Targeting the turnover of oncoproteins as a new avenue for therapeutics development in castration-resistant prostate cancer

Abstract

The current therapeutic armamentarium for castration-resistant prostate cancer (CRPC) includes second-generation agents such as the Androgen Receptor (AR) inhibitor enzalutamide and the androgen synthesis inhibitor abiraterone acetate, immunotherapies like sipuleucel-T, chemotherapies including docetaxel and cabazitaxel and the radiopharmaceutical radium 223 dichloride. However, relapse of CRPC resistant to these therapeutic modalities occur rapidly. The mechanisms of resistance to these treatments are complex, including specific mutations or alternative splicing of oncogenic proteins. An alternative approach to treating CRPC may be to target the turnover of these molecular drivers of CRPC. In this review, the mechanisms by which protein stability of several oncoproteins such as AR, ERG, GR, CYP17A1 and MYC, will be discussed, as well as how these findings could be translated into novel therapeutic agents.

Introduction

Prostate cancer (PCa) is the most common non-cutaneous malignancy in men in the United States and will be responsible for 29,340 deaths in 2018 alone [1–3]. Androgen-deprivation therapy (ADT) is widely used as the first-line therapy for men with metastatic PCa. However, ADT is not curative and most patients relapse with castration-resistant prostate cancer (CRPC). Approved therapeutic options for CRPC include second-generation agents including the Androgen Receptor (AR) inhibitor enzalutamide [4, 5] and the androgen synthesis inhibitor abiraterone acetate [6, 7], immunotherapies like sipuleucel-T [8, 9], chemotherapies including docetaxel and cabazitaxel [10] and the radiopharmaceutical radium 223 dichloride [11, 12]. These agents extend the survival of patients with CPRC by an average of 3–4 months.

Resistance to therapeutic agents in CRPC can be both primary and acquired and results from genetic changes such as amplification, mutation or translocation of driver genes [13]. For example, resistance to androgen deprivation therapy enzalutamide and abiraterone may involve the expression of AR splice variants, such AR-V7 [14], or point mutations like AR-F876L [15] that render the AR immune to these agents, allowing AR signaling to persist. In addition, progression to AR-independent CRPC forms such as neuroendocrine prostate cancer (NEPC) is common. There is thus a clear unmet need to develop novel therapeutic strategies that target the molecular drivers of CRPC.

One novel approach to treat CRPC may be to target the turnover of the molecular drivers of CRPC. In this review, we will provide an overview of the current understanding on how the turnover of putative CRPC driver proteins is regulated and how this could be translated into novel therapeutic agents.

AR

The development and normal physiology of the prostate gland requires androgen receptor (AR) to be activated by androgens and subsequently express key genes that mediate prostate function. The AR is a critical driver of gene expression in PCa through all stages, including therapy-resistant CRPC. AR mutants (AR-mts) and AR variants (AR-Vs) are not seen in primary prostate cancer but emerge after anti-androgen therapy in CRPC [16–18]. AR-Vs and AR-mts can direct the AR transcriptional program in CRPC, without hindrance from anti-androgens. Even in enzalutamide or abiraterone-resistant CRPC, the AR is still the primary molecular driver. Therefore, the AR, AR-mts and AR-Vs remain the primary targets for therapeutic intervention in CRPC.

The human AR gene is located on chromosome X (Xq11-Xq12) and the canonical transcript is encoded by 8 exons. The AR full-length protein (AR-FL) contains 920 amino acids, including three major functional domains: an amino-terminal domain (NTD), a DNA-binding domain (DBD), a C-terminal ligand-binding domain (LBD), and a flexible hinge region (Figure 1). Currently, more than 20 AR-variants (AR-Vs) have been identified in PCa cell lines and prostate cancer clinical specimens [19, 20]. The proteins generated from most AR-Vs (Figure 1) lack the LBD domain, e.g. AR-V1[21], AR-V3[22], AR-V4[22], AR-V7[16], AR-V9[23] and AR-V12 (AR^V567es)[24], which confers androgen-independent constitutive activation. Another AR variant, AR-45, has an intact LBD domain but an altered NTD domain with only seven amino acids [19].

Figure 1. Domain architecture of AR, GR, ERG, CYP17A1 and MYC variants relevant in prostate cancer.

