As a component of the transcriptional repression complex 1 (PRC1), the ring finger protein RING1 participates in the epigenetic regulation in cancer. However, the contributions of RING1 to cancer etiology or development are unknown. In this study, we report that RING1 is a critical negative regulator of p53 homeostasis in human hepatocellular and colorectal carcinomas. RING1 acts as an E3 ubiquitin (Ub) ligase to directly interact with and ubiquitinate p53, resulting in its proteasome-dependent degradation. The RING domain of RING1 was required for its E3 Ub ligase activity. RING1 depletion inhibited the proliferation and survival of the p53 wild-type cancer cells by inducing cell-cycle arrest, apoptosis, and senescence, with only modest effects on p53-deficient cells. Its growth inhibitory effect was partially rescued by p53 silencing, suggesting an important role for the RING1–p53 complex in human cancer. In clinical specimens of hepatocellular carcinoma, RING1 upregulation was evident in association with poor clinical outcomes. Collectively, our results elucidate a novel PRC1-independent function of RING1 and provide a mechanistic rationale for its candidacy as a new prognostic marker and/or therapeutic target in human cancer.
Significance: These results elucidate a novel PRC1-independent function of RING1 and provide a mechanistic rationale for its candidacy as a new prognostic marker and/or therapeutic target in human cancer. Cancer Res; 78(2); 359–71. ©2017 AACR.
In mammals, polycomb group (PcG) proteins consist of two major polycomb repressive complexes (PRC): PRC1 and PRC2 (1, 2). The PRC1 core complex includes RING1/RING1A, RING1B/RNF2, and BMI1, whereas the PRC2 core complex includes EED, EZH2, and SUZ12. PRC1 and PRC2 are known to play an important role as transcriptional repressors in embryonic development, stem cell self-renewal, and cell proliferation through epigenetic modifications of target genes (3, 4). Deregulation of PcG genes is frequently found to be associated with developmental defects and cancer (3, 5–7).
RING1 is a crucial component of the PRC1 and, along with RNF2 and BMI1, acts as an E3 ubiquitin ligase to monoubiquitinate histone H2A at lysine 119 (H2AK119Ub1; refs. 8, 9). Different from the well-studied roles of RNF2 and BMI1 in carcinogenesis (10–12), limited evidence has been provided with regard to the role of RING1 in cancer. An earlier report suggested overexpression of RING1 contributed to cellular transformation by upregulating the expression of proto-oncogenes such as c-jun and c-fos (13). However, homeostatic levels of RING1 were found to vary significantly among different cancer types, including lung cancer, prostate cancer, lymphoma, and kidney cancer (14–20), suggesting a yet unclear role of RING1 in cancer.
In this study, we have uncovered tumor suppressor p53 protein as a novel bona fide substrate of RING1. RING1 directly interacts with, and ubiquitinates p53, which finally leads to its proteasome-dependent degradation. As a result, RING1 depletion results in p53 stabilization, leading to cell-cycle arrest, apoptosis, senescence, and migration inhibition in p53-proficient cells. Moreover, we found that RING1 is highly expressed in hepatocellular carcinoma (HCC) tissues and its expression is associated with poor outcomes.
Materials and Methods
The pCMV-HA-p53, pcDNA3.0-RING1-Flag and deletion or point mutants, pGEX-4T-1-GST-p53, pSJ8-MBP-RING1-6His, and pGL4.17-p21-luc plasmids were constructed in our lab. The siRNAs duplexes were synthesized by RiboBio Co., Ltd. si-RING1-1: 5′-GTGGGAACTGAGTCTGTATGA-3′, si-RING1-2: 5′-CAGATCAGACCACAACGAT-3′. sh-RING1 and sh-p53 (with a sequence of 5′-GACTCCAGTGGTAATCTAC-3′) were cloned into pLKO.1 vector.
