An Allosteric PRC2 Inhibitor Targeting EED Suppresses Tumor Progression by Modulating the Immune Response

Polycomb repressive complex 2 (PRC2) plays important roles in regulating gene expression involved in cell differentiation and development (1). PRC2 consists of 4 core components: enhancer of zeste 2 or 1 (EZH2/1), embryonic ectoderm development (EED), suppressor of zeste 12 (SUZ12), and retinoblastoma-associated proteins 46 and 48 (RBAP46/RBAP48; refs. 2, 3). EZH2 is the methyltransferase catalytic subunit responsible for H3K27 methylation. EED enhances the enzymatic activity of PRC2 via binding with H3K27me3. SUZ12 interacts with all other subunits and contributes to the stability of the complex (4, 5). RBAP46/RBAP48 (also known as RBBP4/7) recognizes histone H3 and H4 (6–8), and might regulate the substrate specificity of PRC2 (9). Besides the canonical PRC2 core members, several cofactors were identified to modulate the PRC2 activity. The best characterized cofactors are Jumonji and AT-rich interaction domain 2 (JARID2) and adipocyte enhancer-binding protein 2 (AEBP2). Structural and functional analysis implicates that both JARID2 and AEBP2 mimic histone H3 tails to bind with PRC2, resulting in global stabilization of the PRC2 complex (5, 10–12) and synergistic stimulation of the PRC2 activity (5, 12–14).

PRC2 is the only known methyltransferase that can specifically mono-, di-, or trimethylate H3 at lysine 27. Only trimethylated H3K27 (H3K27me3) is tightly linked to gene silencing at specific loci through chromatin compaction (6, 7, 10, 15). Global H3K27me3 levels are mediated by 2 conceptually separate mechanisms^___PRC2 enzymatic activity and PRC2 recruitment (16). PRC2 activity is regulated by multiple factors, including allosteric activators described above, various histone-modifiers (H3K4me3, H3K36me2/3; refs. 17–19) and PRC2-interacting partners (DNA and RNA; refs. 16, 20, 21). PRC2 recruitment to chromatin was suggested to be governed by a combination of several molecules and gene-specific transcription factors, resulting in PRC2 enrichment in transcriptionally inactive loci.

EED harbors both a scaffold and an H3K27me3-binding functions. As a scaffolding protein, EED assembles and stabilizes the PRC2 complex. EED binding of H3K27me3 allosterically activates PRC2 and propagates H3K27 methylation in repressive chromatin for gene silencing (22, 23). Structurally, the interaction stimulates the folding of an unstructured region of EZH2 into an alpha helix. The newly formed EZH2 helix in turn stabilizes the SET-I helix, which is part of the substrate binding site of the EZH2 SET domain (4, 22, 24). Furthermore, recent studies indicated EED not only functions as a stimulator to enhance the activity of PRC2 itself, but also serves as an epigenetic exchange factor to coordinate the activities of PRC1 and PRC2. EED recruits PRC1 to H3K27me3 loci and enhances PRC1-mediated H2A ubiquitin E3 ligase activity to further maintain gene silence (25–27).

Dysregulation of PRC2 function is broadly associated with various cancers. Increased activity of PRC2 results in high global levels of H3K27me3, which are linked to different types of cancer, including lymphoma, prostate, breast, and myeloma cancers (28). The importance of PRC2 linked to tumorigenesis has been validated by disrupting the PRC2 enzymatic activity using a number of EZH2 inhibitors (29–31). Several SAM-competitive EZH2 inhibitors are currently in clinical trials (32, 33, NCT02860286, NCT03480646, NCT03603951, NCT02732275, NCT03460977). However, these inhibitors have imperfect pharmacological properties, such as short half-life, moderate to high clearance rate, and low permeability in cell-based assays (34). Furthermore, drug resistance is probably another issue. Thus, development of novel inhibitors against other components of PRC2 has gained attention, as demonstrated by the development of EED inhibitors (33, 35–38). One of EED inhibitors is currently in clinical trial (NCT02000651).