(A). Domain architecture of canonical AR and AR splice variants. NTD, N-terminal domain; DBD, DNA-binding domain; LBD, Ligand-binding domain; CE, Cryptic exon. (B). Domain architecture of GR. (C). Domain architecture of ERG. AD, Activation domain; PNT, Pointed domain; ETS, DNA-binding (Ets) domain. (D). Domain architecture of CYP17A1. P450, P450 superfamily domain. (E) Domain architecture of MYC and MYCN. MB1, Myc box 1; MB2, Myc box 2; MB3, Myc box 3; NLS, nuclear localization signal; bHLH/LZ, basic helix-loop-helix/leucine-zipper.

While expression of AR and AR-Vs are modulated through epigenetic [25], transcriptional [26], and posttranscriptional mechanisms [27], posttranslational regulation of AR and AR-V protein levels may play an important role in modulating AR signaling in PCa.

Regulation of ubiquitination

There are two major intracellular protein degradation systems in eukaryotic cells: the ubiquitin-proteasome system (UPS) and the autophagy-lysosome pathway (ALP), which both play roles in AR protein degradation [28, 29]. In the ubiquitin-proteasome system, three types of enzymes, ubiquitin-activating enzyme (E1), ubiquitin-conjugating enzyme (E2), and ubiquitin ligase (E3), are required for protein ubiquitination: E1 binds to ubiquitin at the C-terminal carboxyl group in an ATP-dependent manner, E2 binds to both activated ubiquitin and E1, and catalyzes the transfer of ubiquitin from E1 to E2, finally, E3 binds to E2 and the protein substrate, and ligates the carboxyl group of the last amino acid (glycine) of ubiquitin to an epsilon-amino group of a lysine residue of the target protein by an isopeptide bond. Ubiquitination of proteins may result in multiple types of ubiquitinated states including mono-ubiquitination or K6-, K11-, K27-, K29-, K33-, K48-, K63- and M1- linked poly-ubiquitin linkage [30]. These distinct ubiquitinated states have different fates: K-48 linked poly-ubiquitinated proteins are subjected to degradation via the proteasome or lysosome, while mono- or polyubiquitination affects the subcellular localization or enzymatic activity [31, 32]. The approximately 1,000 E3 ubiquitin ligases are primarily responsible for the specificity and efficiency of ubiquitination. In addition, deubiquitinating enzymes (or deubiquitinases, DUBs), a group of ~100 proteases, can remove ubiquitin from proteins and counteract the effects of E3 ubiquitin ligase. Together, the activity of ubiquitin ligases and DUBs can finely regulate the ubiquitination state of proteins, and thus their stability [33].

Ubiquitination of AR

Androgen binding to the AR results in ~6-fold increased AR half-life [34]. Direct evidence of AR degradation via the UPS was demonstrated by increased endogenous AR protein levels following treatment with the proteasome inhibitor MG-132 and detection of polyubiquitinated AR [35].

E3 ubiquitin ligases that have been reported to ubiquitinate AR proteins (Table 1), belong to two major families: the HECT (Homologous to the E6AP Carboxyl Terminus) and the RING (Really Interesting New Gene) ubiquitin E3 ligases [36, 37]. Distinct AR domains interact with these E3 ubiquitin ligases: UBE3A (E6-AP) [38], RCHY1 (ARNIP) [39], and CHIP [40] bind to the AR NTD, RNF4 (SNURF) [41] to the AR DBD, while RNF6 [42] and SIAH2 [43] bind to the AR LBD. Some E3 ligases like RNF14 (ARA54) [44, 45] and MDM2 [46] can interact with both the AR NTD and DBD domains. In contrast, the AR domain required for NEDD4 binding has not yet been defined [47]. In addition, the ubiquitin E3 ligase adaptor SPOP recognizes a serine/threonine-rich degron in the hinge domain of AR and AR^V567es for ubiquitination by the SPOP-CUL3-RBX1 E3 ligase complex [48], and ubiquitinates AR-V7 only when full length AR is coexpressed [49]. Predictably, AR-Vs can be targeted by the AR NTD binding E3 ligases like CHIP [50] MDM2 [51] and RCHY1.

Table 1.

