EZH1/EZH2 inhibition enhances adoptive T cell immunotherapy against multiple cancer models

SUMMARY

Tumor resistance to chimeric antigen receptor T cell (CAR-T) and, in general, to adoptive cell immunotherapies (ACTs) is a major challenge in the clinic. We hypothesized that inhibiting the tumor drivers’ methyltransferases EZH2 and EZH1 could enhance ACT by rewiring cancer cells to a more immunogenic state. In human B cell lymphoma, EZH2 inhibition (tazemetostat) improved the efficacy of anti-CD19 CAR-T by enhancing activation, expansion, and tumor infiltration. Mechanistically, tazemetostat-treated tumors showed upregulation of genes related to adhesion, B cell activation, and inflammatory responses, and increased avidity to CAR-T. Furthermore, tazemetostat improved CAR- and TCR-engineered T cell efficacy in multiple liquid (myeloma and acute myeloid leukemia) and solid (sarcoma, ovarian, and prostate) cancers. Lastly, combined EZH1/EZH2 inhibition (valemetostat) further boosted CAR-T efficacy and expansion in multiple cancers. This study shows that EZH1/2 inhibition reprograms tumors to a more immunogenic state and potentiates ACT in preclinical models of both liquid and solid cancers.

In brief

Porazzi et al. highlight how inhibiting the epigenetic regulators EZH2 and EZH1 enhances cancer cell immunogenicity, boosting the effectiveness of CAR- and TCR-engineered T cells across liquid and solid tumor models, which represents a promising strategy for overcoming resistance and improving adoptive cell therapy outcomes.

Graphical Abstract

INTRODUCTION

Adoptive cellular immunotherapies (ACT) have significantly advanced the treatment of hematological malignancies. In particular, five anti-CD19 chimeric antigen receptor T cell (CART19) products have been approved by the US Food and Drug Administration (FDA) for relapsed/refractory (r/r) B cell non-Hodgkin lymphoma (NHL) and leukemia and two anti-BCMA CAR-T for multiple myeloma (MM).^1–5 More recently, two additional products, tumor-infiltrating lymphocytes and TCR-engineered anti-MA-GEA4 T cells, were approved for melanoma⁶ and synovial sarcoma,⁷ respectively. Despite this remarkable success and progress, most patients still experience treatment failure.² The factors contributing to resistance to ACT are numerous and include limited T cell tumor infiltration, the ability of tumor cells to evade T cell recognition and apoptosis, and T cell dysfunction.²

Epigenetic dysregulation is a key factor in the development and progression of cancer, including B cell NHL, MM, sarcomas, and subsets of solid cancers such as prostate and ovarian cancers.^8–12 Therefore, multiple drugs that target the epigenome have been developed and approved for liquid and solid cancers.^13,14 EZH2, a histone methyltransferase, mediates H3K27me3 and drives transcriptional silencing.¹⁵ Somatic gain-of-function mutations of EZH2 are described in ~20% of germinal center (GC)-derived lymphomas, such as follicular (FL) and diffuse-large B cell lymphomas (DLBCL)^16,17; further, EZH2 over-expression has been documented in various liquid and solid tumors.⁸ Indeed, most GC-derived FL, DLBCL, certain T cell lymphomas, MM, sarcomas, and several solid cancers of the prostate and ovary depend on EZH2 for proliferation and survival.^18–20 Notably, in DLBCL/FL, EZH2 alterations contribute to a “cold” tumor microenvironment (TME) and therapy resistance,^21,22 as they impair interactions between neoplastic B cells and T cells, downregulate major histocompatibility complex (MHC) class I/II expression, and increase regulatory T cell (Treg) populations.^18,23–28 Overexpression of EZH2 in solid tumors is associated with aggressive tumor behavior and poor prognosis, due to epigenetic silencing of tumor suppressor genes which drives uncontrolled cell growth and metastasis.^29–33 Interestingly, EZH1, a functional homolog of EZH2, supports cellular plasticity, allowing tumors to adapt and evade immune responses under therapeutic pressure, especially during EZH2 inhibition.^34,35 EZH2 and dual EZH1/2 inhibitors (tazemetostat and valemetostat, respectively), have shown single-agent anti-tumor activity in B cell- and T cell-derived lymphomas and leukemias, sarcomas, and other solid cancers.^15,36 Both tazemetostat and valemetostat have received regulatory approval: tazemetostat for follicular lymphoma and epithelioid sarcoma in the US,^30,37 and valemetostat for T cell leukemia/lymphoma in Japan.³⁸ Building on these findings, we hypothesized that inhibiting EZH2 and EZH1/2 in cancer cells could enhance adoptive T cell therapies by promoting a more immunogenic tumor profile, potentially improving T cell activation, infiltration, and function.

In this study, we evaluated the effects of EZH2 and dual EZH1/EZH2 inhibition on CAR- and TCR T cell therapies across hematologic and solid tumor models. Our findings demonstrate that EZH2 and EZH1/EZH2 inhibition enhance the anti-tumor activity of ACT by increasing tumor immunogenicity and, therefore, sustaining T cell function. These insights provide a foundation for leveraging EZH1/2 inhibition to overcome resistance and improve outcomes in diverse cancer types.

RESULTS

EZH2 inhibition improves the anti-tumor efficacy of CAR-T cells in human B cell lymphoma models

We first explored the potential of EZH2 inhibition with tazemetostat to enhance CART19 immunotherapy, as both therapies are approved for GC B cell lymphomas which often depend on EZH2 for their survival (Figure 1A). To this goal, we selected four GC-derived DLBCL lymphoma cell lines, OCI-Ly18 and Toledo (wild-type, WT EZH2), SU-DHL-4, and Karpas-422 (mut EZH2).¹⁸ We used suboptimal doses of tazemetostat (500 nM for OCI-Ly18, Toledo, SU-DHL-4; 50 nM for Karpas-422) that depleted global H3K27 trimethylation without affecting cell viability or proliferation at short term (~3 days), enabling evaluation of potential combination with CART19 therapy. To determine the most effective treatment sequence, we tested five sequences of administration of tazemetostat and CAR-T (Figures 1B and S1). The most effective sequence (sequence #1) included pre-incubating tumor cells with tazemetostat and continuing exposure to tazemetostat after CART19 administration (Figure 1C). With sequence #1, tazemetostat enhanced CART19-mediated killing of GC-derived DLBCL cell lines at various effector-to-target (E:T) ratios, regardless of their EZH2 mutation status. Notably, it also improved killing in Karpas-422, a cell line resistant to single-agent CART19 (Figure 1C). This finding underscores the need for priming tumor cells but also maintaining EZH2 inhibition during CAR-T anti-tumor effect. Tazemetostat monotherapy reduced H3K27 trimethylation (Figure S2A) but did not significantly impact short-term lymphoma growth at these doses (Figure S2B). From this point onward, unless stated otherwise, all in-vitro experiments will adopt sequence #1 as the standard administration protocol. To confirm these findings genetically, we established a knockdown (KD) of EZH2 in SU-DHL-4 cells (mut EZH2) using short hairpin RNA (Figure 1D). Consistent with the small-molecule inhibition results, EZH2 KD enhanced CART19-mediated killing (Figure 1D).

Figure 1. EZH2 inhibition sensitizes lymphoma cells to CART19 therapy.

(A) Model of action of EZH2 inhibition and CART19 therapy.

(B) Sequence of tazemetostat and CART19 administration (sequence #1).

(C) Luciferase-based killing assay, representative of three, of DLBCL-luc cell lines pretreated for 3 days with 500 nM (OCI-Ly18, Toledo, SU-DHL-4) or 50 nM (Karpas-422) tazemetostat or DMSO and co-cultured with CART19 or untransduced T cells (UTD) (n = 3). Mean ± SEM; one-way ANOVA, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

(D) Left: western blot of EZH2 and H3K27me3 expression in parental vs. shEZH2-SU-DHL-4 cells. Right: luciferase-based killing assay of parental SU-DHL-4 vs.shEZH2-SU-DHL-4 cells with anti-CD19 CART cells for 72 h at a 0.06:1 E:T ratio. Mean ± SEM. One-way ANOVA, *p < 0.05, **p < 0.01, and ***p < 0.001.

