|Home | About | Journals | Submit | Contact Us | Français|
Ets related gene (ERG) is a transcription factor that is overexpressed in 40% of prostate tumors due to a gene fusion between ERG and TMPRSS2. Because ERG functions as a driver of prostate carcinogenesis, understanding the mechanisms that influence its turnover may provide new molecular handles to target the protein. Previously, we found that ERG undergoes ubiquitination and then is deubiquitinated by USP9X in prostate cancer cells to prevent its proteasomal degradation. Here, we identify Tripartite motif-containing protein 25 (TRIM25) as the E3 ubiquitin ligase that ubiquitinates the protein prior to its degradation. TRIM25 binds full-length ERG, and it also binds the N-terminally truncated variants of ERG that are expressed in tumors with TMPRSS2-ERG fusions. We demonstrate that TRIM25 polyubiquitinates ERG in vitro and that inactivation of TRIM25 resulted in reduced polyubiquitination and stabilization of ERG. TRIM25 mRNA and protein expression was increased in ERG rearrangement-positive prostate cancer specimens, and we provide evidence that ERG upregulates TRIM25 expression. Thus, overexpression of ERG in prostate cancer may cause an increase in TRIM25 activity, which is mitigated by the expression of the deubiquitinase USP9X, which is required to stabilize ERG.
Prostate cancer is the most common malignancy in men and the second leading cause of cancer-related death among men in the United States . Fusions of E-twenty-six (Ets) transcription factor genes with androgen-responsive genes , are present in 50-70% of prostate-specific antigen (PSA) screened prostate cancers from populations of predominantly European descent. The most frequently rearranged member of this family is the Ets Related Gene (ERG), which is overexpressed through gene fusion with the 5′ untranslated region of the gene encoding Transmembrane protease, serine 2 (TMPRSS2) in ~40% of prostate tumors [2, 3]. Depletion of ERG by RNAi decreases proliferation and/or invasiveness in prostate cancer cell lines [4, 5], and ectopic expression of ERG in aged mice  or in concert with loss of tumor suppressors in younger mice promotes prostate carcinogenesis [7-9]. Moreover, ERG controls a gene regulatory network related to the development of prostate cancer, and its progression to metastatic disease [10, 11].
The potential role of ERG as a cancer driver and the high incidence of the TMPRSS2-ERG gene fusion in prostate cancer have catapulted this protein into the forefront of new targets for therapeutic intervention. Like most transcription factors, ERG is considered an unconventional drug target due to the absence of an enzymatic activity or ligand-binding domain. Several strategies have been proposed for therapeutic targeting of ERG , such as by inhibiting its synthesis at the transcriptional or post-transcriptional level, or by blocking the interaction of ERG with its genomic targets, however, these strategies have not yielded yet pharmacological agents to test clinically in patients. We recently found that the deubiquitinase USP9X interacts with ERG in VCaP cells , which harbors the TMPRSS2-ERG gene fusion. Using this cell line, as well as other model systems in tissue culture and in vivo, we demonstrated that RNAi knockdown or inhibition of USP9X with the deubiquitinase inhibitor WP1130 efficiently ablates ERG, thus providing a new pharmacological avenue for targeting ERG by accelerating the turnover of the protein.
Our previous findings suggest that ERG stability in prostate cancer is modulated by the interplay of ubiquitination and deubiquitination. Importantly, two recent studies reported that full-length ERG protein is ubiquitinated in a Speckle-Type POZ Protein (SPOP)-dependent manner. Thus, low levels of ERG are maintained unless there are inactivating mutations in the SPOP E3 ubiquitin ligase adaptor protein [14, 15]. According to these studies, loss of the N-terminus of ERG when expressed from the TMPRSS2-ERG fusion gene results in increased stability of the truncated ERG protein due to loss of SPOP-dependent ubiquitination. Since we had previously demonstrated that N-terminally truncated ERG expressed from the TMPRSS2-ERG fusion gene is ubiquitinated and degraded (albeit truncated ERG appears to have a longer half-life than full-length ERG as reported in the two recent studies above), it is likely that there are SPOP-independent ubiquitin ligases that target ERG in fusion-positive prostate tumors. In the present study, we identified TRIM25 as an ubiquitin ligase that targets N-terminally truncated ERG in fusion-positive VCaP cells, and characterized its effects on ERG stability and its expression in prostate cancer.
