaP cells that are derived from a hormone-refractory cancer, possessing an amplified AR gene locus encoding otherwise normal nonmutated AR protein (24), mRNA expression of SUMO-2 and -3 exceeds that of SUMO-1 by 7- and 3-fold, respectively, implying that SUMO-2 is the major SUMO paralog in these cells (see Fig. SA1A in the supplemental material). In keeping with these mRNA data, quantification of SUMO-1 and SUMO-2/3 protein levels using antibodies specific for SUMO-1 or for both SUMO-2 and -3 showed that the VCaP cells contained ≥8 times more SUMO-2/3 than SUMO-1 protein (see Fig. SA1B to D in the supplemental material). Immunoprecipitation coupled to immunoblotting showed that exposure to androgen (testosterone), but not antiandrogen (bicalutamide), induced modification of AR by SUMO-2/3, whereas there were no signs of SUMO-1 in the AR, and the amount of ubiquitin in the receptor showed no clear response to agonist or antagonist exposure in the cells (Fig. 1A). Previous cotransfection-based studies of heterologous cells suggested that the AR is modified more efficiently by SUMO-1 than by SUMO-2 at the steady state (13). The preferential conjugation of SUMO-1 to AR under the conditions where the SUMOs were ectopically expressed at comparable levels appears to have derived from a more rapid deconjugation of AR-SUMO-2 conjugates; ectopically expressed deconjugation-defective SUMO-1-94P and SUMO-2-90P mutants showed no difference in their conjugation to the AR (see Fig. SA2 in the supplemental material).

Fig 1.

Androgen loaded-AR in VCaP prostate cancer cells is modified by SUMO-2/3. (A) VCaP cells were exposed to vehicle (−), testosterone (T, 100 nM), or bicalutamide (B, 10 μM) for 2 h at 37°C as indicated. Whole-cell lysates (Input) were analyzed by immunoblotting with anti-AR and antitubulin (TUB) antibodies, or AR was immunoprecipitated (IP) with an anti-AR antibody or control IgG, and the immunoprecipitates were immunoblotted with anti-SUMO-2/3 (S2/3), anti-SUMO-1 (S1), or antiubiquitin (UB) antibodies. Arrow, unmodified AR; arrowheads, modified AR. (B) Heat stress results in rapid but reversible SUMOylation of endogenous AR in prostate cancer cells. VCaP cells were treated with vehicle, T, or B for 2 h at 37°C, cell dishes were transferred to 43°C for different times (15 to 120 min) as indicated, and whole-cell lysates were analyzed by immunoblotting with anti-AR and antitubulin antibodies. (C) VCaP cells were treated with ligands for 2 h at 37°C, transferred to 43°C for 30 min, and subsequently transferred back to 37°C for 15 or 30 min (Rec., recovery) and analyzed as described for panel B. (D) To identify modifications of AR, the VCaP cell lysates were immunoprecipitated with anti-AR antibody in immunoprecipitation buffer and the precipitates were immunoblotted with anti-AR (mouse monoclonal), anti-SUMO-2/3, anti-SUMO-1, and antiubiquitin antibodies. (E) AR was immunoprecipitated from heat-shocked VCaP cells as described for panel D, but lysis/immunoprecipitation buffer was supplemented with 2% SDS. The inputs and precipitates were immunoblotted with anti-AR and anti-SUMO-2/3 antibodies.

Since recent data suggest that SUMOylation of some target proteins may be linked with their ubiquitylation and proteasome-mediated degradation (20, 46, 50), we tested the effect of proteasome inhibition on the SUMO and ubiquitin modifications of ligand-free, androgen- or antiandrogen-occupied AR in VCaP cells. Interestingly, treatment of the cells with MG132 proteasome inhibitor for 2 h resulted in a massive accumulation of high-molecular-mass anti-AR antibody-immunoreactive forms indicative of ubiquitin or ubiquitin-like modifications (see Fig. SA3 in the supplemental material). Since the blockage of the proteasome function also causes proteotoxic stress due to undegraded proteins, we tested if other types of cell stress, for example, heat shock, have similar effects on the AR pattern. As shown in Fig. 1B, exposure of the cells to heat (43°C) also induced the appearance of high-M_r AR forms. The kinetics of these AR forms at 43°C with the antiandrogen peaked at the 30-min time point, but with the androgen they remained elevated for longer periods (120 min) (Fig. 1B). The reversal of the 30-min hyperthermia-induced modifications was quick, as the modifications disappeared within 15 min at 37°C (Fig. 1C). Also, osmotic (0.3 M NaCl) or heavy metal (200 μM CdCl₂) stress induced similar, although not identical, appearances of high-M_r AR forms (see Fig. SA3 in the supplemental material).