(E) Schematic of the OCI-Ly18 xenograft model. NSG mice harboring OCI-Ly18 xenografts were treated with tazemetostat (200 mg/kg, per os) or vehicle continuously for 8 weeks and infused with UTD or CART19 (3.5 × 10⁶/mouse) 4 days after treatment started.

(F and G) Analysis of tumor progression (F) and survival (G) of NSG mice harboring OCI-Ly18 and treated with tazemetostat (200 mg/kg, per os) or vehicle continuously for 8 weeks and infused with UTD or CART19 (3.5 × 10^6/mouse) 4 days after treatment started. Left: tumor burden over time was measured by caliper. Right: histogram representing tumor volumes assessed at D24 post tumor injection. Mean ± SEM. One-way ANOVA, *p < 0.05. Log rank (Mantel-Cox) test.

(H) Flow cytometry analysis of peripheral blood T cell expansion; one-way ANOVA, *p < 0.05. Also see Figures S1 and S2.

We then sought to validate these results in vivo using a xenograft model of the aggressive EZH2 WT DLBCL cell line OCI-Ly18 that is highly resistant to CART19 (Figure 1E). OCI-Ly18 cells were implanted subcutaneously into immunodeficient NSG mice that were then randomized to receive tazemetostat (200 mg/kg/day, ~7 weeks, orally) or vehicle, followed by a suboptimal dose of CART19 (3.5 × 10⁶ cells/mouse, intravenously [i.v.]) or untransduced (UTD) cells. Tazemetostat combined with CART19 significantly improved tumor control as compared to CART19 or tazemetostat alone (Figures 1F and S2C) and extended mouse survival (Figure 1G). Tazemetostat showed no toxicity to T cells and demonstrated no significant treatment-associated toxicities (weight loss or graft-versus-host disease [GVHD]; Figure S2D), rather, on day 20 post-infusion, mice treated with tazemetostat exhibited significantly increased levels of CART19 cells in the blood (Figure 1H).

EZH2 inhibition increases lymphoma immunogenicity

To investigate how EZH2 inhibition enhances CAR-T T efficacy, we first focused on its effect on cancer cells. We performed RNA sequencing on GFP⁺ OCI-Ly18 and SU-DHL-4 cells (WT and mut EZH2, respectively) pre-treated with tazemetostat or vehicle for 3 days, then exposed to UTD or CART19 with tazemetostat for 48 h (sequence #1). Unsupervised clustering and principal-component analysis (PCA) revealed distinct transcriptional profiles by treatment conditions (Figure S3A). Differentially expressed genes (DEGs) analysis showed tazemetostat alone induced the highest number of DEGs, predominantly upregulated, consistent with EZH2’s repressive role (Figures S3B and S3C). Gene set enrichment analysis (GSEA) of tumor cells exposed to CART19 alone confirmed key mechanisms of CART19-mediated killing, including interferon-gamma/alpha, tumor necrosis factor (TNF) via nuclear factor κB (NF-κB), interleukin (IL)-2-STAT5 signaling, and apoptosis (Figure S3D). Tazemetostat-treated groups (Taz-UTD and Taz-CART19) showed downregulation of cell cycle and proliferative genes (e.g., MYC targets, G2M checkpoint, and E2F targets).^19,39 Tumor cells exposed to Taz-CART19 specifically upregulated genes involved in inflammation (IL2RA), B cell activation/differentiation (e.g., KDM5B and BACH2), adhesion (e.g., OX40L/TNFSF4 and CD80), T cell chemo-attraction (e.g., CXCL9, CXCL10, and CXCL16), and motility (e.g., LMNA, CEACAM1, NRCAM, and EPHB2; Figure 2A). Despite differences in genetic backgrounds, both OCI-Ly18 and SU-DHL-4 cells treated with Taz-CART19 shared upregulation of immunogenic molecules (e.g., CD44, CXCL10, CEACAM1, and EPHB2; Figure S3E).

Figure 2. EZH2 inhibition modulates lymphoma immunogenicity.

(A) Volcano plots of differential expressed genes in OCI-Ly18 and SU-DHL-4 cultured with CART19-tazemetostat vs. CART19-DMSO. Red dots: significantly upregulated genes; blue dots: significantly downregulated genes (log2 fold change ≥ 1 and false discovery rate (FDR) < 0.01).

(B) Schema depicting the study design. 4.0 × 10⁶ OCI-Ly18 cells were implanted into NSG mice on day –14. Mice were treated with tazemetostat (150 mg/kg; per os) or vehicle for two weeks. On day 0, 4.0 × 10⁶ CART19 or UTD were infused via tail vein.

(C) Caliper measurement of tumor growth, tumor resection (day 11 post CART infusion) and downstream applications.

(D) Ex vivo staining immunohistochemistry staining of tumors from UTD vehicle, UTD tazemetostat, CART19 vehicle, and CART19 tazemetostat treated animals. Higher magnification at the bottom. CD3 was quantified as cell count per mm2, and CD19 positivity was quantified as % of area. Scale bar: (Top) 2 or 5 mm; (Bottom) 200 μm.

(E) RNA-seq heatmap of differentially expressed genes between CART19-Vehicle and CART19-Tazemetostat treated animals from in vivo subcutaneous tumors (n = 3 animals/group; 3 replicates/each; fold change > 1.5, q-value < 0.01).

(F) Lumicks avidity assay in DMSO-treated and tazemetostat-treated OCI-Ly18 and SU-DHL-4 cells (left) or in parental vs. SU-DHL-4 EZH2-KD cells (right) bound to a monolayer of CART19 cells when increasing acoustic force was applied (maximum force: 100 pN). Also see Figure S3.

To further explore the effect of EZH2 inhibition in vivo, we employed a CART19-resistant OCI-Ly18 xenograft lymphoma model (Figure 2B) where OCI-Ly18 cells were subcutaneously implanted into NSG mice on day –18 and tazemetostat (150 mg/kg/day, orally) or vehicle were administered starting day –15; CART19 (4.0 × 10⁶ CAR-T⁺ cells/mouse, i.v.) or UTD cells were administered on day 0. This schedule allowed for extended tumor pre-treatment and reprogramming by tazemetostat. On day 11 post-CART infusion, before significant tumor size difference between the CART19 groups, tumors were explanted and analyzed (Figure 2C). At microscopic evaluation, smaller and more necrotic tumors with increased T cell infiltration were observed in animals treated with CART19 and tazemetostat (Figure 2D) as compared to vehicle. Differential gene expression analysis of sorted tumor cells revealed that EZH2 inhibition increased MHC-I/II antigens (HLA-A, HLA-DOA, and HLA-DOB), costimulatory molecules (CXCL9, CXCL10, and CD40), adhesion (ICAM1), B cell activation (FOS and JUN), interferon (IFN) response (STAT1–3 and IRF2/7/9), and p53-mediated apoptosis (FAS, ATF3, CASP3/4/7, MCL1, and MDM2; Figure 2E). These findings suggest that EZH2 inhibition reshapes tumor immunogenicity and modulates inflammation and apoptosis pathways, potentially enhancing tumor-to-CAR-T cell interactions, and ultimately contributing to improved CAR-T efficacy.

To assess the chromatin-level effects of EZH2 inhibition, we conducted ATAC-sequencing on mut EZH2 SU-DHL-4 cells pre-treated with tazemetostat or vehicle for 3 days, then exposed to UTD or CART19 with tazemetostat for 48 h (sequence #1). Genome-wide analysis revealed increased accessibility of genes involved in cell-to-cell adhesion and B cell differentiation in Taz-CART19 vs. DMSO-CART19 (Figure S3F). Taz-CART19 treatment enhanced chromatin opening in regions associated with adhesion molecules (e.g., LAMC1 and LAMA1) and tumor suppressors (e.g., FOXO3 and IL10), while closing chromatin in cell cycle regions (e.g., CDK6). These changes suggest EZH2 inhibition reshapes cell-cell interaction profiles and delays cell cycle progression, consistent with gene expression findings (Figure S3F).

Lastly, we sought to investigate protein changes upon tazemetostat exposure. To this goal, we conducted quantitative flow cytometry to assess CD19 levels on tumor cells after 72 h of treatment. CD19 expression remained unchanged with EZH2 inhibition (Figure S3G). However, the adhesion molecule CD58,⁴⁰ a known EZH2 target, was upregulated in lymphoma cells treated with tazemetostat, as confirmed by flow cytometry (Figure S3H).

