Aberrant Activation of β-Catenin Signaling Drives Glioma Tumorigenesis via USP1-Mediated Stabilization of EZH2

Hyperactivated β-catenin signaling is recognized as an event that occurs frequently in multiple cancer types (1). The canonical Wnt/β-catenin pathway is triggered by Wnt ligands, leading to coactivation of the Frizzled and LRP receptors, and consequent stabilization and translocation of β-catenin into the nucleus (2). Nuclear β-catenin binds to the transcription factor TCF/LEF and thereby activates the gene expression (2, 3). In glioblastoma, early evidences for the involvement of Wnt signaling came from the germline mutations of APC gene (4). Recently, frequent aberrant promoter hypermethylation of Wnt pathway inhibitor genes such as SFRP1, SFRP2, DKK1, and WIF1 have been reported in glioblastomas and WNT3A upregulation has been demonstrated in 81.8% of malignant astrocytic gliomas and in 33.3% of glioblastoma cell lines (5). Our previous study has demonstrated that overexpression of FoxM1 in glioblastoma controls β-catenin nuclear translocation and is required for tumor progression, and therefore, reveals a novel mechanism for β-catenin activation (6). Moreover, dysregulated EGFR signaling, accounting for up to 60% of glioblastoma cases, drives β-catenin signaling activation (7, 8). Thus, aberrant β-catenin activation is prevailing in glioblastoma and is a critical driver for tumorigenesis.

Activated β-catenin signaling is linked to epigenetics to control gene expression (9). For instance, β-catenin interacts with histone H3 arginine 8 methyltransferase (Prmt2) to establish poised chromatin architecture and mark genes for later expression in the development of Xenopus (10). Moreover, β-catenin activation leads to an induction and stabilization of MLL, and the β-catenin–CBP-MLL complex is required to trigger H3K4 trimethylation at promoters of self-renewal genes in tumor-propagating cells (11). However, further study is required to better understand the roles of β-catenin signaling in chromatin remodeling and epigenetic gene regulation.

Enhancer of zeste homolog 2 (EZH2) is a catalytic subunit of the Polycomb group (PcG) complex that participates in transcriptional repression of specific genes by mediating the trimethylation of histone H3K27 (12). EZH2 is frequently overexpressed in glioblastoma and correlated with aggressiveness and advanced tumor progression (13, 14). The oncogenic roles of EZH2 are largely due to its ability to repress the expression of a cohort of downstream tumor suppressor genes through H3K27 trimethylation-mediated epigenetic silencing, including HOXA5, RUNX3, and CDKN1B (12, 15, 16). EZH2 is subjected to proteasome-ubiquitination degradation pathway, and multiple E3 ligases, including Smurf2, β-TrCP, and CHIP, have been identified to mediate EZH2 degradation in different biological contexts (17–19). However, the regulation of EZH2 stability and its implication in PRC2-mediated methyl transferase activity in a cancer-specific context is not fully explored.

In this study, we have demonstrated a critical role of β-catenin signaling in the regulation of EZH2. We found that β-catenin induced the expression of USP1, which then interacted and deubiquitinated EZH2 directly to suppress the ubiquitin–proteasomal degradation of EZH2. Active β-catenin–induced stabilization of EZH2 repressed the downstream tumor suppressor genes expression, and therefore promoted glioma cell proliferation and tumor progression. The β-catenin–USP1-EZH2 axis revealed in our study orchestrates the hyperactivated β-catenin signaling and EZH2-mediated epigenetic gene silencing, which represents a critical mechanism during glioma tumorigenesis.

USP1, USP2, USP3, USP4, USP5, USP8, USP11, USP13, USP14, USP15, USP16, USP18, USP20, USP25, USP26, USP29, USP30, USP36, USP39, USP46, USP48, and USP50 expression plasmids were kindly provided by Dr. Jianhua Yang (Texas Children's Cancer Center, Houston, TX). MYC-EZH2 was kindly provided by Dr. Haojie Huang (Mayo Clinic). MYC-USP1 wild-type and mutant WDR48 were from Dr. Tony T. Huang (New York University School of Medicine, New York, NY). USP1 and EZH2 cDNA were further cloned into pcDNA3-HA vector. The truncated mutants of EZH2 were constructed into pcDNA3-HA vector. MYC-EZH2-K421R was from Dr. Mien-Chie Hung (MD Anderson Cancer Center, Houston, TX), and Y641F site mutagenesis was introduced using the QuikChange site-directed mutagenesis Kit (Agilent Technologies). Lentiviral vector expressing stabilized EZH2 was achieved by cloning the CDS region of MYC-EZH2-K421R/Y641F into a pLVX-IRES-ZsGreen vector (Clontech). Long and short USP1 promoter plasmids were achieved by cloning the USP1 promoter region with corresponding primers into pGL3-basic vector (Promega). Mutant USP1 promoter plasmids were also introduced by the QuikChange site-directed mutagenesis Kit (Agilent Technologies). Lentiviral shRNA plasmids targeting USP1 (#1: TRCN0000342330, #2:TRCN0000342331) and EZH2 (#1: TRCN0000293738, #2: TRCN0000286290) were purchased from Sigma. All plasmid constructions were confirmed by DNA sequencing analysis.