Antibodies, reagents, and cell lines
The anti-RING1 (D2P4D) (#13069), anti-p21 (12D1; #2947), anti-p53 (1C12; #2524), anti-ubiquitin (P4D1; #3936) antibodies were purchased from Cell Signaling Technology. The anti-p53 (DO-1) (#SAB5100001), anti-Flag (#F1804), anti-HA (#H6908), anti-GAPDH (#G8795), anti-β-actin (#A1978) antibodies, anti-Flag M2 Affinity Gel (#A2220), and propidium iodide (PI, #P4170) were purchased from Sigma-Aldrich. Annexin V-PI Staining Kit was from Beijing 4A Biotech Co, Ltd. The anti-HA Affinity Gel (#B23302) was purchased from Biotool. The Glutathione Sepharose 4B (#17075601) was purchased from GE Healthcare. Etoposide (ETO, #S1225) was purchased from Selleckchem. HepG2 and HCT116 cell lines were obtained from China Infrastructure of Cell Line Resources. HCT116 p53−/− cells were gifted from Prof. Bert Volgestein at Johns Hopkins University School of Medicine, Baltimore, MD. HEK293T cell lines were from ATCC in the United States. All of the cell lines were authenticated by STR profiling at December 2016.
Cell transfection and infection
Cell transfection was carried out using Lipofectamine 2000 following the manufacturer's instruction. For lentiviral packaging, psPAX2, pMD2.0G, and pLKO.1-sh-p53, pLKO.1-sh-RING1, or the pLKO.1 scramble vectors were cotransfected into HEK293T cells, and the supernatants harvested at 48 or 96 hours after transfection. Lentivirus infection was transient or followed by screening with appropriate antibiotics to establish cell lines stably expressing the indicated shRNAs. Specifically, for stable knockdown of RING1 or p53, the cells were infected with lentivirus expressing RING1 or p53 shRNA (sh-RING1 or sh-p53) in the presence of 8 μg/mL polybrene (#107689; Sigma). Stable cell lines were selected and maintained at 5 μg/mL puromycin (#P8833; Sigma).
Protein array analyses
PathScan Cancer Phenotype Antibody Array Kit (#14821, Cell Signaling Technology) was used to examine the level of cancer cell-associated proteins in RING1 knockdown cells. The cells were infected with shRNA for 72 hours. Cell extracts were prepared and analyzed following the manufacturer's instruction.
The cells infected with sh-RING1 for 24 hours were seeded in 96-well plates at 1 × 104 cells per well for 24 hours before transfection with the pGL4.17-p21-luc plasmid (100 ng per well) and Renilla (5 ng per well) in triplicate. After 48 hours, the luciferase activity was measured as described (21).
RNA was extracted using TRIzol (#15596026; Invitrogen). The cDNA was synthesized using PrimeScript RT Master Mix (#DRR036A; TaKaRa). The reaction conditions used for PCR set for 35 cycles of denaturation at 94°C for 30 seconds, annealing at 65°C for 30 seconds, and elongation at 72°C for 1 minute. Sequences of the primers were as shown in Table 1.
|Gene||Forward sequence||Reverse sequence|
|Gene||Forward sequence||Reverse sequence|
HepG2 cells were seeded (5 × 105) into 6-well plates. Seventy-two hours after transfection with pcDNA3.0-RING1-Flag or si-RING1-1, cells were treated with 50 μmol/L etoposide for indicated time before harvest and subsequent immunoblotting analyses, using described antibodies.
Cell growth and colony formation assay
Following plasmid transfection or lentivirus infections, cell growth was assessed as described previously (21). For colony formation assay, the cells stably expressing sh-RING1 or sh-p53 were seeded (5 × 103/well) onto 6-well plates. After 3 weeks, the colonies were stained with hematoxylin and counted.