To check whether the high-M_r AR bands in VCaP cells correspond to SUMO- or ubiquitin-modified AR, AR immunoprecipitated from VCaP cells was analyzed by immunoblotting with anti-SUMO-1, anti-SUMO-2/3, and antiubiquitin antibodies (Fig. 1D). According to these analyses, heat shock especially triggers SUMO-2/3 modifications in the AR, whereas SUMO-1 modification(s) or ubiquitylation was affected to a lesser extent. Lysis of cells in an immunoprecipitation buffer containing 2% SDS and subjected to heating at 95°C for 10 min, which should disrupt noncovalent protein-protein interactions, verified that the AR from heat-shocked cells was covalently modified by SUMO-2/3 (Fig. 1E). However, in addition to being covalently modified by SUMO-2/3, AR may bind other SUMOylated proteins in a noncovalent fashion.

To examine the role of the AR's SUMO consensus acceptor lysines K386 and K520 in the modifications induced by cell stress, wild-type AR (wtAR) or a SUMOylation-deficient AR mutant (ARK386,520R) expressing isogenic HEK293 cells was heat shocked and analyzed as described above. As shown in Fig. 2A, the AR mutant devoid of the major SUMOylation sites showed practically no SUMO-1 conjugation and its SUMO-2/3 modifications were also severely compromised, indicating the primary role of the major SUMO acceptor lysines in the cell stress-induced modifications. However, since the mutant still displayed some signs of SUMO-2/3 modification, there seem to be secondary SUMO acceptor lysines in the AR. Also, the ubiquitylation pattern of the AR was influenced by the disruption of the primary SUMOylation sites, as the pattern was shifted to a higher M_r range (Fig. 2A). It is of note that the level of recovery of the receptor in the anti-AR IPs from heat-stressed ARK386,520R cells is about 3.5-fold lower than that from the wtAR cells (cf. the 43°C samples in the upper panel of Fig. 2A). Therefore, the relative amount of ubiquitin in the ARK386,520R was not decreased but was increased compared to that in the wtAR. Accordingly, the SUMOylation mutant showed a faster decay than the wtAR after addition of the protein synthesis inhibitor cycloheximide, with the half-lives of the mutant being ∼6 h (without androgen) and ∼14 h (with androgen) versus ∼14 h (without androgen) and ∼22 h (with androgen) for the wtAR (Fig. 2B). These results indicate that the primary SUMO acceptor sites and SUMOylation of the receptor do not prime the AR for degradation but, if anything, protect the receptor from degradation.

Fig 2.

Heat stress-induced modifications of AR are largely dependent on the receptor's major SUMO acceptor sites. (A) Isogenic HEK293 cells stably expressing the AR or ARK386,520R mutant were treated with vehicle (−), testosterone (T), or bicalutamide (B) for 2 h, followed by exposure to 43°C for 1 h. AR was immunoprecipitated with anti-AR antibody, and the precipitates were analyzed by immunoblotting with anti-AR (mouse monoclonal), anti-SUMO-2/-3, anti-SUMO-1, and antiubiquitin antibodies as indicated. (B) Effect of the protein synthesis inhibitor cycloheximide (CHX) on the levels of wtAR and ARK386,520R in the HEK293 cell lines. Cells were treated with T for 2 h before exposure to CHX for different durations as indicated. Proteins were detected by immunoblotting using anti-AR and anti-GAPDH antibodies. Duplicate immunoblot analyses were performed for detection either conventionally by enhanced chemiluminescence (ECL) (upper panels) or with a Li-COR Odyssey infrared imaging system for precise quantification of AR band intensities. The graph illustrates the amounts of AR and ARK380,520R normalized by the amount of GAPDH in the sample. The points in the graph represent means ± standard deviations (SD) of the results of experiments performed with three samples.

SUMOylated AR is enriched in the nuclear matrix.