Cycloheximide, MG132, and β-catenin siRNAs were purchased from Sigma. Puromycin was purchased from EMD Biosciences. Recombinant human Wnt3α was purchased from R&D Systems. β-catenin inhibitor PRI-724 was from Selleckchem. Dual-Luciferase Reporter Assay System was from Promega. Antibodies used in this study are listed in Supplementary Table S1.

SW1783, U87MG, LN18, HS683, LN229, and HEK293T cell lines were from ATCC. U251MG cell line was from Sigma. The immortalized NHA-E6/E7/hTERT cell line has been described previously (20). These cell lines were grown in DMEM supplemented with 10% bovine calf serum (HyClone). GSC11 cells were isolated from human glioblastoma specimens and cultured as described previously (21). β-catenin^+/+ and β-catenin^−/− mouse embryonic fibroblasts (MEF) were provided by Dr. Xi He (Boston Children's Hospital, Boston, MA) and cultured as described previously (21). Above cells were passaged less than five times after receipt in our laboratory for relevant studies reported here. No cell lines used in this study were found in the database of commonly misidentified cell lines that is maintained by The International Cell Line Authentication Committee and NCBI Biosample. Cell lines were authenticated by short tandem repeat profiling and were routinely tested for Mycoplasma contamination in every 6 months. The latest test date was in May of 2018.

Transfections of plasmid and siRNAs were performed using Turbofect (R0531, Thermo Fisher Scientific) and X-tremeGENE siRNA transfection reagents (04476093001, Roche Diagnostics), respectively. For treatment, cells were plated in multi-well plates and treated with the indicated concentration of cycloheximide, MG132, and Wnt3a for the indicated times.

For lentiviral production, pLKO.1 shRNA vector or pLVX vector, packing (psPAX2) and envelope (pMD2.G) plasmids were cotransfected into HEK293T cells using Turbofect transfect reagent. After 48-hour transfection, glioma cells were infected with viral particles. Stable clones were selected by culturing cells in medium with 2 μg/mL puromycin for 2 weeks.

Transfected cells were treated with 20 μmol/L MG132 for 6 hours. Cells were lysed using RIPA lysis buffer (50 mmol/L Tris-base pH 6.8, 150 mmol/L NaCl, 1% NP-40, 0.5% deoxycholic acid, 0.1% SDS, 10 mmol/L NaF, 10 mmol/L dithiothreitol (DTT), 0.2 mmol/L Na₃VO₄, 1% cocktail protease inhibitors, 1 mmol/L phenylmethylsulfonyl fluoride. Cell lysates were immunoprecipitated using the indicated antibodies and washed three times by RIPA buffer. To exclude nonspecific ubiquitin-modified species from the EZH2 complex, we washed the immunoprecipitates three times using a ubiquitylation wash buffer (50 mmol/L tris-base pH 6.8, 150 mmol/L NaCl, 1% NP-40, 0.5% deoxycholic acid, 1 mol/L urea, 1 mmol/L N-ethylmaleimide, and protease inhibitors).

In vitro deubiquitylation of EZH2 by USP1 was performed as described previously. HEK293T cells were transfected with HA-EZH2 and Flag-ubiquitin and were treated with 20 μmol/L MG132 for 6 hours. HA-EZH2 containing ubiquitylated HA-EZH2 was purified from the SDS-denatured extracts with HA beads, the bound proteins were eluted with 1 mg/mL HA peptides (Sigma) in RIPA buffer after extensive washing. The ubiquitylated EZH2 protein was incubated with recombinant USP1/WDR48 proteins (Ubiquigent Ltd.) in deubiquitylation buffer [20 mmol/L HEPES (pH 8.3), 20 mmol/L NaCl, 100 mg/mL BSA, 500 mmol/L EDTA, 1 mmol/L DTT] at 37°C for 2 hours.

