CBX2 suppresses interferon signaling to diminish tumor immunogenicity via a noncanonical corepressor complex

Significance

We identify that Chromobox 2 (CBX2), a crucial component of polycomb repressive complex (PRC), suppresses tumor immunogenicity and immunotherapy efficacy independent on PRC, but through a CBX2–RACK1–HDAC1 complex, which diminishes the interferon signaling and antigen presentation. Thus, our findings reveal a potential target and biomarker for tumor immunotherapy.

Abstract

Chromobox 2 (CBX2), a crucial component of the polycomb repressive complex (PRC), has been implicated in the development of various human cancers. However, its role in the regulation of tumor immunogenicity and immune evasion remains inadequately understood. In this study, we found that ablation of CBX2 led to tumor growth inhibition, activation of the tumor immune microenvironment, and enhanced therapeutic efficacy of anti-PD1 or adoptive T cell therapies by using murine syngeneic tumor models. By analysis of the CBX2-regulated transcriptional program coupled with mass spectrometry screening of CBX2-interacting proteins, we found that CBX2 suppresses interferon signaling independent of its function in the canonical PRC. Mechanistically, CBX2 directly interacts with RACK1 and facilitates the recruitment of HDAC1, which attenuates the H3K27ac modification on the promoter regions of interferon-stimulated genes, thereby suppressing interferon signaling. Consequently, CBX2 reduces tumor immunogenicity and enables immune evasion. Moreover, a high expression level of CBX2 is associated with immune suppressive tumor microenvironment and reduced efficacy of immunotherapy across various human cancer types. Our study identifies a noncanonical CBX2–RACK1–HDAC1 corepressor complex in suppression of tumor immunogenicity, thereby presenting a potential target and biomarker for tumor immunotherapy.

An immune-activated tumor microenvironment consistently predicts improved clinical outcomes and enhanced efficacy of immunotherapy across various human cancers (1–3). The activation of the tumor immune microenvironment is influenced by multiple factors, with tumor immunogenicity playing a pivotal role (4, 5). Tumor immunogenicity encompasses both antigenicity and adjuvanticity (6). Elevated tumor antigenicity results in increased exposure of tumor antigens through the antigen processing pathway, thereby facilitating recognition and elimination by immune cells, such as CD8+ T cells (7). On the other hand, the adjuvanticity relates to danger molecules that activate innate immune signaling pathways, such as the type I interferon pathway, subsequently attracting immune cells to infiltrate the tumor site (8, 9). Consequently, tumors with high immunogenicity are more likely to be monitored by the immune system, thereby enhancing the antitumor immune response to eradicate tumor cells (7). To evade immune surveillance, tumor cells often selectively silence or mutate these immunogenicity-associated pathways, thereby reducing their antigenicity or adjuvanticity through various mechanisms (10, 11). A recent study indicates that tumor cells restrict H3K4me3 and H3K36me3 modifications on the promoter regions of antigen presentation genes, thereby reducing their expression through the upregulation of mitochondrial electron flow (12). In breast cancer cells, the expression levels of MHC-I genes remain low due to high methylation, thus evading cytotoxic T cell attacks (13). Furthermore, our recent work reveals that RAD21 collaborates with the NuRD complex to deacetylate histones of interferon response genes and suppress gene expression, thereby promoting immune evasion in ovarian cancer (14). Therefore, epigenetic reprogramming serves as a primary mechanism for tumor cells to diminish their immunogenicity.

The polycomb repressive complex (PRC) is implicated in the formation and progression of various cancers, primarily due to its transcription suppression activity through histone methyltransferase or ubiquitination. This complex is typically expressed at a high level in several cancer types (15, 16). The PRC comprises various subunits, including the enhancer of zeste homolog 2 (EZH2) and CBX2, among others (17). EZH2, the enzymatic subunit of the PRC, catalyzes the trimethylation of histone H3 at lysine 27 (H3K27me3), leading to the transcriptional repression of tumor suppressor genes (18, 19), E-cadherin (20), and FOXC1 (21). This repression subsequently promotes tumor progression. Recent studies have demonstrated that EZH2 epigenetically silences Th1-type chemokines, such as CXCL9 and CXCL10, in tumor cells, thereby restricting tumor immunity and the efficacy of immunotherapy (22–24). CBX2 recognizes and binds to the H3K27me modification, facilitating the recruitment of the PRC to target genes, which leads to chromatin compaction and transcriptional repression (25). Additionally, CBX2 promotes cancer progression through the activation of PI3K/AKT signaling, induction of genomic instability, and activation of the Wnt pathway, etc. (25–27). A recent study reports that CBX2 increases M2-like macrophages and decreases its phagocytosis activity in high-grade serous carcinoma (28). SUZ12 plays a critical role in stabilizing the PRC, while RBBP7 is essential for binding to nucleosomes, thereby promoting tumorigenesis (15, 29). Although the role of EZH2 in regulating the tumor immune microenvironment is well documented, the functions of other PRC components remain largely unknown. Moreover, it is unclear whether there are PRC-independent functions of other components.

In this study, we demonstrated that ablation of CBX2 in tumor cells inhibited tumor growth, increased infiltration and activation of immune cells in tumor tissues, and enhanced the therapeutic efficacy of anti-PD1 monoclonal antibody (mAb) or adoptive T cell transfer immunotherapy in the B16 OVA tumor model. Consistently, a higher infiltration and activation of immune cells were observed in Skin cutaneous melanoma (SKCM) patients with low CBX2 expression. We further elucidate that CBX2 suppresses the expression of interferon signaling-associated genes and the antigen processing pathway in tumor cells, consequently reducing their immunogenicity. Specifically, CBX2 associates with RACK1 and forms a repressor complex with HDAC1. This complex attenuates active transcription marker H3K27ac modification on the promoter regions of interferon-stimulated genes (ISGs) and antigen presentation genes, thereby decreasing their expression. This mechanism operates independently of the canonical PRC function. Moreover, we also demonstrate that CBX2 expression levels exhibit an inverse correlation with the tumor immune microenvironment and immunotherapy response across various human cancer types. Collectively, our findings reveal a unique role for CBX2 in the suppression of tumor immunogenicity and suggest its potential as a target and biomarker for precise cancer immunotherapy.