To reveal the subcellular localization of endogenous AR in relation to SUMO-2/3 in VCaP cells and whether the AR and the SUMO-2/3 are targeted to cellular substructures, we analyzed their localization by immunocytochemistry and confocal microscopy. Confocal imaging of nonstressed VCaP cells revealed that the endogenous SUMO-2/3 was enriched in the nuclear periphery of VCaP cells, whereas the holo-AR was more evenly distributed throughout the nucleus but colocalized with the SUMO-2/3 to a small extent (Fig. 3A). Interestingly, heat stress resulted in the appearance of SUMO-2/3 in bigger, discrete intranuclear granules which also contained AR. A portion of the AR in heat-stressed cells colocalized with promyelocytic leukemia (PML) protein, suggesting that at least some of the AR- and SUMO-2/3-containing granules corresponded to the PML bodies (Fig. 3A) that have been shown to contact the nuclear matrix and to be enriched in SUMOylation pathway components (11, 28).

Fig 3.

SUMOylated AR is enriched in the nuclear matrix of prostate cancer cells. (A) Partial colocalization of AR and SUMO-2/3 in nuclear bodies in response to heat stress. VCaP cells grown on coverslips in the presence of testosterone were either kept at 37°C or exposed to 43°C for 30 min. Cells were fixed and immunostained with anti-AR and anti-SUMO-2/3 or anti-PML antibodies using fluorescein isothiocyanate (FITC)-, Rhodamine Red X-, or cyanine 5-labeled secondary antibodies. Cell images were captured separately by using Zeiss LSM 700 confocal microscope (FITC at 488-nm, rhodamine at 555-nm, and cyanine 5 at 639-nm excitation) and merged as indicated. In addition, a representative Z-stacked maximum projection merge of AR and SUMO-2/3 staining of a 43°C- and T-exposed cell nucleus showing nuclear bodies is displayed. (B) Biochemical fractionation of VCaP cells showed that SUMOylated AR was enriched in the nuclear matrix. Cells were treated with vehicle (−), testosterone (T), or bicalutamide (B) for 2 h. After 30 min at 37°C or 43°C, cells were fractionated to nucleocytoplasm (lanes 2), chromatin (lanes 3), and nuclear matrix (lanes 4) and analyzed by immunoblotting with indicated antibodies. Input, whole-cell lysate (lanes 1). LAM, anti-lamin B1; POL II, polymerase II. (C) VCaP cells were treated with ligands as described above, and whole-cell lysates or the nuclear matrix fractions were prepared after 30 min at 43°C and following 30 min of recovery at 37°C. The samples were immunoblotted with anti-AR, anti-lamin B1, and anti-GAPDH antibodies. (D) Nuclear matrix preparations of HEK293 cells expressing AR or the ARK386,520R mutant treated with vehicle (−), T, or B for 120 min followed by exposure to 43°C or 37°C for 60 min were analyzed by immunoblotting as indicated. Total intensities of the anti-AR immunoreactive bands were quantified by use of a Li-COR Odyssey infrared imaging system, and the numbers below the lanes depict the amount of each receptor form relative to the vehicle-treated cells at 37°C (set to a value of 1). (E) The effect of heat stress on AR nuclear distribution was studied in COS-1 cells transiently expressing EGFP-AR. At 24 h after transfection, the cells were treated with 10 nM R1881 for 2 h and exposed to 37°C (nonstressed) or 43°C for 30 min. Rec., heat-stressed cell returned to 37°C for 15 min. A representative cell nucleus is shown for each condition.

To complement biochemically the preceding microscopic studies, we examined the effect of heat stress on the association of SUMOylated AR in different cellular fractions. To this end, VCaP cells were sequentially fractionated to the nucleocytoplasm (nonionic detergent-soluble fraction), chromatin fraction (released by DNase I and 0.25 M ammonium sulfate), and insoluble nuclear matrix (remaining after 2 M NaCl extractions) (9). Immunoblotting analyses of these fractions interestingly revealed that the SUMOylated AR was enriched in the nuclear matrix fraction (Fig. 3B). Heat stress also increased the amount of non-SUMOylated AR in the nuclear matrix. Under nonstressed conditions, the compartmentalization of the AR in the VCaP cell nuclear matrix was largely promoted by the androgen, with the antiandrogen being less efficient in targeting of the receptor to the nuclear matrix. Alleviation of the cell stress which reversed the AR SUMOylation normalized the amount of AR in the nuclear matrices to the levels of nonstressed cells (Fig. 3C). Comparison of the relative amounts of wtAR and ARK386,520R in the nuclear matrix fractions of HEK293 cells prior and after heat shock revealed that, interestingly, heat had a more pronounced nuclear matrix-tethering effect on the SUMOylation mutant AR in both the absence and presence of androgen (Fig. 3D). This result suggests that the heat stress-induced SUMOylation that may involve SUMO chain formation possibly sustains the nucleoplasmic solubility of the AR.