Here we report the discovery of a potent EED inhibitor, BR-001, which disrupts EED-H3K27me3 interaction. Co-crystal crystallography shows that BR-001 binds to the same pocket in EED as H3K27me3. BR-001 decreases H3K27 methylation and has potent antiproliferative activity in vitro. BR-001 exhibited excellent efficacy against Karpas422 and Pfeiffer xenograft models, efficiently repressed H3K27me3 in tumors. Additionally, BR-001 also showed anti-tumor effect in the CT26 syngeneic mouse model. The antitumor efficacy mainly relied on CD8⁺ T-cell–mediated immunity. Collectively, BR-001 exerts its antitumor activity through 2 mechanisms: direct inhibition of the activity of PRC2 and modulation of immune response in tumor microenvironment. Of note, modulation of the tumor immune microenvironment by EED inhibitor BR-001 probably depends on specific types of cancer.

All cell lines used in the experiments were purchased from Cobioer. All cells have been conducted Mycoplasma testing using MycoAlert Mycoplasma Detection Kit from Lonza and are grown in a 37°C incubator at 5% CO₂. Karpas 422 (from ECACC), WSU-DLCL2 and HPB-ALL (from DSMZ), Pfeiffer, SNU-16 (from ATCC), and CT26 cells (from KCLB) are cultured in RPMI1640 medium with 10% FBS. KATO III (from ATCC) and 697 cells (from DSMZ) are cultured in medium IMDM with 20% and 10% FBS, respectively. U87MG cells (from ATCC) are cultured in MEM medium with 10% FBS and NEAA plus NaP. LN229 cells (from ATCC) are cultured in medium DMEM with 10% FBS.

The potency of compounds was tested using AlphaScreen competition assay. Briefly, the compounds were 3-fold serial dilutions in DMSO and further diluted 133.3 folds using assay buffer (25 mmol/L HEPES, pH 8.0, 0.5% BSA, 0.02% Tween-20 and 50 mmol/L NaCl). Five microliters of the serially diluted compounds was transferred to each well of a 384-well plate (ProxiPlate 384 plus plate; PerkinElmer) containing 10 μL of 30 nmol/L His-tagged EED (1–441aa) protein and 37.5 nmol/L biotinylated H3K27me3 peptide (19–33aa). The reaction mixture was incubated at room temperature for 30 minutes. The 4× working solution (20 μg/mL) of AlphaScreen detection beads mix (PerkinElmer) was freshly prepared by mixing nickel chelate acceptor beads and streptavidin donor beads in a 1:1 ratio into the buffer described above. Five microliters of beads mixture was immediately added to the plate. The plate was then incubated in the dark at room temperature for 1 hour and was read on Spectramax i3 (Molecular Devices) at emission 570 nm and excitation 680 nm. The signal at 680 nm was used to quantify compound inhibition and normalized with DMSO and negative control (no EED). The data were then fit to a dose–response equation using the GraphPad Prism6 to get the IC₅₀ values.

Thermal shift assay was performed using a real-time PCR instrument from Bio-Rad. The 25 μL of reaction mixture containing 2.5 μg EED full-length protein, 50 to 100 μmol/L BR-001 and 1:1000 diluted SyproOrange dye in buffer (0.1 M HEPES, pH 7.5, and 150 mmol/L NaCl; triplicate per sample) was incubated at room temperature for 30 minutes. SyproOrange was purchased from Invitrogen and 5X working concentration in the reaction. Melting temperature (T_m) of EED protein was measured using the following program: 25°C for 2 minutes, 25°C for 30 seconds, and increase 1°C every 30 seconds till to 90°C. The T_m values were obtained from the midpoint of the transition.