Nuclear proteins were extracted using the CelLytica NuCLEAR Extraction Kit (Sigma) according to the manufacturer's instructions. The extraction of total proteins was performed using immunoprecipitation lysis buffer (25 mmol/L Tris-HCl pH 7.4, 150 mmol/L NaCl, 1% NP-40 nonyl phenoxypolyethoxylethanol, 1 mmol/L ethylenediaminetetraacetic acid, protease, and phosphatase inhibitors), and immunoprecipitation or immunoblotting with corresponding antibodies was performed as described previously (21).

Total RNA was isolated using TRIzol reagent (Invitrogen) from cells and cDNA was synthesized using the iScript Reverse Transcription Supermix (Bio-Rad). Two-step RT-PCR was performed using SYBR Select Master Mix (Life Technologies) and 7500 RT-PCR System (Life Technologies) according to the manufacturer's instructions. The primer sequences used for PCR are listed in Supplementary Table S2. The expression of housekeeping gene and GAPDH in each sample was used as an internal control. All quantitative PCR results were from three independent assays.

The chromatin immunoprecipitation (ChIP) assay was performed using the ChIP Kit (Cell Signaling Technology #9002). Cells were cross-linked with formaldehyde and subjected to ChIP assay according to the manufacturer's instructions. The ChIP DNA was subjected to quantitative PCR or electrophoresed on a 1.5% agarose gel. The primers used for PCR were listed in Supplementary Table S2.

For immunofluorescence, cells cultured on coverslips were fixed with 4% paraformaldehyde in PBS for 10 minutes, followed by permeabilization with Triton X-100 for 15 minutes. Cells were then blocked by 10% goat serum in PBS for 2 hours, and incubated with the indicated antibody overnight at 4°C, followed by incubation with Alexa Fluor 488 goat anti-mouse IgG or Alexa Fluor 594 goat anti-rabbit IgG (Invitrogen) for 1 hour at room temperature. Coverslips were mounted on slides using antifade mounting medium with DAPI. Immunofluorescence images were acquired on a deconvolutional microscope (Zeiss). For each channel, all images were acquired with the same settings.

For IHC staining, tissue slides (Biomax) were deparaffinized and rehydrated through an alcohol series. After microwave antigen retrieval, indicated antibodies were incubated with the slides overnight at 4°C. Staining was performed using the EnVision Kit (DAKO). The quantification of the staining was evaluated according to the percentage of cells with positive nuclear staining and to the staining intensity as described previously (21).

Glioma cells or MEFs were transfected with the indicated USP1 promoter reporters combined with siRNAs, Wnt3α, or β-catenin inhibitor. The USP1 promoter activity in these cells was normalized by cotransfecting a β-actin/Renilla luciferase reporter (pRL-TK) containing a full-length Renilla luciferase gene. The luciferase activity in the cells was quantified using a Dual Luciferase Assay System (Promega) according to the manufacturer's instructions.

Cell proliferation assay was conducted using XTT (Sigma) or CCK8 (Dojindo) according to the manufacturer's instructions. SW1783 cells expressing indicated constructs (2 × 10³) and GSC11 cells expressing indicated constructs (8–12 × 10³) were seeded into 96-well plate and cultured in corresponding medium for the indicated numbers of days. Then, the cells were incubated with XTT or CCK8 for 3 hours. The absorbance was measured at 450 nm using a microplate reader. Cell viability assay was conducted using CCK8 (Dojindo) according to the manufacturer's instructions.

SW1783 or GSC11 cells expressing indicated constructs were harvested, washed with PBS, fixed with 75% ethanol overnight at 4°C, and then incubated with RNase at 37°C for 30 minutes. The cell nuclei were stained with propidium iodide for an additional 30 minutes and then analyzed by flow cytometry. The results are presented as the percentage of cells in each phase.

All mouse experiments were approved by the Institutional Animal Care and Use Committee of The University of Texas MD Anderson Cancer Center. GSC11 cells (5 × 10⁵), or SW1783 cells (1 × 10⁶) were injected intracranially into 6- to 8-week-old nude (nu/nu) mice (6 mice for each group) as described previously (21). The mice were euthanized 4 weeks (GSC11), or 6 weeks (SW1783) after glioma cells injection and the brains were removed, fixed in 4% formaldehyde, and embedded in paraffin. Tumor formation and phenotype was determined by histologic analysis of hematoxylin and eosin–stained sections. Tumor volume was calculated using the formula V = (π/6) × a² × b, where a and b are the shortest diameter and longest diameter, respectively. For glioma cell–injected mice survival assay, another 8 mice for each group were injected intracranially. Animals were humanely killed when they were moribund; the remaining mice were killed 90 days (GSC11-implanted mice) or 120 days (SW1783-implanted mice) after glioma cells injection.

