Increasing evidence links deregulation of the ubiquitin-specific proteases 22 (USP22) deubitiquitylase to cancer development and progression in a select group of tumor types, but its specificity and underlying mechanisms of action are not well defined. Here we show that USP22 is a critical promoter of lethal tumor phenotypes that acts by modulating nuclear receptor and oncogenic signaling. In multiple xenograft models of human cancer, modeling of tumor-associated USP22 deregulation demonstrated that USP22 controls androgen receptor accumulation and signaling, and that it enhances expression of critical target genes coregulated by androgen receptor and MYC. USP22 not only reprogrammed androgen receptor function, but was sufficient to induce the transition to therapeutic resistance. Notably, in vivo depletion experiments revealed that USP22 is critical to maintain phenotypes associated with end-stage disease. This was a significant finding given clinical evidence that USP22 is highly deregulated in tumors, which have achieved therapeutic resistance. Taken together, our findings define USP22 as a critical effector of tumor progression, which drives lethal phenotypes, rationalizing this enzyme as an appealing therapeutic target to treat advanced disease. Cancer Res; 74(1); 272–86. ©2013 AACR.
Developing new means to treat aggressive tumors is of critical clinical importance. To date, there are few metrics that identify the subset of tumors that will progress to therapeutic resistance, and means to treat later stages of disease remain elusive. However, a gene signature was identified that predicts death from disease across multiple solid tumors; this “Death-from-Cancer” signature is composed of 11 genes and predicts for disease recurrence, formation of metastasis, and/or response to therapy (1). Most of the signature impinges on known cancer-associated pathways (e.g., mitosis, growth factor signaling), and have established roles in promoting protumorigenic phenotypes. Intriguingly, the deubiquitylating (DUB) enzyme USP22, one of >50 ubiquitin-specific proteases (USP) that cleave ubiquitin (Ub) moieties from target substrates (2), is a key element of the signature with poorly defined functions in human cancer.
Although understanding of USP22 function remains scant, recent studies provide preliminary insight into its cellular activities. Initially, USP22 was identified as conserved component of human SAGA (Spt–Ada–Gcn5 acetyltransferase) transcriptional regulatory complex (3, 4) through which USP22 modulates transcription of genes through mono-deubiquitylation of histones H2A and H2B (5, 6). In addition, USP22 promotes stability of multiple cancer-associated protein targets through deubiquitylation (e.g., TRF1 and SIRT1; refs. 7, 8), and influences oncogene accumulation (e.g., BMI1; 10). Alternatively, USP22 utilizes deubiquitylation as a means to modulate protein function and promote cell proliferation (9), and increased USP22 expression is associated with poor outcomes in multiple cancers (11–16). Importantly, USP22 proved to be required for c-MYC (referred to here as MYC) function and potentiates MYC-mediated oncogenic cell transformation in a subset of cancers (4). Breast cancer and prostate adenocarcinoma frequently display altered MYC signaling throughout disease progression (17, 18), and in a subset of these tumor types, MYC deregulation acts in concert with hormone-regulated transcription factors (e.g., the estrogen receptor and the androgen receptor, AR) to promote malignant phenotypes. In breast cancer, MYC deregulation occurs in ∼40% of tumors and is a downstream effector of estrogen receptor signaling that promotes estrogen-induced cell proliferation and endocrine-therapy resistance (19, 20). In models of prostate adenocarcinoma, high MYC drives tumorigenesis (21), whereas in human disease, elevated MYC (mRNA and protein) occurs early, and MYC gene amplification correlates with both disease progression and poor survival (22). MYC is positioned downstream of androgen receptor and can promote prostate adenocarcinoma cell growth in the absence of androgens, termed castration resistance (23). Importantly, based on mouse model studies, MYC-dependent prostate tumor formation requires androgen receptor signaling (21). Thus, MYC is a known effector of breast cancer and prostate adenocarcinoma progression, frequently acting in concert with estrogen receptor and androgen receptor.
Although resistance to hormone therapy in breast cancer can occur intrinsically by lack of estrogen receptor expression (20), prostate adenocarcinoma is exquisitely dependent on androgen receptor signaling at all stages; thus, ablation of androgen receptor activity (known as androgen deprivation therapy, ADT) is the first line of therapeutic intervention for disseminated disease. Resistance to ADT occurs as a result of inappropriately restored androgen receptor activity (termed castration-resistant prostate adenocarcinoma, CRPC) through multiple mechanisms, most frequently via enhanced accumulation of androgen receptor (24). Heightened androgen receptor expression alone is sufficient to drive CRPC in xenograft models, and high androgen receptor is robustly associated with increased risk of death from prostate adenocarcinoma (25). Based on genome-wide analysis in prostate adenocarcinoma, androgen receptor binding patterns are enriched for MYC binding motifs in CRPC (26), suggesting that androgen receptor and MYC act in concert to promote CRPC progression. Thus, developing means to cooperatively target androgen receptor and MYC would be of significant clinical benefit.