Cell cycle and apoptosis analysis
The cells were infected with sh-RING1 for 72 hours, and cell-cycle distribution was performed as described previously (21). Apoptosis was determined by Annexin V/PI Apoptosis Detection Kit. In drug treatment groups, the cells were infected with sh-RING1 for 24 hours, followed by etoposide (50 μmol/L) treatment for another 48 hours.
Senescence-associated β-galactosidase staining
Senescence-associated β-galactosidase (SA-β-gal) staining was performed and quantified as described before (21).
Migration capacity was measured by transwell and wound healing assay. The transwell filter (8-μm pores; Millicell Cell Culture Inserts) was used according to the instructions. Hematoxylin was used to stain the cells that migrated to the lower side of the top chamber. For wound-healing assay, the cells infected with sh-RING1 were seeded into 6-well plates, and confluent cell monolayers were wounded by manually scraping the cells, washed with PBS, and further cultured in medium without FBS for the indicated time. Migration ability was represented by the percentage of the wound-healing area after normalization to control using ImageJ.
Generation of RNF2 KO cells using CRISPR/Cas9 system
A CRISPR/Cas9 system was used to establish RNF2-deficient HEK293T cells. Single guide RNA (sgRNA) was generated by online CRISPR Design tool (http://crispr.mit.edu/). The sgRNA sequences were cloned into pX260-U6-Chimeric-BB-CBh-hSpCas9 plasmid and transfected into HEK293T cells. After being selected with 5 μg/mL puromycin, the cells were seeded into 96-well plates for monoclonalization. Colonies derived from a signal cell were detected for gene expression. The sgRNA sequences of RNF2 were 5′-AATTCACTGTGTAGACTTCG-3′ and 5′-ATCATCACAGCCCTTAGAAG-3′.
Immunoblotting, immunofluorescence, and immunoprecipitation
Standard immunoblotting (IB) was performed with indicated antibodies and scanned with ImageJ software for normalization where needed as described before (21). For immunofluorescence analysis, the cells were seeded on glass coverslips and fixed with 4% paraformaldehyde for 20 minutes at room temperature and stained with indicated antibodies. The cell nuclei were stained with DAPI. For immunoprecipitation (IP) analysis, the cells were lysed in Triton X-100 lysis buffer (150 mmol/L NaCl, 50 mmol/L Tris, and 1% Triton X-100, pH 7.5), and incubated with 5 μL respective antibodies and 20 μL protein G plus-agarose or 20 μL Flag-agarose beads overnight at 4°C. The immunocomplexes were subsequently washed with lysis buffer and subjected to IB.
RING1 and p53 proteins were purified as described previously (22). Purified GST–p53 fusion protein bound to the Glutathione Sepharose 4B was incubated with cell lysates. MBP-RING1-His fusion protein bound to the Amylose resin was incubated with purified His-p53. All the interactions were rotated incubated at 4°C for 6 hours. After washing with Triton X-100, the Sepharose beads bound proteins were separated by SDS/PAGE and IB.
In vivo ubiquitination assay was performed as described before (23). Briefly, HEK293T cells were transfected with the indicated plasmids or siRNA for 72 hours, followed by MG132 treatment for 6 hours before harvested and lysed. IP was carried out with anti-HA-beads. After rotated incubation, the anti-HA beads were washed with RIPA lysis buffer and boiled in 1× SDS-PAGE sample buffer, followed by IB analyses using indicated antibodies. The in vitro autoubiquitination assay was carried out using an Ubiquitination Kit from Enzo Life Science (#BML-UW9920-0001) as described previously (24).
Xenograft tumor model
Female BALB/c nude mice (6-week-old) were purchased from SPF Biotechnology Co. Ltd. The animal study was approved by our Institute's Animal Care and Use Committee. The HepG2-GFP cell population with stable RING1 knockdown was amplified, mixed with matrigel (50%) and followed by injection subcutaneously into the mice (1 × 106/per mice). Tumor volumes were measured twice a week and calculated. The whole-body imaging of tumor-bearing mice was monitored twice a week by the FluorVivo Model 100 (INDEC BioSystems). All mice were sacrificed after 18 days, and the tumor tissues were collected, photographed, and weighted.
