RNF2/Ring1b negatively regulates p53 expression in selective cancer cell types to promote tumor development

Abstract

Large numbers of studies have focused on the posttranslational regulation of p53 activity. One of the best-known negative regulators for p53 is MDM2, an E3 ubiquitin ligase that promotes p53 degradation through proteasome degradation pathways. Additional E3 ligases have also been reported to negatively regulate p53. However, whether these E3 ligases have distinct/overlapping roles in the regulation of p53 is largely unknown. In this study, we identify RNF2 (ring finger protein 2) as an E3 ligase that targets p53 for degradation. The E3 ligase activity of RNF2 requires Bmi1 protein, a component of the polycomb group (PcG) complex. The up-regulation of p53 does not affect RNF2 expression. Unlike Mdm2, RNF2 only degrades p53 in selective cell lines, such as those from germ-cell tumors. The knockdown of RNF2 induces apoptosis, which can be rescued through the reduction of p53 expression. Moreover, the down-regulation of RNF2 expression in germ-cell tumors significantly reduces tumor cell growth, while the simultaneous down-regulation of both genes restores tumor cell growth in vitro and in tumor xenograft models. Furthermore, a reverse correlation between RNF2 and p53 expression was detected in human ovarian cancer tissues. Together, these results indicate that RNF2 is an E3 ligase for p53 degradation in selective cells, implicating RNF2 as a therapeutic target to restore tumor suppression through p53 in certain tumor cells.

Under deleterious conditions or during DNA damage, p53 activity will be up-regulated to inhibit cell cycling or to promote apoptosis, two major events associated with the tumor suppressor function of p53 (1, 2). These two activities ensure that an individual organism will either initiate the repair of damaged cells or eliminate them. During normal cell growth, p53 is maintained at extremely low levels through various negative regulatory mechanisms. One of these mechanisms occurs through the 26S proteasome in which a number of ubiquitin-protein ligases are involved. Poly- and monoubiquitin can be covalently conjugated to p53 through the lysine amino acid residues (K48 versus K63) of ubiquitin (3, 4). The principal E3 ligase for p53 is Mdm2, which itself is a product of p53-inducible genes. The importance of the p53–Mdm2 autoregulatory loop was clearly demonstrated in a study in which the embryonic lethality induced through an Mdm2 deletion was rescued through the simultaneous knocking-out of the p53 gene (5, 6). Additional studies showed that Mdm2 also affects the transcriptional potential and subcellular distribution of p53 (7–11), highlighting the complex nature of p53 regulation. Furthermore, subsequent investigations uncovered ∼20 additional E3 ligases that regulate p53 activity, including COP1, Pirh2, ARF-BP1, CHIP, TOPORS, Synoviolin, CARP1, CARP2, and Trim24, which negatively regulate p53 stability through degradation, and ICP0, CUL7, MSL2, WWP1, Ubc13, and E4F1, which affect p53 subcellular localization, transcription activity, tetramerization, and so forth (3, 12, 13). These findings suggest that both Mdm2-dependent and Mdm2-independent mechanisms are involved in regulating p53 activity. However, whether these ligases act redundantly and regulate p53 in specific cell types or function under particular contexts remains unclear. Moreover, although p53 orthologs and the structural conservation of p53 exist across invertebrates, the Mdm2 gene might have been lost during evolution in some invertebrates, such as Caenorhabditis elegans and Drosophila melanogaster, suggesting that other factors (including other E3 ligases) might be involved in controlling p53 activities in these animals. For example, Bonus is a homolog of TRIM 24, a p53 ligase in vertebrates. The results from loss-of-function studies indicated that Bonus is critical for maintaining p53 activity in Drosophila, suggesting that Bonus might be an evolutionarily conserved E3 ligase for p53 (14). The existence of multiple E3 ligases for p53 strongly suggests that specialization among them must be required for controlling p53 at multiple levels in different regulatory programs.