Finally, to functionally validate the hypothesis that enhanced killing resulting from EZH2 inhibition was chiefly due to an improved interaction between tumor cells and CAR-T, we investigated the avidity of CART19 cells and tumor cells using the z-Movi platform (LUMICKS).⁴¹ Avidity, influenced by several factors, including the CAR affinity, adhesion molecules and costimulatory proteins, determines the strength of interactions at the T cell-target cell interface, affecting CAR signaling, cytokine secretion, proliferation, and persistence.^42,43 Indeed, tazemetostat pre-treatment increased CART19 avidity to both OCI-Ly18 and SU-DHL-4 cells, indicating enhanced binding efficiency (Figure 2F). These findings were further validated in SU-DHL-4 cells with EZH2 KD via short hairpin RNA (shRNA), reinforcing the role of EZH2 inhibition in strengthening tumor-CAR-T interactions (Figure 2F).

EZH2 inhibition improves CAR-T cell expansion and activation upon tumor encounter

EZH2 is also expressed by T cells, where it regulates their activation, proliferation, and differentiation while suppressing pro-inflammatory cytokines.⁴⁴ We therefore sought to investigate the effect of EZH2 inhibition on CAR-T cells. Prior studies suggest EZH2 inhibition can prevent recovery from CAR-T exhaustion in a tonic signaling CAR-T model.⁴⁵ Others suggest that adoptively transferred T cells pretreated with tazemetostat showed increased antitumor activity in the context of murine (GC)-lymphomas and melanoma models.^46,47

We first studied the effect of tazemetostat on CAR-T cells when tumor cells were pre-treated with tazemetostat and then both CAR-T and tumor cells were exposed to the drug during the killing assay (sequence #1). First, we assessed tazemetostat’s impact on CART19 activation, long-term tumor control, expansion, and memory formation. In vitro, tazemetostat significantly increased CAR-T cytokine secretion (IFN-γ, TNF, and IL-2) (Figure 3A), long-term killing capacity, and promoted expansion when co-cultured with tumors that were pretreated with tazemetostat (Figure 3B). Furthermore, flow cytometry analysis revealed an increased effector memory fraction in CART19 cells treated with tazemetostat (Figures 3C and S4A). Conversely, exposure of CART cells only to tazemetostat did not lead to enhanced function, as previously shown (FigureS1, sequences #4 and #5). To evaluate CAR-T functionality over time, we performed a rechallenge experiment. In this setting, CART19 cells were cultured with SU-DHL-4 cells, either untreated or pre-treated with tazemetostat. Untreated tumor cells were reintroduced at 4, 7, and 11-day intervals to trigger CAR-T dysfunction. Of note, in this model, CART19 combined with tazemetostat maintained higher cytotoxicity and expansion over time as compared to CART19 alone that showed diminished killing and inferior proliferation (Figures 3D and S4B).

Figure 3. EZH2 inhibition enhanced CART19 cytotoxicity and induced a more naive/early memory phenotype.

(A) ELISA assay on supernatants (IFN-γ, TNF, and IL-2) collected after 48 h of co-culturing OCI-Ly18 and SU-DHL-4 (pretreated with 500 nM tazemetostat or DMSO for 3 days) with CART19 or UTD (E:T = 0.25:1, n = 3). One-way ANOVA, *p < 0.05, **p < 0.01.

(B) Left: long-term killing assay by CellCyteX of OCI-Ly18-GFP or SU-DHL-4-GFP exposed to UTD or CART19 (E:T ratio 0.125:1) in the presence of DMSO ortazemetostat (500 nM) over 150 h. Total fluorescence intensity was measured over time Mean ± SD. One-way ANOVA, ***p < 0.001. Right: flow cytometric absolute quantification of T cells collected after 7 days of long-term killing assay. 3 donors presented with different colors. Mean ± SEM. One-way ANOVA, **p < 0.01.

(C) Flow cytometric analysis of CART19 memory phenotypes collected after 7 days of long-term killing assay.

(D) Re-challenging assay. Luciferase-based killing assay, representative of three, of SU-DHL-4-luc cells pretreated 3 days with 500 nM tazemetostat or DMSO and co-cultured with CART19 or untransduced T cells (n = 3). Additional SU-DHL-4 cells were reintroduced at intervals of 4, 7, and 11 days. Mean ± SEM. One-way ANOVA, **p < 0.01, ***p < 0.001, and ****p < 0.001.

(E) Left: schema depicting the study design as per Figure 2A. Right: scRNA sequencing and UMAP of all clusters among CART19-vehicle and CART19-tazemetostat populations purified from tumor infiltrating T cells. Bottom, bar graph indicating the distribution of cells into different clusters.

(F) Volcano plot displaying differential expressed genes in all clusters among CART19 vehicle and CART19 tazemetostat populations purified from tumor infiltrating T cells. Blue dots: significantly upregulated genes; red dots: significantly downregulated genes (log2 fold change ≥ 1 and FDR < 0.01).

(G) Left: schematic of the xenograft model, as per Figure 1E. Right: scRNA sequencing and UMAP of all clusters among CART19 vehicle and CART19 tazemetostat populations purified from peripheral blood at day 20. Bottom, bar graph indicating the distribution of cells into different clusters.

(H) Luciferase-based killing assay of SU-DHL-4-luc cells pretreated 3 days with 500 nM tazemetostat or DMSO and co-cultured with WT CART19 vs. EZH2-KOCART19 (n = 3) cells for 72 h (E:T ratio 0.06:1). Mean ± SEM. One-way ANOVA, ***p < 0.001. Also see Figures S4 and S5.

Building on the finding that tazemetostat enhances CART19 efficacy through tumor reprogramming, we investigated the profile of tumor-infiltrating CART19 cells in vivo. Tumors were pretreated with suboptimal doses of tazemetostat or vehicle for 15 days before CART19 infusion, enabling tumor cell reprogramming via EZH2 inhibition (Figure 3E) and on day 11 post-CAR-T infusion, tumor-infiltrating T cells were sorted for single-cell RNA sequencing (scRNA-seq) (Figure 3E). Tazemetostat-treated tumors exhibited greater T cell infiltration (Figure S4C) and Uniform Manifold Approximation and Projection (UMAP) analysis revealed seven distinct T cell subpopulations based on transcriptomic signatures, including clusters enriched for cytotoxicity (e.g., IFNG and GZMB) and cell cycle (TOP2A and PCNA) markers. In both groups, tumor-infiltrating T cells showed high CD8A expression, indicating a predominance of cytotoxic T cells in the TME (Figures S4D and S4E). GSEA of CART19 cells from tazemetostat-treated tumors revealed upregulation of activation and cytotoxicity genes (e.g., GZMB, PRF1, IFNG, and CD69) and downregulation of exhaustion markers (e.g., LAIR2) compared to vehicle-treated controls (Figures 3F and S4F). These findings suggest that EZH2 inhibition enhances T cell infiltration and promotes a tumor-reactive, cytotoxic profile in intratumoral CART19 cells.