The results presented above suggest that the AR that was targeted to the nuclear matrix under conditions of stress can move back to the nucleoplasm and/or chromatin upon relief of the stress. To investigate the reversibility of nuclear compartmentalization of AR, we expressed EGFP-tagged androgen-occupied AR in COS-1 cells and monitored green fluorescence by confocal microscopy in fixed cells. As shown in Fig. 3E, heat stress resulted in redistribution of the EGFP-AR from fine nuclear speckles to bigger and discrete nuclear granules which, however, dispersed back to a more specked pattern soon after the cells were returned to normal conditions. The heat stress-induced change in the nuclear localization pattern of the AR was paralleled by the enrichment of the EGFP-AR in the nuclear matrix fraction (data not shown). These data strongly suggest that the stress-induced sequestration of the AR to the nuclear matrix is a reversible phenomenon.

SUMOylation sites influence the intranuclear mobility of AR.

The data presented above imply that SUMOylation of AR is linked to the intranuclear compartmentalization of the AR. To test whether the SUMOylation influences the kinetics of the AR, we compared the intranuclear mobility of EGFP-tagged wtAR to that of the SUMOylation-deficient ARK386,520R mutant by employing FRAP (fluorescence recovery after photobleaching) analyses in living isogenic HEK293 cells exposed to androgen or antiandrogen. The measured average half-recovery times for the wtAR were in line with those in the literature and showed that the mobility of the antiandrogen-loaded receptor (2.1 s ± 0.1) was clearly more rapid than that of the agonist-bound AR (3.9 s ± 0.1) (Fig. 4) (26). Interestingly, disruption of the AR's primary SUMOylation sites had a small but significant retarding effect on the mobility of the receptor in the presence of androgen.

Fig 4.

SUMOylation influences the nuclear mobility of the AR. FRAP analysis was performed in isogenic HEK293 cells stably expressing EGFP-AR or EGFP-ARK386,520R exposed to androgen (R1881 [R], 10 nM) or antiandrogen (bicalutamide [B], 10 μM). Representative images of cell nuclei before bleaching (prebleach), immediately after the bleach pulse (bleach), and during recovery (2 s, 4 s, and 8 s) are shown. Columns represent the half-recovery times ± standard errors of the means (SEM) (n = 50). ***, P ≤ 0.001 for the difference between wild-type and mutant AR in the presence of androgen.

Cell stress detaches AR from the chromatin.

The increase in the amount of AR in the nuclear matrix compartment after cell stress implied that the heat stress can repress AR-dependent gene expression and remove the AR from the chromatin. To test these notions, we employed RT-qPCR and quantitative chromatin immunoprecipitation (qChIP) assays to monitor the androgen-induced expression of TMPRSS2, SPOCK1, and S100P and the occupancy of androgen-loaded AR in regulatory regions of these genes in nonstressed and heat-stressed VCaP cells. Measurement of the mRNA levels of these AR target genes showed that 30-min exposure of the cells to 43°C decreased their levels of mRNA expression, albeit the genes (whose androgen induction kinetics differed) showed differences in their inhibition kinetics and recovery from the stress (see Fig. SA4 in the supplemental material). In line with the cell fractionation data, the heat shock resulted in a marked release of AR from the chromatin of these genes (Fig. 5). When the stressed cells were allowed to recover at 37°C, the AR was capable of returning to the chromatin.

Fig 5.

Effect of heat stress and recovery from the stress on the chromatin binding of AR in prostate cancer cells. Loading of AR onto TMPRSS2, SPOCK1, and S100P regulatory regions at different time points after addition of testosterone (T), exposure to 43°C for 30 min, and recovery at 37°C was monitored by using qChIP assays in VCaP cells. qChIP assays with normal IgG monitored nonspecific binding. Results are shown as fold increases compared to the normal IgG-precipitated samples and represent the means ± SD of the results of three experiments.