Each experiment was repeated three times or more. If not mentioned, all data were presented as mean ± SEM of three independent experiments, and the two-tailed Student t test was used to compare two groups for independent samples. A Pearson correlation test was performed to analyze the relationships between active β-catenin, USP1, and EZH2. Survival analysis was conducted using the Kaplan–Meier model with a two-sided log-rank test. The results for statistical significance tests are included in each figure legend. P < 0.05 was considered as statistically significant.

Our previous study has shown that the Wnt/β-catenin signaling pathway plays a critical role in sustaining glioma stem cells and glioma tumorigenesis (6). In an attempt to test the effect of β-catenin on EZH2 expression, we detected EZH2 protein levels in β-catenin wild-type (β-catenin^+/+) and β-catenin knockout (β-catenin^−/−) MEFs. We found that EZH2 was reduced in β-catenin^−/− MEFs compared with β-catenin^+/+ MEFs, which led to decreased level of histone H3K27me3 in β-catenin^−/− MEFs (Fig. 1A). Accordingly, overexpression of the constitutively active β-catenin–S33Y mutant upregulated EZH2 levels in both SW1783 and normal human astrocytes (NHA; Fig. 1B). This result was further confirmed in the patient-derived GSC11 cells using the β-catenin siRNAs (Fig. 1C). However, β-catenin had no obvious effect on EZH2 mRNA levels (Supplementary Fig. S1A–S1C), suggesting that β-catenin regulates EZH2 expression posttranscriptionally.

Some studies have shown that EZH2 is regulated by the ubiquitin-proteasome–dependent degradation pathway (17–19), which has been confirmed in our study (Supplementary Fig. S1D and S1E). To determine whether β-catenin regulates EZH2 protein stability, we measured the abundance of EZH2 in β-catenin^+/+ and β-catenin^−/− MEFs treated with cycloheximide. We found EZH2 was greatly destabilized in the absence of β-catenin (Fig. 1D). Accordingly, overexpression of the active β-catenin–S33Y promoted the stability of EZH2 protein (Fig. 1E). We next investigated the effect of β-catenin on the ubiquitination of EZH2, and found that overexpression of β-catenin–S33Y reduced EZH2 ubiquitination in SW1783 cells (Fig. 1F). These results together indicate that β-catenin stabilizes EZH2 protein by inhibiting its ubiquitination.

As a reverse procedure of ubiquitination, protein deubiquitination critically regulates protein turnover. To identify the deuibiquitinases (DUB) responsible for reversing EZH2 polyubiquitination, we screened a panel of 23 DUBs, which usually function as oncogenes or show high expression in central nervous system. We found that EZH2 protein level was upregulated by deubiquitinases USP1, USP11, USP16, and USP29 in HEK293T cells under cycloheximide treatment (Supplementary Fig. S2A). Because DUBs may regulate EZH2 expression indirectly through mechanisms other than the ubiquitin–proteasome pathway, we thus examined the effect of DUBs on EZH2 ubiquitination. Of those four DUBs, only USP1 significantly reduced the ubiquitination of EZH2 protein (Supplementary Fig. S2B), indicating that USP1 may act as a deubiquitinase of EZH2.

We next assessed the effect of USP1 on EZH2 protein stability. Depletion of USP1 in U87MG and GSC11 cells decreased EZH2 protein level as determined by Western blot analysis (Fig. 2A). In addition, overexpression USP1 in HEK293T and SW1783 cells significantly upregulated EZH2 protein level as determined by immunofluorescent staining (Fig. 2B). Accordingly, EZH2 was almost abolished in USP1-depleted cells as determined by immunofluorescent staining (Supplementary Fig. S2C). However, USP1 had no effect on EZH2 mRNA levels (Supplementary Fig. S2D and S2E). Moreover, depletion of USP1 promoted EZH2 degradation in both U87MG and GSC11 cells, whereas USP1 overexpression increased EZH2 stability (Fig. 2C and D). Consistently, the expression of EZH2 was positively correlated with USP1 expression in glioma cell lines, but not with WD-repeat protein WDR48, a binding partner of USP1 (Fig. 2E; ref. 22). These findings indicate that USP1 contributes to the stabilization of EZH2 in glioma cells.