E3 ubiquitin ligases and DUBs for AR

Protein	Subfamilies	Interaction domain In AR	Functions	References
UBE3A(E6-AP)	HECT	AF1(NTD)	Co-activator	38
NEDD4	HECT	N.A.	Stability	47
RNF14(ARA54)	RING finger	652–920(DBD LBD) FXXLF motif	Transcriptional activity	44
RCHY1(ARNIP)	RING finger	11–172(NTD)	NA	39
MDM2	RING finger	NTD and DBD, Ser213	Stability, Transcriptional activity	46, 51
RNF4(SNURF)	RING finger	DBD	Transcriptional activity	41
RNF6	RING finger	Thr850	Transcriptional activity	42, 52
CHIP(STUB1)	U-box	NTD 220–270	Stability Transcriptional activity	40
SIAH2	RING finger	LBD	Stability Transcriptional activity	43
SPOP	Speckle-type poxvirus and zinc finger (POZ)	Hinge region	Stability Transcriptional Mediator	48
USP7	ubiquitin-specific protease (USP/UBP) superfamily	NA	Chromatin binding	74
USP10	ubiquitin-specific protease (USP/UBP) superfamily	NA	Co-activator	67
USP12	ubiquitin-specific protease (USP/UBP) superfamily	NA	Deubiquitination, Stabilization	71
USP14	ubiquitin-specific protease (USP/UBP) superfamily	NA	Deubiquitination, Stabilization	77
USP26	ubiquitin-specific protease (USP/UBP) superfamily	NA	Deubiquitination, Stabilization	69

The specificity of ubiquitination can be regulated by posttranslational modifications (PTMs), which can be placed directly on the substrate to modulate the binding of an E3 ligase, or on the E3 ligase itself to modulate its enzymatic activity [36]. Ser-213 and Ser-791 phosphorylation is necessary for AR degradation by MDM2 E3 ligase [46], while Thr-850 phosphorylation may stabilize AR by recruiting RNF6 [52]. Phosphorylation of AR Ser-578 by PAK6 (p21-activated kinase 6) can also promote the association between AR and MDM2 to activate MDM2-mediated proteasomal degradation of AR [53]. These findings suggest that MDM2 is a key E3 ubiquitin ligase for phosphorylation-mediated AR degradation.

Importantly, the interaction of distinct E3 ligases with AR and AR-Vs may have vastly different effects. RNF6 stabilizes AR and enhances AR transcriptional activity via K6/K27-linked polyubiquitination at AR-K845 and AR-K847 [42]. MDM2 ubiquitinates AR on K311 and results in rapid AR degradation [54]. RNF14 [55, 56] and RNF4 [41, 57] promote AR transcriptional activity, however, it remains unclear whether this involves AR ubiquitination. RNF4 also facilitates AR nuclear import and reduces its nuclear export [58]. E6-AP possesses two independent and distinct functions, as a transcriptional coactivator of AR and as an ubiquitin ligase of degradation of AR protein, which may reflect two different ubiquitination mechanisms by E6-AP [38]. SIAH2 induces K48-linked polyubiquitination of corepressor NCOR1-bound transcriptionally inactive AR for ubiquitin-dependent degradation and facilitates the recruitment of p300-bound transcriptionally active AR to increase the expression of several AR target genes to promote the growth of CRPC [43, 59]. Importantly, CHIP [50, 60] and MDM2 [46, 51] both promote the degradation of AR and AR-V7, and therefore their activation may present a promising therapeutic strategy to target AR-V7 in CRPC.

Considering their AR stimulatory effects and their upregulated expression in CRPC, RNF6 and SIAH2 may represent effective therapeutic targets. In fact, knockdown of RNF6 and SIAH2 inhibited the growth of CRPC xenograft tumors [42, 43]. Also, tumor growth of CRPC cells where the RNF6-targeted AR K845 was mutated was markedly reduced compared to the wildtype AR cells, which indicates that ubiquitination of K845 by RNF6 is critical for ADT-resistant prostate cancer cell growth [42]. Siah2^−/−;TRAMP mice, exhibited increased castration sensitivity which coincided with reduced expression of a specific subset of AR target genes [43]. Also, SIAH2-induced degradation of NCOR1-bound AR may facilitate the subsequent recruitment of p300-bound AR (transcriptionally active) to increase the expression of selective AR target genes, such as the master regulator of lipid metabolism, SREBF1, to promote the growth of CRPC [43].