The global H3K27me3 levels in cultured cells or tumor nodules were detected by ELISA assay. Briefly, 5,000 cells were seeded in each well of 96-well plates and treated with the compounds immediately. DMSO was used as a control. After 72 hours treatment, PBS-washed cells were lysed in 100 μL 0.4 N HCl buffer for 2 hours at 4°C with gently shaking. The cell lysate was then neutralized by adding 80 μL neutralization buffer consisting of 0.5 M Na₂HPO₄ (pH 12.5), protease inhibitor cocktail (1:100), and 2.5 mmol/L DTT, mixed well by agitating. Ten microliters (for H3K27me3 analysis) or 2.5 μL (for H3 analysis) of cell lysate was transferred to 2 Optiplate-384 HB plates (PerkinElmer), respectively. The final volume was adjusted to 50 μL with PBS. The assay plates were incubated at 4°C overnight with gently shaking for ELISA detection (triplicate per sample).

Tumor samples were homogenized in 0.4N HCl buffer. 20 to 30 mg tumor tissues in 400 μL of 0.4N HCl were smashed by JXFSTPRP-64 (JingXin) under conditions of 68 Hz, 120 seconds. After 1 hour incubation at 4°C, the tumor tissue homogenate was centrifuged for 15 minutes. One hundred microliters of supernatant was transferred to a new tube and neutralized with 80 μL neutralizing buffer as described above but without DTT. Total protein in supernatant was quantified by Pierce BCA protein Assay Kit (Thermo Fisher Scientific). 30 ng protein/well (for H3K27me3 analysis) or 7.5 ng protein/well (for H3 analysis) was transferred to 2 Optiplate-384 HB plates (PerkinElmer), respectively. The final volume was adjusted to 50 μL with PBS. The assay plates were incubated at 4°C overnight with gently shaking for ELISA detection (triplicate per sample).

To perform ELISA assay, the assay plates were washed with TBST buffer for 5 times, blocked with TBST containing 5% BSA at room temperature for 1 to 2 hours. Then 30 μL per well primary antibody (Supplementary Table S1) was added to the plates and incubated at room temperature for 1 hour. After 5 times of wash with TBST, 30 μL of the secondary antibody (Supplementary Table S1) was added to each well and incubated at room temperature for another 1 hour. After wash, 30 μL of enhanced chemiluminescence substrate (ECL; Yeason) was added to each well. The signal was detected using SpectraMax i3x (Molecular Devices). The data were then fit to a dose–response equation using the GraphPad Prism6 to get the IC₅₀ values.

A total of 10,000 cells per well were plated onto 96-well round bottom plates (Corning, #3799), and treated with BR-001 at the indicated concentrations. Every 3 to 4 days, viable cell number in DMSO control was detected using Vi-CELL (Beckman Coulter). The same density of cells was seeded back to a new plate. The split cell number for BR-001 treated groups was determined by the ratio calculated from DMSO control. At day 13, 40 μL of CellTiter-Glo reagent from Promega Corporation was added to the plates. The cell viability was detected by read luminescence signal via SpectraMax i3x (Molecular Devices). The IC₅₀ was calculated using GraphPad Prism6.

Hanging drop vapor diffusion method was used for crystallization, with the crystallization well containing 3.6 M NaCOOH, 16% glycerol, and 0.1M Bis-Tris, pH 6.0, and a drop with a 1:1 volume of EED protein and crystallization solution. The crystal was soaked with crystallization solution containing 0.5 mmol/L BR-001 for 24 hours. 30% glycerol was used as cryoprotectant for flash frozen in liquid nitrogen. Diffraction data were collected at beamline BL17U in Shanghai Synchrotron Radiation Facility (SSRF). The data set was processed using HKL3000.

The protein structure of EED in complex with ligand EED226 (PDB: 5WUK) was used as a search model for molecular replacement using Phaser. The ligand BR-001 were manually built in COOT and the structures was further refined by Phenix. The final structures were checked using the MolProbity server. The statistics of structure refinement and the quality of the final model were summarized in Supplementary Table S2. Protein structure figures were presented with PyMol.