Previously, we identified ERG interaction partners by performing co-immunoprecipitation studies of endogenous ERG in VCaP cells , but no ubiquitin ligases were identified that reduced ERG stability. Possible factors contributing to our inability to identify enzymes with the targeted activity include the following: 1) the interaction of such proteins with ERG may be transient in nature;2), the binding site of the ubiquitin ligase may be masked by the antibody used for immunoprecipitation; 3) the protein may be expressed at too low a level; 4) the immunoprecipitation may be too inefficient to detect the endogenous protein. To circumvent these issues we have now purified epitope-tagged recombinant ERG (ERG-V5) from HEK293T cells. The purified protein was used as bait for pulldown of ERG binding partners from whole cell extracts of VCaP cells. After elution of ERG-V5 and associated proteins, and separation by PAGE (Figure (Figure1A),1A), we identified putative ERG interacting proteins by mass spectrometry. Among these putative ERG-binding proteins were 11 ubiquitin ligases (Figure (Figure1A,1A, Supplementary Table 1). To determine which of these proteins were interacting nonspecifically and which were associated with regulation of ERG stability, we systematically knocked down each ubiquitin ligase transcript in VCaP cells using endoribonuclease-prepared siRNA (esiRNA) [16-18] and assessed the level of ERG protein levels. Only the knockdown of Tripartite motif-containing protein 25 (TRIM25) resulted in an increase in ERG protein levels (Figure (Figure1B)1B) as expected from a ubiquitin ligase that targets ERG for degradation.
We then examined the effect of inactivating TRIM25 using esiRNA, which efficiently reduced TRIM25 protein levels in VCaP cells (Figure (Figure1C).1C). We validated the specificity of the TRIM25 knockdown by testing two siRNAs that target independent regions of the TRIM25 transcript. Treatment of cells with both siRNAs resulted in efficient knockdown of TRIM25 and increase in levels of ERG (Figure (Figure1D).1D). Furthermore, we demonstrate that knockdown of TRIM25 with the same siRNAs causes increased protein levels of ERG in 22Rv1 prostate cancer cells (Supplementary Figure 1A) that express ERG-V5 from a stably integrated expression vector .
TRIM25 functions as an ubiquitin E3 ligase, and has been implicated in the regulation of innate immunity by mediating ubiquitination of RIG-I (Retinoic-acid-inducible gene-I) , a protein involved in interferon synthesis in response to viral infection. TRIM25 has also been shown to be regulated by estrogen receptor alpha in breast cancer and to mediate ubiquitination and degradation of two transcription factors: KLF5  and ZFHX3 . TRIM25 has not been implicated in prostate cancer biology, and no gain- or loss-of-function mutations for this gene have been reported for prostate cancer. We found that TRIM25 knockdown significantly increased migration of VCaP cells (Supplementary Figure 1B), which is inhibited by ERG knockdown . Based on the identification of TRIM25 as an ERG-binding protein and on the observation that TRIM25 knockdown causes an increase in ERG protein levels in VCaP cells, we hypothesized that TRIM25 functions as an ERG ubiquitin ligase, which initiates a pathway leading to ERG degradation in fusion-positive prostate cancer.
Next we validated the physical interaction between ERG and TRIM25 by co-immunoprecipitating TRIM25 with endogenous ERG in VCaP cells (Supplementary Figure 1C). TRIM25 was also pulled down with GST-ERG from whole VCaP cell extract (Supplementary Figure 1D). Epitope-tagged ERG (ERG-HA) also co-immunoprecipitated with TRIM25-Flag (Supplementary Figure 1E).
The common TMPRSS2-ERG fusion proteins that lack 39 (ΔN39) or 99 (ΔN99) amino acids from the N-terminus of ERG were interacting with TRIM25, as we found that both truncated proteins co-immunoprecipitated with TRIM25 (Figure (Figure2A).2A). To identify the domains of ERG and TRIM25 that are required for their interaction, we performed co-immunoprecipitation experiments with deletion mutants expressed in HEK293T cells (Figure (Figure2B).2B). TRIM25 contains an N-terminal Really Interesting New Gene (RING) finger domain, a coiled-coil domain (CCD) and a C-terminal SPla and the RYanodine Receptor (SPRY) domain. The RING domain is essential for simultaneously recruiting ubiquitin-conjugating enzymes and substrates , the CCD domain may be essential for multimerization [23, 24], and the SPRY domain is required for substrate recruitment .