In addition to genomic alterations, epigenetic changes have been increasingly associated with tumor development. Cross-talk between genetic and epigenetic regulations has also been documented. The polycomb group (PcG) proteins have long been considered as one of the epigenetic regulators for gene silencing during embryonic development and tissue homeostasis in adult life (15, 16). Polycomb proteins exist in at least two multimeric nuclear protein complexes: polycomb repressor complex 1 (PRC1) and PRC2 (17). The core human PRC1 consists of Polycomb (PC), polyhomeotic (PH), Bmi1, Ring1a and Ring1b (Rnf2) (18), while PRC2 contains EED, EZH2, and SUZ12 (19). The gene regulation through PC complexes involves the trimethylation of histone H3 at lysine 27 catalyzed through the activity of PRC2 (20), which in turn recruits the PRC1 complex to chromatin to mediate histone H2A monoubiquitination (21). As the primary E3 ligase responsible for H2A modification in the PRC1 complex, Rnf2 plays important roles in both early development and ES cell maintenance (22). In addition to its monoubiquitination activity, the PRC1 complex also has poly-ubiquitination activity. For example, PRC1 poly-ubiquitinates Geminin, a DNA replication inhibitor, to sustain adult hematopoietic stem cell activity (23). Protein structure analysis revealed that within the PRC1 complex, Bmi1 and RNF2 heterodimerize via their N-terminal RING domains to form an active E3 ubiquitin ligase (24–26). The overexpression of Rnf2 has also been documented in clinical gastrointestinal tumors and lymphomas (27). Despite these findings, the mechanisms of Rnf2 in tumor formation remain poorly understood.

Here, we report that p53 is an endogenous binding partner for Rnf2 and that Rnf2 functions as an E3 ubiquitin ligase for p53. Rnf2 ubiquitinates and promotes p53 degradation both in vitro and in vivo. Moreover, the ubiquitination activity of Rnf2 requires the presence of Bmi1. In contrast to reported E3 ligases for p53, Rnf2 only regulates p53 levels in selected tumor cell types. Furthermore, in a xenograft mouse model, the reduction of Rnf2 significantly reduced tumor cell growth, while the simultaneous depletion of p53 restored tumor growth. Our study not only reveals a p53 E3 ligase but also provides a potential therapeutic target in therapies for tumors in which the p53 level is specifically regulated through Rnf2.

Results

Identification of Rnf2 as a Negative Regulator of p53 in Selective Cells.

Previous studies have indicated that Rnf2 expression is increased in a large variety of tumors (27). As a first step to characterize the molecular mechanisms of Rnf2 in tumor formation, we attempted to identify the binding partners of this protein. In a yeast two-hybrid screen using Rnf2 as the bait, we identified p53 as a major binding partner. We confirmed the interaction of these proteins in Tera-1 cells, a human testicular germ-cell-tumor-derived cell line, by coimmunoprecipitation (Fig. 1 A and B).

Fig. 1.

Rnf2 negatively regulates p53 stability. (A and B) Rnf2 interacts with p53 in vivo. Tera-1 cell lysates were subject to immunoprecipitation with anti-IgG, anti-p53 (A), or anti-Rnf2 (B) antibodies. The immunoprecipitates were subsequently blotted with the indicated antibodies. (C) Rnf2 interacts with p53 in vitro. p53 was purified from bacteria and incubated with GST or GST–Rnf2 coupled to GSH–Sepharose. The proteins retained on the Sepharose beads were subsequently blotted with the indicated antibodies. (D and E) The shRNA-mediated depletion of Rnf2 increases p53 protein levels and induces p53 target gene activation. Tera-1 cells were transfected with the indicated shRNAs. After 72 h, the proteins and mRNA were extracted and subjected to Western blot or quantitative RT-PCR. The error bars represent the SEM of triplicate experiments. ***P < 0.001 two-tailed Student t test. (F) Overexpression of Rnf2 suppresses p53 protein level. HCT116 cells transfected with the indicated constructs were left untreated or treated with MG132. The proteins were extracted and subjected to Western blot analysis. (G and H) Rnf2 depletion increases p53 stability. Tera-1 cells transfected with the indicated shRNA constructs were treated with CHX (0.1 mg/mL) and harvested at the indicated time. The cell lysate was subjected to Western blot analysis. (G) Immunoblots of p53 and Rnf2. (H) Quantification of the p53 protein levels relative to beta-actin.