RNA was isolated from the cultured cells using E.Z.N.A. Total RNA Kit I (Omega Bio-Tek, Cat# R6834-02) according to manufacturer's instructions. The cells were treated with DMSO or BR-001 at various concentrations for 2 days before total RNA extraction. The tumor tissues were homogenized in the buffer from kit using JXFSTPRP-64 under the same conditions as described above, then centrifuged at 4°C for 15min. Isolated RNA was quantified using the NanoDrop One (Thermo Scientific). Gene expression was analyzed using One Step RT-PCR Kit (Bio-Rad, Cat# 172551). 200 ∼ 500 ng total RNA was added to the reaction mix following the product protocol. Quantitative real-time PCR (qRT-PCR) was performed in a CFX384 Touch Real-Time Detection System (Bio-Rad). The qRT-PCR primers were listed in Supplementary Table S2. Relative quantified RNA was normalized with USF1 or GAPDH housekeeping gene.

Total protein from the cultured cells or tumor samples were extracted and subject to process western blotting. Isolation of protein from the cultured cells was done by lysing the cells using RIPA buffer (Thermo Fisher Scientific; Catalog No. 89901) supplemented with a protease inhibitor cocktail. Protein in tumor tissues was extracted by homogenizing. Briefly, 30 to 50 mg tumor tissue in 240 to 400 μL RIPA buffer was smashed using JXFSTPRP-64 under the same conditions as described previously. After the centrifuge, protein concentration in the supernatant was determined using a Pierce BCA Kit (Thermo Fisher Scientific).

Thirty micrograms of total protein was separated on 10% to 12% SDS-PAGE, then transferred to PVDF membrane (Bio-Rad, Catalog No. 1620239). The membrane was incubated with primary antibody (Supplementary Table S1) in blocking buffer (TBST with 5% non-fat powdered milk) at 4°C for overnight. Then the membrane was incubated with the secondary antibody (Supplementary Table S1) for 1 hour at room temperature after wash. The protein was visualized by enhanced chemiluminescence (Pierce ECL Western Blotting Substrate; Thermo Fisher Scientific) and detected using ChemiDoc Imaging Systems (Bio-Rad).

The study is compliant with all relevant ethical regulations regarding animal research. The efficacy studies shown in Fig. 4D and E was conducted in Crown Bioscience Inc., and the efficacy study shown in Fig. 5D was conducted in Pharmaron. All experiments conducted were performed in female NOD/SCID mice in the Association for Assessment and Accreditation of Laboratory Animal Care certificated facility. All the procedures and protocols related to animal handling, care, and the treatment in the study were approved by the Institutional Animal Care and Use Committee of Crown Bioscience or Pharmaron.

Flow cytometry single-cell suspensions of tumors were prepared and analyzed by Pharmaron. Additional information is available in the Supplementary Materials and Methods.

To inhibit the PRC2 activity, we designed EED-binding ligands that can disrupt the EED–H3K27me3 interaction. Using a recently published allosteric PRC2 inhibitor EED226 as a starting point (35), we performed scaffold hopping of the core structure followed by side chains optimization. Such medicinal chemistry effort led to a new class of pyrimidone compounds, exemplified by BR-001 (Fig. 1A). In a competition binding assay, BR-001 showed a dose-dependent displacement of H3K27me3 from EED with an IC₅₀ value of 4.5 nmol/L. As a negative control, an EZH2 inhibitor EPZ6438 did not show any displacement activity (Fig. 1B). Furthermore, BR-001 significantly increased the thermal stability of EED, as evidenced by 6 to 10°C increase in melting temperature (Fig. 1C) in thermal shift assay. BR-001 had no activity against 371 wild-type kinases (Supplementary Table S3). These results suggest that BR-001 is a selective EED binder to compete against H3K27me3 binding.

EED is a core component of PRC2, which catalyzes H3K27me3 formation. Targeting EED or EZH2 inhibition can cause similar effect on H3K27 methylation (36). We evaluated the effect of BR-001 on the cellular level of H3K27me3 using a cell-free ELISA assay. Karpas422 cells were treated with BR-001 for 72 hours and lysed in HCl. The levels of H3K27me3 in the cell lysate were measured. As shown in Fig. 1D, BR-001 significantly reduced the cellular H3K27me3 level and also exhibited a strong antiproliferative effect on the cells (Fig. 1E). Our results suggest that BR-001 effectively inhibits H3K27 methylation and proliferation of Karpas 422 cells.