Flag-tagged TRIM25 deletion mutants lacking the RING domain did not co-immunoprecipitate HA-tagged ERG (Figure (Figure2C),2C), suggesting that this domain is critical for ERG recruitment. Deletion of the SPRY domain also reduced the amount of immunoprecipitated ERG. We validated these findings with a pulldown experiment, for which we expressed recombinant GST-tagged TRIM25 variants in E. coli for pulldown of ERG from whole VCaP cell extract (Figure (Figure2D).2D). We also performed co-immunoprecipitation experiments with V5-tagged ERG deletion mutants to identify ERG domains that are required for the interaction with TRIM25. Deletion of the C-terminal activation domain (CAD) of ERG (Figure (Figure2E)2E) nearly abolished co-immunoprecipitation of TRIM25, suggesting that this domain is essential for TRIM25 binding. Finally, we demonstrated in GST pulldown experiments that both the CAD domain of ERG and the RING domain of TRIM25 are sufficient to pull down TRIM25 (Figure (Figure2F)2F) or ERG (Figure (Figure2G)2G) from whole VCaP cell extract. These findings suggest that the RING domain of TRIM25 and the CAD domain of ERG are the most critical domains for mediating the physical interaction between these proteins.
The physical interaction of ERG with TRIM25 and the increased ERG protein levels after TRIM25 depletion suggest that TRIM25 may be an ERG-specific ubiquitin ligase and a major determinant of the rate of turnover of ERG. To test this hypothesis, we first analyzed the effect of TRIM25 inactivation on the stability of ERG in VCaP cells. We performed a time-course experiment examining the effect of knockdown of TRIM25 on ERG protein levels in cycloheximide-treated cells. Cells were transfected with siRNA against TRIM25 or a non-targeting control siRNA 72 hours before initiating cycloheximide treatment. TRIM25 depletion markedly increased the stability of ERG (Figure 3A, 3B).
Next, we tested the role of TRIM25 as a putative ERG-specific ubiquitin ligase using a cell-based ubiquitination assay. Because of low plasmid transfection efficiency and slow growth of VCaP cells, we analyzed the impact of TRIM25 loss on ubiquitination of ERG in HeLa cells (Figure (Figure3C).3C). We knocked out TRIM25 expression using CRISPR-Cas9 mediated gene editing (Figure (Figure3D)3D) . Recombinant HA-ubiquitin and ERG-V5 were expressed in two independent HeLa cell knockout clones, as well as in parental HeLa cells and ERG-V5 was immunoprecipitated from cell lysates. The levels of ubiquitinated ERG-V5 was analyzed by immunoblotting using an HA antibody. Levels of polyubiquitinated ERG-V5 were markedly decreased in TRIM25 knockout cells. We verified the specificity of the ubiquitination assay by comparing V5 immunoprecipitates from HeLa cells expressing HA-ubiquitin only, or ERG-V5 in the presence or absence of HA-ubiquitin. In both cases, no ubiquitination signal was present in the HA immunoblot (Supplementary Figure 2A). Also, the V5 antibody markedly enriched ubiquitinated ERG-V5 as compared to IgG (Supplementary Figure 2B).
To further corroborate the role of TRIM25 as an ERG ubiquitin ligase we tested if TRIM25 overexpression increased ERG ubiquitination. For that purpose we expressed TRIM25-Flag or an empty vector together with full-length ERG-V5 and HA-ubiquitin in HEK293T cells. TRIM25-Flag expression markedly increased the ubiquitination of ERG-V5 (Figure (Figure3E).3E). We also examined the effect of TRIM25 overexpression on the ubiquitination of ERG-ΔN39-V5 and ERG-ΔN99-V5. We found that TRIM25-Flag overexpression markedly increased polyubiquitination of both N-terminally truncated variants of ERG, suggesting that TRIM25 mediates polyubiquitination of full-length, as well as truncated, ERG variants.