Results

CBX2 Suppresses Antitumor Immunity in both Mouse Tumor Models and Human Cancer Patients.

CBX2 has been implicated in the development of various human cancers. To determine its role in the regulation of tumor immunogenicity and immune evasion, we initially validated its tumor-promoting phenotype using a mouse tumor model. Specifically, we inoculated B16 OVA cells (a mouse melanoma cell line overexpressing the ovalbumin antigen) that either expressed sgCbx2 (CBX2 knockout cells) or a scramble control (wild-type cells, sgSCR) into a syngeneic mouse tumor model. The results showed that CBX2 depletion significantly inhibited tumor growth in immune-competent C57BL/6 mice bearing B16 OVA tumors (Fig. 1 A–C and SI Appendix, Fig. S1A). Consistently, similar results were observed in the MC38 (mouse colon cancer cell) tumor model (SI Appendix, Fig. S1 A–D). Notably, there were no significant differences in tumor growth or tumor weight when the same type of tumor cells was inoculated into T cell–deficient nude mice (SI Appendix, Fig. S1 E and F). Furthermore, flow cytometry analysis revealed a significant increase in levels of tumor-infiltrating immune cells, such as CD8+ T cells, but not CD4+ T cells, in mice bearing Cbx2-KO B16 OVA tumors comparing to that of sgSCR B16 OVA tumors (Fig. 1 D–F). The levels of T cell activation markers IFNγ, and GZMB in tumor-infiltrating CD8+ T cells were significantly elevated in sgCbx2 B16 OVA tumors compared to those in sgSCR tumors (Fig. 1 G–J). A similar pattern of immune infiltration was observed in Cbx2-knockout MC38 tumors comparing to that of sgSCR expressing MC38 tumors (SI Appendix, Fig. S1G). These findings suggest that the depletion of CBX2 enhances immune cell infiltration and activates the tumor immune microenvironment. Consistent with our observations in the mouse models, Gene Set Enrichment Analysis (GSEA) of Gene Ontology (GO) terms applied to transcriptional data from skin cutaneous melanoma (SKCM) patient samples in The Cancer Genome Atlas (TCGA) also reveal that antitumor immune signaling pathways, such as T cell-mediated cytotoxicity and antigen presentation, are enriched in the CBX2 low expression group (SI Appendix, Fig. S1 H–J). Furthermore, the immune score, immune cell infiltration, and T cell cytotoxic signaling in most SKCM patient tumor tissues exhibit a negative correlation with CBX2 expression (Fig. 1 K–M). Collectively, these data suggest that CBX2 suppresses tumor immunity in both human cancer patients and mouse tumor models.

Fig. 1.

CBX2 ablation inhibits tumor growth and activates the tumor immune microenvironment. (A–C) B16 OVA tumor cells with sgSCR, sgCbx2-1, and sgCbx2-2 were inoculated on B6 mice (0.5 million/mouse), then the tumor volume was measured and recorded, and the tumor weight and photographs were recorded on day 14 (n = 10). (D–J) B16 OVA tumor cells with sgSCR, sgCbx2-1, and sgCbx2-2 were inoculated on B6 mice (0.5 million/mouse), and the proportion of immune cells in B16 OVA tumor tissue was analyzed on day 16 (n = 6). (K) The correlation between CBX2 expression levels and immune scores, as well as immune cell infiltration, in SKCM samples. (L) The GO-Term of enrichment negative regulated pathways in patients with SKCM from TCGA cohorts based on the CBX2 mRNA level. (M) GSEA enriches the T cytotoxicity and inflammatory signaling in patients with SKCM from TCGA cohorts based on the CBX2 mRNA level. Tumor volume is shown as mean ± SD. ns = no significance, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, in two-way ANOVA with Bonferroni’s posttest (A), in one-way ANOVA with Bonferroni’s posttest (B and D–J).

CBX2 Represses Interferon Signaling and Antigen Presentation Pathways.

Our in vivo results indicate that CBX2 suppresses antitumor immunity. To elucidate the underlying mechanisms, we conducted RNA sequencing (RNA-Seq) analysis on Cbx2-KD B16 OVA cells. Clustering of 1,387 differentially expressed genes (DEGs) (P < 0.05, fold change > 2) revealed that the majority of DEGs were significantly up-regulated following the loss of CBX2, suggesting that CBX2 may function as a repressor of gene expression (SI Appendix, Fig. S2A). Notably, GSEA demonstrated an enrichment of up-regulated genes in pathways related to innate immunity, including antiviral signaling and IFN signaling pathways (Fig. 2 A–C). Among the genes regulated by the innate immune system, ISGs exhibited the most significant alterations following CBX2 depletion (Fig. 2D). Consistently, pathways related to innate immunity, particularly interferon (IFN) signaling, were up-regulated in SKCM patients with low CBX2 expression (Fig. 2E and SI Appendix, Fig. S2B). Moreover, compared with other PRC components, the CBX2 expression is notably inversely correlated with IFNa signaling across multiple cancer types (SI Appendix, Fig. S2C). qRT-PCR analysis further validated that Cbx2 knockdown, mediated by siRNA, led to increased expression of a subset of ISGs, including Cxcl9, Cxcl10, and Ifit3 (Fig. 2F and SI Appendix, Fig. S2D). This upregulation of ISG expression was confirmed by knocking down Cbx2 with shRNA in B16 OVA and MC38 cells (SI Appendix, Fig. S2 E and F). Accordingly, the expression levels of ISGs such as chemokines showed inverse correlation with CBX2 expression levels in SKCM patients (SI Appendix, Fig. S2G). In addition, we observed that CD8+ T cells migrated more efficiently toward the culture medium from IFNγ-pretreated shCbx2 B16 OVA cells in an in vitro transwell assay to that of the culture medium from control B16 OVA cells (Fig. 2G). GSEA also indicated that the antigen processing pathway was up-regulated in Cbx2-knockdown B16 OVA cells (Fig. 2H). Using qRT-PCR and flow cytometry analysis, we confirmed that siRNA-mediated Cbx2 knockdown increased the expression of B2m and the MHC-I complex (Fig. 2 I and J). Moreover, GSEA further revealed an enrichment of up-regulated genes involved in antigen processing and presentation in cancer patients with low CBX2 expression (SI Appendix, Fig. S2H). Collectively, our findings indicate that CBX2 suppresses the interferon signaling pathway and the antigen processing pathway, thereby reducing tumor immunogenicity.