Deubiquitination of AR

Protein ubiquitination can be reversed by ubiquitin-specific peptidases or deubiquitinases (DUB), which are proteases that hydrolytically cleave ubiquitin or ubiquitin-like proteins from their substrates. Deubiquitination affects the stability, localization and activity of its target proteins [61–66]. Several DUBs have been shown to affect AR transcriptional activity and stability. For example, USP10 stimulates AR transcriptional activity in the presence of its synthetic agonist R1881 [67]. In addition, USP10 induced deubiquitination of mono-ubiquitinated H2A.Z, a variant of histone H2A results in its H2A.Z release from the chromatin of the AR target genes and their subsequent AR-mediated transcriptional activation [68]. USP26, interacts with AR in the nucleus upon DHT treatment and increased the transcription of AR target genes by deubiquitinating poly-ubiquitinated AR in HEK293 and LNCaP cells [69]. USP26 also deubiquitinates and stabilizes Mdm2 [70] thus underlining the complex regulation of AR transcriptional activity. USP12 in the presence of its cofactors Uaf-1 and WDR20 deubiquitinates AR to enhance AR protein stability and transcriptional activity[71] USP12 in complex with Uaf-1 and WDR20 also abrogates phosphorylation of Ser213 of AR by deubiquitinating Akt phosphatases PHLPP and PHLPPL. This results in decreased Akt activity, which further stabilizes AR by reducing the recruitment of the E3 ligase MDM2 to AR [72].

USP7 deubiquitinates full-length AR, but not AR-V7, in prostate cancer cells, which suggests that the LBD domain of AR may be critical for USP7 recruitment. Interaction of USP7 with AR and its deubiquitination is androgen-dependent. A mechanistic model for this androgen-dependence would be that bound androgen induces nuclear translocation of AR where it is bound by USP7, which is predominantly localized in the nucleus. Changes in the conformation of the LBD induced by androgen may further facilitate this interaction. Functionally, USP7 may play a key role for prostate cancer cell proliferation by stabilizing AR and modulating AR binding to chromatin [73]. The USP7 inhibitor P5091 reduces protein levels of both full-length AR and AR-V7, which may be because USP7 deubiquitinates and thereby stabilizes the heterodimer of AR-V7 and AR [74].

USP14, a proteasome-associated DUB negatively regulates proteasome activity by cleaving proteasome-bound ubiquitin chains [75, 76]. USP14 inhibits AR ubiquitination and subsequent degradation in both prostate and breast cancer cells [77, 78]. USP14 knockdown or treatment with USP14 inhibitor IU1 induced G0/G1 cell cycle arrest and suppressed cell proliferation in AR-positive/ER-negative breast cancer cells and androgen-responsive prostate cancer cells.

Autophagic degradation of AR

Although AR is predominantly degraded through the ubiquitin-proteasome system, autophagic degradation of AR is observed in lysosomes of prostate cancer cells. Treatment of LNCaP cells with the autophagy inhibitors 3-methyladenine (3-MA) and NH₄Cl, and knockdown of LC3, a key component of the autophagy machinery inhibited hypoxia-induced AR protein degradation. Hypoxia promotes AMPK induced-Ser403 phosphorylation of p62, a stress inducible intracellular protein that binds polyubiquitinated proteins, which increases p62 and AR interaction and facilitates AR lysosomal degradation [79]. Thus AR degradation is predominantly regulated by the UPS under normoxic conditions and by autophagy under hypoxic conditions.

Regulation of ERG turnover

TMPRSS2-ERG (ETS-related gene) gene fusion is found in ~40–50% of prostate cancers and results in androgen dependent overexpression of the oncogenic transcription factor ERG [80]. ERG promotes cell migration and invasion [81]. Importantly, ERG expression is observed in CRPC after initial repression after ADT in treatment-naïve prostate cancer [82] and high levels of ERG expression may contribute to taxane resistance in CRPC [83].

Full-length ERG protein contains 486 amino acids and includes an ETS DNA-binding domain. At least 17 TMPRSS2-ERG splice variants and two predominant fusion breakpoints resulting in the deletion of the first 39 (ERG∆39) or 99 (ERG∆99) amino acids of full-length ERG [84]. SPOP is a component of E3 ubiquitin ligase complex that mediates ubiquitination and degradation full length ERG through the recognition of a 5-aa degron (⁴²ASSSS⁴⁶) in the NTD of ERG. SPOP interacts with the TMPRSS2-ERG product ERG∆39 but not ERG∆99 because it lacks this degron. However, compare to full length ERG, the interaction between SPOP and ERG∆39 is significantly reduced, which may be due to degron masking as a result of conformational changes. This interaction with SPOP is restored by casein kinase I δ-mediated serine phosphorylation in the degron [85, 86].