We next detected the interaction between USP1 and EZH2. USP1 and EZH2 proteins colocalized in nucleus in glioma cells (Supplementary Fig. S2C). In addition, USP1 coimmunoprecipitated with EZH2 in HEK293T cells ectopically expressing both proteins (Supplementary Fig. S3A). Moreover, reciprocal coimmunoprecipitation experiments showed that endogenous USP1 and EZH2 interacted with each other in U87MG and GSC11 cells (Fig. 3A). To map the protein region in EZH2 that mediates its interaction with USP1, we constructed plasmids expressing a series of deletion mutants of EZH2 and found the protein residues 310-484 of EZH2 containing a SANT domain is required for its interaction with USP1 (Fig. 3B).

We further evaluated the effect of USP1 on EZH2 ubiquitination. Overexpression of wild-type USP1 notably reduced the ubiquitination of EZH2 (Fig. 3C, lane 3), and this effect was enhanced by further expressing WDR48, an USP1-binding partner enhancing the activity of USP1 (Fig. 3C, lane 4). However, overexpression of WDR48 alone did not significantly reduce the endogenous ubiquitination of EZH2 (Fig. 3C, lane 7). Moreover, overexpression of USP1-C90S, a catalytically inactive form of USP1 (23), did not alter EZH2 ubiquitination either in the presence of WDR48 or not (Fig. 3C, lane 5 and lane 6). Accordingly, depletion of endogenous USP1 using two independent shRNAs promotes EZH2 ubiquitination in GSC11 and U87MG cells (Fig. 3D; Supplementary Fig. S3B). To address the question whether USP1 deubiquitinates EZH2 directly, ubiquitinated EZH2 was purified from HEK293T cells and incubated with USP1 and WDR48 proteins. We found that USP1/WDR48 decreased EZH2 ubiquitination, and the efficiency was increased with the amount of USP1/WDR48 proteins that were added to the reactions (Fig. 3E). Thus, these results strongly support that USP1 interacts with and deubiquitinates EZH2 directly.

We sought to determine whether USP1 expression was regulated by β-catenin in glioblastoma. First, we surveyed the β-catenin, TCF4, and USP1 mRNA expression levels in glioblastoma utilizing The Cancer Genome Atlas (TCGA) datasets. We found USP1 mRNA levels were linearly correlated with β-catenin and TCF4 mRNA levels in glioblastoma, respectively (Supplementary Fig. S4A). The linear correlation between β-catenin and USP1 expression levels was observed in all glioblastoma subtypes (Supplementary Fig. S4B). Moreover, transfection of β-catenin–S33Y significantly increased USP1 mRNA level in SW1783 and NHA cells (Fig. 4A and B). This result was further confirmed by depleting β-catenin in U87MG cells (Fig. 4C) and GSC11 cells (Supplementary Fig. S4C). Accordingly, USP1 mRNA level was decreased in β-catenin^−/− MEFs compared with β-catenin^+/+ MEFs (Supplementary Fig. S4D). These results suggest that β-catenin activates USP1 transcription.

We next searched the promoter region of USP1 and found one potential LEF/TCF consensus–binding sequence around 2,000-bp upstream the transcription start site of USP1. In addition, ChIP assay showed that β-catenin bound to USP1 promoter region and the binding capability was enhanced under Wnt3a treatment (Fig. 4D). To further confirm the transcriptional regulation of USP1 by β-catenin on, we constructed two USP1 promoter reporters, Pro-L and Pro-S, of which, Pro-S lack the upstream TCF4-binding element (Fig. 4E). Reporter gene assays showed that Pro-L was actually more responsive to Wnt3a and β-catenin signal compared with Pro-S (Fig. 4F–H; Supplementary Fig. S4E). Moreover, we also constructed promoter reporter plasmids harboring mutations of the TCF4-binding site using the empirical algorithm (24). The result showed that β-catenin inhibitor PRI-724 could significantly inhibit the wild-type promoter activity, whereas it showed no impact on mutant promoter reporters (Fig. 4I). Also, the reporter gene activities of the wild-type promoter in β-catenin^−/− MEFs was significantly lower than that in β-catenin^+/+ MEFs, whereas no great change of the reporter gene activities driving by the mutant promoter was observed between β-catenin^+/+ and β-catenin^−/− MEFs (Supplementary Fig. S4F).