Co-occupancy of the holo-AR and SUMOylation pathway components on the chromatin.

Interestingly, qChIP assays with anti-SUMO-2/3 antibody revealed that androgen treatment of VCaP cells increased the level of SUMO-2/3 in the AR-binding regions of the three AR target genes in parallel with the binding of the AR (Fig. 6A). Moreover, re-ChIP assays in which chromatin was sequentially immunoprecipitated with anti-SUMO-2/3 and anti-AR antibody confirmed the simultaneous presence of SUMO-2/3 and AR in these chromatin regions (Fig. 6B). The latter results suggest that a portion of chromatin-bound AR is either conjugated by SUMO-2/3 or closely associated with other SUMOylated proteins.

Fig 6.

Co-occupancy of the holo-AR and SUMOylation pathway components on the chromatin. (A) Effect of androgen on the occupancy of SUMO-2/3 on the TMPRSS2, SPOCK1, and S100P regulatory regions as assessed by qChIPs. **, P ≤ 0.01; *, P ≤ 0.05 for the difference between vehicle (−)- and androgen (T)-treated cells. (B) Re-ChIP assays in which chromatin was sequentially immunoprecipitated with anti-SUMO-2/3 and anti-AR antibody. Gray lines depict the background levels in control re-ChIP assays in which normal IgG was used instead of the anti-AR antibody. *, P ≤ 0.05 for the difference between vehicle- and androgen-treated cells. (C) Occupancy of PIAS1 on the TMPRSS2, SPOCK1, and S100P regulatory regions as assessed by qChIPs with anti-PIAS1 antibody. *, P ≤ 0.05 for the difference between vehicle- and androgen-treated cells. (D) Effect of PIAS1 depletion on the AR binding on the TMPRSS2, SPOCK1, and S100P regulatory regions as assessed by qChIPs with anti-AR antibody. ***, P ≤ 0.001 for the difference between the cells transfected with nonspecific target siRNA (siSCR) and with PIAS1 siRNA (siPIAS1) in the presence of androgen.

Analysis of mRNA and protein expression in VCaP cells showed that PIAS1 is the major PIAS protein in these cells, with its mRNA being upregulated by androgen (see Fig. SA5 in the supplemental material). qChIP assays also showed evidence for androgen-enhanced recruitment of PIAS1 to the same chromatin regions (Fig. 6C). Interestingly, depletion of PIAS1 by RNAi had a target-selective effect on the chromatin binding of the holo-AR; it significantly attenuated the binding of AR on SPOCK1 and showed a similar trend on TMPRSS2 but slightly increased the receptor occupancy on S100P (Fig. 6D). These results suggest a functional interaction between the SUMOylation and the holo-AR on the chromatin.

SUMOylation affects the dynamics of AR-chromatin interactions and the AR activity in a target gene-selective fashion.

Previous studies addressing the regulatory role of AR SUMOylation sites in transcriptional regulation have been carried out merely with ectopic reporter genes (13, 32). To test the effect of the SUMOylation sites in a genuine chromatin environment, we compared the levels of androgen induction of AR target genes in the isogenic HEK293 cell lines expressing the wild-type AR or the AR SUMOylation mutant by measuring levels of androgen-regulated mRNAs using RT-qPCR. Interestingly, S100P and SPOCK1 mRNAs, for example, accumulated to higher levels in the ARK386,520R- than in the wtAR-expressing cells, but the cells did not differ with respect to androgen regulation of MAFB mRNA, and in the case of FOXO4 mRNA, the AR mutant-expressing cells actually showed compromised androgen induction (Fig. 7A). In line with the S100P and SPOCK1 expression data, the SUMOylation-deficient AR was more extensively loaded onto the S100P and the SPOCK1 regulatory regions after androgen induction as assessed by qChIP assays (Fig. 7B). On MAFB, the wtAR and ARK386,520R showed comparable levels of loading, whereas FOXO4 was more efficiently occupied by the ARK386,520R mutant. Moreover, heat shock more drastically decreased the occupancy of ARK386,520R than of the wtAR on the S100P and the SPOCK1 chromatin, whereas these two receptor forms did not markedly differ in their responses to the heat stress on the FOXO4 and the MAFB chromatin. These data confirm that SUMOylation modulates the AR function on chromatinized templates. Our results also suggest that the modulatory role of SUMOylation is dependent on the AR target gene and that the modulation is reflected in the interaction dynamics of the AR on the target gene chromatin.