SPOP is frequently inactivated by somatic mutations in PCa [87], and these mutations attenuates the degradation of full-length ERG in PCa [85], but TMPRSS2ERG fusion gene products and SPOP degron-deleted ERG is ubiquitinated and degraded with a half-life of ~4–6 hours (compared to ~30 minutes for full-length ERG), indicating that other E3 ubiquitin ligases can mediate ERG ubiquitination. We identified TRIM25, an E3 ligase that regulates the turnover of TMPRSS2-ERG gene products [88]. TRIM25, a RING-finger E3 ubiquitin ligase recognizes the C-terminal domain in ERG protein through its RING domain. TRIM25 polyubiquitinates and targets ERG, ERG∆39 and ERG∆99 for degradation in PCa cells. Interestingly, ERG positively regulates TRIM25 expression by binding to the TRIM25 gene promoter and provides a negative feedback mechanism to maintain physiological ERG protein levels [88].

We also identified the DUB USP9X as a key determinant of ERG turnover in PCa [89]. USP9X binds to the ETS domain in ERG protein; therefore, USP9X deubiquitinates and stabilizes full-length ERG and TMPRSS2-ERG fusion gene products. The pan-selective DUB inhibitor WP1130, which also inhibits USP9X, as well as knockdown of USP9X were found to inhibit the biological effects of ERG in cells, such as migration, invasion and cell proliferation in ERG-positive PCa cells and xenografts. Thus, the depletion of ERG by USP9X inhibition may be a promising approach for targeted therapy of PCa. Recently, peptidomimetic inhibitors that bind to the ETS domain of ERG and target it for degradation were developed [90]. Although the precise mechanism for peptidomimetic induced ERG degradation has not been elucidated, we hypothesize that the inhibitory peptides likely prevent the interaction between ERG and USP9X interaction, and thus cause increased ERG degradation.

Regulation of GR turnover

Glucocorticoid receptor (GR, NR3C1), a member of the steroid hormone receptors family, shares 81% sequence homology with the androgen receptor (NR3C4) [91]. Previous work revealed significant overlap in both the genomic binding sites and the genes regulated by AR and GR in prostate cancer [92]. In ADT-sensitive prostate cancer, GR expression is suppressed by AR. However, following ADT or antiandrogen treatment, increased GR expression is observed, which further enhances the expression of AR-targeted genes required for the growth of CRPC cells [92–95]. Therefore, GR also represents a potential therapeutic target in CRPC.

GR has the same domain structure as AR [96]. However, unlike AR, GR ligands, such as glucocorticoids, do not only activate GR activity by triggering homodimerization and nuclear translocation, but they also repress GR mRNA and protein levels [97, 98]. For unliganded cytoplasmic GR, the chaperone HSP90 is a key interaction partner required for GR protein stability. Inhibition of HSP90 with geldanamycin results in reduced GR protein levels [99]. Degradation of GR by the ubiquitin-proteasome system is evident, because accumulated GR protein and increased polyubiquitination of GR is observed following treatment with proteasome inhibitors [100–102]. Also, there is a protein degradation motif (PEST sequence) present in the NTD of GR, and mutation of lysine 426 to alanine in this motif enhances the transcriptional activity of GR [100].

Four E3 ubiquitin ligases (CHIP, MDM2, E6-AP and FBXW7) interact with GR. CHIP abrogates ligand binding to GR, and promotes its ubiquitination and degradation [101, 103]. However, in CHIP^+/+ and CHIP^−/− mouse embryonic fibroblasts (MEFs) treated with the Hsp90 inhibitor geldanamycin, the degradation rate of GR is essentially identical, which indicates that CHIP is not the main determinant for GR degradation [104]. MDM2 also promotes GR ubiquitination and degradation after treatment with dexamethasone in hepatoma cells [105] or 17β-estradiol in breast cancer [106]. However, the degradation of GR mediated by MDM2 is dependent on the presence of a p53-GR complex and overexpression of only MDM2 in CHIP^−/− MEFs does not affect GR protein levels [104]. E6-AP increases GR transcriptional activity without affecting its turnover [106, 107]. FBXW7, a substrate-recognition factor of SCF (SKP1, CUL1 and F-box protein) complex ubiquitin ligases [108] , ubiquitinates the GR variant GRα and triggers its degradation in T-cell acute lymphoblastic leukemia (T-ALL) cells, which requires glycogen synthase kinase 3β-mediated GR Ser404 phosphorylation [109]. Also, FBXW7 mutations common in primary T-ALL cause both decreased GRα ubiquitination and degradation, and thus result in increased GRα protein levels, which presents the molecular mechanism for the enhanced sensitivity to glucocorticoids in FBXW7-depleted or FBXW7-mutated T-ALL cells [109].