We further assessed the effect of β-catenin on USP1 and EZH2 protein expression. As indicated in Fig. 2E, the expression of β-catenin in different glioma cell lines showed good positive correlation with that of USP1 and EZH2. Both USP1 and EZH2 were upregulated under Wnt3a treatment in SW1783 and NHA cells (Fig. 4J). Accordingly, depleting β-catenin in U87MG cell using shRNA could significantly downregulate the USP1 and EZH2 protein expression (Fig. 4K). Furthermore, EZH2 expression was decreased in β-catenin^−/− compared with β-catenin^+/+ MEFs, and overexpression of USP1 in β-catenin^−/− MEFs restored the EZH2 level (Supplementary Fig. S4G). Taken together, these results demonstrate that β-catenin transcriptionally activates USP1 expression to promote EZH2 protein level in glioma.

By catalyzing histone H3 lysine 27 trimethylation (H3K27me3), EZH2 represses a series of tumor suppressor genes associated with tumorigenic properties of glioblastoma, including CDKN1B, RUNX3, and HOXA5. We therefore assessed the effect of USP1 on EZH2-mediated gene expression. Depletion of USP1 significantly decreased the recruitment of EZH2 on the promoters of CDKN1B, RUNX3, and HOXA5, which led to decreased H3K27me3 levels on these promoters (Fig. 5A and B). We then measured the mRNA abundance of the EZH2 targets in the β-catenin/USP1/EZH2 context. In GSC11 cells, depletion of β-catenin/USP1/EZH2 increased the mRNA levels of those genes (Fig. 5C). This result was further confirmed in SW1783 cells overexpressing Wnt3a (Fig. 5D). Accordingly, the protein levels of p27 (a protein coded by CDKN1B gene), RUNX3, and HOXA5 were all upregulated by depletion of β-catenin/USP1/EZH2 in GSC11 cells (Fig. 5E), and this result was further confirmed in SW1783 cells overexpressing Wnt3a (Fig. 5F). Thus, these results together demonstrate that the β-catenin–USP1-EZH2 axis critically regulates target gene expression by controlling the H3K27me3 levels.

To further test the effect of USP1–EZH2 axis on the target protein expression, we constructed an EZH2 variant, EZH2-421R/641F, which is resistant to the proteasome-mediated protein degradation (17, 18). We designate this variant as ND-EZH2 (nondegradation EZH2). Compared with the wild-type EZH2, ND-EZH2 was insensitive to cycloheximide treatment (Fig. 5G). Considering that the residue 421 of EZH2 is within the 310-484 region that mediates the interaction with USP1 (Fig. 3B), it is probably that USP1 stabilizes EZH2 by reversing β-TrCP/Smurf2–mediated EZH2 ubiquitination. Moreover, overexpression of ND-EZH2 rescued the effect of USP1 depletion on the expression of RUNX3, HOXA5, and p27 (Fig. 5H). These results support that USP1 promotes EZH2–mediated gene repression by stabilizing EZH2.

We further investigated the roles of the β-catenin–USP1-EZH2 axis in glioma cell proliferation and tumorigenesis. Depletion of USP1/EZH2 in SW1783 cells abolished the effect of Wnt3a overproduction on cell proliferation, and reconstituted expression of the ND-EZH2 variant reversed this effect (Fig. 6A; Supplementary Fig. S5A). Moreover, depletion of β-catenin/USP1/EZH2 in GSC11 cells markedly decreased cell proliferation, and reconstituted overexpression of ND-EZH2 almost fully reversed the effect of USP1 depletion (Fig. 6B; Supplementary Fig. S5B). In addition, expression of the catalytically inactive EZH2 mutant (EZH2 ΔSET) rescued the effect of USP1 depletion on GSC11 proliferation, although to a less extent compared with restoring the full-length EZH2 (Supplementary Fig. S5C). This result indicates that the noncatalytic function of EZH2 plays an important role in glioma cell proliferation, and the regulation of USP1 toward EZH2 is mediated by both EZH2′s catalytic-dependent and -independent functions.

We next explored to determine whether the effect of β-catenin/USP1/EZH2 on cell proliferation was caused by the changes in cell-cycle distribution. As we expected, Wnt3a overproduction in SW1783 cells significantly decreased the proportion of G₀–G₁ phase cells and increased G₂–M phase cells (Supplementary Fig. S5D). This effect was abolished by USP1 depletion, which was rescued by the reconstituted expression of ND-EZH2 (Supplementary Fig. S5D). Accordingly, depletion of USP1/EZH2 in GSC11 cells markedly induced G₀–G₁ phase arrest, and reconstituted overexpression of ND-EZH2 reversed the effect of USP1 depletion in cell-cycle distribution (Supplementary Fig. S5E).