IHC analysis of human tumor tissue array
IHC was performed in human liver cancer tissue microarrays (purchased from Shanghai Biochip in Shanghai, China) containing 90 pairs of clinical hepatocellular carcinoma with adjacent normal tissues to compare in situ expression of RING1 as described previously (21). The staining intensity was classified into five groups with increasing staining intensity from marginal (±) to the strongest (++++; ref. 25). The paired human liver tumor tissues were collected from Qilu Hospital in Jinan, China, with written informed consents obtained from all the patients based on the Declaration of Helsinki. The study was approved by Research Ethics Committee of Qilu Hospital and our Institute's Ethics Committee.
All experiments were repeated at least three times. Statistical analysis was performed using Student t test (between two groups) or one-way ANOVA analysis (within multiple groups) for data comparisons. For IB analysis, one representative result from at least three experiments was shown. Statistical analysis was performed using SPSS 13.0, and differences between the groups were identified as statistically significant at three levels: P < 0.05, P < 0.01, and P < 0.001.
P53 protein expression is regulated by RING1
We first used two siRNAs targeting RING1 to suppress the protein expression in HepG2 cells and HCT116 cells. Both siRNAs efficiently knocked down RING1 expression compared with control group, with si-RING1-1 showing more pronounced effect (Supplementary Fig. S1A). This siRNA sequence was thus used to construct lentiviral sh-RING1 vector for stable expression in the following work, which markedly depleted endogenous RING1 in both cells at 72 hours postinfection (Fig. 1A). To unbiasedly identify possible molecular targets of RING1, a protein antibody array was employed to assess the effects of RING1 knockdown on the static levels of 19 cancer cell–associated proteins, from which p53 level to be apparently upregulated upon RING1 knockdown in both HepG2 and HCT116 cell lines (Fig. 1B). Further IB analyses using two siRNAs targeting RING1 cells confirmed the p53 upregulation in these cells (Fig. 1C). Consistently, p21, a well-known target of p53, was also increased at both mRNA and protein levels (Fig. 1C; Supplementary Fig. S1B). In addition, RING1 knockdown has little effect on p53 mRNA level in both HepG2 and HCT116 cells, as shown by RT-PCR analysis (Fig. 1D). Together, these findings suggested a role for RING1 in negatively regulating the static level of wild-type p53 proteins, apparently at posttranscriptional level.
RING1 negatively regulates p53 protein stability with or without genotoxic stress
Next, to determine whether RING1 might regulate the stability of p53, a cycloheximide (CHX)-based chase experiment was performed to examine the half-life of p53 protein in the presence of RING1 depleted or at endogenous level. CHX (50 μg/mL) was added at 72 hours post-infection of sh-RING1 to block protein synthesis, and static levels of endogenous p53 proteins were then checked by IB analysis. Upon RING1 knockdown, the half-life of p53 in both HepG2 and HCT116 cells was significantly extended, compared with that of the control groups, whereas RING1 overexpression caused marked reduction in p53 protein level (Fig. 2A and B). When the proteasome activity was blocked by the proteasome inhibitor MG132 (20 μmol/L) for 6 hours, the decrease in p53 protein level upon RING1 overexpression could be reversed (Fig. 2B). This suggested that RING1 could negatively regulate p53 protein stability in a proteasome-dependent manner.
Given the fact that p53 is stabilized upon DNA damage-induced stress, we also checked whether RING1 regulates p53 expression in response to DNA damage. As shown in Fig. 2C, when HepG2 cells were treated with DNA damaging agent etoposide (ETO), ectopic RING1 expression significantly decreased ETO-induced upregulation of p53 protein and the downstream hallmarks of DNA damage response such as phosphorylated p53, p21, and γH2AX (Fig. 2C, left). Conversely, RING1 depletion upregulated p53 protein level as well as the hallmarks for DNA damage response upon the ETO exposure (Fig. 2C, right). This finding suggested that RING1 destabilizes endogenous p53 proteins in cells with or without genotoxic stress.