Given these observations and the need to develop additional targets to manage advanced disease, it was hypothesized that USP22 may be positioned to control fundamental oncogenic signaling nodes implicit to prostate adenocarcinoma initiation and/or progression. Findings herein demonstrate that USP22 predicts for prostate adenocarcinoma disease outcome, promotes CRPC phenotypes, and is necessary for CRPC tumor maintenance by controlling signaling events dually regulated by androgen receptor and MYC. Key observations demonstrate, for the first time, that USP22 expression promotes activation of targets genes coordinately regulated by androgen receptor and MYC, which is maintained in the absence of androgens or the presence of androgen receptor antagonists, through enhanced androgen receptor protein accumulation. Most strikingly, USP22 deregulation induces androgen-independent androgen receptor recruitment to target gene regulatory loci and subsequent expression of a CRPC-associated gene profile, and supports cell growth and proliferation in the absence of androgens. Conversely, depletion of USP22 dramatically downregulates androgen receptor protein levels and abrogates basal and dihydrotestosterone (DHT)-stimulated androgen receptor activity in both ADT-sensitive and CRPC cells. Finally, USP22 is significantly upregulated in CRPC tumor samples compared with primary tumors, and is requisite for CRPC xenograft tumor growth. In sum, these studies identify USP22 as a regulator of oncogene expression and activity; a novel driver of progression to CRPC by virtue of the ability to regulate androgen receptor levels, androgen receptor output, and coordination of androgen receptor/MYC signaling.
Materials and Methods
Cell culture and treatments
LNCaP and C4-2 cells were maintained in IMEM supplemented with 5% FBS (heat-inactivated FBS). 22Rv1 cells were maintained in Dulbecco's Modified Eagle Medium supplemented with 10% FBS. Media was supplemented with 2 mmol/L l-glutamine and 100 units/mL penicillin–streptomycin. For hormone-deficient conditions, phenol red-free media was supplemented with charcoal dextran-treated serum. DHT was used at 1 nmol/L for 16 hours, unless otherwise noted. Casodex was used at 20 μmol/L, cycloheximide (CHX) at 10 μg/μL (Fisher), epoxomicin at 10 μmol/L (Santa Cruz), MG132 at 25 μmol/L (Santa Cruz). Cell lines were not cultured longer than 6 months after receipt from the original source of American Type Culture Collection.
Gene expression analysis
mRNA was analyzed by quantitative real-time (qRT)-PCR as described (27) using primers described in Supplementary Table S1.
Cell growth assays
Cells were treated with 1 μg/mL doxycycline or USP22 inhibitor then seeded at equal densities in hormone proficient or depleted conditions, as indicated. Cells were harvested at indicated time points and cell number was determined using trypan blue exclusion and a hemacytometer. Media and treatments were refreshed every 72 hours.
Cells were treated and seeded in hormone proficient or depleted conditions, as indicated, and labeled with BrdU (Invitrogen) 2 hours before harvest. Cells were fixed in 100% ethanol, stained with FITC-conjugated anti-BrdU antibody (BD Biosciences), and processed using FACS Calibur (BD Biosciences).
Chromatin immunoprecipitation analysis
Cells were cultured in hormone-depleted conditions for 72 hours and chromatin immunoprecipitation (ChIP) analyses and quantitative PCR (qPCR) were performed as previously described (27) using primers described in Supplementary Table S1.
Immunoblotting and protein stability
Cells were seeded in hormone proficient or hormone-deficient conditions and treated as specified. Protein isolation and immunoblotting were conducted as previously described. Antibodies used to detect proteins were: AR-N20 (previously described; ref. 27), USP22 (3), MYC (Cell Signaling), GAPDH (Santa Cruz), and β-actin (Santa Cruz). Protein stability was analyzed by seeding equivalent cell number and 24 hours later treating with 10 μg/μL CHX (Fisher Scientific) for indicated time course. To analyze contributions of proteasomal degradation, LNCaP cells were infected with indicated shRNA-containing lentivirus and selected with puromycin for 5 days, including culture in hormone-deficient media for 72 hours, stimulation with 1 nmol/L DHT, and 25 μmol/L MG132 or 10 μmol/L epoxomicin treatment for final 8 hours.
Lentiviral shRNA plasmids corresponding to USP22-1 (XM_042698.6-914s1c1; Sigma), USP22-2 (XM-042698.6-2196s1c1; Sigma), and luciferase (SHC007; Sigma) were obtained from TRC library and used as previously published (4). For siRNA-mediated depletion, USP22 SMARTpool oligonucleotides (M-006072-01) were transfected with Dharmafect (Dharmacon) and incubated for 72 hours. For inducible USP22 depletion, shRNA sequences targeting USP22 used for transient depletion were annealed and cloned into pENTR/d-topo and packaged into virus using the techniques mentioned above. Tet-Repressor (TetR) expression constructs were purchased from Invitrogen as part of the Virapower system and positive populations were selected using Blasticidin.