To further investigate the effects of EZH2 inhibition on CART19 cells in the blood, we analyzed circulating CART19 cells in the peripheral blood using our initial model, where the combination of tazemetostat and CART19 improved tumor control and extended survival (Figure 3G). Increased CART19 expansion was observed in the peripheral blood of tazemetostat-treated animals (Figure 1H). At scRNA-seq, UMAP analysis of transcriptomic profiles revealed seven clusters with distinct signatures. CD8A expression was predominant in cluster 3, while CD4 expression characterized other clusters (Figure S4G). Clusters 1 and 2, more abundant in tazemetostat-treated animals, were enriched for genes associated with activated naive/early memory states,⁴⁸ effector function, apoptosis resistance, and glycolysis.⁴⁹ In contrast, clusters 0 and 4, prevalent in vehicle-treated animals, expressed genes linked to terminal effector phenotypes and the transforming growth factor β (TGF-β)/Smad3 pathway, which suppresses CD4 T cell proliferation and co-stimulation (Figures 3G and S4H). These findings suggest that, in the presence of tazemetostat, CART19 cells at the tumor site exhibit an effector phenotype, while circulating CD4⁺ T cells maintain a more naive/early memory state, in line the increased persistence.⁵⁰

Finally, we studied the effects of tazemetostat directly and exclusively on CAR-T cells. First, we measured the proliferation of CAR-T exposed to increasing concentration of tazemetostat and seeded on a layer of irradiated OCI-Ly18 cells. In this context, EZH2 inhibition did not affect CART19 proliferation (Figure S5A). Moreover, as previously shown (Figure S1) therapeutic sequences #4 and #5 that included CAR-T pre-exposure to tazemetostat did not lead to significant improvement of CAR-T functionality. Further, incorporating tazemetostat during CART19 manufacturing did not affect CART19 viability, expansion, or killing capacity (Figures S5B–S5D). We then generated EZH2 knockout (KO) CART19 cells using CRISPR-Cas9 (Figure S5E). Of note, EZH2-KO CART19 cells demonstrated comparable killing capacity to WT CART19 cells in vitro (Figure 3H). Furthermore, when SU-DHL-4 cells were pre-treated with tazemetostat and exposed to WT or EZH2-KO CART19 cells (sequence #1), tazemetostat still enhanced the cytotoxic activity of both groups similarly (Figure 3H).

All together, these findings suggest that the primary benefit of EZH2 inhibition by tazemetostat is on tumor cells, reshaping them to be more immunogenic that, as a result, promotes better activation of CART19 cells. Nevertheless, EZH2 inhibition did not compromise the anti-tumor effect of CAR-T cells.

EZH2 and EZH1/EZH2 inhibition improves the anti-tumor efficacy of CAR-T- and TCR T cell-based therapies against multiple liquid and solid tumors

We sought to explore whether EZH2 inhibition could broadly enhance the efficacy of CAR-T immunotherapies across multiple CAR-T products, given the well-documented role of EZH2 dysregulation in both liquid and solid cancers.²⁰ To address this, we first generated CAR-T cells targeting CD22 and CD79b for B cell lymphoma, acknowledging the pressing clinical need for alternative targets to address therapy resistance and antigen escape in B cell malignancies.² Remarkably, EZH2 inhibition significantly improved CAR-T-mediated killing across these models (sequence #1, Figure 4A).

Figure 4. EZH2 and EZH1/EZH2 inhibition sensitizes hematological and solid tumor models to immunotherapies.

(A) Luciferase-based killing assay of SU-DHL-4-luc cells pretreated 3 days with 500 nM tazemetostat or DMSO and co-cultured with UTD, CART19, CART79b, or CART22 (n = 3) cells for 72 h (E:T ratio 0.06:1), in the presence of tazemetostat/DMSO. Mean ± SEM. One-way ANOVA, ***p < 0.001, and ****p < 0.001.

(B) CellCyteX long-term cytotoxicity assays. Total fluorescence intensity parameter was measured over time in the multiple myeloma cell line RPMI-8226-GFP.Cells were pretreated with 250 nM tazemetostat or DMSO for 3 days and then exposed to anti-BCMA-CART cells or UTD at 0.5:1 E:T ratio. Co-cultures were monitored for further 150 h. Mean ± SD. One-way ANOVA, ***p < 0.001.

(C) Killing assay of AML cell lines (KG-1 and THP-1), pretreated with 5 μM Taz or DMSO for 3 days and then exposed to UTD or CART33 in the presence of vehicle (DMSO) or tazemetostat (72 h) at multiple E:T ratios. Mean ± SEM. One-way ANOVA, *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.001.

(D) Impedance-related Cell Index (CI) parameter over time of HER2+ SKOV3 (ovarian cancer) and PC3 (prostatic adenocarcinoma) cell lines were pretreatedwith 5 μM tazemetostat or DMSO for 3 days and then exposed to anti-HER2-CART cells or UTD (E:T ratio 1:1). Co-cultures were monitored for further 120 h. Mean ± SD. One-way ANOVA, *p < 0.05, ***p < 0.001.

(E) Killing assay by incucyte live-cell analysis system of mCherry+ A673 (Ewing sarcoma) cell line. A673 cells were pretreated with 5 μM tazemetostat or DMSO for 3 days and then exposed to anti-LOXHD1-TCR-T cells at 3:1 E:T ratio, in presence of tazemetostat or DMSO (n = 3). Images (left) taken at different time points before and after adding anti-LOXHD1-TCR T cells. Mean ± SD. One-way ANOVA, *p < 0.05.

(F) Luciferase-based killing assay, representative of three, of DLBCL-luc cell lines pretreated 3 days with 100 nM (OCI-Ly18, Toledo, SU-DHL-4) or 50 nM (Karpas-422) valemetostat or DMSO and co-cultured with CART19 or untransduced T cells (E:T ratio: 0.06:1; n = 3) for additional 3 days. Valemetostat was maintained during the killing assays. Mean ± SEM; one-way ANOVA, **p < 0.01, ***p < 0.001, ****p < 0.0001.

(G) Re-challenging assay. Luciferase-based killing assay, representative of three, of SU-DHL-4-luc cells pretreated 3 days with 100 nM valemetostat or DMSO and co-cultured with CART19 or untransduced T cells (n = 3). Additional SU-DHL-4 cells were reintroduced at intervals of 4, 7, and 11 days. Mean ± SEM. One-way ANOVA, *p < 0.05, ***p < 0.001, and ****p < 0.001.

(H) CellCyteX long-term cytotoxicity assays. Total fluorescence intensity parameter was measured over time in the multiple myeloma cell line RPMI-8226-GFP.Cells were pretreated with 10 nM valemetostat or DMSO for 3 days and then exposed to anti-BCMA-CART cells or UTD at 0.5:1 E:T ratio. Co-cultures were monitored for further 150 h. Mean ± SD. One-way ANOVA, ***p < 0.001.

(I) Bioluminescence analysis of tumor progression in NSG mice harboring intaosseous RPMI-8226-luc xenografts, treated with valemetostat (100 mg/kg, o.s) orvehicle, starting one week post tumor implant and continued for 7 weeks; mice were infused with UTD or anti-BCMA-CART (0.8 × 10⁶/mouse) 5 days after treatment started.

(J) Flow cytometry analysis of peripheral blood T cell expansion, one-way ANOVA, ****p < 0.0001.

(K) Luminex-analysis of serum IFN-γ. Mean ± SEM. Also see Figures S6–S8.

In the context of hematological malignancies, EZH2 is a critical driver of MM progression, contributing to uncontrolled cell cycle regulation, resistance to apoptosis, and the aggressive behavior of MM cells.⁵¹ To evaluate whether EZH2 inhibition could enhance the efficacy of MM-specific CAR-T therapies, we tested the combination of tazemetostat with anti-BCMA CAR-T cells. Notably, EZH2 inhibition significantly enhanced the long-term killing capacity of anti-BCMA CAR-T cells (sequence #1, Figure 4B).

Since EZH2 plays a role in acute myeloid leukemia (AML) cell differentiation⁵² and sensitization to other therapies,²² we also assessed anti-CD33 CAR-T cells with EZH2 inhibitors in AML (KG-1 and THP-1). Also in this model, EZH2 inhibition led to enhanced CAR-T in vitro killing (sequence #1; Figure 4C). These findings collectively highlight the potential of EZH2 inhibition to broadly potentiate CART therapies across a range of hematological malignancies.

To assess the potential impact of EZH2 inhibition beyond CART immunotherapy of hematological cancers, we investigated EZH2 inhibition in solid tumors that are known to be dependent on EZH2, such as HER2+ ovarian³² and prostatic^29,33 cancers, and sarcoma.^30,31 Cancer cells (SKOV-3 and PC3) were pre-treated with tazemetostat, and the drug was also maintained during the HER2-targeted CAR-T cell-mediated killing (sequence #1); we observed significant enhancement of CAR-T killing at long term (Figure 4D). Furthermore, given tazemetostat’s FDA approval for epithelioid sarcoma, we extended our study to the A673 sarcoma cell line and engineered TCR T cells with anti- HLA-A*02:01-LOXHD1 TCR antigen.⁵³ Also in this model, tazemetostat significantly enhanced TCR T cells mediated killing of cancer cells (Figure 4E). Hence, EZH2 inhibition holds promise for a broad spectrum of solid tumors and diverse adoptive T cell immunotherapies.