To determine whether the interaction between Rnf2 and p53 is direct, we generated and purified recombinant Rnf2 and p53 and tested their direct binding in vitro. Purified His-p53 bound to GST–Rnf2 in a cell-free system, indicating a direct interaction between Rnf2 and p53 (Fig. 1C). Next, we used an immunoprecipitation approach to map the regions of p53 required for Rnf2 binding through combining various fragments of p53 with full-length Rnf2 in an in vitro translated system. The results revealed that the DNA binding domain of p53 (AA94–AA292) contributes to the Rnf2–p53 interaction (Fig. S1A). Together, these results indicate that Rnf2 is a unique p53 binding partner and suggest that Rnf2 might regulate p53 activity in vivo. Rnf2 functions as an E3 ubiquitin ligase to monoubiquitinate substrates, such as H2A (21). More recently, Rnf2 was also shown to poly-ubiquitinate some of its substrates. For example, Rnf2 mediates PGC-1α degradation in cells under metabolic stress to p53-mediated starvation response in vivo (28). To investigate whether the binding of Rnf2 regulates p53 activity, we used Rnf2-specific short hairpin RNAs (shRNAs) to reduce the Rnf2 protein level in Tera-1 cells. The knock-down of Rnf2 using either of two independent shRNAs resulted in marked increases in the levels of endogenous p53 protein, without affecting the expression of p53 messenger RNA (mRNA) (Fig. 1 D and E and Fig. S1B). A number of genes, including p21, Mdm2, Bax, and Puma, were positively regulated at the transcriptional level through p53 (29). Consistent with this result, the mRNA levels of the target genes were greatly increased in Rnf2 knock-down cells (Fig. 1E). The p53-mediated up-regulation of the target genes was further confirmed through the reversion of the increased gene transcriptions upon the depletion of p53 in Rnf2 knock-down cells (Fig. 1E).

To further determine the role of Rnf2 in regulating p53 levels, we overexpressed Rnf2 in Tera-1 cells and the colon cancer cell line HCT116, which has greatly decreased endogenous p53 protein levels. As a negative control, the overexpression of an Rnf2 mutant, lacking the Ring domain, did not affect the level of p53 (Fig. 1F and Fig. S1C). Moreover, the reduction of the p53 expression could be blocked using the proteasome inhibitor MG132, suggesting that Rnf2 regulates p53 levels through proteasome-dependent degradation. We also examined the degradation kinetics of p53 in the presence of the protein translation inhibitor cycloheximide (CHX) in Tera-1 cells. The rate of p53 protein degradation was clearly reduced in Rnf2 knockdown cells (Fig. 1 G and H). To extend this observation of the negative Rnf2-mediated regulation of p53, we repeated the Rnf2 RNAi experiments in additional cell lines. Although the p53 expression increased when the Rnf2 expression was down-regulated in Tera-1, Tera-2, MCF-7 cell lines, and mouse embryonic fibroblasts (MEFs), the p53 expression was not affected in transfected HEK 293 cells, HeLa cells, or the human ovary malignant teratoma cell line PA-1 (Fig. 2 A and B and Fig. S2A). The knockdown of Rnf2 in Tera-1 cells markedly increased p53 expression in the nucleus, while the overexpression of Rnf2 in Tera-1 and HCT116 cells reduced the nuclear p53 protein levels (Fig. 2 A, C, and D). In contrast, Rnf2 overexpression in HEK 293 cells did not affect p53 levels. As a control, Mdm2 effectively degraded p53 in HEK 293 cells (Fig. 2E). These results demonstrate that, in contrast to the primary p53 E3 ligase Mdm2, Rnf2 only down-regulates p53 in specific cell lines, suggesting that Rnf2 has nonredundant functions in regulating p53-mediated activities in selective tissues.

Fig. 2.

Rnf2 regulates the p53 levels in selective cells. (A and B) Rnf2 knockdown increases p53 protein levels in Tera-1 cells but not in HEK 293 cells. Tera-1 and HEK 293 cells transfected with the indicated constructs were treated with puromycin (1 μg/μL). After 72 h, the cells were harvested and fractionated into cytosolic and nuclear fractions. The fractions were subsequently blotted with the indicated antibodies. Lamin B, a nuclear marker, and tubulin, a cytosolic marker, were used as loading controls. (C) Rnf2 knockdown causes nuclear p53 up-regulation. Tera-1 cells transfected with the indicated constructs were fixed and stained as described. (D and E) Rnf2 overexpression represses nuclear p53 protein levels in specific cell types. HCT116 and HEK 293 cells transfected with the indicated constructs were harvested and fractionated. The cellular fractions were subsequently blotted with the indicated antibodies. LE, long exposure; SE, short exposure.

Rnf2 Is an E3 Ligase That Ubiquitinates p53.