RING1 directly binds to p53 to ubiquitinate p53 in vivo and in vitro
As RING1 itself is an E3 Ub ligase (26), we speculated that RING1 might directly interact with, and conjugate Ub to p53, which could finally lead to faster p53 degradation. As shown in Fig. 3A, coimmunoprecipitation (co-IP) analysis indicated that a significant fraction of Flag-tagged RING1 and HA-tagged p53 could form a complex in HEK293T cells. Further immunofluorescence microscopy analyses indicated that endogenous RING1 and p53 proteins, visualized by respective antibodies, were colocalized in the nuclei of HepG2 and HCT116 cells (Fig. 3B). Supportively, endogenous binding between RING1 and p53 proteins was observed in HCT116 cells by co-IP analysis with RING1 antibody (Supplementary Fig. S2A).
Furthermore, in an in vitro GST pull-down assay using the recombinant GST-p53 protein and whole-cell lysates from HEK293T cells expressing RING1-Flag, RING1 bound with p53 (Fig. 3C). Pull-down assays with recombinant MBP-RING1-His and His-p53 proteins further confirmed the direct binding of the proteins to each other in vitro (Supplementary Fig. S2B), suggesting a direct interaction between RING1 and p53.
To check whether RING1 did promote p53 ubiquitination, RING1 was overexpressed in HEK293T cells that were pretreated with MG132 to block proteasome-dependent degradation, which seemed to markedly induce p53 polyubiquitination (Fig. 3D). Conversely, RING1 knockdown decreased the ubiquitination of p53 (Supplementary Fig. S2C). In a reconstituted in vitro ubiquitination assay that included E1, UbcH6 (E2), ATP and Ub, RING1 was also found to ubiquitinate p53 (Fig. 3E). Altogether, these data clearly suggested that RING1 not only directly interacts with p53 protein, but is also capable of conjugating poly-Ub chains onto p53 both in vitro and in vivo.
E3 ubiquitin ligase activity of RING1 is dependent on its RING domain
We went on to map the p53-interacting regions in RING1. As shown in Fig. 4A, co-IP experiments were performed with full-length p53 and different deletion mutants of Flag-tagged RING1 that were coexpressed in HEK293T cells. Deletion of RING-domain–containing area of aa48-88 (RING1-Δ1) significantly reduced the interaction of RING1 with p53. Consistently, RING domain deletion totally abolished the ubiquitination state of p53 compared with the full-length RING1, or either of two other deletion mutants examined (Fig. 4B). To further address the role of RING domain, a same point mutation within the RING domain as reported before, which is not capable of binding with E2 ligase (27), was used to compare the ubiquitination activity of RING1 with that of the wild type. As shown in Fig. 4C, the I50A mutation greatly reduced the E3 ligase activity of RING1, without affecting the binding ability with p53 (Supplementary Fig. S2D). Together, this suggests the N-terminal RING domain in RING1 is critical for its binding and ubiquitin ligase activity towards p53.
RNF2 is not required for RING1-mediated poly-ubiquitination of p53 in vivo
RNF2 was previously shown to ubiquitinate p53 (27). As a paralog of RING1, RNF2 can also form a complex with RING1, so we next checked whether RNF2 is involved or required for the E3 ligase activity of RING1 toward p53. It has first come to our attention that the shRNA for RING1 used in above sections might also interfere with RNF2 expression. However, as shown in Supplementary Fig. S3A, the knockdown effect of sh-RING1 specifically limited to RING1, thus suggesting that the above observed RING1-mediated ubiquitination of p53 is unlikely an artifact. Furthermore, two independent HEK293T-derived RNF2 knocked out cell lines (RNF2−/−) were constructed with the CRISPR/cas9 technique (Supplementary Fig. S3B). We used #2 cell line to demonstrate that coexpressed RING1-Flag and HA-p53 were able to form complex that survived the co-IP procedures (Supplementary Fig. S3C), and RING1 interacted with p53 in the absence of RNF2 and induced p53 degradation independently of RNF2 (Fig. 4D).