There are no reports of GR protein turnover in prostate cancers. However, due to the emerging role of GR in CRPC, further research may be warranted as GR clearly presents an interesting target for CRPC therapy.

Regulation of CYP17A1 turnover

CYP17A1 (Cytochrome P450 17A1), a cytochrome P450 superfamily of enzymes with both 17α-hydroxylase activity and 17,20-lyase activity, is required for the synthesis of steroids such as progestins, mineralocorticoids, glucocorticoids, estrogens, and androgens [110]. Considering the central role of the androgen-AR signaling axis, anti-androgens and androgen synthesis inhibitors are the two main classes of drugs for ADT. Because of the essential role of CYP17A1 in androgen synthesis, development of specific and potent inhibitors has been at the forefront of drug development. Several potent inhibitors have been developed to treat CRPC, such as abiraterone acetate [6, 7], galeterone [111] and orteronel [112]. Though direct inhibitors that block the enzymatic activity of CYP17A1 are preferred, therapeutic strategies that aim at CYP17A1 degradation may present an effective alternative, especially in the case of abiraterone resistance [113].

CYP17A1 contains 508 amino acids, and although the mechanism of the regulation of CYP17A1 turnover remains unknown, several factors have been identified to affect its stability. Binding of CYP17A1 ligands such as progesterone (a substratetype ligand) or testosterone (a product-type ligand), enhance CYP17A1 stability against proteolysis induced by gonadotropin-induced down-regulation of rat testicular CYP17 [114]. Phosphorylation of CYP17A1 by protein kinase A stimulates its degradation [115]. Furthermore, the ubiquitin-mediated degradation of other members of the cytochrome P450 family such CYP3A4 [116] and CYP2E1 [117] are elucidated and may give insight into how CYP17A1 turnover is regulated.

Regulation of MYC turnove

The Myc family of transcription factors comprises c-Myc (or MYC), L-Myc (MYCL), and N-Myc (MYCN). Amplification of the MYC locus is frequently observed in many cancer types [118, 119], and therefore MYC is considered as a major therapeutic target. In prostate cancer, MYC is commonly amplified [120, 121] and the expression of MYC protein is positively correlated with the tumor proliferation rate [122]. Previous work demonstrated that MYC overexpression can initiate prostate carcinogenesis, mediate genomic instability and drive metastatic prostate cancer in conjunction with loss of Pten and/or p53 in genetically engineered mouse models (GEMMs) [123, 124]. Overexpression of MYCN is observed in approximately 40% of NEPCs (neuroendocrine prostate cancer), but only 5% of prostate adenocarcinomas [125]. MYCN was shown to induce EZH2-mediated transcription to drive NEPC or transform human prostate epithelial cells to prostate adenocarcinoma and NEPC with activated AKT1 [126, 127]. Amplification of MYCL is found in approximately 25% of localized prostate adenocarcinomas [128]. However, an oncogenic function of MYCL in prostate cancer has remained unclear.

Full-length MYC or MYCN contain 439 amino acids or 464 amino acids, respectively and both proteins localize in nucleus as transcription factors. MYC protein contains five domains/functional elements: MB1 (Myc box 1), MB2 (Myc box 2), MB3 (Myc box 3), NLS (nuclear localization signal), and bHLH/LZ (basic helix-loophelix/leucine-zipper). After cleavage by calpain, 298 amino acids N-terminal fragments of both proteins are exported to the cytoplasm from the nucleus to promote cytoplasmic reorganization, differentiation and cell migration by promoting α-tubulin acetylation [129, 130]. The half-life of MYC protein is ~20–30 minutes and this high rate of turnover is regulated by the ubiquitin-proteasome systems [131]. Previous work has identified several ubiquitin E3 ligases to regulate MYC protein stability, including Skp2, FBXW7, CHIP, TRUSS, HectH9, FBXL14 and SPOP.