Fig. 2.

CBX2 suppresses I-IFN signaling, antigen presentation, tumor immunogenicity, and immunotherapy response. (A) GO Term enrichment pathways in B16 OVA cells with Cbx2-KD. (B and C) GSEA enriches the innate immune signaling and IFNB1 signaling in B16 OVA cells with Cbx2-KD. (D) Heatmap of the ISGs mRNA in B16 OVA cells with control and Cbx2-KD. Colors from blue to red indicate the expression level from low to high. (E) GSEA enriches IFNβ signaling in patients with SKCM from TCGA cohorts based on the CBX2 mRNA level. (F) The mRNA expression of ISGs in B16 OVA cells with control and Cbx2-KD after IFNγ (10 ng/mL) treatment was measured by qRT-PCR. 18S rRNA was the housekeeping gene control. (G) In vitro transwell measured the migration of CD8+ T cells toward supernatant from B16 OVA cells with control and Cbx2-KD after IFNγ (10 ng/mL) treatment. (H) GSEA enriches the antigen presentation signaling in B16 OVA cells with Cbx2-KD. (I and J). The mRNA expression of B2m or MHC-I expression in B16 OVA cells with control and Cbx2-KD after IFNγ treatment was measured by RT-qPCR or flow cytometry. 18S rRNA was the housekeeping gene control. (K) B16 OVA tumor cells with control and Cbx2-KD were pretreated with IFNγ (10 ng/mL) for 24 h and then cocultured with OT-I cells for additional 24 h. The production of IL2 and IFNγ in the supernatant from OT-I T cells after cocultured was measured by the ELISA. (L) The cells were treated as K described and then cocultured with B3Z cells for additional 24 h. The LacZ activity of B3Z T cells after cocultured was measured. (M) The cells were treated as L described. The LacZ activity of B3Z T cells after cocultured was measured. (N) B16 OVA tumor cells with control and Cbx2-KD were pretreated with IFNγ (1 ng/mL) for 24 h and then cocultured with activated OT-I cells for additional 24 h. The live cells were stained by crystal violet. (O) Tumor growth curves of shSCR and shCbx2 B16 OVA tumors received PBS vehicle, anti-PD1 mAb, or OT-I treatment. Anti-PD1 (5 mg/kg) was i.p. administered on days 6, 9, and 12 after tumor inoculation. OT-I cells (2 million/mouse) were i.v. administrated on day 10 post tumor inoculation. n = 6 for each group. Tumor volume is shown as mean ± SD. ns = no significance, *P < 0.05, **P < 0.01, ***P < 0.001, in one-way ANOVA with Bonferroni’s posttest (F, G, and I–N), in two-way ANOVA with Bonferroni’s posttest (O).

CBX2 Ablation Induces T Cell Activation and Enhances Tumor Responsiveness to Immunotherapy.

Given the positive correlation between IFN signaling activity and the antigen processing pathway with tumor immunogenicity, we subsequently investigated the potential association of CBX2 with tumor immunogenicity and T cell activation. To this end, we employed a coculture system utilizing B16 OVA cells and OVA-specific primary CD8+ T cells from OT-I mice, as well as a CD8 T cell hybridoma named B3Z, which carries an IL-2 promoter-driven β-galactosidase (LacZ) reporter. The levels of the T cell–derived cytokines IL-2 and IFNγ in the supernatant were significantly elevated in primary OT-I T cells when cocultured with B16 OVA cells treated with Cbx2-targeted siRNA or shRNA, compared to that of T cells cocultured with B16 OVA cells treated with scramble control siRNA or shRNA (Fig. 2K and SI Appendix, Fig. S2 I and J). Correspondingly, CBX2 depletion markedly enhanced IL-2 promoter activity reflected by the significantly increased LacZ activity (Fig. 2L). Furthermore, the proportion of CD8+ T cells expressing the effector molecules such as CD69, IFNγ, and TNFα was also increased in Cbx2 shRNA–transduced B16 OVA cells following coculture (SI Appendix, Fig. S2K). Furthermore, the reexpression of Cbx2 in siRNA-expressing B16 OVA cells mitigated the heightened activation of cocultured T cells (Fig. 2M). Similarly, CBX2 overexpression in B16 OVA cells also resulted in a reduction of cocultured T cell activation (SI Appendix, Fig. S2L). Subsequently, we investigated the cytotoxic effects of OT-I cells over B16 OVA cells within a coculture system. Following coculture with activated OT-I T cells, the number of live tumor cells, stained with gentian violet, was significantly reduced to a greater extent in Cbx2-KD cells upon IFNγ stimulation in B16 OVA cells (Fig. 2N and SI Appendix, Fig. S2M). Finally, we administered anti–PD1 monoclonal antibody (mAb) or adoptive transfer of activated OT-I T cells to mice bearing shCbx2 or scramble B16 OVA tumors. The results demonstrated that knockdown (KD) of Cbx2 in tumor cells significantly enhanced the antitumor efficacy of both anti-PD-1 mAb and OT-I T cells, as evidenced by significantly reduced tumor growth (Fig. 2O). Collectively, these findings indicate that elevated expression of CBX2 in tumor cells suppresses T cell activation and diminishes tumor response to immunotherapy.

CBX2 Represses ISGs by Attenuating the H3K27ac Level.