Knockdown of RING1 inhibits the proliferation and survival of cancer cells in a p53-dependent manner
To further investigate the possible functional consequences of RING1-mediated ubiquitination of p53, the effects of RING1 knockdown on the cell proliferation or survival were assessed. RING1 knockdown dramatically inhibited the proliferation of HepG2 or HCT116 cells, compared with that of the control (Fig. 5A), which was also consistent with the data from further colony formation assay (Fig. 5B).
As cellular senescence constitutes an another major tumor-suppressing mechanism in cancer, we next asked whether RING1 knockdown might affect senescent phenotype using the established senescence-associated β-galactosidase (SA-β-Gal) staining assay (21). As shown in Fig. 5C, significant induction of senescence was observed, as 3- to 7-fold more cells were positively stained in the RING1-deficient group showed than in the controls. FACS analysis also indicated that, upon RING1 depletion, increased cell subpopulations were found to retain in G1 phase (increased from 60.0% to 68.4% in HepG2 cells, and from 53.1% to 58.9% in HCT116 cells), suggesting that RING1 deficiency might at least cause partial G1 arrest (Supplementary Fig. S4A).
However, RING1 knockdown in HCT116 p53−/− cells caused a significantly less inhibitory effect on the proliferation and survival, when compared those with p53 wild-type cells (Fig. 5A and B). Moreover, little changes of senescence or G1 arrest was observed in HCT116 p53−/− cells upon RING1 depletion (Fig. 5C; Supplementary Fig. S4A). Consistently, RING1 knockdown caused little effect on p21 expression in HCT116 p53−/− cells (Supplementary Fig. S4B). These data clearly suggested a prominent role of p53 in RING1 actions to regulate cell cycle, proliferation, or senescence. Supportively, knockdown of p53 partially and significantly rescued the growth inhibitory effect of RING1 knockdown in HepG2 cells (Fig. 6A; Supplementary Fig. S4C).
Furthermore, by FACS analysis after PI/Annexin V staining, 8.3% or 11.9% of HepG2 or HCT116 cells, respectively, were found to undergo apoptosis upon RING1 knockdown, in comparison with 1.96% or 4.05% in the respective control groups (Fig. 6B). In addition, when cells were subjected to both RING1 knockdown and ETO treatment, significantly more apoptosis were observed with both HepG2 and HCT116 cells (Fig. 6B), suggesting that RING1 depletion did potentiate cellular response to DNA damage response. Markedly more proliferation was recorded with HepG2 cells overexpressing Flag-tagged RING1, when compared to the cells expressing RING1 endogenously, consistently suggesting a proliferation-promoting function of RING1 (Fig. 6C).
Knockdown of RING1 inhibits the migration ability of cancer cells
Metastasis is a key feature of cancer and p53 also plays a pivotal role in regulating metastasis (28). We further used transwell and wound healing assay to investigate the effect of RING1 on tumor migration capacity. In Fig. 6D, RING1 depletion significantly reduced the migration capacity within 72 hours by transwell assay, and the wound healing ability reduced to 31.5% and 14.7% at 24 and 72 hours, respectively, in HepG2 cells (Supplementary Fig. S5A). A similar result was found in HCT116 cells (Supplementary Fig. S5A).