We solved the co-crystal structure of EED in complex with BR-001 at a 2.0-Å resolution (Fig. 2A; Supplementary Table S4). Hydrogens were not added during the structure refinement and deuterium of BR-001 were not distinguishable in the density map. BR-001 bound to the induced aromatic cage of EED. Multiple interactions between EED and BR-001 are observed: R365 forms pi-cation interaction with benzene ring; the amide side chain of N194 forms multipolar interaction with fluorine and Y365, Y148, F97 form pi-pi stack with the central ring (Fig. 2B). Compared with the apo structure, cage residues undergo conformation change upon binding with BR-001, including Y365 and R367 (Fig. 2C). Together, BR-001 binds with EED in the aromatic cage by inducing local conformation changes.

Most DLBCL cell lines were reported to carry gain-of-function mutations of EZH2 and were phenotypically more sensitive to EZH2 inhibitors than the cells with wild-type EZH2. We tested BR-001 activity in 3 DLBCL cells with EZH2 Y641 mutations. In line with previous reports, BR-001 exhibited good potency against 3 cell lines (Fig. 3A and B). To assess whether BR-001 also has antiproliferation effects in cells with high-level expression of EZH2, we evaluated BR-001 activity against gastric, glioma, and lymphoblastic leukemia cell lines with high or low expression of EZH2. Most of the cell lines did not significantly respond to PRC2 inhibition, indicating no correlation between cellular EZH2 expression level and their phenotypic sensitivity to EED inhibitor, BR-001 in the tested cells.

A recent report demonstrated that MLL1 expression level and H3K27ac upregulation are positively correlation with resistance to EZH2 inhibition (39). In agreement with the report, our results showed that the cells that were sensitive to BR-001 expressed low level of MLL-1, whereas BR-001-resistant cells had relative high level of MLL-1 and H3K27Ac (Fig. 3C). KATO III was an exceptional cell line, which expressed MLL-1 at a high level and H3K27ac at a low or undetectable level, was sensitive to BR001. Moreover, no increase of H3K27ac was observed when KATO III and sensitive cells were treated with BR-001. In contrast, H3K27ac level was increased in the insensitive cells tested (Fig. 3D). The levels of H3K27me3 in all cell lines tested were decreased upon BR-001 treatment (Fig. 3D; Supplementary Fig. S1A and S1B). Taken together, the results suggest that H3K27ac upregulation mainly contribute to the resistance to PRC2 inhibition.

We further examined the gene expression changes in DLBCL cells. Six known PRC2 target genes were evaluated in both Karpas 422 and Pfeiffer cells that are sensitive to BR-001 (Fig. 4A). Dose-dependent induction of gene expression was observed in both cell lines (Fig. 4B and C). Overall, the changes of gene expression in Karpas 422 is much larger than that in Pfeiffer cells. The mechanism of variation of expression level of genes in the 2 cell lines will warrant further investigations. The results demonstrate that the global loss of H3K27me3 following inhibition of PRC2 by BR-001 is associated with transcriptional activation of PRC2 target genes.

PRC2 inactivation effectively inhibited DLBCL cells with EZH2 Y641 mutation in vitro and markedly suppressed the tumor growth of EZH2 mutant DLBCL xenografts in immunocompromised mice (35, 36). We chose 2 sensitive cell lines (Karpas422 and Pfeiffer) to establish xenograft models. 40 or 100 mg/kg twice daily of BR-001 were orally dosed to mice at day 14 postimplantation. The tumor volume was measured every 3 or 4 days. BR-001 showed a dose-dependent efficacy in both xenografts. 100 mg/kg doses of BR-001 induced 85% tumor growth inhibition (TGI) in Karpas 422 and 96% TGI in Pfeiffer models at day 36 posttreatment. The body weight in both models was not significantly affected (Fig. 4D and E; Supplementary Fig. S2A and S2B). Levels of H3K27me3 and PRC2 target genes in tumors were analyzed at the last day of treatment. BR-001 caused substantial decreases of H3K27me3 in dose-dependent manner (Fig. 4F and G) and upregulated the PRC2 target genes (Fig. 4H and I).