TMAs were stained for USP22 by using USP22 polyclonal antibody diluted 1:150 (NBP1-49644 Novus) with detection via LEICA polyvision+ (PV6119; LEICA Microsystems). Briefly, unstained 5 μm sections were cut from paraffin TMA blocks; slides were deparaffinized by standard techniques, steamed for 25 minutes in sodium citrate buffer, cooled for 5 minutes, blocked with peroxidase blocking solution for 5 minutes, incubated with the primary antibody for 45 minutes at room temperature, and incubated with secondary PowerVision+ rabbit antibody for 30 minutes.
All procedures involving mice were performed in accordance with Thomas Jefferson University IUCUC protocols. Seven-week-old SCID mice (NCI Frederick) were surgically castrated and 7 days later 2.75 × 106 cells in 100 μL total saline/Matrigel (BD Biosciences) were injected subcutaneously into the flank. When tumors reached ∼100 to 150 mm3 mice were administered 2 mg/mL doxycycline in sucrose-supplemented water. Doxycycline water was refreshed every 4 days. Tumor volume was measured with calipers.
All results were analyzed using the 2-tailed Student t test (adjusted for variance) or Mann–Whitney test. For all analyses, P < 0.05 was deemed significant.
In “Death-from-Cancer” gene signature, USP22 selectively predicts for prostate adenocarcinoma patient survival
The “Death-from-Cancer” 11-gene signature predicts for disease recurrence, metastasis, and therapeutic failure in multiple cancers; in prostate adenocarcinoma, the signature predicted for recurrence after therapy and decreased 5-year survival (1). Analysis of individual components revealed that 10 of the 11 genes are altered in prostate adenocarcinoma (28); thus, patient survival was stratified for individual components. As shown in Fig. 1A, 6 of 10 genes demonstrate no significant, independent predictive power; by contrast, elevated levels of BUB1, KNTC1, mKi67, or USP22 alone associated with poor survival. Functional analyses of these 4 reveal a distinct pattern, in that BUB1 (a regulator of mitotic spindle checkpoints), KNTC1 (involved in chromosome segregation), and Ki67 (a known marker of active mitotic cycling) are each associated with enhanced cellular proliferation. These findings are consistent with previous reports linking enhanced cellular proliferation to poor outcome (29). However, USP22, which was also a robust marker of poor outcome, has no known role in the cell cycle.
To assess putative consequences, cooperative events were investigated. Analysis of primary and metastatic prostate tumors indicated that USP22 deregulation co-occurred with both androgen receptor and MYC upregulation within the same tumor (Fig. 1B, left, P = 0.04 and 0.002, respectively). When survival of patients within this dataset was analyzed, USP22 combined with androgen receptor perturbation statistically predicted for poor outcome (Fig. 1B, right). These data demonstrate that a significant proportion of prostate adenocarcinoma specimens harbor aberrant USP22, androgen receptor, and MYC expression. This is significant, as androgen receptor upregulation drives the CRPC phenotype, MYC is a known prostate adenocarcinoma oncogene, and USP22 regulates MYC transcription. Therefore, it was critical to gain further insight into the role of USP22 in hormone-dependent cancer.
USP22 enhances androgen receptor activity and promotes bypass of androgen receptor antagonists
Because high USP22 predicted for poor outcome and was altered in concert with androgen receptor and MYC, this disease state was modeled by stable upregulation of USP22 (LN-USP22) in ADT-sensitive prostate adenocarcinoma cells. Compared with control (LN-Vec) cells, androgen receptor activity in the absence of androgen was low, as expected. By contrast, USP22 deregulation induced marked enhancement of ligand-independent androgen receptor activity, determined by analyses of multiple, clinically relevant androgen receptor target genes (Fig. 2A). Levels of induction were similar to that observed with DHT in control cells (Fig. 2A). Moreover, USP22 and DHT acted cooperatively to further enhance androgen receptor activity (Fig. 2A). These data demonstrate that USP22 potentiates both ligand-dependent and ligand-independent androgen receptor function.
Based on previous findings that increased androgen receptor activity is sufficient to bypass the response to androgen receptor antagonists (30), the ability of the direct androgen receptor antagonist Casodex to inhibit androgen-mediated androgen receptor activity in the context of USP22 deregulation was determined. Casodex abrogated androgen-mediated androgen receptor activity in control cells (compared with DHT-stimulated conditions). However, androgen receptor activity was resistant to Casodex in LN-USP22 cells (1.28-, 4.05-, and 6.21-fold increases of KLK3, TMPRSS2, and FKBP5 expression, respectively, compared with control; Fig. 2A). These data demonstrate that tumor-associated USP22 elevation has the capacity to activate androgen receptor in the absence of ligand, and render androgen receptor function refractory to Casodex. Thus, USP22 not only enhances androgen receptor activity in the absence of ligand, but USP22 thwarts the effects of androgen receptor antagonists.