To investigate the functions of the β-catenin–USP1-EZH2 axis in tumor formation in vivo, we intracranially injected glioma cells into nude mice. SW1783 cells, which derived from lower grade glioma (grade III astrocytoma), did not show tumor formation in nude mice (Fig. 6C; Supplementary Fig. S5F). However, overexpression of Wnt3a in the cells significantly promoted tumor growth, and the effect was largely reduced by USP1/EZH2 depletion (Fig. 6C; Supplementary Fig. S5F). Moreover, overexpression of ND-EZH2 further reversed the effects of USP1/EZH2 depletion in tumor formation (Fig. 6C). These results indicate that Wnt/β-catenin–USP1-EZH2 axis may play an important role in the malignant transition from glioma to glioblastoma. Accordingly, all mice injected with GSC11 control cells developed tumors with characteristic glioblastoma features. In contrast, depletion of β-catenin/USP1/EZH2 eliminated tumor formation, and the effect of USP1 depletion on tumor formation was rescued by the reconstituted expression of ND-EZH2 (Fig. 6D; Supplementary Fig. S5G). Furthermore, IHC staining of mice brain tumor tissues confirmed that β-catenin/USP1–induced EZH2 stabilization was retained in these tumors (Supplementary Fig. S5H and S5I).

We next evaluated the effect of β-catenin/USP1/EZH2 axis on the survival of glioma-bearing mice. Overexpression of Wnt3a in SW1783 cells shortened the survival time of the implanted mice compared with the control, whereas depletion of USP1/EZH2 eliminated this effect (Fig. 6E; Supplementary Fig. S5J). Notably, reconstituted expression of ND-EZH2 further reversed the effect of USP1 depletion (Fig. 6E). Moreover, depletion of β-catenin/USP1/EZH2 significantly extended the survival time of GSC11-glioblastoma–bearing mice, whereas reconstituted expression of ND-EZH2 fully reversed the effect of USP1 depletion on mouse survival (Fig. 6F; Supplementary Fig. S5K).

Considering the remarkable roles of Wnt3a-EZH2 on glioma cell proliferation and tumorigenesis, we further investigated the genes that were involved in WNT3A-EZH2 pathway using IPA analysis (Ingenuity Pathway Analysis, Qiagen). The 330 experiment-validated genes downstream WNT3A (Supplementary Fig. S6A and Excel S1, sheet 1) and 746 genes downstream EZH2 (Supplementary Fig. S6B and Excel S1, sheet 2) were built and grown based on the Ingenuity Knowledge Database. Sixty-five coregulated genes downstream WNT3A and EZH2 were further screened out by “comparison” algorithm of the IPA software (Supplementary Fig. S6C and Excel S1, sheet 3) and among them, 45 genes were associated with “cell proliferation of tumor cell lines” (Supplementary Fig. S6D and Excel S1, sheet 4). Taken together, these results consistently demonstrate that the β-catenin–USP1-EZH2 axis plays a critical role in glioma cell proliferation and tumorigenesis.

To determine the clinical relevance of our findings, we assessed the β-catenin/USP1/EZH2 protein expression in serial sections of 30 grade II and III astrocytoma and 50 glioblastoma specimens. Expression level of USP1 was significantly correlated with nuclear β-catenin and EZH2 levels, respectively (Fig. 7A and B, top, r = 0.77, P < 0.0001; bottom, r = 0.80, P < 0.0001). We next assessed whether the expression levels of nuclear β-catenin, USP1, and EZH2 correlated with the grade of glioma malignance in those specimens. We found that the expression levels of nuclear β-catenin, USP1, and EZH2 were significantly higher in glioblastoma than in lower-grade astrocytoma (Fig. 7C and D). Moreover, from the TCGA dataset, higher USP1 mRNA expression levels predicted worse survival outcomes in patients with glioma, and so did EZH2 and CTNNB1 (Fig. 7E). These results suggest that β-catenin–USP1–EZH2 axis is functional in human gliomas.

As a classical view, activation of the β-catenin signaling enables β-catenin's nuclear translocation to form a transactivation complex with the TCF/LEF1 family of transcription and activates downstream gene expression. In this study, we demonstrate a previously unrecognized role of β-catenin signaling, which stabilizes EZH2 and promotes the enrichment of H3K27me3 on target promoters, and thus mediates transcriptional gene silencing. Our study reveals a novel function for the dysregulated β-catenin signaling during tumorigenesis.