Previous studies have indicated that the PRC, particularly EZH2, modulates tumor immunogenicity by epigenetically silencing the interferon signaling pathway through increasing H3K27 trimethylation (23). This observation led us to investigate whether CBX2 is involved in the regulation of H3K27 trimethylation. Unexpectedly, H3K27 trimethylation levels remained unchanged in CBX2-depleted B16 OVA or MC38 cells compared to that of control cells, indicating a methylation-independent mechanism by which CBX2 regulates gene expression (Fig. 3A and SI Appendix, Fig. S3A). CBX2 depletion still enhanced MHC-I expression under GSK343 (an EZH2-specific inhibitor) treatment condition (SI Appendix, Fig. S3 B and C), suggesting a regulatory mechanism distinct from the EZH2-centered PRC2 complex. Notably, we observed a significant increase in the H3K27 acetylation (H3K27ac) level in CBX2-depleted B16 OVA and MC38 cells compared to control cells (Fig. 3A and SI Appendix, Fig. S3A). Moreover, CBX2 depletion did not further enhance H3K27 acetylation or MHC-I expression under TSA (a pan-HDAC inhibitor) treatment condition (SI Appendix, Fig. S3 D and E), indicating that CBX2 regulates MHC-I expression by suppressing H3K27 acetylation. In addition, CUT&Tag analysis of H3K27 acetylation revealed a significant increase in the H3K27 acetylation binding motif in Cbx2-KD B16 OVA cells (Fig. 3B and SI Appendix, Fig. S3 F and G). The majority of the increased binding sites were located around promoter regions, whereas the decreased peaks were found in intronic regions, suggesting a potential role of CBX2 in gene transcription regulation (Fig. 3C and SI Appendix, Fig. S3H). Furthermore, KEGG pathway analysis of these increased peaks indicated significant enrichment in signal transduction, viral infection, and immune system pathways (Fig. 3D). GO analysis further revealed that the signaling pathways associated with Type I Interferon production and immune activation were significantly enriched following the loss of Cbx2, corroborating our RNA-seq results (Fig. 3E). Additionally, using the Integrative Genomics Viewer, we observed a marked increase in the mRNA expression peaks of ISGs such as Ddx3x, Stat1, Irf9, Oasl1, and Psme1/2 in Cbx2-knockdown cells. Such increase was accompanied by elevated binding peaks of H3K27 acetylation (Fig. 3F and SI Appendix, Fig. S3I). Using ChIP-qPCR and CUT&Tag qPCR analyses, we further confirmed that Cbx2 knockdown increased the H3K27 acetylation level at the promoters of Ddx58, B2m, Isg15, Cxcl9, and Cxcl10 (Fig. 3G and SI Appendix, Fig. S3 J and K). Collectively, these findings suggest that CBX2 represses ISG expression by attenuating H3K27ac rather than increasing the H3K27me3 modification level.

Fig. 3.

CBX2 decreases the H3K27ac in the promoter regions of ISGs. (A) The expression of H3K27ac, H3K27me3, and CBX2 in control and Cbx2-KD B16 OVA cells was detected by WB. β-actin was used as the loading control. (B) Heatmap of the H3K27ac CUT&Tag–seq signal peaks in control and Cbx2-KD B16 OVA cells. (C) Genomic distribution of increased H3K27ac peaks in Cbx2-KD B16 OVA cells. (D) KEGG enrichment of increased H3K27ac peaks in Cbx2-KD B16 OVA cells. (E) GO enrichment of increased H3K27ac peaks in Cbx2-KD B16 OVA cells. (F) Genome browser tracks of H3K27ac CUT&Tag -Seq and RNA-Seq at genomic loci of Ddx3x, Stat1, and Irf9 in control or Cbx2-KD B16 OVA cells. (G) ChIP-qPCR analysis of H3K27ac occupancy at genomic loci of Cxcl9 and Cxcl10 in Cbx2-KD and control B16 OVA cells mediated by shRNA. ns = no significance, ****P < 0.0001, in one-way ANOVA with Bonferroni’s posttest (G).

CBX2 Interacts with RACK1 to Attenuate the H3K27ac Level.