We then hypothesized that combining EZH1 with EZH2 inhibition could further enhance CAR-T therapy due to EZH1’s compensatory role when EZH2 is inhibited and EZH1 expression in non-proliferating cells.⁵⁴ We found that in human lymphoma models (OCI-Ly18, Toledo, SU-DHL-4, and Karpas-422), EZH1/EZH2 inhibitor valemetostat reduced H3K27 tri-methylation levels (Figure S6A) but did not significantly inhibit lymphoma growth except at high doses (Figure S6B). We then selected a suboptimal dose of valemetostat and observed that it significantly enhanced CART19 tumor killing (Figure 4F). Among various valemetostat and CART19 administration protocols (sequences #1–#5; Figures 4F and S7), pre-incubation of tumor cells and continuous drug presence was the most effective regimen (Figure 4F, sequence #1). Unlikely with the single inhibitor tazemetostat, simultaneous administration of CART19/Val to tumor cells showed a trend toward increased killing and the pre-incubating CART19 cells with valemetostat followed by continuous treatment during killing (sequence #5) enhanced anti-tumor effects, suggesting dual EZH1/EZH2 inhibition provides greater and faster effect than EZH2 inhibition alone on tumor and T cells.^55,56 In direct comparisons using SU-DHL-4 cells, valemetostat outperformed tazemetostat in enhancing CART19 efficacy (Figure S8A). We then evaluated the valemetostat effect on CART19 serial killing efficiency and resistance to exhaustion in a rechallenge experiment. CART19 cells were cultured with SU-DHL-4 cells, either untreated or pre-treated with valemetostat. Untreated tumor cells were reintroduced at 4, 7, and 11-day intervals to assess CART19 cytotoxicity over time. CART19 cells killing in the presence of valemetostat achieved superior initial and sustained tumor killing compared to vehicle controls, demonstrating enhanced cytotoxicity without compromising sustained cytotoxic potential (Figure 4G). Additionally, CART19 cells expanded across multiple rechallenge rounds, with greater proliferation observed in the presence of valemetostat, indicating that EZH1/EZH2 inhibition did not compromise CART19 prolonged activation and a durable expansion response (Figure S8B).

Lastly, to examine the impact of dual EZH1/EZH2 inhibition on other CAR-T products and diseases, we tested valemetostat in combination with anti-BCMA CAR-T cells against the MM cell line RPMI-8226. A suboptimal dose of valemetostat (Figure S8C), was able to significantly enhance the long-term killing efficiency of CARTBCMA cells (Figure 4H). This enhancement was more pronounced than with the single EZH2 inhibitor, as shown in long-term killing assays (Figure S8D), underscoring the potential of dual EZH1/EZH2 inhibition to potently boost CAR-T efficacy across various tumor types. Lastly, we established an intraosseous orthotopic RPMI-8226 xenograft model of MM. Three days post intraosseous tumor implant, mice started receiving valemetostat (100 mg/kg, per os, daily for 5 weeks) or vehicle. One-week post-RPMI-8226 implant, mice were randomized to receive CARTBCMA (0.8 × 10⁶/mouse, i.v.) In this model, continuous administration of valemetostat improved CARTBCMA tumor control, promoted T cell expansion in the blood and demonstrated no significant treatment-associated toxicities (weight loss or GVHD) (Figures 4I, 4J, and S8E). Additionally, animals treated with the CARTBCMA/valemetostat combination exhibited elevated serum levels of the cytotoxic cytokine IFN-γ, suggesting an enhanced activation and cytotoxic response from the T cells (Figure 4K).

In summary, we studied the role of EZH2/EZH1 inhibition in multiple preclinical models of human cancer treated with ACT. In all the models tested, EZH2/EZH1 inhibition led to enhanced CAR- and TCR-T anti-tumor effect. In lymphoma models, EZH2 inhibition modulated the immunogenicity of the lymphoma cells making them more immunogenic and more effective in activating CAR-T cells.

DISCUSSION

In this study, we demonstrate that inhibiting EZH2 or simultaneously EZH1/EZH2 significantly enhances the anti-tumor efficacy of CAR- and TCR-ACT across diverse cancer models, including liquid and solid tumors. Using the clinically approved selective EZH2 inhibitor tazemetostat and dual EZH1/EZH2 inhibitor valemetostat, we show that these treatments reprogram tumor cells to become more immunogenic and vulnerable to immune-mediated killing, regardless of their EZH2 mutational status.

Our results indicate that the changes induced in tumor cells by EZH1/2 inhibition represent a crucial step in facilitating more efficient CAR-T-mediated tumor clearance. Indeed, it was essential to pre-treat the tumor and maintain the presence of EZH1/2 inhibitors throughout the CAR-T-killing process. Interestingly, EZH2 KO in CART19 alone did not improve tumor killing, indicating that inhibition at the tumor level is an essential prerequisite of this combination. Indeed, previous studies have shown that EZH2 inhibition reprograms tumor cells, enhancing their immunogenicity by upregulating antigen presentation, T cell co-stimulation, and chemotactic pathways.^18,23 We show here that this reprogramming sensitized tumor cells to CAR-T killing, promoting tumor recognition, infiltration, and sustained cytotoxicity.

Because EZH1/EZH2 broadly act on gene expression through H3K27me3, it is not surprising that multiple mechanisms contribute to the final effect, in addition to the known role of EZH2 in regulating cancer cell proliferation.^18,39 The effect of EZH2 inhibition of lymphoma cells relies on the reprogramming of the tumor cells, in part by upregulating genes involved in inflammatory response, B cell activation/differentiation (e.g., BATCH2), T cell co-stimulation (OX40L, CD80, and EPHB2), T cell attraction (CXCL9, CXCL10, and CXCL16), antigen presentation genes (HLA-A, HLA-DOA, HLA-DOB, and HLA-E), and IFNG-regulated genes (STAT1, IRF2, IRF7, and IRF9), thus potentiating the tumor-to-CAR-T cell interaction and response to apoptosis. For instance, upregulation of OX40L on tumor cells improves OX40 signaling in CAR-T cells, reducing apoptosis and boosting activation, expansion, and persistence.^57–59 Similarly, CXCL9 and CXCL10, critical for T cell trafficking, are increased in our in vivo model, facilitating CAR-T infiltration.⁶⁰ These chemokines, under direct EZH2 repression, have been shown to enhance CAR-T therapies in ovarian and colon cancer models.^61,62 Additionally, EZH2 inhibition combined with CAR-T therapy upregulates interferon response genes, further activating CAR-T cells and overcoming immunotherapy resistance.^33,63 Together, these pathways enhanced CAR-T infiltration and sustained cytotoxicity, addressing key barriers to CAR-T efficacy.

Applying an EZH2 or EZH1/EZH2 inhibitor may also have a significant effect on CAR-T cells. In our study, tazemetostat and valemetostat when combined with CAR-T immunotherapy, in vitro and in vivo, did not impair CART19 killing and expansion. On the contrary, the presence of EZH2 or EZH1/EZH2 inhibitor led to increased expansion, enhanced cytotoxicity and ability to sustain multiple rounds of tumor serial killing. Consistent with our findings, a recent study explored the effects of transient EZH2 inhibition using tazemetostat during T cell expansion in vitro. The study showed that temporary EZH2 suppression effectively delays T cell exhaustion and preserves stemness without impairing proliferation.⁴⁷ It is important to note that our modulation of EZH2 is performed within the framework of clinically approved CAR-T cell products (CART19 and CARTBCMA), and not in tonically active and exhausted CAR-T models.⁴⁵ This transient inhibition, importantly, avoids the potential negative effects associated with complete EZH2 depletion in T cells, suggesting that controlled modulation of EZH2 during T cell preparation could improve therapeutic outcomes. In our study, in vivo scRNA-seq revealed that tazemetostat treatment increased naive and early memory CD4⁺ T cells in the peripheral blood while reducing terminally differentiated and exhausted CD4⁺ T cells compared to controls. Tazemetostat preserved stem-like memory signatures (e.g., TCF7 and LEF1), which are linked to long-term CAR-T persistence and function.⁴⁸ It also enriched CD4⁺ cells tolerant to oxidative stress and apoptosis, while promoting CD8⁺ T cell cytotoxicity and reducing exhaustion markers at the tumor site.⁶⁴ By sustaining less differentiated T cells with enhanced longevity, tazemetostat supports durable CAR-T responses, which is particularly critical for overcoming relapse.