Because USP22 is necessary for MYC function (4), the ability of USP22 modulation to influence genes coregulated by both androgen receptor and MYC was determined. Ornithine decarboxylase (ODC) gene expression is induced by MYC (31) and androgen receptor (32), and is overexpressed in prostate adenocarcinoma (33). Accordingly, ODC mRNA increased 2-fold in response to DHT in control cells, and was abrogated by Casodex (Fig. 2B). LN-USP22 cells expressed significantly higher ODC in the absence of DHT. In addition, DHT stimulation in USP22-upregulated cells promoted a ∼4-fold increase in ODC expression above DHT-stimulated conditions (Fig. 2B), which was significantly sustained in LN-USP22 cells upon Casodex treatment (although compared with the androgen receptor targets examined in Fig. 2A, Casodex showed a relatively more pronounced inhibitory effect). Given these results, the impact on additional MYC target genes that are unaffected by androgens, MTA1 and BAG1, was assessed. As expected, DHT had no significant impact on expression; however, little effect on gene expression was observed by USP22 upregulation (Supplementary Fig. S1A), indicating that the effects of USP22 on MYC in this tumor type may be restricted to genes coregulated by androgen receptor and MYC, and that USP22 may be required for but not sufficient to alter MYC activity in prostate adenocarcinoma cells.
Next, the putative mechanisms by which USP22 alters transcriptional output were determined. Consistent with previous reports, LN-USP22 cells expressed similar levels of MYC (compared with control) in the absence or presence of DHT (Supplementary Fig. S1B). As expected, androgen receptor protein levels in control cells were enhanced by DHT, attributed to the known capacity of androgen to stabilize androgen receptor (Fig. 2C, left, compare lanes 1 and 3). Strikingly, USP22 upregulation in the absence of exogenous DHT enhanced androgen receptor levels similar to that observed with DHT in control cells (Fig. 2C, left, compare lanes 1, 2 and 2, 3). Furthermore, DHT and USP22 acted in concert to further increase androgen receptor expression above that seen in DHT-stimulated or USP22 high cells alone (Fig. 2C, left, compare lanes 3 and 4). This increased androgen receptor was not a result of increased androgen receptor gene expression, as mRNA levels were not increased by USP22 (Fig. 2C, right). These data put forth the provocative hypothesis that USP22 increases the activity of androgen receptor through altering androgen receptor protein levels, thus identifying a novel role for USP22 in modulating steroid receptor function.
USP22 promotes castration-resistant phenotypes
The observations that USP22 upregulation is sufficient to promote ligand-independent androgen receptor expression/activity, and induce Casodex resistance are clinically relevant, as these attributes reflect key biochemical characteristics of CRPC. To determine if USP22 is sufficient to promote the transition to CRPC, additional molecular readouts were assessed. First, recruitment of androgen receptor to well-characterized androgen receptor occupied regions (AROR) of clinically relevant androgen receptor target genes in the absence of hormone was quantified by ChIP-qPCR. As expected, androgen-deprivation of control cells was coincident with low androgen receptor occupancy (∼0.3% input; Fig. 2D). Strikingly, USP22 deregulation significantly increased androgen receptor occupancy at known ARORs (1.15–2.6% input; Fig. 2D), but not in control regions of the KLK3/PSA (“EF” region; Supplementary Fig. S1C). These results indicate that enhanced USP22 promotes androgen receptor binding and androgen receptor–dependent transcription in the absence of ligand, suggesting that USP22 mediates castrate-resistant androgen receptor activity.
Second, the ability of USP22 to promote gene signatures strongly associated with CRPC was assessed. Although UBE2C expression was unchanged, CDC20 and CDK1 (34) were both significantly increased to levels seen in other models of CRPC (Fig. 2E). In addition to upregulation of known androgen receptor target genes, the CRPC-specific signature also includes androgen-repressed genes associated with polycomb group protein pathways involved in development and differentiation (35). Consistent with the premise that USP22 drives CRPC-associated androgen receptor activity, expression of OPRK1, SI, MET, and DDC were significantly downregulated under castrate conditions in LN-USP22 cells, relative to control conditions (Fig. 2E, right). Together, these data demonstrate that USP22 regulates ligand-independent androgen receptor residence at target gene loci and promotes androgen receptor–driven CRPC gene profiles, which may have specificity for USP22 perturbation, further implicating USP22 as an independent effector of aggressive tumor phenotypes.
USP22 regulates proteasome-dependent androgen receptor degradation
Because the deubiquitylase function of USP22 can putatively protect substrates from degradation, the impact of USP22 expression on androgen receptor stability was initially measured using CHX. Androgen receptor stability was enhanced by USP22, whereby the androgen receptor half-life was extended from 14 hours (LN-Vec) to 18 hours (LN-USP22; Fig. 3A). To further define the mechanism by which USP22 regulates androgen receptor, previously characterized shRNAs against USP22  or control (luciferase, shLUC) were used in the presence or absence of proteasome inhibitors. First, depletion of USP22 significantly reduced androgen receptor protein (∼73%) compared with control (Fig. 3B, compare lanes 3, 4, and 7, 8). In addition, treatment with 2 different proteasome inhibitors MG132 (Fig. 3B, left) or epoxomicin (Fig. 3B, right) rescued androgen receptor levels by ∼3-fold. These data suggest that USP22 functions to enhance androgen receptor stability and promote inappropriate castration-resistant androgen receptor signaling through proteasome-dependent regulation of androgen receptor levels.