To elucidate the mechanism by which CBX2 reduces the H3K27ac level, we conducted a mass spectrometry-based proteomic analysis using anti-CBX2-immunoprecipitated proteins. The analysis revealed several potential CBX2-interacting proteins, including cytosol cytoskeleton tubulin family, ribonucleoprotein RPS family, core components of the PRC, such as RBBP4, and scaffold protein RACK1(SI Appendix, Fig. S4A). Based on the ranking, function, and subcellular localization of these identified proteins, and recently described antiviral function of RACK1 (30), we focused on RACK1 as a potential partner for CBX2 in regulating H3K27 acetylation (Fig. 4A). RACK1, a member of the tryptophan-aspartate repeat (WD-repeat) family of proteins, is known to interact with the ribosomal machinery, various cell surface receptors, and nuclear proteins (31). The coimmunoprecipitation assay demonstrated a robust interaction between FLAG-tagged CBX2 and HA-tagged RACK1 when coexpressed in 293 T cells (Fig. 4B). Furthermore, we confirmed the physical interaction between endogenous CBX2 and RACK1 proteins in both B16 OVA and MC38 cells (Fig. 4C and SI Appendix, Fig. S4B). Colocalization of CBX2 and RACK1 was observed within the nuclei of both B16 OVA and MC38 cells by confocal immunofluorescence microscopy (Fig. 4D and SI Appendix, Fig. S4C). These findings suggest that CBX2 interacts with RACK1 to form a complex. Moreover, knockout of Rack1 by sgRNA resulted in an increased expression of a subset of ISGs, including Cxcl9, Cxcl10, Cxcl11, and Ifit1 (Fig. 4E and SI Appendix, Fig. S4 D–F). Knockout or knockdown of Rack1 in tumor cells also significantly enhanced IL-2 promoter activity in cocultured T cells (Fig. 4F and SI Appendix, Fig. S4G). Moreover, RACK1 depletion significantly inhibited tumor growth in immune-competent B6 mice bearing B16 OVA tumors (Fig. 4G). Consistently, a significant increase in the H3K27ac level was observed in Rack1-knockout B16 OVA cells (Fig. 4H). In addition, ChIP-seq analysis of H3K27 acetylation revealed an increase in the H3K27ac binding motif in Rack1-KO B16 OVA cells (Fig. 4I). Integrative analysis of the CUT&Tag assay from Cbx2 KD and ChIP-seq from Rack1 KO B16 OVA cells revealed that 2,736 genes were potential direct targets of the CBX2–RACK1 complex (SI Appendix, Fig. S4H). Notably, GO analysis showed enrichment of up-regulated genes in immune-related pathways, including immune response-activating signaling and leukocyte chemotaxis signaling (Fig. 4J). We also observed an increase of binding peaks of H3K27ac on promoters of ISGs such as Stat1, Irf9, and Ddx3x (Fig. 4K and SI Appendix, Fig. S4I). Furthermore, ChIP-qPCR analysis revealed a significantly increased occupancy of H3K27ac on the promoters of Cxcl9 and Cxcl10 in Rack1 KO B16 OVA cells (SI Appendix, Fig. S4J). Additionally, RACK1 expression was higher in tumor tissues compared to normal tissues across various human cancer types (SI Appendix, Fig. S4K). The expression levels of interferon-associated chemokines such as CXCL9 and CXCL10 were inversely correlated with RACK1 expression in SKCM patients (SI Appendix, Fig. S4L). Additionally, the expression level of RACK1 showed an inverse correlation with the Immune Score in SKCM patient tissues (SI Appendix, Fig. S4M). To further elucidate the core interaction domain between RACK1 and CBX2, we constructed various CBX2 domain deletion mutants. The results indicated that the amino acid 1 to 60 domain of CBX2 is crucial for its interaction with RACK1 in overexpressed 293 T cells or B16 OVA cells with Flag-CBX2 expression (Fig. 4 L and M). Moreover, only the wild-type CBX2, but not the 1 to 60 domain deletion mutant, was able to rescue shCbx2-expressing B16 OVA cells, thereby attenuating the increased expression of Cxcl9 and the activation of cocultured T cells (Fig. 4 N–P). Collectively, these findings indicate that CBX2 interacts with RACK1 to modulate tumor immunogenicity.

Fig. 4.

CBX2 interacts with RACK1 to trigger the H3K27ac modification. (A) Mass spectrometry detects the peptide from RACK1 in anti-CBX2 Ab coimmunoprecipitation proteins. (B and C) Co-IP and immunoblot analysis of the interaction of CBX2 and RACK1 in extracts of HEK293T cells transfected with HA-RACK1 and Flag-CBX2 (B) or in extracts of B16 OVA cells (C). (D) Flag-CBX2 expressed B16 OVA cells were stained with anti-Flag (green) and anti-RACK1 (red) antibodies to detect CBX2 and RACK1 via confocal microscopy. (Scale bars, 5 μm.) (E) The mRNA expression of Cxcl9 and Cxcl10 in B16 OVA cells with control and Rack1-KO after IFNγ (10 ng/mL) treatment was measured by qRT-PCR. 18S rRNA was the housekeeping gene control. (F) B16 OVA tumor cells with control and Rack1-KO were pretreated with IFNγ for 24 h and then cocultured with B3Z cells for an additional 24 h. LacZ activity of B3Z T cells after cocultured was measured. (G) B16 OVA cells expressing sgSCR or sgRack1 were subcutaneous inoculated on B6 mice (0.5 million/mouse); then, the tumor volume was measured and recorded. n = 6 for each group. (H) The expression of H3K27ac, H3K27me3, and RACK1 in control or Rack1-KO B16 OVA tumor cells was detected by WB. β-actin was used as the loading control. (I) Heatmap of the H3K27ac ChIP–seq signal peaks in B16 OVA cells with control or Rack1-KO. (J) GO enrichment of overlapped increased H3K27ac peaks in Rack1-KO and Cbx2-KD B16 OVA cells. (K) Genome browser tracks of H3K27ac ChIP-Seq at genomic loci of Stat1 and Irf9 in control or Rack1-KO B16 OVA cells. (L and M) Co-IP and immunoblot analysis of the interaction of CBX2 and RACK1 in extracts of HEK293T cells transfected with HA-RACK1 and Flag-CBX2 (L) or in extracts of B16 OVA cells transfected with Flag-CBX2 (M). (N) The expression of CBX2 truncation rescued in B16 OVA cells with endogenous Cbx2-KD was detected by WB. β-actin was used as the loading control. (O and P) Measure the expression of Cxcl9 in B16 OVA cells or LacZ activity of B3Z T cells after cocultured with B16 OVA cells after IFNγ (10 ng/mL) treatment. Tumor volume is shown as mean ± SD. ns = no significance, *P < 0.05, **P < 0.01, ***P < 0.001, in one-way ANOVA with Bonferroni’s posttest (E, F, O, and P), in two-way ANOVA with Bonferroni’s posttest (G).

RACK1 Functions as an Essential Scaffold Facilitating the Interaction between CBX2 and HDAC1, thereby Reducing the H3K27ac Level.