Fig 7.

SUMOylation modulates AR activity in the chromatin environment. (A) Comparison of the androgen induction of S100P, SPOCK1, MAFB, and FOXO4 in wild-type AR- and AR SUMOylation mutant ARK386,520R-expressing isogenic HEK293 cell lines. Cells were treated with vehicle (−) or androgen (R, 10 nM) for 24 h as indicated, and AR target gene mRNAs were analyzed by RT-qPCR as described in Materials and Methods. The relative mRNA levels represent the means ± SD of the results of three independent experiments. (B) The effect of heat shock on the occupancy of the AR and the ARK386,520R on S100P, SPOCK1, FOXO4, and MAFB regulatory regions 2 h after addition of androgen and exposure to 43°C for 1 h was analyzed using qChIP assays. The results represent the means ± SD of the results of three independent experiments.

DISCUSSION

SUMOylation of transcription factors, including nuclear receptors, has often, but not always, been linked to transcriptional repression (6, 49). However, SUMOylation functions also in constitutive transcription and during activation of inducible genes (34). The mechanistic explanations for the SUMO modification-linked repression include interaction of the SUMOylated protein with the SUMO-binding DAXX (death domain-associated protein) corepressor, histone deacetylases (HDACs) and demethylases, and other corepressors or coregulators, such as GPS2, ARIP4 and CoREST1, and heterochromatin proteins (23, 29, 30, 33, 43, 51, 53). However, in the case of AR, the mechanisms underlying the apparent SUMO-dependent repression of the receptor have remained elusive, as, e.g., HDAC-linked corepressor activity or DAXX does not associate with its repression (13).

In this work, we have addressed the role of SUMOylation in the stability, chromatin binding kinetics, and movement of the AR in the context of endogenous proteins in human prostate cancer cells or isogenic HEK293 cells stably expressing AR or its SUMOylation-deficient form. We show that the endogenous AR in normally growing VCaP prostate cancer cells is indeed modified by SUMO-2/3, with the stimulation provided by AR agonists rather than by antagonists. Our sequential ChIP assays showed also that the chromatin-bound AR was also modified by SUMO-2/3 to a certain extent. SUMOylation of AR is reversible (13); in particular, the modification by SUMO-2 seems to be highly dynamic and to result in rapid turnover. Thus, even though only a small percentage of the androgen-bound AR is modified by SUMO-2 (3) at a steady state in VCaP cells, a large AR pool can go through rapid cycles of SUMOylation and deSUMOylation, which is likely to have a significant impact on the AR function. These notions are supported by our ChIP and FRAP data, showing that the AR SUMOylation sites can have a significant impact on the intranuclear mobility and the chromatin loading of the receptor.

The role of SUMOylation is also reflected in the transcriptional function of AR at endogenous target loci, as the ARK386,520R mutant exhibited differential transcriptional activity on some, albeit not all, target genes. However, SUMOylation does not simply repress the AR activity on all target genes in HEK293 cells. For example, activation of the MAFB locus by androgen was insensitive to AR SUMOylation, whereas at the FOXO4 locus, the ARK386,520R mutant showed severely compromised transcriptional activity. Also, our preliminary results from comparative genome-wide expression profiling of the isogenic AR- and ARK386,520R-expressing cell lines imply that the AR SUMOylation sites exert no general repressive effect but that they are in most cases required for the receptor's full transcriptional activity. AR lysine residues 386 and 520 have not been shown to be targeted by posttranslational modifications other than SUMOylation; however, such targeting remains a formal possibility. Also, in the case of GATA-1, the SUMOylation-deficient protein resulting from the K-to-R mutation was considerably less active than the wild-type protein in inducing expression of several genes controlling hematopoiesis, with the attenuated activity being linked to decreased chromatin occupancy at selected loci. These loci resided in distinct subnuclear compartments compared to the SUMOylation-independent target loci (22). Similarly, the SUMOylation-defective microphthalmia-associated transcription factor (MITF) mutant that predisposes to melanoma and renal carcinoma clearly differed from its wild-type counterpart on some, but not all, MITF target genes, with the mutant signature being related to cell growth, proliferation, and inflammation. Again, the SUMOylation also influenced MITF occupancy on its target loci (1, 55). Notably, elimination of orphan nuclear receptor SF-1 (steroidogenic factor 1) SUMOylation in mice did not lead to a simple gain of SF-1 phenotype or increased levels of SF-1 target genes but resulted in abnormal Hedgehog signaling and endocrine development (21). The observed abnormalities may be linked to altered recognition of a group of target genes that are sensitive to modification of SF-1 (3). For most of the SUMO-sensitive target genes studied in the mouse model, interestingly, the SUMOylation-defective SF-1 allele prevailed even in the presence of wild-type SF-1 (21).