USP22 depletion suppresses ligand-dependent and castration-resistant androgen receptor expression and activity
As androgen receptor signaling is required for disease maintenance and progression, the ability to reduce androgen receptor expression and activity would be of clinical benefit. As expected, in control cells, DHT promoted significant androgen receptor accumulation (Fig. 4A, lanes 1 and 3). In contrast, USP22 depletion reduced androgen receptor protein levels in the absence of androgen, and inhibited DHT-induced androgen receptor expression, within models of therapy-sensitive disease (Fig. 4A, compare lanes 2 and 4). In addition, siRNAs targeting USP22 resulted in loss of androgen receptor protein (Supplementary Fig. S2) and reduced androgen receptor function. In androgen-deprived conditions, USP22 depletion modestly decreased basal androgen receptor activity. Androgen stimulation of ADT-sensitive cells increased KLK3, TMPRSS2, FKBP5, and ODC gene expression 2.5-, 4.7-, 8.6-, and 4-fold, respectively; however, USP22 depletion significantly blunted DHT effects (decrease of 30%, 72%, 72%, 82%, and respectively; Fig. 4B). It was further queried whether USP22 suppression could alter androgen receptor activity in the castrate setting. As shown, in CRPC cells, USP22 depletion suppressed androgen receptor protein accumulation in both the androgen-stimulated and deprived conditions (Fig. 4C, compare lanes 1, 2 and 3, 4). Thus, the effect of USP22 downregulation on androgen receptor levels is retained in CRPC. With regard to function, CRPC cells exhibited robust androgen receptor activity in the absence of ligand. Consistent with the impact on androgen receptor levels, androgen receptor activity was suppressed (Fig. 4D), indicating that USP22 is both sufficient and necessary for CRPC-derived androgen receptor activity. Although CRPC cells retain substantive androgen receptor activity in the absence of androgen, the receptor does remain responsive to DHT stimulation. Exemplifying this, control cells stimulated with DHT showed increased KLK3, TMPRSS2, FKBP5, and ODC gene expression (1.6, 2.0, 17.2, and 11.9-fold; Fig. 4D). Similar to results in ADT-sensitive cells (Fig. 4B), USP22 depletion abrogated DHT-induced androgen receptor activity in CRPC (Fig. 4D). Combined, these data illustrate the potential for targeting androgen receptor in both early-stage and advanced disease, underpinned by the dramatic requirement of USP22 to maintain androgen receptor levels and activity.
Another mechanism capable of contributing to transition to CRPC is expression of constitutively active androgen receptor splice variants (AR-SV), which lack the ligand-binding domain and, as such, are unresponsive to androgen receptor antagonists (36–38). Notably, when USP22 was upregulated in CRPC model systems expressing both androgen receptor full length (AR-FL) and AR-SVs, levels of both species were significantly enhanced (Fig. 5A). USP22 further increased ligand-independent androgen receptor transcriptional activity, based on KLK3/PSA expression, compared with control (Fig. 5B). To determine if loss of USP22 could impact AR-SV expression, and represent a novel mechanism to target constitutively active androgen receptor, USP22 expression was depleted by shRNA (shUSP22-2), which resulted in significantly reduced protein expression of both AR-FL and AR-SVs in androgen-deprived and DHT-stimulated conditions (Fig. 5C). In addition, in both culture conditions, USP22 depletion diminished androgen receptor activity (Fig. 5D). Combined, these data suggest that USP22 expression is required for androgen receptor expression in ADT-sensitive prostate adenocarcinoma and CRPC models, and represents a novel target that can modulate the expression of both full length and constitutively active androgen receptor.
USP22 promotes ligand-dependent and castrate-resistant prostate adenocarcinoma cell growth
To determine the biological impact of USP22 deregulation, the impact on cellular outcomes was determined. First, using hormone-proficient conditions, USP22 enhanced the rate of cell-cycle progression, evidenced through increased BrdU incorporation (Fig. 6A, left). Second, USP22 significantly increased cell growth in the presence of androgen (Fig. 6A, right). Third, and most critically, USP22 robustly promoted cell growth and proliferation in the absence of androgen. As shown in Fig. 6B, control cells significantly reduced BrdU incorporation upon hormone deprivation (compare Fig. 6A), whereas, USP22 upregulation induced a 2.7-fold increase in BrdU incorporation and substantially enhanced cell proliferation rates in the absence of androgen. Thus, USP22 promotes biological and biochemical castration resistance, thus defining a novel mechanism of CRPC progression.
Given the ability of USP22 to enhance androgen receptor accumulation, androgen receptor activity, and CRPC, the biological impact of a model of tetracycline-inducible shUSP22 was developed in therapy-sensitive prostate adenocarcinoma cells. As shown, doxycycline (Dox) decreased USP22, resulting in marked loss of androgen receptor (Fig. 6C, top), attenuated cell-cycle progression (determined by BrdU incorporation; Fig. 6C, top right, middle), and significantly suppressed cell doubling (Fig. 6C, bottom). These data collectively demonstrate that USP22 is critical for cell-cycle progression and cell growth in ADT-sensitive prostate adenocarcinoma.