Given the observed increase in H3K27ac levels in both RACK1- and CBX2-ablated cells, we further investigated the regulatory role of the CBX2–RACK1 complex on this histone modification. Previous studies have identified the histone deacetylase (HDAC) family proteins as the key regulators of H3K27 acetylation (32). Our prior findings also indicated that Cbx2 knockdown did not further elevate H3K27 acetylation in the presence of TSA treatment (SI Appendix, Fig. S3D). Additionally, a recent work has demonstrated that RACK1 interacted with and stabilized HDAC1 protein (32). Based on this body of evidence, we first confirmed the nuclear localization of CBX2, RACK1, and HDAC1 proteins (SI Appendix, Fig. S5A), and further validated the protein–protein interaction between RACK1 and HDAC1 in B16 OVA cells (Fig. 5A). However, the expression levels of HDAC1 and RACK1 remain unchanged in Cbx2 knockdown cells compared to that of control B16 OVA cells (Fig. 5B). Moreover, CBX2 ablation did not affect the interaction between HDAC1 and RACK1 (Fig. 5C). Notably, CBX2 physically interacted with HDAC1 in both B16 OVA and MC38 cell lines (Fig. 5D and SI Appendix, Fig. S5B). Furthermore, confocal immunofluorescence microscopy analysis revealed a colocalization of CBX2 and HDAC1 within the nuclei of both B16 OVA and MC38 cells (Fig. 5E and SI Appendix, Fig. S5C). More importantly, the interaction between CBX2 and HDAC1 was abolished in Rack1 knockout cells, suggesting that RACK1 serves as an essential scaffold adaptor for the CBX2–HDAC1 interaction (Fig. 5F and SI Appendix, Fig. S5D). Strikingly, chromatin binding of both HDAC1 and RACK1 was significantly reduced in Cbx2 knockdown cells (Fig. 5G). A consistent reduction in HDAC1 occupancy at the promoters of ISGs was observed in Cbx2 knockdown cells, as determined by ChIP-qPCR analysis (Fig. 5H and SI Appendix, Fig. S5E). Taken together, these findings demonstrate that RACK1 is indispensable for the CBX2–HDAC1 interaction, which is crucial for the regulation of H3K27ac and gene expression levels (Fig. 5I).

Fig. 5.

RACK1 works as the essential scaffold for the CBX2–HDAC1 complex to perform deacetylation. (A) Co-IP and immunoblot analysis of the interaction of HDAC1 and RACK1 in extracts of B16 OVA cells. (B) The expression of HDAC1, CBX2, and RACK1 in control and Cbx2-KD B16 OVA tumor cells was detected by WB. β-actin was used as the loading control. (C) Co-IP and immunoblot analysis of the interaction of the HDAC1 and RACK1 in extracts of Cbx2-KD or control B16 OVA cells. (D) Co-IP and immunoblot analysis of the interaction of HDAC1 and CBX2 in extracts of B16 OVA cells. (E) Flag-CBX2 expressed B16 OVA cells were stained with anti-Flag (green) and anti-HDAC1 (red) antibodies to detect CBX2 and HDAC1 via confocal microscopy. (Scale bars, 10 μm.) (F) Co-IP and immunoblot analysis of the interaction of the HDAC1 and CBX2 in extracts of Rack1-KO or control B16 OVA cells. (G) The distribution of HDAC1, CBX2, and RACK1 in control and Cbx2-KD B16 OVA tumor cells was detected by WB. β-actin was used as the whole cell loading control. GAPDH was used as cytosol protein control, and Histone H3 was used as chromatin-binding protein control. (H) ChIP-qPCR analysis of HDAC1 occupancy at genomic loci of Cxcl9 and Cxcl10 in Cbx2-KD and control B16 OVA cells mediated by shRNA. (I) Schematic model for the role of CBX2 in suppressing tumor immunogenicity. **P < 0.01, ***P < 0.001, in one-way ANOVA with Bonferroni’s posttest (H).

CBX2 Exhibits an Inverse Correlation with the Tumor Immune Microenvironment Markers and the Efficacy of Immunotherapy.

To validate the clinical relevance of our findings, we conducted a systematic assessment of the correlation between the expression level of CBX2 and other PRC components with tumor immune microenvironment markers, such as the Tertiary lymphoid structure (TLS) signature score, CD8⁺ T cell infiltration, IFNγ response signaling, and Tumor-Associated Macrophage (TAM) M2 infiltration, across various cancer types by analyzing TCGA data and TIGER database (33). Compared to other subunits of the PRC, CBX2 exhibits notably inverse correlations with TLS signature score, CD8⁺ T cell infiltration, and IFNγ response signaling across multiple cancer types (Fig. 6 A and B and SI Appendix, Fig. S6 A and B). Notably, CBX2 expression exhibited a significant inverse correlation with the Lymphocyte infiltration, T cell-inflamed gene expression profile (GEP) (2), cytolytic activity, and IFNγ response across multiple cancer types (Fig. 6 C–F). Conversely, a significant positive correlation was observed between CBX2 expression and myeloid-derived suppressor cell infiltration (SI Appendix, Fig. S6C). Kaplan–Meier survival analysis using the TCGA data demonstrated that elevated CBX2 expression correlated with poorer overall survival across various cancer types (Fig. 6G and SI Appendix, Fig. S6D). Furthermore, by retrieving and reanalyzing the existing clinical immunotherapy trial data, we found that the melanoma (34), lung squamous cell carcinoma (33), and renal cell carcinoma (RCC) patients (35) exhibiting low CBX2 expression responded more favorably to anti-PD1 immunotherapy (Fig. 6H and SI Appendix, Fig. S6 E and F). Collectively, these clinical data suggest that CBX2 exerts a detrimental effect in tumors, predicting an adverse tumor immune microenvironment and a diminished response to immunotherapy.

Fig. 6.

CBX2 suppresses tumor immune microenvironment and immunotherapy response. (A and B) Correlation between the expression of PRC subunits and the TLS Signature (A) or CD8+ T cell levels (B) in pan-cancer tissue samples from TCGA cohorts. Left: The heatmap illustrates the correlation between specific immune signatures and selected genes across various cancer types. Right: The bar chart indicates the number of cancer types exhibiting a statistically significant positive or negative correlation with each gene. (C–F) Correlation between CBX2 expression levels and lymphocyte infiltration (C), T cell-inflamed GEP (D), cytolytic activity (E), and IFN-gamma response signature score (F) in pan-cancer tissue samples from TCGA cohorts. (G) Kaplan–Meier survival analysis demonstrating overall survival times in patients from the TCGA-BRCA (n = 1,075), TCGA-SKCM (n = 98), and TCGA-LGG (n = 506) cohorts, stratified by CBX2 expression levels. (H) Kaplan–Meier survival analysis of overall survival times in clear cell RCC patients receiving immunotherapy, stratified by CBX2 expression levels.