Previous studies revealed that Rnf2 acts as an E3 ubiquitin ligase (21, 23, 28). Various E3 ligases regulate p53 activity through direct degradation, changes in subcellular localization, or the alteration of transcription activity (3, 10, 12). We investigated the possibility of Rnf2 to directly degrade p53 through the proteasome pathway. First, we examined whether Rnf2 can ubiquitinate p53. To this end, we overexpressed Rnf2 in HCT116 cells and detected the endogenous p53 ubiquitination levels in the presence of the proteasome inhibitor MG132. The Rnf2 overexpression indeed markedly induced nuclear p53 poly-ubiquitination (Fig. 3A and Fig. S3 A and B). The Ring domain of Rnf2 and phosphorylation at S41 are indispensable for the activity, as Rnf2ΔRing, Rnf2^I53A (a Ring domain mutation, lacking the ability to interact with E2 ligase) and Rnf2^S41A mutants (30, 31) failed to increase the ubiquitination levels of p53 (Fig. 3 A and D and Fig. S3A). Conversely, the down-regulation of Rnf2 reduced the basal level of p53 ubiquitination (Fig. 3C). In contrast, Rnf2 overexpression in HEK 293 cells did not up-regulate p53 ubiquitination levels (Fig. 3B), which is consistent with the previous observation that Rnf2 failed to reduce the p53 protein levels in these cells.

Fig. 3.

Rnf2 ubiquitinates p53 both in vivo and in vitro. (A and B) Rnf2 promotes p53 ubiquitination in vivo. HCT116 and HEK 293 cells transfected with the indicated constructs were treated with MG132 for 4 h before harvest. p53 was immunoprecipitated with anti-p53 polyclonal antibodies and immunoblotted with monoclonal anti-HA or anti-p53 antibodies. (C) Rnf2 depletion reduces p53 ubiquitination in vivo. Tera-1 cells transfected with the indicated constructs were treated with MG132 for 4 h before harvest. p53 was immunoprecipitated with anti-p53 polyclonal antibodies and immunoblotted with monoclonal anti-HA or anti-p53 antibodies. (D) Phosphorylation-deficient Rnf2 lacks E3 ligase activity. HCT116 cells transfected with the indicated constructs were treated with MG132 for 4 h before harvest. p53 was immunoprecipitated with anti-p53 polyclonal antibodies and immunoblotted with monoclonal anti-HA or anti-p53 antibodies. (E) Rnf2 ubiquitinates p53 in vitro. ³⁵S-labeled p53 was incubated with purified Rnf2 or Rnf2^I53A in the presence of Bmi1 or Bmi1Δ1–44 in vitro. The reactions were subsequently analyzed through autoradiography.

Although Rnf2 ubiquitinates p53, these results did not distinguish the direct or indirect ubiquitination of p53 through Rnf2. To resolve this issue, we investigated the ubiquitination activity of Rnf2 toward p53 using a cell-free system. Bmi1 was previously demonstrated to be essential for the E3 ligase activity of the PRC1 complex (32). We used immuno-purified Rnf2 or Rnf2^I53A translated in vitro and preincubated these samples separately with recombinant Bmi1. Similar to previous reported studies, the purified Rnf2, but not catalytically deficient Rnf2^I53A, promotes p53 ubiquitination in the presence of Bmi1 in vitro (Fig. 3D). The in vitro ubiquitination is less profound compared with that observed in vivo, suggesting that additional factors might be required for efficient ubiquitination and are lacking in the in vitro system. We also observed that the Bmi1 knockdown in HCT116 cells blocked Rnf2-induced p53 ubiquitination (Fig. S3C). These results indicate that Rnf2, in the presence of Bmi1, can directly ubiquitinate p53 in vitro and is required for the ubiquitination in vivo.

Rnf2 Negatively Regulates p53-Mediated Biological Functions in Tera-1 Cancer Cells.

Rnf2 shows increased expression in a high percentage of various tumors compared with normal tissue counterparts (27). Because we identified Rnf2 as a direct negative regulator of p53, we postulated that Rnf2 might act as an onco-protein in specific tissue types to antagonize p53 functions and to promote tumor formation. To test this hypothesis, we first examined whether Rnf2 is essential for tumor cell proliferation in vitro, by knocking down Rnf2 in cultured Tera-1 cells. The cell proliferation results indicated that Rnf2 is indispensable for Tera-1 cell growth (Fig. 4A and Fig. S4). The requirement of Rnf2 for colony formation was also confirmed in a separate tumor cell growth assay (Fig. 4 B and C).