As a critical regulator of cell-cycle progression, stem cell maintenance, and tumorigenesis, the regulation of EZH2 expression and activity is intensively investigated. EZH2 mRNA expression is activated by the pRB–E2F pathway and c-Myc, but repressed by p53, leading to its overexpression in tumors (25–27). Moreover, EZH2 protein is phosphorylated by multiple protein kinases to regulate its H3K27 methyltransferase activity during cell-cycle progression and oncogenesis (28, 29). Recently, EZH2 protein has been shown to be degraded through the ubiquitin–proteasome pathway by multiple E3 ubiquitin ligases under different biological context (17–19). As a reverse procedure of ubiquitination, protein deubiquitination is a tightly controlled process catalyzed by deubiquitinating enzymes and critically regulates protein turnover (21, 30). During preparation of this manuscript, two other deubiquitinases, USP21 and ZRANB1, have been reported to stabilize EZH2 in different biology contexts. Chen and colleagues identify that USP21 can deubiquitinate and stabilize EZH2 to promote cell proliferation and metastasis in bladder carcinoma (31); Zhang and colleagues find that ZRANB1 is an EZH2 deubiquitinase and regulates EZH2 function in breast cancer through both catalytic activity–dependent and -independent manner (32). Herein, our study identifies USP1 as a novel deubiquitinase, which stabilizes EZH2 during glioma tumorigenesis, and demonstrates that the USP1-EZH2 axis plays a critical role in the malignant transformation of glioma cells.

A few studies have previously reported about the cross-talk between EZH2 and the Wnt/β-catenin signaling. On the one hand, EZH2 physically interacts with β-catenin and serves as a transcription coactivator to drive the expression of genes that are commonly targeted by Wnt signaling (33–35). On the other hand, EZH2 acts as the catalytic subunit of PRC2 that represses Wnt antagonist genes expression and thus causes hyperactivated β-catenin signaling (36). Our present study demonstrates that β-catenin signaling, in turn, induces EZH2 stabilization and promotes EZH2-mediated H3K27 trimethylation and gene silencing. Together, these findings thus indicate a β-catenin–EZH2 regulatory feedback loop, which may be a prevailing mechanism in oncogenesis.

The deubiquitinase USP1 plays a key role in cell-cycle progression and genome integrity (37, 38). Some recent studies have reported the oncogenic functions of USP1 in multiple cancers. USP1 deubiquitylates the inhibitors of DNA-binding (ID) proteins and promotes stem cell–like characteristics in osteosarcoma (39), whereas USP1 inhibitor pomozide promotes ID1 degradation and inhibits leukemic cells growth (40). Moreover, USP1-deficient mice are resistant to Ras-driven skin tumors by elevating Fancd2 ubiquitination (41). During the preparation of this manuscript, two additional studies have reported the roles of USP1 in glioblastoma stem cell maintenance and proneural glioma cell survival (42, 43). Our study here identifies the epigenetic modifier EZH2 as a new substrate of USP1. We find that overexpression of USP1 promotes EZH2-mediated repression of downstream tumor suppressor genes and glioblastoma tumorigenesis, which thus expands our understanding of the functional mechanism of USP1 in tumorigenesis.

In summary, our study demonstrates a β-catenin–USP1-EZH2 axis orchestrating the hyperactivated β-catenin signaling and EZH2-mediated epigenetic gene silencing, which represents a critical mechanism for glioma tumorigenesis. Thus, targeting the axis may provide a new therapeutic strategy for glioblastoma.

No potential conflicts of interest were disclosed.

Conception and design: L. Ma, K. Lin, A. Zhou

Development of methodology: L. Ma, K. Lin, T.T. Huang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L. Ma, K. Lin, G. Chang, C. Yue, Q. Guo, S. Zhang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Ma, K. Lin, Z. Jia

Writing, review, and/or revision of the manuscript: L. Ma, K. Lin, A. Zhou

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Ma, K. Lin, A. Zhou

Study supervision: A. Zhou, S. Huang

Other (provided reagents and conceptual advice): Y. Chen

We thank Dr. Jianhua Yang (Texas Children's Cancer Center, Houston, TX) for providing DUB-expressing plasmids, Dr. Mien-Chie Hung (MD Anderson Cancer Center, Houston, TX) for providing MYC-EZH2-K421R plasmid, and Dr. Xi He (Boston Children's Hospital, Boston, MA) for providing β-catenin^+/+ and β-catenin^−/− MEFs. This work was supported in part by US NIH grants R01CA182684, R01CA201327, R01CA220236, and P50 CA127001-06.

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.