To assess whether EZH2 inhibition also acts directly on CAR-T cells, we pretreated CAR-T with tazemetostat before the functional assays but did not observe a beneficial effect. We also exposed CAR-T to tazemetostat during manufacturing and did not observe major changes in CAR-T function. Lastly, we directly KO EZH2 in CAR-T and showed no improvement in anti-tumor function. These results are in line with a recent paper showing that tazemetostat did not damage T cells and did not impact manufacturing and functionality of patient T cells.⁴⁶ Therefore, we concluded that the main mechanism by which EZH2 and EZH1 inhibition enhances CAR-T function by directly modulating the tumor.

Targeting EZH2 in combination with immunotherapy might be a compelling strategy for solid tumors, where EZH2 acts as an oncogene²⁰ and immunotherapies are challenging due to factors such as immune evasion, poor T cell trafficking, and limited persistence.⁶⁵ In this study, we further evaluated combining EZH2 inhibition with CAR-T and TCR-T therapies in solid tumor models where EZH2 acts as an oncogene. Notably, tazemetostat, the first epigenetic drug approved for solid tumors, is indicated for metastatic/advanced Ewing sarcomas not eligible for resection.³⁰ Pretreatment with tazemetostat sensitized HER2+ prostate and ovarian cancer cell lines to apoptosis by anti-HER2 CAR-T cells. Additionally, in an Ewing sarcoma model, tazemetostat pretreatment enhanced the cytotoxicity of LOXHD1-specific TCR-T cells.⁵³

In summary, our research unveils an approach to augment adoptive T cell therapies through the utilization of EZH1/EZH2 inhibitors. Moreover, it offers mechanistic insights into the role of EZH2 and EZH1/EZH2 in regulating tumor cells and T cells during immunotherapy. These findings lay the foundation for genetic engineering advancements in next-generation CAR-T products, either by modulating their epigenetic modifiers or engineering CAR-T cells to deliver epigenetic modulators directly to cancer cells. Lastly, our work is promising for enhancing T cell therapies’ efficacy in solid cancers, where CAR/TCR-T cells have historically fallen short.

Limitations of the study

Our study is limited by the lack of a fully functional TME in the human xenograft models used. Recent studies highlighted EZH2’s critical role in regulating immunosuppressive cells within the TME.^66–68 Nevertheless, in a companion publication, we showed that genetic modulation or pharmacological inhibition of EZH2 reprogrammed the TME in clinically relevant murine models of lymphoma and enhanced CAR-T therapy.⁴⁶ Moreover, this study does not address the sequential contribution of different pathways altered by EZH1/EZH2i at the tumor level. For example, to directly establish the role of EZH2 in regulating antigen presentation within our models, future studies utilizing TCR-KO murine models will be critical. Furthermore, this study is not powered to capture the potential side effect of this combination. While single-agent studies of tazemetostat and valemetostat report a promising safety profile,^37,69,70 we demonstrated that these agents could enhance CAR-T expansion and function, potentially leading to enhanced cytokine-release syndrome. To this goal, phase 1–2 clinical trials are ongoing to assess the safety of this approach (NCT05934838 and NCT05994235).⁴⁶

RESOURCE AVAILABILITY

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Marco Ruella, MD (mruella@upenn.edu).

Materials availability

The cell lines generated in this study are available from the lead contact with a completed materials transfer agreement.

Data and code availability

Bulk RNA-seq, scRNA-seq, and ATAC-seq data have been deposited on the NCBI Gene Expression Omnibus via GSE265799, GSE267074, GSE266249, and GSE285897.

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Cells and generation of genetically engineered lymphoma cell lines

The study utilized four different cell lines of Germinal Center Diffuse Large B-cell Lymphoma (GBC-DLBCL): OCI-Ly18 [wt EZH2], Toledo [wt EZH2], SU-DHL-4 [mut EZH2; Y641F], and Karpas-422 [mut EZH2; Y641N]. All cell lines were transduced with a lentivirus encoding click beetle green (CBG) and green fluorescent protein (GFP). For EZH2 knockdown experiments, the EZH2-targeting shRNA-475 (TRCN0000010475 [475]) from the RNA Consortium (GE Dharmacon) was used as described in Porazzi et al.²² SU-DHL-4 cells were infected with lentivirus by three cycles of spinoculation (1,000 g, 45 min, 37°C) with subsequent incubation at 37°C for 24 h. Then, cells were selected with 0.5 μM puromycin for 48–72 h. One human HER-2+ ovarian cancer cell line was used (ovarian adenocarcinoma: SK-OV-3). One human HER-2+ prostate cancer cell line was used (prostate adenocarcinoma: PC-3). Two CD33⁺ acute myeloid leukemia (AML) cell lines were used (KG-1, THP-1). One multiple myeloma (MM) cell line was used: RPMI-8226. One Ewing Sarcoma cell line was used (A673). Unless otherwise specified, all cell lines were cultured in R10 media (Roswell Park Memorial Institute medium 1640 (RPMI; Gibco; Cat#11875–085) supplemented with 10% fetal bovine serum (FBS, Gibco; Cat# 16140–071), 1% penicillin and 1% streptomycin (Gibco; Cat# 15140–163), 1% GlutaMAX supplement (Gibco; Cat# 35050–079, and 1% HEPES (Gibco; Cat# 15630–130) in a 37°C incubator with 5% CO₂. RPMI-8226 was cultured in R20 media (supplemented with 20% FBS). All cell lines were authenticated by short tandem repeat (STR) analysis and tested for mycoplasma using a MycoAlert Plus Mycoplasma Detection Kit (Lonza; Cat# LT07–710). SK-OV-3 were grown and cultured in “DMEM10” (Dulbecco’s Modified Eagle Medium (DMEM) + 10% FBS, 1% Penicillin/Streptomycin, 1% HEPES, 1% GlutaMAX), and maintained at a confluency of ~60–70% under standard conditions.

Immunodeficient mice and xenograft models

Six- to 8-week-old NSG (NOD SCID gamma) mice were obtained from the Stem Cell and Xenograft Core at the University of Pennsylvania. NSG mice were housed in pathogen-free conditions. The OCI-Ly18-Luc+ subcutaneous model was established as our lab has described previously.⁸¹ Tumor volume (mm³) = ½ (L × W ²), where L (mm) is the major axis of the tumor and W (mm) is the minor axis perpendicular to L. Tumor size was monitored each week via caliper and bioluminescence. In the early treatment/survival model, 4 × 10⁶ OCI-Ly18 cells were implanted under the skin of the right flank. Tazemetostat (200 mg/kg) or vehicle were delivered via oral gavage daily, starting 3 days after implanting the tumors and continuing for ~8 weeks. Mice were randomized based on bioluminescence and infused with 3.5 × 10⁶ CAR19+ T cells via tail vein one week after tumor inoculation. Mice were imaged weekly after placing mice on their left side (left lateral recumbent position) using the IVIS Lumina S5. In the stress-test model, 4 × 10⁶ OCI-Ly18 cells were implanted under the skin of the right flank. Tazemetostat (150 mg/kg) or vehicle was delivered via oral gavage daily, starting 3 days after implanting the tumors. 14 days after implantation, subcutaneous tumors had reached a size ~20–40 mm³. Mice were then randomized and infused with 4.0×10⁶ CAR19+ T cells via tail vein. For the multiple myeloma xenograft model, RPMI-8226-Luc+ cells were injected intraosseously (1 × 10⁶/mouse) into six- to 8-week-old NSG (NOD SCID gamma) mice. Tumor progression was monitored each week via bioluminescence. Valemetostat (100 mg/kg) or vehicle were delivered via oral gavage daily, starting 7 days after implanting the tumors and continuing for ~8 weeks. Mice were randomized based on bioluminescence and infused with 0.8 × 10⁶ CARTBCMA+ T cells via tail vein two weeks after tumor inoculation. Mice were imaged weekly using the IVIS Lumina S5.