It has been reported that disrupting PRC2 function has immunomodulatory activity. Highly expressed PRC2 components was inversely associated with the expression of CD4, CD8, and subsequent production of Th1-type chemokines in human colon cancer (40). In melanoma cancer cells, PRC2 inhibition promoted expression of PRC2 target genes associated with tumor immunogenicity (41). To investigate whether BR-001 was able to upregulate immunogenicity-related genes in colorectal cancer cells, we examined the expression of the target genes and cell proliferation ability in murine CT26 cells that were exposed to BR-001. BR-001 had no antiproliferative effect on the CT26 cells, while a reference compound, 5-FU (principally as a thymidylate synthase inhibitor) showed activity against the cells (Fig. 5A). Next, we evaluated mRNA level of immune-related genes including the member of the major histocompatibility complex, chemokine and receptor. Interestingly, we found only the mRNA level of CXCL10 was greatly increased in treated cells compared with the DMSO-treated control cells (Fig. 5B). The fold change of CXCL10 gene expression in the 10 μmol/L BR-001 treated CT26 cells was similar to that in 2000 U IFNγ-stimulated cells (Fig. 5C).

CXCL10 is a chemoattractant for activated T cells and plays an important role in recruitment of effector T cells into tumor sites (41, 42). The above in vitro result suggests a hypothesis that BR-001 might be able to repress CT26 tumor growth in vivo through stimulating CXCL10 expression. To further test this hypothesis, we assessed antitumor efficacy of the EED inhibitor BR-001 in a CT26 syngeneic model. Treatment of the xenograft with BR-001 showed significant tumor repression with TGI value of 59.3% at a 30 mg/kg dose (Fig. 5D). The body weight was not significantly affected (Supplementary Fig. S2C). BR-001 caused substantial increase of CXCL10 at both mRNA and protein levels in the CT26 tumor, whereas EZH2 level remained the same (Fig. 5E and F). These results correlate well with our in vitro data. Furthermore, we found that the TNFα level was moderately increased in the BR-001-treated CT26 tumors, but decreased in the CT26 cells (Fig. 5E and B). The discrepancy between the CT26 tumor and CT26 cells is likely due to the difference between in vitro and in vivo environment.

Inhibition of PRC2 by BR-001 was sufficient to cause the CT26 tumor suppression. To investigate whether the combination of BR-001 with anti-PD-1 could further improve antitumor activity, we conducted the combination therapy in the CT26 xenograft model. Co-administration of 30 mg/kg (twice daily) of BR-001 and 10 mg/kg (twice a week) of anti-PD-1 did not further reduced the tumor size when compared with treatment with BR-001 as a single agent at the dose of 30 mg/kg (Fig. 5D). Thus, no synergy effect was observed in BR-001 and anti-PD-1 combination. The result is consistent with a previous report that EZH2 inhibition plus immunotherapy did not provide any advantage in MC-38 colon tumor (41).

To assess the mode of action of PRC2 inactivation-caused TGI, we analyzed tumor-infiltrating lymphocytes (TIL) from fresh tumor samples. Compared with vehicle control, BR-001 treatment significantly increased the counts of CD8⁺ T cells in TILs (Fig. 5G and H). Treg cell counts in TILs remained the same between treatment and control groups. Therefore, the ratio of CD8⁺ T cells to Treg cells in BR-001 treated groups was increased (Fig. 5I and J). The level of CD11b⁺ Gr-1⁺ MDSCs was also higher in the BR-001–treated tumor tissues (Supplementary Fig. S3A). As controls, natural killer cells and tumor-associated microphage and white blood cells (Supplementary Fig. S3B, S3C, and S3D) did not change. Total T cells and CD4⁺ T cells did not significantly affect by BR-001 treatment either (Supplementary Fig. S3E and S3F). Collectively, these results indicated that PRC2 inhibition by BR-001 has an immunomodulatory effect in CT26 colon carcinoma. Highly expressed CXCL-10 effectively recruited CD8⁺ T cells to tumor positions, resulting in increase of the ratio of CD8⁺ T cells to Treg cells. It should be noted that all tumor samples for gene, protein and TILs analyses were performed after the last day of BR-001 dosing.