The decreased levels of polyubiquitinated ERG upon TRIM25 knockout and the increase in ERG ubiquitination upon overexpression of TRIM25 suggest that TRIM25 is an E3 ubiquitin ligase that targets ERG. To directly demonstrate this activity for TRIM25, we tested if ERG is ubiquitinated by TRIM25 in vitro. For this purpose, we incubated E1 and E2 conjugating enzymes, 6xHis-Ubiquitin and full length ERG-V5 (immunoprecipitated from HEK293T) with purified GST-tagged (expressed in E. coli) wildtype TRIM25 (GST-TRIM25) or a TRIM25 variant that lacked the RING domain. When incubated with GST-TRIM25 we observed ERG polyubiquitination, while no polyubiquitinated ERG-V5 was observed in the presence of the RING-domain deletion mutant of TRIM25 (Figure (Figure3F).3F). TRIM25 also polyubiquitinated ERG-ΔN39-V5 and ERG-ΔN99-V5. Collectively, these results establish that TRIM25 is an ubiquitin ligase of full-length ERG as well as the N-terminally truncated ERG variants that originate from the TMPRSS2-ERG gene fusion in prostate tumors.
Since TRIM25 was shown to mediate both K48-  and K63-linked  ubiquitination we sought to further characterize the mechanism of ERG ubiquitin chain linkage mediated by TRIM25. For this purpose, we used 6xHis-Ubiquitin K48 and 6xHis-Ubiquitin K63 (i.e., ubiquitin mutants where all lysine residues except K48 or K63 are substituted for arginine). We found an increase in polyubiquitinated ERG for both 6xHis-Ubiquitin K48 and 6xHis-Ubiquitin K63 (Supplementary Figure 2C). Since K48 chain linking generally targets proteins for proteasomal degradation, these results are consistent with the premise that TRIM25 promotes proteasomal degradation of ERG. In this experiment, we also found that the C-terminal activation domain of ERG was critical for ubiquitination in vitro.
We examined the expression of TRIM25 protein by immunohistochemical analysis (Figure (Figure4A)4A) in 209 prostate cancer tumor samples that were placed on a tissue microarray generated at the University of Michigan. Significantly higher levels of TRIM25 were present in the tumors that expressed ERG (p = 3.89e-07) than in the ERG-negative tumors (Figure (Figure4B).4B). This observation may appear to be counterintuitive because high TRIM25 protein expression would be expected to result in lower ERG protein levels. However, we note that the deubiquitinase USP9X, which removes ubiquitin from ERG, is also expressed at higher levels in ERG-positive tumors and correlates with the levels of ERG protein in these tumors . Furthermore, we demonstrate here that catalytically active USP9X (but not catalytically inactive USP9X in which serine was substituted for cysteine at position 1566 ) reverts TRIM25-mediated ubiquitination of ERG in vitro (Supplementary Figure 2D). Thus, the expression of TRIM25 in prostate cancer cells may not result in reduced ERG protein levels when USP9X is also expressed.
Surprisingly, we found a positive correlation between the transcript levels of TRIM25 and ERG in ERG-positive prostate tumors (Figure (Figure4C)4C) in the RNA-seq data from The Cancer Genome Atlas (TCGA) consortium. This positive correlation between ERG and TRIM25 mRNA expression suggests that the transcription factor ERG may regulate TRIM25 mRNA expression. Several lines of evidence support this hypothesis: First, we found high ERG occupancy in the vicinity of the transcription start site of TRIM25 when we analyzed ChIP-seq data generated for ERG in VCaP cells (Figure (Figure4D).4D). We validated the enrichment of ERG at the TRIM25 gene promoter in VCaP cells with a chromatin immunoprecipitation and quantitative PCR assay, which demontrated 5.3-fold greater enrichment for ERG occupancy (Figure (Figure4E).4E). Thus, ERG appears to be bound to the TRIM25 gene promoter. We found that ERG knockdown with two independent siRNAs resulted in 42% or 58% reduction of TRIM25 expression in VCaP cells (Figure (Figure4F).4F). And, when we analyzed the impact of ectopic ERG expression in 22Rv1 cells that do not harbor any Ets gene fusions we found that ERG-V5 expression increased TRIM25 expression by 53% (Figure (Figure4G).4G). These results suggest that ERG may positively regulate TRIM25 expression, which could present a negative feedback mechanism to maintain physiological ERG protein levels. Our previous discovery of USP9X as an ERG-stabilizing deubiquitinase  suggests that reduction of ERG protein levels by TRIM25-mediated proteasomal degradation is prevented by expression of USP9X in fusion-positive prostate cancer cells.