USP22 expression is elevated in CRPC and is required for CRPC growth
Based on the data above, suppression of USP22 in models of ADT-sensitive prostate adenocarcinoma and aggressive CRPC decreased the androgen receptor signaling axis (Figs. 4 and 5). In addition, based on human gene expression and cell models, USP22 upregulation is associated with decreased patient survival (Figs. 1 and 2). To further interrogate the profile of USP22 alterations during tumor progression, USP22 expression was analyzed using clinical specimens of primary prostate adenocarcinoma and CRPC. As shown, USP22 was detected in both the cytoplasm and nuclei, and was expressed in surrounding stroma, in low-grade Gleason 6 prostate adenocarcinoma (Fig. 7A). In higher grade Gleason 8 prostate adenocarcinoma, USP22 expression was enriched in the nuclei of epithelia, in addition to cytoplasm and fibroblasts, although the overall staining intensity was not significantly increased between primary prostate adenocarcinoma groupings (Gleason 5/6 and 7/8/9; Fig. 7A). In CRPC, USP22 expression was highly detectable and enriched in the nucleus (Fig. 7A). When quantified, USP22 detection was significantly increased in CRPC, compared with primary tumors grouped by Gleason score (Fig. 7A). These findings suggest that USP22 increases as a function of disease progression and represents a critical mediator of progression to CRPC.
To determine whether USP22 could serve as a putative therapeutic target in advanced disease, CRPC xenografts expressing Dox-inducible shRNA directed against USP22 were utilized. As shown, Dox treatment resulted in decreased USP22 expression and corresponding reduction in androgen receptor expression (Fig. 7B, left), reduced the proliferative capacity (40% control to 0.6% BrdU-positive shUSP22 cells Fig. 7B, middle), and suppressed cell doubling (Fig. 7B, right). Given the robust in vitro biological response, the ability of USP22 depletion to inhibit CRPC growth in vivo was assayed by recapitulating the castration-resistant environment. For these studies, immunocompromised mice were castrated and 7 days later injected with CRPC cells expressing Dox-inducible shUSP22 or parental controls. As expected, given the ability of CRPC cells to proliferate in the absence of androgen, tumors were established and monitored for growth to 100 to 150 mm3. At that time, mice were randomized to Dox-supplemented drinking water or control. Before Dox administration, both cell models demonstrated similar percentage of tumor rate take (∼70%), and grew at similar rates. As shown, USP22 depletion resulted in tumor growth suppression (Fig. 7C). These findings strongly support the contention that not only is USP22 a driver of CRPC, but that established CRPC tumors are reliant on sustained USP22 expression and activity. Combined, these data herein identify USP22 as a master regulator of androgen receptor stability and activity that drives prostate cancer progression.
Given that the androgen receptor signaling axis is paramount for early stage prostate adenocarcinoma and development of lethal CRPC, understanding novel mechanisms of androgen receptor maintenance and uncovering new targets to thwart androgen receptor–mediated tumor growth is critical in advancing patient treatment options. This study presents novel data demonstrating the ability of a single “Death-from-Cancer” signature gene, USP22, to control signaling pathways requisite for prostate cancer cells. Key findings show that (i) increased USP22 predicts for poor patient outcome, (ii) USP22 is required for androgen receptor accumulation and activity, and maintained activity in the presence of antagonist treatment, (iii) USP22 promotes a castration-resistant transcriptional profile and associated therapeutic resistance, and (iv) USP22 expression is increased in CRPC tumor samples and is necessary for CRPC androgen receptor activity and tumor growth. Together, these results support the provocative hypothesis that USP22 activity is essential for androgen receptor expression, activity, and CRPC tumor growth.
USP22 controls androgen receptor stability and resultant prostate adenocarcinoma progression
The findings that USP22 regulates androgen receptor levels and activity, and promotes CRPC phenotypes are of significance, as restored androgen receptor activity is the major mechanism for transition to incurable CRPC (24). Findings herein demonstrate that USP22 is indispensible for castration-resistant androgen receptor expression and activity, cell proliferation, and tumor growth. USP22 promotes CRPC through enhanced ligand-independent and androgen-stimulated androgen receptor transcriptional activity and sustained androgen receptor activity in the presence of antagonists, concomitant with androgen receptor protein accumulation. These results are divergent with a previous report suggesting that USP22 functions as an androgen receptor transcriptional coactivator but does not influence protein expression (3). One important distinction is that previous findings are largely based on reporter assays analyzing USP22-mediated androgen receptor transactivation in Drosophila and human embryonic kidney (HEK) cells, and androgen receptor expression perturbation under conditions of ectopic overexpression of androgen receptor and USP22 in HEK cells. By contrast, the studies herein clearly demonstrate the importance of USP22 as a modulator of endogenous androgen receptor in human prostate adenocarcinoma. Furthermore, USP22 mediates androgen receptor expression through a proteasome-dependent mechanism, because modeling clinically relevant USP22 upregulation results in an increased androgen receptor protein half-life and proteasome inhibition rescued the decreased androgen receptor expression following USP22 depletion. Using multiple approaches to monitor changes in androgen receptor ubiquitylation, levels of this modification were not reproducibly changed in response to USP22 knockdown (Supplementary Fig. S3), suggesting that androgen receptor may be an indirect target. Alternatively, degradation could be driven by USP22-mediated mono-deubiquitylation and not detectable in the experimental assays. Similarly, several proteins (e.g., PAX3 and cyclinB1) can be monoubiquitylated and sufficiently targeted to the proteasome for degradation (39, 40). It is also possible that USP22 is indirectly regulating degradation in a proteasome-dependent, ubiquitin-independent mechanism. Such modes of regulation have precedent; for example, REGγ activates the proteasome and directly recruits substrates for degradation in the absence of ubiquitin (e.g., p21CIP1, p16INK4a; ref. 41, and SRC/AIB1; ref. 42). NQO1 associates with proteasome subunit 20S and p53 to regulate access to the proteasome in an NADH-dependent mechanism, and subsequently controls p53 degradation (43). Identification of potential endogenous USP22 targets is currently in progress.