RING1 knockdown suppresses the growth of tumor xenograft in mice
We then assessed the possible effects of RING1 knockdown on tumor growth in vivo, by using a HepG2 tumor xenograft mouse model. In Fig. 6E, the growth of the tumor deficient in RING1 was significantly suppressed, as judged by whole-body florescence-imaging of the tumor-bearing mice. Final imaging of all the groups on the 18th day since implanting of tumor cells along with tumor mass was shown as well (Supplementary Fig. S5B). Consistent with above observed effect of RING1 depletion on cell proliferation, these data from mouse model clearly suggested that knockdown of RING1, which certainly stabilized tumor suppressor p53, significantly inhibits tumor proliferation both in vitro and in vivo.
RING1 is upregulated in clinical hepatocellular carcinoma specimens compared with adjacent liver tissues
With IHC assay with human hepatocellular carcinoma tissue array containing 90 cases of paired specimens, as shown in two representative images in Fig. 7A, much stronger staining for RING1 was observed in the tumors (left) compared with that in adjacent tissues (right). On the basis of the intensities in IHC staining signals, all samples were classified into five groups with increasing intensities, with the strongest staining (++++) for RING1 in approximately 50% tumor specimens whereas 30% in adjacent tissues (Fig. 7B). RING1 was also assessed at both protein and mRNA levels in eight paired tissues obtained from cancer patients. As shown in Fig. 7C, both protein and mRNA of RING1 expression were significantly higher in tumor tissues than that of adjacent ones. Therefore, it seemed that the static levels of RING1 were predominantly higher in tumors when compared to adjacent tissues.
Consistently, bioinformatics analysis with hepatocellular carcinoma data retrieved from Oncomine database also indicated significantly higher abundance of RING1 in the tumor tissues than in controls (Fig. 7D; ref. 29). Meanwhile, the patients of higher levels of RING1 (P < 0.001) were also recorded with poorer 5-year posttreatment survival, whereas RING1 abundances were also positively correlated with the Barcelona Clinic Liver Cancer Stage progression (P < 0.05; Fig. 7D). All these data strongly suggested that RING1 abundance might be closely associated with liver cancer progression and poor prognosis.
Tumor suppressor p53 plays a pivotal role in the maintenance of genome integrity, cell metabolism, and cellular responses to stresses, and primarily undergoes proteasome-mediated degradation (30). In this study, we have identified RING1 as a novel E3 ligase to ubiquitinate p53 and target it for degradation, based on the observations: (i) p53 and RING1 protein expression is inversely correlated either with or without genotoxic stress (Figs. 1 and 2); (ii) RING1 negatively regulates p53 protein half-life (Fig. 2); (iii) By interacting with p53, RING1 promotes its polyubiquitination and degradation in vivo and in vitro (Figs. 3 and 4). Our work has thus added RING1 to the already existed list of E3 Ub ligases for p53 ubiquitination, including MDM2, COP1, and Pirh2 (30–34). It is both intriguing and important to understand how the activities of these E3 Ub ligases are delicately orchestrated to control the homeostasis and functionality of p53 in both tumoral and nontumoral settings. It is interesting to note that, different from some E3 Ub ligases like MDM2 whose expression was transcriptionally regulated by p53, our data indicated that expression of human RING1 gene was not affected by p53 (Supplementary Fig. S6A).
Previous work has established that PcG proteins are critically involved in regulating tumorigenesis by regulating expression of target genes through both polycomb-dependent and -independent manners (5). By identifying p53, along with TOP2α (35), as another non-histone substrate for the E3 Ub ligase activity of RING1, our work has significantly deepened the current understanding about the non-polycomb function of RING1 in cancer. Previously, RNF2, another crucial component of PRC1, has been shown to regulate the homeostatic level of p53, by either directly degrading p53, or promoting MDM2-mediated p53 ubiquitination in cancer cells (27, 36). In this work, our data have clearly indicated that RNF2, although a paralog of RING1 that may form complex with RING1, is not absolutely required for the E3 Ub ligase activity of RING1 towards p53 in the cells (Supplementary Fig. S3C; Fig. 4D). Obviously, one could not exclude the possibility that other components such BMI, which was shown to promote RING1- and RNF2-mediated ubiquitination (27, 35, 37), could also modulate the E3 Ub ligase activity of RING1 towards p53 in vivo.