Fig. 4.

Rnf2 regulates p53-mediated cell growth arrest and apoptosis. (A) Rnf2 regulates cell growth. Tera-1 cells transfected with the indicated constructs were trypsinized and plated after puromycin treatment. The cell numbers were subsequently counted at the indicated times. (B and C) Colony-formation assay was performed with Tera-1 cells transfected with control shRNA or Rnf2 shRNA. (B) Tera-1 cells transfected with the indicated constructs were plated at a density of 1×10⁶ cells/plate, and the clone numbers were quantified at 3 wk after puromycin treatment. (C) Quantification of colonies formed in B. The error bars represent the SEM of triplicate experiments. **P < 0.01 two-tailed Student t test. (D) Down-regulation of Rnf2 causes G1 arrest in Tera-1 cells. Tera-1 cells transfected with the indicated constructs were treated with puromycin for 3 d. The cells were harvested, and the cell-cycle profiles were determined through FACS analysis. (E and F) Rnf2 knockdown increases apoptosis in Tera-1 cells. Tera-1 cells were transfected with the indicated constructs. After 3 d of puromycin treatment, the apoptotic cells were visualized and counted. (E) Apoptotic cells were detected through TUNEL staining. (F) Quantification of apoptotic cells detected using FACS analysis. The error bars represent the SEM of triplicate experiments. **P < 0.01; *P < 0.05 two-tailed Student t test.

A number of mechanisms might underlie the roles of Rnf2 in tumor cell proliferation. Because Rnf2 directly targets p53 for degradation, we investigated the requirements of Rnf2 for cell cycle arrest and apoptosis, two major events regulated through p53. Indeed, the down-regulation of Rnf2 caused G1 arrest in Tera-1 cells (28% G0/G1 in Rnf2 shRNA-transfected cells compared with 17% G0/G1 cells in control shRNA-treated cells) (Fig. 4D). More importantly, silencing Rnf2 and p53 together through RNA interference showed a marked reversal in G1 arrest (Fig. 4D). Furthermore, we also examined whether Rnf2 affects p53-dependent apoptosis. Terminal transferase dUTP nick-end labeling (TUNEL) and Annexin V staining were used to detect apoptotic cells. Rnf2 depletion greatly increased apoptosis in Tera-1 cells, while the coexpression of p53 shRNA reversed the number of apoptotic cells to the level of untreated cells (Fig. 4 E and F). Collectively, these results demonstrate that Rnf2 antagonizes p53 functions in Tera-1 cells.

Rnf2 Is Required for Ovarian Cancer Growth.

Our results strongly indicate that Rnf2 plays key roles, through p53, in controlling testicular cancer cell proliferation. To further investigate whether Rnf2 is required for other types of tumors, particularly those derived from the reproductive system, we also explored whether Rnf2 regulates p53 in ovarian cancer cells. As expected, p53 was greatly up-regulated, even when Rnf2 was partially depleted in two ovarian cancer cell lines, A2780 and HO-8910, in which p53 gene is not mutated (Fig. 5A). The depletion of Rnf2 in A2780 suppressed the cancer cell proliferation, while p53 knockdown restored the cell growth rate resulting from the loss of Rnf2 (Fig. S5 B–E). We further evaluated Rnf2 expression levels in clinical ovarian tumor tissues using two different Rnf2 antibodies. Tissue array analysis revealed that Rnf2 is up-regulated in 88% of the ovarian serous cystoadenomas, 90% of ovarian mucinous carcinomas, and 60% of clear cell carcinomas. In contrast, Rnf2 expression was barely detectable in normal ovarian tissues (Fig. 5B and Table S1). Interestingly, we observed a reverse correlation between Rnf2 overexpression and low p53 expression in over 60% (44/72) of ovarian serous cystoadenomas and 90% (9/10) of ovarian mucinous carcinomas (Table S2), suggesting that Rnf2 might negatively control p53 levels in these cancer cells. The down-regulation of p53 expression in these cells is unlikely caused by Mdm2 overexpression, as most of the adjacent sections showed no Mdm2 up-regulation compared with normal tissue (Table S3).

Fig. 5.