Skp2 was the first E3 ligase identified to regulate MYC ubiquitination and stability [132, 133]. Skp2 interacts with MB2 and bHLH/LZ domains of the MYC protein. Interestingly, Skp2 can also activate the transcriptional activity of MYC, which indicates that K48 is not only lysine-linkage to MYC by Skp2. FBXW7 as component of the SCF^Fbxw7 E3 ubiquitin ligase complex promotes MYC ubiquitination and turnover [134, 135]. The interaction between FBXW7 and MYC is dependent on phosphorylation of threonine-58 (T58) by glycogen synthase kinase 3 (GSK3) of the MB1 domain in MYC protein. Recently, a small molecule, Oridonin, was reported to trigger FBXW7-mediated MYC degradation by activating Fbxw7 and GSK3 [136], and another compound, Guttiferone K (GUTK) can accelerate c-MYC protein degradation through stabilization of FBXW7 and further curb cell cycle re-entry of quiescent prostate cancer cells [137]. CHIP was shown to reduce MYC stability and transcriptional activity in a T58 mutation independent manner in glioma cells [138]. However, the interaction domain for CHIP in MYC has not been identified. TRUSS (tumor necrosis factor receptor-associated ubiquitous scaffolding and signaling protein), a receptor for the DDB1-CUL4 ubiquitin ligase complex, regulates ubiquitination and degradation of both MYC and MYCN [139]. The interaction of TRUSS with MYC requires both the Nterminal and C-terminal MYC domains. In the prostate cancer cell line LNCaP, reduced expression of TRUSS may drive overexpression of MYC protein [139]. HectH9 (HUWE1) is an E3 ligase for ubiquitination of MYC [140]. Unlike other E3 ligases, HUWE1 only affects the transcriptional activity of MYC by adding K63-linked polyubiquitin chains to MYC. Interestingly, HUWE1 appears regulate MYCN stability by mediating K48-linked polyubiquitination [141]. Therefore, HUWE1 may be a potential therapeutic target in NEPC, where MYCN is frequently overexpressed. FBXL14 is a newly identified E3 ligase that appears to ubiquitinate MYC and regulate its stability in glioblastoma multiforme (GBM) [142]. Recently, SPOP was identified as a substrate adapter of an E3 ligase complex in prostate cancer, which promotes MYC ubiquitination and degradation [143]. This study found that only wild-type SPOP, but not the prostate cancer-associated variants SPOP^F102C or SPOP^F133V, recruit MYC to trigger its degradation.

Importantly, several deubiquitinases were reported to stabilize MYC protein levels. USP28 was the first deubiquitinase identified to stabilize MYC by counteracting the function of FBXW7 [144, 145]. USP36 was shown to stabilize MYC by reversing the effect of FBXW7 [146, 147]. In neuroblastoma, USP7 (HAUSP) can deubiquitinate and thereby stabilize MYCN [64]. Considering the overexpression of MYCN in NEPC, inhibition of HAUSP deubiquitinase activity may be a promising therapeutic strategy for NEPC treatment. Recently, USP13 was reported to antagonize FBXL14-mediated MYC ubiquitination in glioblastoma [142] and USP22 was found to positively regulate MYC stability and tumorigenic activity in breast cancer [148]. Although the interactions between MYC and some E3 ligases and deubiquitinases, such as CHIP, FBXL14, USP13 and USP22, are not identified yet in prostate cancer, the similar molecular mechanism enable them as probable targets for MYC degradation in prostate cancer.

The regulatory complexity of MYC turnover is apparent from the observation of multiple E3 ligases and deubiquitinases regulating its stability in different types of cancers. In CRPC, MYC overexpression may be one key mechanism for resistance to anti-androgen therapy [149]. Therefore, combination of ADT with MYC depletion, which may be achieved by targeting its turnover, could be an effective therapeutic strategy in MYC-positive CRPC.

Regulation of the turnover of other oncoproteins

Though the precise molecular mechanisms of progression from early-stage prostate cancer to CRPC are still an area of active research, a large multi-institutional study of CRPC tumors has comprehensively analyzed the landscape of somatic genomic alterations and genomic differences between primary prostate cancer and CRPC [150]. This study identified frequent focal amplifications of AR and putative oncogenes such as CCND1, PIK3CA and PIK3CB, indicating that these genes may play important roles in CRPC and could be therapeutic targets. Cyclin D1 (encoded by CCND1) is an important positive cell-cycle regulator for G1 to S transition and is overexpressed in many types of cancers. The E3 ligase complex SCF^{Fbx4/aB-crystallin} and FBXW8 complexes have been reported to ubiquitinate and mediate cyclin D1 degradation via the UPS [151, 152]. The deubiquitinase USP2 stabilizes cyclin D1 by preventing ubiquitination-mediated degradation [153].

The PI3K/Akt/mTOR pathway is also altered in CRPC [150, 154]. Inhibitors of this pathway alone or in combination with AR signaling inhibitors show therapeutic efficacy in preclinical CRPC models [155, 156]. The catalytic subunit PIK3CA of PI3K is polyubiquitinated via K29-linkage for proteasomal degradation by the E3 ligase NEDD4L [157]. Both E3 ligases TRAF6 [158] and Skp2 [159] regulate Akt activity through K63-linked ubiquitination while CYLD promotes deubiquitination of Akt [160].