Our data demonstrated that depletion of RING1 results in the suppression of cell proliferation and survival, causing G1-phase arrest, apoptosis, and senescence, in dependence of endogenous wild-type p53 (Figs. 5 and 6A). Given the fact that p53, upon activation, suppresses tumor through regulating the expression of a set of genes in controlling cell-cycle arrest, apoptosis, and senescence (38), our findings have strongly suggested that the biological functions of RING1 could be, at least partially, mediated by p53 pathway. Certainly, this does exclude the possibility that, on some occasions, RING1 depletion might lead to cell-cycle arrest independent of p53 function, especially when RING1, along with BMI1, was reported to promote ubiquitination of the chromatin at the promoters of specific genes to stimulate cell-cycle progression (39). In addition, the PcG proteins were also shown to promote carcinogenesis through bypassing cellular senescence due to the transcriptional repression of the INK4b-ARF-INK4a locus (40). Therefore, apart from the direct involvement of p53, there is possibility that these epigenetic changes might also contribute to the observed G1 arrest upon RING1 depletion. This may also account for the partial rescue of growth inhibition upon p53 and RING1 double knockdown in HepG2 cells, compared with RING1 depletion alone (Fig. 6A).
We made novel observation that RING1 depletion markedly abolished the migration ability of both cancer cells (Fig. 6D; Supplementary Fig. S5A), indicating RING1 might be involved in the regulation of tumor metastasis. Despite the possibly involved role of p53 during the metastasis regulation (28), another piece of work suggested the possible involvement of Snail, a master regulator of epithelial–mesenchymal transition and metastasis in cancer, based on observations that RING1, along with RNF2, is capable of interacting with Snail and is required for Snail-mediated target gene repression and cell migration in pancreatic cancer cells (41).
Notably, high levels of RING1 proteins were widely observed in hepatocellular carcinoma specimens but not in the normal tissues, suggesting that potential association between RING1 amplification and hepatocellular carcinoma pathology, based on Oncomine database analysis (Fig. 7). It is interesting to note that, high RING1 levels seem to be more enriched in tumors with wild-type p53 according to the data retrieved from Oncomine database (Supplementary Fig. S6B). In light of our data that overexpression of RING1 indeed promoted the proliferation of liver cancer cells (Fig. 6C), it is possible that RING1-targeted therapeutic interventions might help restore the function of p53 to suppress tumor. Moreover, as RING1 destabilizes endogenous p53 protein level upon genotoxic stress and RING1 depletion potentiates DNA damage responses (Figs. 2C and 6B), RING1-targeted therapeutic strategy may also be employed in combination with DNA-damaging agents to potentiate the antitumor effect of chemotherapy.
Taken together, here we have presented strong evidence that RING1 might serve as a yet another E3 Ub ligase for p53 in regulating human cancer. RING1 is thus emerging as a potential biomarker and target for cancer prognosis and treatment.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: J. Shen, R. Hu, Z. Wang
Development of methodology: J. Shen, X. Shao, Y. Yang, Z. Wang
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Shen, P. Li, X. Liu, Z. Wang
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Shen, P. Li, X. Shao, Y. Yang, Q. Yu, R. Hu, Z. Wang
Writing, review, and/or revision of the manuscript: J. Shen, Q. Yu, R. Hu, Z. Wang
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Feng, Z. Wang
Study supervision: R. Hu, Z. Wang
This project is sponsored by the National Natural Science Foundation of China (nos. 81572752, 81373438, 81321004 to Z. Wang) and CAMS Innovation Fund for Medical Sciences (2016-I2M-2-002 to Z. Wang). We thank Prof. Tan Jing in Zhongshan University (Guangzhou, China) for helpful discussions in the project.
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.