Recent work has revealed that SUMO-2-conjugated proteins are accumulated in response to proteasome inhibitors, suggesting that SUMO-2 conjugation and ubiquitylation are in some cases linked (41). SUMO-2 modification can target certain proteins for subsequent ubiquitylation and degradation by proteasomes (20, 44, 46), and SUMO-targeted E3 ubiquitin ligases bridge these processes. SUMOylation and ubiquitylation are, for example, dynamically linked in the regulation of ETS domain transcription factor PEA3. SUMOylation is required for its maximal activity on target gene promoters as well as its ubiquitylation, promoting its degradation (8). Thus, although SUMOylated PEA3 is unstable, SUMOylation appears to play a positive role in PEA3-mediated transcriptional activation.

In the case of AR, disruption of the major SUMOylation sites actually decreases the half-life of the receptor, suggesting that the sites contribute to AR proteostasis. Interestingly, inhibition of proteasome function (using MG132 for 2 h) in VCaP prostate cancer cells resulted in accumulation of more SUMO-2/3-conjugated AR than SUMO-1- or ubiquitin-conjugated AR forms to a marked extent. This occurred in parallel with a marked increase in overall formation of cellular SUMO-2/3 conjugates. However, since various types of short-term proteotoxic stresses, such as heat shock, high salt concentrations, and heavy metal exposure of the cells, rapidly resulted in similar accumulations of AR-SUMO-2/3 conjugates, our results do not support the straightforward notion that the SUMO-2/3 conjugate formation primes the AR for degradation. It is also of note that the AR in VCaP cells is relatively stable, with half-times without androgen or with antiandrogen being 6 h and with agonist being 19 h.

Rapid conjugation of SUMO-2/3 to proteins in response to cell stresses, such as heat, osmotic, hypoxic, and oxidative stress, in COS-7, neuronal, and HeLa cells has also been previously reported (7, 39, 48, 54). Interestingly, a list of heat shock-regulated SUMO-2 targets obtained by proteomic analyses of HeLa cells was enriched for nuclear proteins involved in chromatin remodeling, DNA repair, transcription, and RNA processing and trafficking. Even though heat stress in HeLa cells resulted in a net increase in cellular SUMO-2 conjugation, the response was a substrate-specific one, with some substrates showing reduced and some unaltered SUMOylation (7). Similarly to our VCaP cell experiments, the heat shock-induced SUMO-2 modification of proteins in HeLa cells and that of SUMO-2/3 in COS-7 cells was generally reversed by deSUMOylation during the recovery phase (39). The acute and proteasome function-independent changes in protein modification by SUMO-2/3 in response to cell stress, and the importance of SUMO-2/3 in cell survival after heat shock and recovery from ischemia, suggest that stress-induced SUMOylation plays a role in protein quality control. It is likely to be one of the early cellular defense mechanisms against stress-induced proteotoxicity and protein aggregation (7, 48, 54). The finding that SUMOylation attenuates Parkinson disease-associated α-synuclein aggregation and cytotoxicity strongly supports the notion that SUMOylation promotes the solubility of aggregation-prone proteins (19). Interestingly, a SUMOylation level of as little as 10% was shown to be sufficient to cause a dramatic delay in the aggregate formation of α-synuclein. In Kennedy's disease, in turn, expansion of the polyglutamine tract in the N-terminal domain of AR causes aggregate formation of the receptor. In keeping with the protein solubility-enhancing role of SUMOylation, overexpression of SUMO-3 attenuated the aggregate formation of ectopically expressed polyglutamine track-expanded AR in HeLa cells in a fashion that was dependent on the major AR SUMOylation sites but, interestingly, independent of the transcriptional activity of AR (27). The mechanism underlying the protein solubility-promoting effects of covalently conjugated SUMOs is currently unclear. One plausible mechanism is that SUMOylation cooperates with chaperone proteins, such as HSP70. Interestingly, preexposure of VCaP cells to a mild heat shock that increases cellular HSP70 protein levels blunts the conjugation of SUMO-2/3 to AR during a repeated stress (M. Rytinki, unpublished observations), suggesting that the stress-induced chaperone protein system and SUMOylation are at least to some extent linked.