Regulation of Rnf2 in ovarian carcinoma cell growth depends on p53. (A) Rnf2 depletion up-regulates p53 protein levels in human ovarian carcinoma cell lines (A2780 and HO-8901 bearing wild-type p53). (B) Immunohistochemical staining of Rnf2 in normal ovarian tissues and ovarian carcinoma. (C–F) Effects of Rnf2 depletion on A2780 cell growth and tumor development. A2780 cells stably expressing the indicated shRNAs (GFP-labeled) were injected into nude mice. After 4 wk of injection, the animals were euthanized and analyzed for tumor growth. (C) Protein expression in A2780 cells stably expressing the indicated shRNAs. (D, Upper) Representative photographs captured with visible light of the animals corresponding to each treatment group at day 20 after tumor cell injection. (Lower) Noninvasive visualization of GFP-tagged tumor cells by whole-body fluorescence imaging showing a significant reduction in tumor size when Rnf2 is depleted. (E) Comparative analysis of the localized tumor growth; bars, SE. **P < 0.01; ***P < 0.001, statistically significant compared with the control shRNA group. (F) Estimated tumor weight for the indicated groups at day 20 after injection. **P < 0.01, statistically significant compared with the control or rescue groups.

We established that Rnf2 is critically required for ovarian cancer cell proliferation in vitro and that Rnf2 overexpression is correlated with tumor formation (Fig. 5B and Fig. S5 A and B). However, direct evidence that Rnf2 is required for tumor cell survival in vivo has not yet been provided. Thus, we next investigated the in vivo activity of Rnf2 on the growth of A2780 tumors in a xenograft model. The suppression of Rnf2 activity through RNA interference largely inhibited the growth of xenograft tumors (Fig. 5 D–F). The reduction of tumor volumes/masses was dependent on p53 function, as knocking down both Rnf2 and p53 expression reversed tumor growth to the level of the control group (Fig. 5 D–F). These results strongly indicate that Rnf2 is a p53-dependent key player in ovarian tumor progression.

Discussion

This study showed that (i) Rnf2, a binding partner for p53, ubiquitinates p53 for degradation; (ii) the negative Rnf2-mediated regulation of p53 affects cell cycle progression and apoptosis controlled by p53; (iii) Rnf2 targets nuclear p53 for degradation in specific cell types, such as testicular cancer cells and ovarian tumors; and (iv) Rnf2 is overexpressed in most clinical ovarian tumors. Thus, this study illustrates a unique activity for the PcG complex in cancer development.

The PcG genes have been implicated in embryonic development and stem cell self-renewal. As a core member of the PRC1 complex, Rnf2 is up-regulated in various tumors. However, the mechanism of Rnf2 in cancer development remains poorly understood. Previously, Rnf2 was shown to repress the transcription of p16Ink4a, which mediates Rnf2 activities in cancer cells (22, 33, 34). However, in the present study, Rnf2 directly degraded p53 through the proteasome pathway. We also showed that Rnf2 targets nuclear p53 for degradation in testicular cancer cells and ovarian tumors. The regulation of p53 through Rnf2 affects cell cycle progression and apoptosis controlled by p53. To confirm its role in vivo, we also showed that Rnf2 is overexpressed in most clinical ovarian tumors and is required for tumor growth in a mouse xenograft model.

Approximately 20 E3 ubiquitin ligases for p53 have been identified (12). Thus, overlapping/redundant roles for these E3 ligases have been proposed; however, the contributions of the individual ligases to the specific physiological/pathological processes remain undetermined. We showed that Rnf2 cooperates with the E2 enzyme UbcH5c to mediate the attachment of ubiquitin to p53 in the presence of Bmi1 (Fig. 3E). Notably, p53 ubiquitination in the in vitro assay is less efficient compared with that in vivo (Fig. 3). We suspect that additional components might be required for efficient p53 polyubiquitination mediated by Rnf2. Similar observations have been previously reported. In the presence of Bmi1, Rnf2 activity was enhanced to monoubiquitinate H2A (32). Rnf2 was also shown to promote PGC-1 alpha degradation and to indirectly regulate p53 functions in cells responsive to stress (28). The results obtained in the present study expand the current understanding of the molecular functions of PcG complex 1. The ability of Rnf2 to regulate p53 through direct ubiquitination adds another E3 ligase to the array of known negative regulators for p53. Unlike Mdm2, Pirh2, or other previously identified E3 ligases for p53, Rnf2 is not a transcriptional target of p53 (Fig. S2D). Moreover, Rnf2 represses nuclear p53 levels in a cell-type-specific manner. In contrast to Mdm2, the overexpression/depletion of Rnf2 did not affect p53 degradation in HEK 293 cells, although the association of Rnf2 and p53 was detected in these cells (Fig. S2B). Similarly, cell lines, including PA-1 and HeLa cells, are insensitive to Rnf2 depletion. Conversely, in addition to Tera cells, MCF-7, HO-8910, and A2780 are also responsive to Rnf2 depletion, which resulted in the up-regulation of p53. The Rnf2-mediated degradation of p53 in specific cell lines could be achieved through various mechanisms. For example, the unique functional composition of the PRC1 complex might exist in the reactive cells, while p53 degradation-resistant cells might lack key functional components of the PRC1 complex, or these components might be inaccessible to p53. However, no obvious negative correlation was observed between Rnf2 and p53 expression levels in some cancer cell lines (Fig. S2C), suggesting that the expression of Rnf2 is not a determining factor in the regulation of p53 levels in these cells.