Epigenetic deregulations are associated with aberrant transcription and gene function and consequently drive cancer development. Increased methylation activity of PRC2 has been frequently found in a wide variety of cancerous tissue types. In addition, accumulated evidence suggests that PRC2 also plays a somewhat antagonistic role in the antitumor immune response. PRC2 silences Th1-type chemokines to suppress CD8⁺ T cell migration. Upregulation of EZH2 or other PRC2 components confers resistance to immunotherapy (40, 41, 43–46). Therefore inhibition of PRC2 in combination with checkpoint inhibitors might be an attractive strategy for cancer therapy.

Here we reported the discovery of a potent, selective EED inhibitor, BR-001. BR-001 has good phenotypic activities in vitro and in vivo, and also exhibits anti-tumor efficacy in the CT26 syngeneic model. BR-001 treatment leads to higher expression levels of CXCL10 in CT26 cells and tumors. In support our studies, previous reports show that PRC2 components and H3K27me3-mediated the TH1-type chemokine gene silencing in ovarian and colon cells (40, 43). Thus removal of H3K27me3 increases CXCL-10 expression. Cytokine TNFα in CT26 tumor is also moderately upregulated upon BR-001 treatment. TNFα is a multifunctional cytokine playing key roles in inflammation and immunity. One function is to remodel the tumor microenvironment by promoting the activation of cytotoxic CD8⁺ T cells (47, 48). Our data show that TNFα strongly stimulates CXCL10 expression in CT26 cells (Fig. 5C), suggesting that TNFα plays a role in modulating immune response through triggering CXCL10 expression in CT26 carcinoma tumor.

The syngeneic study is encouraging and may provide an opportunity for immune-oncology therapy. In our study, BR-001 and anti-PD-1 combination did not provide any advantage over BR-001 alone in the CT26 xenograft (Fig. 5D). This is another evidence to show that PRC2 inhibition plus immunotherapy did not benefit for colon carcinoma. Synergistic antitumor activity is only observed in tumors that PRC2 members are upregulated upon immunotherapy. In contrast, the tumors that are bare or absent upregulation of PRC2 does not benefit from the addition of PRC2 blockade to immunotherapy (41, 49). Hence, the cancers characterized by increased PRC2 members after immunotherapy could be suitable for combination of EED/PRC2 inhibitors with immune checkpoint inhibitors.

Overall, our data suggested that BR-001 is a potent EED inhibitor with 2 antitumor mechanisms for therapy: direct inhibition by inactivation of PRC2 and modulation of the tumor microenvironment by increasing CD8⁺ T-cell tumor infiltration.

No potential conflicts of interest were disclosed.

Conception and design: H. Dong, Y. Chen, B. Zou

Development of methodology: S. Liu, X. Zhang, Y. Chen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): X. Zhang, S. Chen, L. Kang, Y. Chen, Y. Liu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H. Dong, S. Liu, X. Zhang, S. Chen, L. Kang, Y. Chen, Y. Liu, B. Zou

Writing, review, and/or revision of the manuscript: H. Dong, S. Liu, Y. Chen, H. Zhang, B. Zou

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Chen, Y. Chen, S. Ma, X. Fu

Study supervision: H. Dong, B. Zou

We would like to thank staff at SSRF for X-ray diffraction data collection. We are grateful to researchers at Crown Bioscience Inc. and Pharmaron who support our in vivo pharmacology studies and FACS analysis. We thank Joseph M. Salvino and reviewers for critical and constructive suggestions of the article.

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.