Animals were continuously monitored for signs of disease progression and overt toxicity, such as xeno-Graft-Versus-Host-Disease (GVHD), as evidenced by >20% loss in body weight, excessive fur loss, diarrhea, conjunctivitis, disease-induced hind limb paralysis. Animals were also continuously monitored for signs of distress affecting gait or normal posture. NIH Guidelines were followed for animal care and use. All experimental protocols and endpoints, including tumor volume thresholds of up to 4.2 cm³ for the OCI-Ly18 model, were approved by the University of Pennsylvania Animal Care and Use Committee (IACUC).

METHOD DETAILS

Lentiviral vector production, CAR-T and TCR T Cell manufacturing

Replication-defective, third-generation lentiviral vectors were produced using HEK293T cells. Approximately 8×10⁶ cells were plated in T150 culture vessels in standard culture media and incubated overnight at 37°C. 18–24 h later, cells were transfected using a combination of Lipofectamine 2000 (116 μL, Invitrogen; Cat#11668–019), pVSV/G or pCocal (7 μg), pRSV/Rev (18 μg), pGag/Pol (18 μg) packaging plasmids and 15 μg of expression plasmid (CART19, FMC63⁷²; CAR22-short⁷⁶; CAR79b; CART33, GO⁷⁴; CART-HER2, 4D5⁷⁵; CARTBCMA⁷³; TCR-LOXHD1 (HLA-A*02:01)). Lipofectamine and plasmid DNA were diluted in 4 mL Opti-MEM media (Gibco; Cat# 31985–070) before transfer into lentiviral production flasks. At both 24 and 48 h following transfection, culture media was isolated and concentrated using high-speed ultracentrifugation (8,500 rpm overnight or 25,000 rpm for 2.5 h). All CAR constructs were composed of a scFv, 4–1BB costimulatory domain, and CD3ζ costimulatory domain, unless otherwise noted. Human CD4⁺ and CD8⁺ T cells from healthy donors were sourced from the University of Pennsylvania Human Immunology Core, mixed 1:1 and stimulated with CD3/CD28 stimulatory beads (Thermo Fisher) at a bead-to-cell ratio of 3:1. Lentiviral vectors (lentiviral vectors encoding CAR) were then introduced to the stimulated T cell cultures at a multiplicity of infections ranging between 1 and 2. Stimulatory beads were removed on day 6 and expansion was monitored daily and frozen when cells reached ~350 fL. CAR positivity was tested by flow cytometry using an anti-CD19-PE antibody (Novartis) or anti-G4S-PE linker antibody (Cell Singaling #38907). Synthetic TCR DNA vector constructs were synthesized (TWIST Bioscience, San Francisco, CA) as previously described⁸² to include TCRα and TCRβ chains separated by a T2A sequence. TCRα and TCRβ constent domains were codon altered to be resistant to Cas9 protein riboprobes targeting endogenous TRAC and TRBC1/TRBC2. TCR-engineered were generated as previously described.⁸² Briefly, human primary CD8⁺ T cells were activated and expanded using anti-CD3/CD28 antibody–conjugated paramagnetic microbeads (Life Technologies); after 16h, T cells were transduced with lentiviral vector particles and maintained in media supplemented with IL-7 (5 ng/ml; R&D Systems) and IL-15 (5 ng/ml;R&D Systems). On day 5, T cells were edited for endogenous TCR using CRISPR/Cas9 and cultured in media supplemented with IL-7 and IL-15 for a total of 14 days followed by rapid expansion method for an additional 14 day culture period. Engineered TCR expression was assessed by p-HLA multimer staining.

Manufacturing of primary EZH2-knockout CAR T cells

CD4⁺ and CD8⁺ cells were combined at a 1:1 ratio and used for electroporation. CRISPR-Cas9 sgRNA specific for EZH2 (CAAGGGCACGAACTGTCACA)⁷¹ were chemically synthesized (Integrated DNA Technologies). Ten μg sgRNA were premixed with 10 μg of TrueCut Cas9 Protein v2 (Invitrogen; Cat# A36499) for 10 minutes at room temperature to form a ribonucleoprotein (RNP) complex prior to electroporation. 5×10⁶ T cells in 100 μL of the buffer provided with the P3 Primary Cell 4D-Nucleofector X Kit L (Lonza; Cat# V4XP-3024) were mixed with the RNP complex and subsequently electroporated using the pulse code EO-115 in a 4D-Nucleofector (Lonza; Cat# AAF-1002B). After electroporation, T cells were incubated at 37°C for 24 h and subsequently activated using CD3/CD28 Dynabeads (Gibco; Cat# 40203D) at a ratio of 3 beads/cell. The following day, a CAR19–41BB-CD3z lentiviral vector was added to stimulated cultures at a multiplicity of infection (MOI) of 1, as described in the previous section (CAR-T and TCR T Cell Manufacturing). Beads were removed at Day 5 of stimulation. Cells were counted every other day using a Moxi GO II (Orflo) until growth kinetics and cell size demonstrated they had rested from stimulation. All T cells were initially grown with 20 ng/mL of supplemental cytokines IL-7 and IL-15 that was decreased to 0 ng/mL by the end of the expansion. For EZH2, genomic DNA was isolated 5 days post electroporation via the Qiagen DNEasy Blood and Tissue Kit. PCR was performed to generate amplicons of CRISPR-edited exon (Primers: Forward: GACTCTTGGCTTTAACGCATTCC Reverse: GTCCTTGACCATTCCAATAGC). The amplicon was purified and submitted for Sanger sequencing; subsequently, editing efficiency (KO Score) was assessed via TIDE (Tracking of Indels by Decomposition).

EZH1/2 inhibitors

The EZH2 inhibitor tazemetostat (EPZ6438, Cayman Chemicals, #16174) was resuspended in DMSO for a stock solution of 5 mM. The EZH1/2 inhibitor valemetostat (DS-3201, Selleck Chemicals, #S8926) was resuspended in DMSO for a stock solution of 5 mM. For in vivo administration, tazemetostat was dissolved in the vehicle solution 0.5% hydroxypropyl-methylcellulose, 0.1% tween-80, and administered at 200 or 150 mg/kg once a day by oral gavage. For in vivo administration, valemetostat tosylate was dissolved in sterile water, and administered at 100 mg/kg once a day by oral gavage.

Western Blot analysis

Protein extracts from DLBCL cell lines (OCI-Ly18, Toledo, SU-DHL-4 and Karpas-422) were prepared in RIPA buffer (Research Products International; #R26200–250.0) and protease inhibitor cocktail (Cell Signaling Technology, #5871S). Protein samples (20μg/sample) were resolved on SDS-PAGE gels (4–20% Mini-PROTEAN^® TGX^™ Precast Protein Gels, #4561094), transferred to a 0.2μm nitrocellulose membrane, blocked with Intercept^® (PBS) Blocking Buffer (# 927–70001, LI-COR Biosciences) and incubated with the following primary antibodies (β-Actin, Cell Signaling Technology, #3700S; EZH2, Cell Signaling Technology, #5246; Tri-Methyl-Histone H3 (H3K27me3), Cell Signaling Technology, #9733) at 4C overnight (O/N) followed by secondary antibody (LI-COR Biosciences, Goat anti-Rabbit IgG Secondary Antibody, #926–32211; or IRDye^® 680LT Goat anti-Mouse IgG2b-Specific Secondary Antibody, #926–68052) at room temperature. Proteins were visualized using a chemiluminescence instrument and analyzed using ImageJ software.

Cytotoxicity assays

Bioluminescence-based cytotoxicity assays

Cell lines (OCI-Ly18, Toledo, SU-DHL-4, Karpas-422) were engineered to express CBG/GFP, and cell survival was measured using bioluminescent quantification. D-luciferin potassium salt (Gold Biotechnology; Cat# 115144–35–9) was added to the cell cultures at a final concentration of 15 μg/mL and incubated at 37°C for 10 minutes, after which bioluminescence signals were captured using a BioTek Synergy H4 imager and analyzed using BioTek Gen5 software. To determine the efficacy of tumor elimination, the results were normalized to control cells treated with DMSO/UTD.