We have identified TRIM25 as an ERG-binding ubiquitin ligase in prostate cancer cells. We demonstrated that TRIM25 knockdown results in increased ERG stability in TMPRSS2-ERG expressing prostate cancer cells. Using several biochemical assays we show that TRIM25 mediates the polyubiquitination of full-length ERG as well as N-terminally truncated ERG. In conclusion, our work provides novel insights in the regulation of ERG protein stability in prostate cancer.
ERG-V5 was ectopically expressed in HEK293T cells, immunoprecipitated with a V5 antibody, and used as bait for pulldown of ERG-binding proteins from VCaP cell lysates. Eluted proteins were separated by SDS-PAGE. For Mass spectrometry, PAGE gel slices were digested with trypsin, and high performance liquid chromatography tandem mass spectrometry (HPLC/MS/MS) analysis of tryptic peptides was performed with a Thermo Fusion Lumos mass spectrometer (Thermo) coupled to an Ultimate 3000 RSLC-Nano liquid chromatography systems (Dionex).
Full-length TRIM25-FLAG, ERG-V5, ERG-HA, deletion variants of ERG-V5 and TRIM25-Flag, and/or HA-ubiquitin expression plasmids were transfected into HEK293T or HeLa cells, and immunoprecipitation was performed with V5, FLAG or HA antibodies.
GST-TRIM25 and GST-TRIM25-ΔRING were expressed in E. coli BL21(DE3), and purified with glutathione sepharose beads and glutathione elution. For GST pulldown, whole VCaP cell extracts were incubated with GST or GST-TRIM25 variants, and eluted proteins were detected by immunoblot analysis.
VCaP cells transfected with siRNA against TRIM25 or a non-targeting control were treated with cycloheximide, or DMSO. Immunoblot analysis of TRIM25, ERG, and GAPDH expression was performed for several time points over 8 hours.
We used the GeneArt® CRISPR Nuclease (OFP Reporter) Vector Kit (Thermo Fisher Scientific) according to the manufacturer's instructions to generate TRIM25 CRISPR nuclease constructs, which were transfected into HeLa cells with Effectene transfection reagent (Qiagen). TRIM25-knockout HeLa cells clones were identified by immunoblot analysis.
TRIM25-knockout HeLa cells grown in DMEM with 10% FBS were co-transfected with ERG-V5 and HA-ubiquitin expression plasmids. Immunoprecipitation with a V5 antibody was performed as described above. For Western blot analysis antibodies against TRIM25, HA, V5, and GAPDH were used.
ERG-V5, ERG-ΔN39-V5 and ERG-ΔN99-V5, was overexpressed in HEK293T cells and immunoprecipitated. These purified proteins were incubated with recombinant GST-TRIM25 variants purified from E. coli, human recombinant UBE1, UbcH5a and 6xHis-Ubiquitin (all three proteins were purchased from Boston Biochem).
Immunohistochemical staining was performed as previously described (Strand, 2012; Ma, 2010) for ERG (1:100, Biocare Medical) and TRIM25 (1:100, Abcam). ERG and TRIM25 expression were evaluated on a previously described tissue microarray containing 209 cases of localized prostate cancer arrayed in triplicate . All patients underwent radical prostatectomy at the University of Michigan Health System as primary monotherapy without neoadjuvant hormonal or radiation therapy. This radical prostatectomy series is part of the University of Michigan Prostate Cancer Specialized Program of Research Excellence Tissue Core. Tumor cells with ERG and/or TRIM25 staining were scored manually per tissue core by a reviewer who was blinded to the clinical data.
We are grateful to the McDermott sequencing and mass spectrometry cores at UT Southwestern for technical assistance. We thank Helen Hobbs for critical reading of the manuscript.
CONFLICTS OF INTEREST
The authors have no conflicts to disclose.
This study was supported by a grant of the National Institutes of Health (R01CA200787), the CPRIT Scholar in Cancer Research Award (R1002) of the Cancer Prevention and Research Institute of Texas and the John L. Roach Endowment in Biomedical Research to R.K.
This paper has been accepted based in part on peer-review conducted by another journal and the authors' response and revisions as well as expedited peer-review in Oncotarget.