Discussion

Tumor cells diminish their immunogenicity through various mechanisms to evade immune surveillance and facilitate tumor progression (36). A considerable body of clinical evidence suggests that tumor immunogenicity plays a pivotal role in determining the antitumor immune response and the therapeutic efficacy of immunotherapies, including anti-PD-1 and adoptive T cell therapy, across various human cancer types. High tumor immunogenicity is consistently associated with a “hot” tumor immune microenvironment (1, 3). Consequently, identifying unknown strategies to reduce immunogenicity may offer potential targets and biomarkers for enhancing the efficacy of immunotherapy. In this study, we demonstrated that CBX2 ablation inhibits tumor growth and activates tumor immune microenvironment in mouse models. We also conducted a systematic investigation into the correlation between the tumor immune microenvironment with the expression levels of CBX2 and other PRC components across various human cancers. Our findings reveal that CBX2, unlike other components, exhibits an inverse correlation with the tumor immune microenvironment markers and clinical outcomes in most cancer types, suggesting a unique role for CBX2 in regulating tumor immunity independent of the PRC function. Recent studies reported that CBX2 represses the expression of RBL2, an mTORC1 inhibitor, thereby driving breast cancer progression (37), and reprograms the epigenetic landscape of p38 MAPK-associated regulatory sites to promote AML progression (38). Additionally, another recent study has reported that CBX2 induced metabolic reprogramming by stimulating the Warburg effect, further contributing to breast cancer progression (39). Consistent with clinical analysis data, our results indicate that CBX2 forms a transcriptional corepressor complex to suppress its downstream targets by reducing the active histone marker H3K27ac, without increasing the repressive marker H3K27me3. Therefore, we identified a unique role for CBX2 in epigenetic reprogramming through H3K27ac modification.

Previous studies have reported that CBX2 functions as an oncogene, suppressing the expression of tumor suppressor genes and inducing metabolic reprogramming (27, 38, 39). Additionally, prior research has demonstrated that EZH2 epigenetically silences Th1-type chemokines, such as CXCL9 and CXCL10, in tumor cells, thereby suppressing antitumor immunity and immunotherapy (22, 23). A recent study reports that CBX2 increases M2-like macrophages and decreases its phagocytosis activity in high-grade serous carcinoma (28). In this study, our RNA-seq results indicated that CBX2 suppresses the expression of ISGs and antigen presentation genes. Consistently, the activation of the interferon signaling pathway is significantly enriched in SKCM patients exhibiting low CBX2 expression. Furthermore, CUT&Tag sequencing and ChIP-qPCR results indicate that CBX2 reduces H3K27ac modification at regulatory sites associated with the interferon signaling and antigen processing pathways. Consequently, our findings demonstrate that CBX2 represses tumor immunogenicity, given that the interferon signaling and antigen processing pathways are critical factors in the assessment of tumor immunogenicity. Consistently, knockout of CBX2 inhibits tumor growth and enhances the infiltration and activation of immune cells within tumor tissue. Furthermore, the ablation of CBX2 induces the activation of T cells and enhances T cell cytotoxic effector functions in vitro. Importantly, it increases the therapeutic efficacy of anti-PD1 therapy and adoptive T cell immunotherapy in vivo. Therefore, our results identify CBX2 as a potential target and biomarker for precise cancer immunotherapy.

In this study, we demonstrate that CBX2 suppresses H3K27ac but not H3K27me3 modification. Histone acetylation is directly modulated by the Histone Acetyltransferase or HDAC family proteins, which induce histone acetylation and deacetylation, respectively. Furthermore, our results show that knockdown of Cbx2 did not further enhance H3K27ac levels or the tumor immunogenicity increased by TSA treatment, suggesting that CBX2's regulation of H3K27ac is dependent on the HDAC activity. Unexpectedly, the positive hits from mass spectrometry analysis using CBX2 coimmunoprecipitation proteins did not contain members of the HDAC family, suggesting an indirect interaction. We propose RACK1, which was among the top-ranked positive hits and has been reported to promote tumorigenicity (40), as a candidate for further investigation. A recent study indicates that RACK1 can form a complex with HDAC1, thereby maintaining HDAC1 protein stability and its deacetylation function (32). Consistent with this, our results demonstrated that reducing RACK1 levels in tumor cells increased H3K27ac modification and tumor immunogenicity. CUT&Tag sequencing and ChIP-seq also demonstrated that CBX2 and RACK1 coregulate the H3K27ac modification on the promoters of ISGs. Moreover, the chromodomain (CD) of CBX2, known for its interaction with H3K27me3, is essential for the interaction between CBX2 and RACK1. Contrary to previous reports, our findings indicate that CBX2 does not influence the protein stability of either RACK1 or HDAC1, nor affects their interaction. Recent studies have demonstrated that RACK1 functions as a scaffold for protein interactions, signal transduction, and flavivirus replication (30). For instance, RACK1 serves as a bridge for active PKCβII and its phosphorylated substrates, facilitating PKC signaling transduction (41, 42). It also functions as a mediator for the cross-talk between the PKC and MAPK pathways, leading to PKC phosphorylation of JNK when stimulated with various PKC isoforms (α, β, γ) upon exposure to stimuli (43). Notably, our data indicate that knockout of Rack1 completely abolishes the interaction between CBX2 and HDAC1, suggesting that RACK1 serves as an indispensable scaffold for this interaction. In addition, we further demonstrated that knockdown of Cbx2 significantly decreased the chromatin binding and histone modification activity of HDAC1, suggesting that the CBX2/HDAC1/RACK1 complex plays a crucial role in initiating this modification. Several complexes, such as the NuRD complex, CBP/p300, and SWI/SNF complex, have been reported to participate in the regulation of H3K27ac (44–47). Our findings reveal a noncanonical CBX2/RACK1/HDAC1 complex that reduces the H3K27ac level to regulate expression of genes related to tumor immunogenicity.