The mechanisms underlying the acute changes in the SUMO-2/3 conjugate levels in response to cell stress are still elusive. It is possible that cell stress concomitantly affects both the forward and the reverse reactions of the SUMOylation pathway, altering the balance of SUMOylation cycles toward conjugate formation. Attenuation of SENP activity was detectable in extracts of heat-shocked VCaP cells assessed by conjugation of SUMO-2 vinyl sulfone in vitro (S. Kaikkonen, unpublished data). Depletion of PIAS1 was accompanied in turn by a modest reduction in levels of SUMOylated proteins, including the AR.

Heat shock treatment of VCaP prostate cancer cells resulted in rapid inactivation of AR activity, as illustrated by repression of AR target gene expression and detachment of the receptor from its binding sites on the chromatin. Interestingly, the removal of the AR from the chromatin was paralleled by targeting and sequestration of the receptor to a “biochemically insoluble” nuclear matrix fraction which at least in part corresponded to the PML bodies that have been shown to be enriched in SUMOylation pathway components (11, 28). In agreement with the latter notion, immunoblotting of the VCaP cell nuclear matrix revealed the presence of UBC9, PIAS1, PIAS2, PIAS3, and SENP1, -2, -3, and -5 in this fraction (M. Rytinki and S. Kaikkonen, unpublished results). Similarly, SUMOylated orphan nuclear receptor LHR-1 (liver receptor homolog) has been shown to be compartmentalized in the PML bodies/nuclear matrix in a dynamic fashion (4). PML nuclear bodies were suggested to function as dynamic storage sites controlling availability of transcription factors to chromatin. Association of nuclear receptors, including the AR, with chromatin is a dynamic and cyclic process (12, 16). It is likely that at least a portion of the AR is continuously cycling between the chromatin and nuclear matrix and that heat shock alters the dynamics of the process so that the recycling to the chromatin is prevented. Previously, interaction of glucocorticoid receptor with the nuclear matrix was found to be dynamic and regulated by an ATP-driven process (45). In the case of LHR-1, SUMOylation was suggested to tether the orphan receptor in the nuclear matrix, as release of LHR-1 from the nuclear bodies correlated with its deSUMOylation. It is possible that “basal” SUMOylation cycles play a role in the agonist-promoted association of the AR with the nuclear matrix, as the agonist was inert in inducing the nuclear matrix compartmentalization of the SUMOylation-defective AR. However, since heat shock had a more pronounced nuclear matrix-tethering effect on the AR mutant, SUMOylation per se does not seem to play a major role in targeting the AR to the nuclear matrix under stress conditions but sustains the nucleoplasmic solubility of the AR. In line with this result, the SUMOylation-defective AR was also found to be less mobile than the wtAR in the nucleoplasm. Balanced SUMOylation/deSUMOylation may thus at least in part control the chromatin/nuclear matrix portioning of the AR. Interestingly, since prostate cancer progression is related to profound changes in nuclear matrix protein expression patterns (2, 31), these changes may in turn influence the balance of levels of AR in different nuclear compartments.

Taken together, our results suggest that the SUMOylation is not merely linked to repression of AR activity but is more generally involved in the activity cycles of the AR, affecting its nuclear mobility, solubility, and stability in the nucleus and on the chromatin. The androgen-induced SUMOylation cycles of the AR are suggested to be linked to the normal activity of the holo-AR in the nucleus and chromatin binding, whereas the cell stress-induced SUMO-2/3 modifications are likely to sustain AR protein homeostasis in the nucleus. Since prostate transformation is linked to cell stress (42), the stress-induced SUMO-2/3 modifications may also contribute to deregulated androgen signaling during prostate carcinogenesis.