We also investigated whether the regulation of Rnf2 phosphorylation plays a role in its E3 ligase activity toward p53. The phosphorylation of ⁴¹Ser through p38/MAPK was previously considered as critical for Rnf2 functions (31). Our results showed that a mutated Rnf2 (S41A) did not ubiquitinate p53, suggesting that the E3 ligase activity is dependent on the phosphorylation of ⁴¹Ser (Fig. 3D). We used immuno-purified Rnf2 and Bmi1 in in vitro ubiquitination experiments (Fig. 3E). The nature of the in vitro translation system prevents us from excluding other residual factors in the extract from participating in the ubiquitination reaction. Further analysis is required to completely address the regulation of Rnf2 E3 activity toward p53.

Because p53 is the major protector of genome integrity, p53 activation in cancer cells can trigger tumor regression, whereas the misregulation of the negative regulators of p53 plays a role in tumor formation/progression (35–37). Indeed, small molecules targeted to the interface of p53 and major negative regulators of p53, such as Mdm2, have been examined for the inhibition of tumor progression through the restoration of the apoptotic function of p53 (38, 39). Similar to Mdm2, through its regulation of p53, Rnf2 promotes tumor progression. In this study, we revealed the up-regulation of Rnf2 in a high percentage of human ovarian carcinoma samples. The p53 down-regulation in these samples was correlated with Rnf2 up-regulation. Surprisingly, p53 mutations did not affect the Rnf2-mediated degradation of p53, as the overexpression of Rnf2 can ubiquitinate four p53 mutants isolated from human tumors (Fig. S5F). This result suggests that Rnf2 negatively regulates both wild-type and mutated p53. We observed that Rnf2 might bind to multiple regions in the DNA binding domain of p53, which might partially explain why single point mutants do not significantly diminish the interaction between p53 and Rnf2. The significance of the degradation of mutant p53 through Rnf2 remains unknown. Previous studies have shown that the increased expression of Bmi1 and Rnf2 are associated with tumor cell transformation in a number of tumors (27, 40). Our results indicate that Rnf2 is required for tumor progression in a p53-dependent manner. We propose that in normal cells, Rnf2 is not significantly involved in regulating p53 activity, while in some tumor cells, the overexpression of Rnf2 antagonizes p53 through degradation and thus functions as an oncogene. Our findings might provide an additional option for therapeutic intervention, which might restore the p53 response in selective tumor cells.

Materials and Methods

The Institute of Genetics and Developmental Biology Animal Care and Ethics Committee approved all experimental procedures involving the immunodeficient mice. The experiments were conducted according to the principles expressed in the Declaration of Helsinki. BALB/c Nude mice were maintained under pathogen-free conditions and used at 6–10 wk of age. To analyze localized ovarian tumor growth in vivo, 1×10⁶ GFP-expressing A2780 cells stably transfected with the indicated shRNA constructs were injected s.c. in both rear flanks (n = 16 tumors per experimental condition). The in vivo imaging of tumor cells was performed using an IVIS Spectrum fluorescence light system (Caliper), and the light imaging was captured with a Canon EOS 600D camera. The volume of the s.c. xenograft was calculated as V = L × W²/2, where L and W stand for tumor length and width, respectively.

Detailed descriptions of other experimental procedures are provided in the SI Text.