In conclusion, this study presents strong evidence that CBX2 functions as a suppressor of tumor immunogenicity and serves as a potential prognostic marker for poor survival outcomes and reduced efficacy of immunotherapy, independent of the PRC. Our findings also introduce preliminary evidence for a noncanonical regulatory complex, CBX2/RACK1/HDAC1, which epigenetically silences interferon signaling genes and antigen presentation genes. Furthermore, we have identified a role for CBX2 in the suppression of tumor immunogenicity and the antitumor immune response, which contributes to formation of an immunosuppressive tumor microenvironment. The ablation of CBX2 enhances the therapeutic efficacy of anti-PD1 therapy and adoptive T cell therapy. Our findings reveal a unique function and complex mechanism of CBX2 in epigenetic reprogramming of target genes through H3K27ac modification, which suppresses tumor immunogenicity. Together, our findings provide a potential target and biomarker for precise cancer immunotherapy.

Materials and Methods

Mice and Reagents.

Six to eight-week-old female C57BL/6 J and BALB/c Nude mice were purchased from Beijing Vital River Animal Technology, Inc. OT-I mice were obtained from Jackson Laboratory. All the mice were maintained under specific pathogen–free conditions.

The B16 OVA cells were constructed by stably expressing OVA cDNA on B16 cells (C57BL/6 mouse melanoma) and HEK293 cell lines were obtained from ATCC. MC38 (C57BL/6 mouse colon adenocarcinoma) was provided by Yang Xuanming (Shanghai Jiaotong University, China). B3Z hybridoma cells were kindly gifted by Nilabh Shastri (Johns Hopkins University). All cell lines were tested and maintained Mycoplasma-free. The cells were maintained either with DMEM (Invitrogen) supplemented with 10% FBS and 1% penicillin–streptomycin or RPMI 1640 (Invitrogen) supplemented with 1% penicillin–streptomycin and 10% FBS in a humidified atmosphere at 37 °C and 5% CO₂.

GSK343 (A3449) and TSA (A8183) were purchased from Apexbio Inc. IFNγ (315-05) was from PeproTech; IFNβ (8234-MB) was from R&D SYSTEMS. Anti-mouse PD1 antibody (BE0146) was from BioXcell Inc.

Real-Time PCR and RNA-Seq.

The tumor cells were treated with IFNγ or IFNβ for 4 h. The total RNA was isolated using TRIzol (Invitrogen, 15596018) according to the manufacturer’s instructions. RNA was reverse transcribed using the HiScript III RT SuperMix for qPCR (Vazyme, R323-01). Real-time PCR was performed using the SYBR Premix Kit (MIKX, MK900-10) and analyzed using the Bio-Rad CFX96 thermal cycler. The primer sequences used for the investigated mouse genes were as follows:

18 s rRNA: FP: TTCCGATAACGAACGAGACTCT; RP: TGGCTGAACGCCACTTGTC;

Cxcl9: FP: TCCTTTTGGGCATCATCTTCC; RP: TTTGTAGTGGATCGTGCCTCG;

Cxcl10: FP: CCAAGTGCTGCCGTCATTTTC; RP: GGCTCGCAGGGATGATTTCAA;

Cxcl11: FP: GGCTTCCTTATGTTCAAACAGGG; RP: GCCGTTACTCGGGTAAATTACA;

Ifit1: FP: TACAGGCTGGAGTGTGCTGAGA; RP: CTCCACTTTCAGAGCCTTCGCA;

Ifit3: FP: GCTCAGGCTTACGTTGACAAGG; CTTTAGGCGTGTCCATCCTTCC;

B2m: FP: CAGTATGTTCGGCTTCCCATTC; RP: TTCTGGTGCTTGTCTCACTGA

For RNA-seq, total RNA was extracted by the RNeasy Mini Kit (Qiagen), according to the manufacturer’s instructions. The cDNA library construction and sequencing were made by Gene Denovo Biotechnology Co. (Guangzhou, China) with Illumina HiSeq2500 (150 bp, paired-end). RNA sequencing was performed on a NovaSeq sequencer (Illumina).

Western Blot and Co-IP.

The procedures for protein sample preparation from cell culture, protein quantification, western blot, and data analyses were performed as previously described (48). The following antibodies were used for western blot analyses: β-actin (Santa Cruz, catalog sc8432), Flag (Sigma, catalog A8592), HA (Roche, 12013819001), CBX2 (CST, catalog 18687), RACK1 (CST, catalog 5432), HDAC1 (CST, catalog 34589), EZH2 (CST, catalog 5246), GAPDH (Santa Cruz, catalog sc32233), H3K27ac (CST, catalog 8173), and H3K27me3 (CST, catalog 9733). For Co-IP, the cell lysate supernatants were incubated with the indicated antibody overnight at 4 °C, followed by the addition of Protein A/G beads (Pierce, catalog 20421). For IP with anti-Flag, anti-Flag beads (Sigma, catalog A2220) were used. Protein bands were visualized by chemiluminescence using an ECL detection kit (Thermo Scientific, 32106).

Statistics.

Data were analyzed using the GraphPad Prism software, V.5. Comparisons between two groups were analyzed using a two-tailed unpaired Student’s t test. Comparisons between multiple groups were analyzed using one-way ANOVA with Bonferroni’s posttest or two-way ANOVA with Bonferroni’s posttest for tumor growth study. Statistical significance was defined as a P value less than 0.05. The log-rank test was used for univariate survival analyses and showed as the Kaplan–Meier plot.

More detailed information on experimental procedures is available in SI Appendix, Materials and Methods.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

The raw data for RNA sequencing, ChIP-seq, and CUT&Tag reported in this manuscript are available at the Genome Sequence Archive (GSA) database of the National Genomics Data Center (NGDC) (https://ngdc.cncb.ac.cn/gsa/browse/CRA018316 and https://ngdc.cncb.ac.cn/gsa/browse/CRA020675) (49, 50). The data authenticity of this article has been validated by uploading the key raw data onto the Research Data Deposit platform and approved by Sun Yat-sen University Cancer Center Data Access/Ethics Committee (https://www.researchdata.org.cn/Search.aspx?k=RDDB2025288759) (51). All other data are included in the manuscript and/or SI Appendix.