WNT signaling is another important pathway with a role in CRPC [150, 161]. Importantly, genetic changes of the CTNNB1 gene that result in increased β-catenin activity are found in 12% of CRPC specimens [162]. The E3 ubiquitin ligases Ozz-E3 [163], SCF^β-TrCP [164], EDD [165], Mindbomb 1 (MIB1) [166], c-Cbl [167], RNF20 [168], E6AP [169], Mule [170] and the deubiquitinases USP47 [171] and USP9X [172] were found to affect the ubiquitination and stability of β-catenin. In addition, the upregulated members of noncanonical WNT pathway may also function in antiandrogen resistance in advanced prostate cancer, such as Wnt5a [173], FZD4 [174] and RSPO2 [150]. ZNRF3/RNF43 E3 ubiquitin ligases have been reported to promote Dishevelled (DVL)-dependent ubiquitination and degradation of FZD4 [175] and RSPO2 could directly inhibit the activity of ZNRF3/RNF43 to enhance Wnt signaling [176, 177].

Members of the ETS family such as FLI1, ETV4 and ETV5 that are overexpressed by recurrent gene fusions as well as other proteins such as Heat shock protein 27 (Hsp27), Heat shock protein 90 (Hsp90), poly (ADP-ribose) polymerase, aurora kinase A present potential targets for treatment of CRPC [178]. Though small molecule inhibitors that directly inhibit the enzymatic activity or that target ligandbinding proteins, an in-depth understanding of the mechanism of protein turnover may provide the foundation for the development of alternative therapies that target such oncogenic proteins in CRPC. This is critical in particular for transcription factors that lack an enzymatic activity or ligand-binding domain.

Pharmacological targeting of oncoprotein turnover

The critical role and dysregulation of protein turnover in cancer makes E3 ubiquitin ligases, deubiquitinases and the UPS interesting targets for cancer therapy. Currently, several drugs targeting the UPS have been approved by the FDA for treatment of multiple myeloma such as thalidomide (Immunoprin) [179], an inhibitor for the E3 ligase complex factor cereblon (CRBN) and bortezomib (Velcade) [180], a proteasome inhibitor, which highlights the potential for drugs targeting the UPS. Multiple small molecule inhibitors targeting UPS have shown therapeutic potentials in preclinical cancer models, e.g. nutlins, a group of cis-imidazoline analogs, are MDM2 antagonists and activate the p53 pathway by blocking MDM2-induced p53 degradation [181]. The MDM2 antagonist Nutlin-3 was also shown to reduce AR expression and inhibit cell proliferation in LNCaP cells [182]. This is likely caused by Nutlin-3 mediated p53 activation and subsequent p53-mediated repression of AR expression [183], which counteracts the stabilization of AR protein by inhibiting MDM2. P5091 [74] , a USP7 inhibitor and IU1 [77], a USP14 inhibitor have demonstrated antitumorigenic effects in prostate cancer by reducing AR protein levels. WP1130 (Degrasyn) is a pan-selective deubiquitinase inhibitor for USP5, USP9X, USP14, UCHL1, and UCH37. Our previous work demonstrated that WP1130 mediates ERG degradation and inhibited growth of ERG-positive tumors by inhibiting USP9X in prostate cancer [89]. Guttiferone K (GUTK), a bioactive polycyclic polyprenylated acylphloroglucinol, was shown to stabilize FBXW7 to promote MYC protein degradation in prostate cancer cells [137].

The development of specific agents for E3 ligases and deubiquitinases is still challenging. The combination of structural biology, high throughput screening, and biochemical analyses will advance the understanding of detailed molecular mechanism of the interaction of E3 ligases, deubiquitinases with their substrates and the development of novel small molecule inhibitors.

Figure 2.

Model of the regulation of AR turnover by UPS in prostate cancer.

Figure 3.

Model of regulation of ERG turnover by UPS in prostate cancer.

Figure 4.

Model of regulation of MYC turnover by UPS in prostate cancer.

Highlights:

1. There are few effective targeted therapies for castration-resistant prostate cancer.

2. Targeting protein turnover of oncogenic proteins may provide a new avenue for therapeutic development.

3. There are actionable ubiquitin ligases and deubiquitinases that control the turnover of oncoproteins in CRPC.