These findings are among the first to identify cancer-associated molecules that alter androgen receptor stability. Selected E3 ligases [e.g., MDM2, CHIP (C-terminus of Hsp70-interacting protein), and Siah2] can control androgen receptor protein levels through addition of ubiquitin moieties and subsequent proteasomal degradation, but the link to human disease is uncertain. MDM2 can target androgen receptor for degradation in an AKT-dependent manner (44), but there is limited evidence demonstrating MDM2 deregulation in prostate adenocarcinoma. CHIP ubiquitylates androgen receptor (45), but it is unclear if CHIP expression is altered in prostate adenocarcinoma. Finally, Siah2 targets only chromatin-bound androgen receptor for degradation (46), consistent with previous studies demonstrating the requirement of ubiquitylation and turnover of nuclear receptors for efficient transcription. The link between DUBs and androgen receptor is even less well characterized. USP26 can deubiquitylate androgen receptor in the presence of androgen (47), whereas USP10 and 2A-DUB both regulate androgen receptor transcriptional activity but do not alter protein stability (48, 49). Thus, the studies here identify USP22, which is predictive for prostate adenocarcinoma disease outcome and is increased in advanced incurable patient samples, as a DUB that promotes androgen receptor accumulation and prostate adenocarcinoma progression.
USP22 regulates AR-SVs
Multiple mechanisms induce resurgent androgen receptor activity and resultant CRPC formation following hormone therapy, including androgen receptor overexpression/amplification, somatic androgen receptor mutations, posttranslational modification and cofactor perturbations, intracrine androgen synthesis, and AR-SVs (24). Here, constitutively active AR-SVs, which lack a ligand-binding domain and are not inhibited by clinically approved androgen receptor antagonists, also require USP22 for accumulation (Fig. 5). Given the clinical implications for AR-SV-positive tumors, intensive efforts are ongoing to elucidate AR-SV regulation. The androgen receptor ligand-binding domain contains 2 lysine residues (845 and 847) that promote degradation upon ubiquitylation, suggesting that regulation of AR-SV expression is independent of these canonical proteasomal degradation targets (50). Because USP22-mediated DUB activity can modulate transcriptional elongation (5) and H2B monoUb can affect exon skipping (51), it is possible that USP22 could be involved in mediating androgen receptor alternative splicing. Regardless of the mechanism, studies here show that disrupting USP22 expression or activity dramatically reduces AR-SV production and/or accumulation, thus providing one mechanism by which USP22 may control CRPC transition and presenting a potential means to target the heretofore nontargetable AR-SVs.
USP22 selectively controls MYC in prostate adenocarcinoma
The concept that USP22 controls androgen receptor activity is complemented by observations that USP22 is required for expression of genes coregulated by both androgen receptor and MYC. Prior reports showed that USP22 is recruited via SAGA to MYC binding sites and therein modulates MYC output. Further studies reveal that USP22 function is highly selective, such that it is required for activation of a subset of p53 target genes (4, 8). Studies herein suggest that USP22-mediated MYC regulation in prostate adenocarcinoma is likely gene selective, based on USP22 increasing gene expression of androgen receptor/MYC coregulated target (ODC) but not of multiple MYC targets. Whether MYC occupancy is altered at coregulated sites in response to tumor-associated USP22 deregulation remains an open question that will be revealed by genome-wide analyses. The concept that USP22 is required for genes coregulated by androgen receptor and MYC is intriguing, given the clinical importance of androgen receptor-MYC crosstalk. It is of interest that androgen receptor and MYC both function to promote expression of target genes required for growth and proliferation, and are intricately interconnected. In mouse models of prostate adenocarcinoma, MYC-driven tumor formation is reduced by castration or androgen receptor antagonists (52). Androgen stimulation increases MYC mRNA levels, which is suppressed in response to Casodex (23), and androgen receptor promotes MYC activation indirectly through mediating expression of the ETS fusion protein, TMPRSS2:ERG (53). ODC is one characterized gene dually regulated by androgen receptor and MYC. However, MYC binds to promoters of numerous androgen receptor target genes (54), suggesting that USP22 could regulate a unique, cell-specific gene signature that is critical for androgen receptor–MYC coordination.
USP22 deregulation in human malignancies
Although these studies identify a key role for USP22 in prostate adenocarcinoma, USP22 expression is predictive for poor outcome in numerous malignancies. For example, USP22 is required for hepatocellular carcinoma (HCC) and bladder cancer growth (55, 56) is elevated in advanced melanoma (57), and independently predicts for poor prognosis in colorectal carcinoma (12), invasive breast cancer (11), gastric cancer (58), esophageal and oral squamous cell carcinoma (15, 16), and papillary thyroid carcinoma (13). Although these outcomes could occur via USP22-mediated MYC/CyclinD2 and/or BMI-1–mediated modulation (10), androgen receptor activity promotes tumorigenesis in several of these tumor types, including bladder (59), HCC (60), and breast carcinoma (61). In contrast to the studies reported here for prostate adenocarcinoma, previous studies stopped short of identifying molecular targets that mediate USP22-dependent tumor phenotypes. In breast cancer, androgen receptor is a driver of disease in the molecular apocrine subtype, wherein androgen receptor drives a subset of classical estrogen receptor–responsive genes mediated by FOXA1 (62), and promotes a signaling pathway that activates β-catenin, HER3, and HER2 (61). Intriguingly, a feed-forward pathway was found within this tumor type, whereby androgen receptor activates MYC, which binds to promoters of androgen-responsive genes to promote sustained androgen receptor activity (54). Similar to prostate adenocarcinoma, cell growth is responsive to Casodex treatment (62) and USP22 depletion reduces androgen receptor expression (data not shown), suggesting that targeting USP22 in this breast cancer subset could be advantageous. Further investigation into the role of USP22 in molecular apocrine breast cancer is ongoing. In sum, USP22 represents a biomarker for advanced malignancies and can influence androgen receptor expression in multiple tumor scenarios, suggesting that USP22 inhibition could have implications in numerous tumor types.
USP22 as a therapeutic target
Despite a growing body of literature demonstrating that USP-family members are implicated in multiple diseases, including cancer, development of active site inhibitors is challenging (63). USP7 destabilizes multiple tumor suppressors, and as such is associated with deregulating signaling pathways altered in cancer (64). Several inhibitors have been generated that selectively inhibit USP7 in a reversible manner. However, no agents that specifically target USP enzymatic activity are in clinical trials. The potential benefit of inhibiting USP22 is of putative clinical relevance, as restored androgen receptor activity is the major mechanism for transition to incurable CRPC. As shown herein, USP22 promotes and is required for the maintenance of CRPC; conversely, USP22 depletion blunts cell growth in models expressing mutated androgen receptor, constitutively active androgen receptor, and androgen receptor amplification. USP22 depletion acted in concert with castration to elicit an enhanced biochemical and cellular response, and based on data presented herein, targeting USP22 should also act in concert with the recently approved CYP17A inhibitor abiraterone acetate, which targets intracrince androgen synthesis (65, 66). On balance, USP22 would be predicted to enhance therapeutic options for treatment of CRPC.
In summary, the present first-in-field observations identify USP22 as a major effector of androgen receptor levels, androgen receptor output, androgen receptor–MYC coordination, and the transition to CRPC. Robust in vitro and in vivo data support the concept that USP22 is not only a major driver of disease progression, but that therapeutic targeting of USP22 would be of likely high clinical benefit.
Disclosure of Potential Conflicts of Interest
S.B. McMahon is employed (other than primary affiliation; e.g., consulting) as a consultant in CellCentric Ltd. S.B. McMahon is a consultant/advisory board member of CellCentric Ltd. No potential conflicts of interest were disclosed by the other authors.
Conception and design: R.S. Schrecengost, J.L. Dean, J.F. Goodwin, M.J. Schiewer, S.B. McMahon, K.E. Knudsen
Development of methodology: R.S. Schrecengost, J.L. Dean, J.L. Hicks, A.M. DeMarzo
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.L. Dean, J.F. Goodwin, M.J. Schiewer, M.W. Urban, T.J. Stanek, R.C. Birbe, R.A. Draganova-Tacheva, T. Visakorpi, A.M. DeMarzo
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R.S. Schrecengost, J.L. Dean, R.C. Birbe, A.M. DeMarzo, S.B. McMahon, K.E. Knudsen
Writing, review, and/or revision of the manuscript: R.S. Schrecengost, J.L. Dean, J.F. Goodwin, M.J. Schiewer, R.C. Birbe, T. Visakorpi, K.E. Knudsen
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): R.T. Sussman, J.L. Hicks
Study supervision: K.E. Knudsen
The authors thank members of the K. Knudsen laboratory for input and commentary.
This work is supported by NIH grants (R01 CA099996, R01 ES016675-06, and R01 CA159945 to K.E. Knudsen, F32 CA156989-01 to R.S Schrecengost), a Prostate Cancer Foundation Mazzone Challenge Award to K.E. Knudsen, and by the Department of Defense Prostate Cancer Research Program Award No. W81XWH-10-2-0056 and W81XWH-10-2-0046 Prostate Cancer Biorepository Network (PCBN).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.