Flow-cytometry-based (extended coculture; long term killing assays) cytotoxicity assays

CAR T cells were combined with target cancer cells at a given effector: target ratio, and cocultures were evaluated for absolute count of T cells (anti CD3-APC, Beckman-Coulter #IM2467) and cancer cells (GFP+) by flow cytometry using Flow-Count Fluorospheres (Beckman-Coulter) at two- or three-day intervals. Cocultures were maintained at an absolute concentration of 1 × 10⁶ cells/mL.

CellCyte X-based cytotoxicity assays

72 hours before co-culture, GFP+ tumor cells (OCI-Ly18, SU-DHL-4, RPMI-8226) were treated with tazemetostat, velemetostat, or DMSO. GFP+ tumor cells (4,000 cells/well in U-bottom 96-well plates; 200 μL R10 media) were then co-cultured with untransduced T cells or CAR T cells, with or without tazemetostat or velemetostat. Plates were briefly spun (100g × 1 min) to collect cells at the bottom, and GFP+ tumor cell growth was monitored using a CellCyteX equipped with a 43 objective. Wells were imaged every 8 hours for the indicated time period, and GFP intensity in each well was quantified over time using ImageJ and expressed as total fluorescence intensity.

IncuCyte-based cytotoxicity assays

Seventy-two hours before co-colture, cells were treated with tazemetostator DMSO; 24 hr prior to coculture with CAR T cells, GFP+ adherent cell lines (SK-OV-3 and PC-3) were seeded at 4,000 cells/well in a 96-well plate (100 μL R10 media). After 24 hr, cancer cells were treated with untransduced T cells, or CAR T cells with or without tazemetostat. Each condition was plated in at least triplicate. Cells were imaged at 3–4-hour intervals using the IncuCyte SX5 Live-Cell Analysis System and a 20X objective for < 160 hr with data collection by the Incucyte ZOOM software. GFP+ count, intensity, and total confluence were recorded at each time point. One-way ANOVA with post hoc Tukey tests was used to determine statistical significance for each experimental effector T cell condition relative to untransduced T cells at the experimental endpoint.

xCELLigence real-time apoptotic cell death analysis

Real-time apoptotic cell death analysis (live cell imaging with cellular impedance) was performed to assess extended cytotoxic activity using the xCELLigence Real Time Cell Analysis eSight system (ACEA Biosciences). Seventy-two hours before co-colture, cells were treated with tazemetostat or DMSO; 24 h prior to coculture with effector T cells, mCherry+ cell line (A673) was plated (1 × 10⁴ cells/well) and allowed to adhere for 24 h. Effector T cells were added at an E:T ratio of 3:1. Time lapse video monitoring was performed with acquisition of brightfield and red (mCherry) every hour for 96 h. Concurrent cell index (relative cell impedance) was monitored every 15 min. Data were normalized to the maximum total integrated intensity or cell index value immediately following effector-cell plating. Shaded lines reflect the mean of replicate wells ± SD.

CellTrace CFSE proliferation assays

T cells were resuspended in PBS with CFSE stain (1:1000 dilution), at a concentration of 1e6 per mL, and incubated for 15 min at 37 C. T cells were washed with 10 mL R10 media, resuspended, and plated in coculture with irradiated cancer cells (OCI-Ly18) with or without increasing doses of tazemetostat. Flow cytometry was used to trace multiple generations using dye dilution.

Multiplexed immunohistochemistry and image analysis

Explanted tumor specimens (from sub-cutaneous OCI-Ly18 tumor models) were fixed in 10% formalin at room temperature for 48 h. Specimens were washed in PBS, stored in 70% ethanol at 4°C and processed and embedded in paraffin to create formalin-fixed paraffin-embedded (FFPE) blocks. FFPE tumor specimens were sectioned at 5μm onto positively charged glass microscope slides and baked at 56°C overnight. Automated multiplexed immunohistochemistry (mIHC) was performed on a Ventana Discovery Ultra automated slide staining system (Roche). Slides were deparaffinized at 70°C for 24 minutes. Following deparaffinization, heat-induced epitope retrieval was performed using an EDTA-based preconditioning solution (Discovery CC1, Roche, cat# 950–500) for 64 minutes at 100°C. Endogenous peroxidases were neutralized with Discovery Inhibitor (Roche, cat# 760–4840), and S Block (Roche, cat# 760–4212) was applied prior to the addition of antibodies to prevent non-specific binding. Slides were incubated with primary and secondary antibodies prior to chromogenic signal detection using the reagents and antibodies listed in STAR Methods Table. Tissues were counterstained with hematoxylin and bluing reagent. Whole slide scans of stained tissues were acquired at 40X using an Aperio CS2 scanner (Leica Biosystems). Tumor regions were identified and analyzed in Visiopharm Integrator System (VIS) software (Version 2023.09). CD3 was quantified as cell count per mm² and CD19 positivity was quantified as % of area.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
β-Actin (8H10D10) Mouse mAb	Cell Signaling Technology	Cat#3700; RRID: AB_2242334
EZH2 (D2C9) XP® Rabbit mAb	Cell Signaling Technology	Cat#5246; RRID: AB_10694683
Tri-Methyl-Histone H3 (Lys27) (C36B11) Rabbit mAb	Cell Signaling Technology	Cat#9733; RRID: AB_2616029
800CW Goat anti-Rabbit IgG Secondary Antibody	LICOR Biology	Cat#925–32211; RRID AB_2651127
IRDye® 680LT Goat anti-Mouse IgG2b-Specific Secondary Antibody	LICOR Biology	Cat# 926–68052; RRID AB_2783644
GAPDH (D16H11) XP® Rabbit mAb	Cell Signaling Technology	Cat#5174S; RRID: AB_10622025
G4S Linker-PE (E7O2V) Rabbit mAb	Cell Signaling	Cat#38907; RRID: AB_3626304
HQ-HRP	Ventana	Cat#760–4820; RRID: AB_3068525
IgG Goat anti-Rabbit-NP	Ventana	Cat#760–4817; RRID: NA
IgG Goat anti-Rabbit-HQ	Ventana	Cat#760–4815; RRID: AB_2811171
NP anti-Mouse-AP	Ventana	Cat#760–4827; RRID: NA
IgG Goat anti-Rabbit-HRP	Ventana	Cat#760–4311; RRID: AB_2811043
Human CD3-Unconjugated	Ventana	Cat#790–4341
Human CD3-Alexa Fluor® 700	Biolegend	Cat#317340; RRID: AB_2563407
Human CD3-APC	Beckman Coulter	Cat#IM2467; RRID: AB_130788
Human CD3-PE	Beckman Coulter	Cat#IM1282U; RRID: AB_10640418
Human CD5-APC	Biolegend	Cat#985302; RRID: AB_2892520
Human CD19-Unconjugated	Cell Signaling Technology	Cat#90176S; RRID: AB_2800152
Human CD19-PE	Novartis Pharmaceuticals	N/A
Mouse CD31-Unconjugated	Cell Signaling Technology	Cat#77699; RRID: AB_2722705
Human CD45RA-PB	Biolegend	Cat#A82946; RRID: AB_2904200
Human CD45-FITC	Beckman Coulter	Cat#A07782; RRID: AB_10645157
Mouse CD45-AF700	Biolegend	Cat# 103127 RRID: AB_493714
Human CD58-APC	Beckman Coulter	Cat#IM3701; RRID: N/A
Human CD197(CCR7)-PC7	Beckman Coulter	Cat#B46025; RRID: N/A
Human Whitlow/218 Linker-PE	Cell Signaling Technology	Cat#62405; RRID: AB_3626306
ViaKrome 808 Fixable Viability Dye	Beckman Coulter	Cat#C36628; RRID: N/A
Human TruStain FcX™ (Fc Receptor Blocking Solution)	Biolegend	Cat#422302; RRID: AB_2818986
TruStain FcX™ (anti-mouse CD16/32) Antibody	Biolegend	Cat#101320; RRID: AB_1574975
Quantum™ Simply Cellular® anti-mouse lgG	Bangs Laboratories, Inc.	Cat#815B
Biological samples
Primary human T cells from healthy donors	University of Pennsylvania’s Human Immunology Core	N/A
Chemicals, peptides, and recombinant proteins
D-luciferin potassium salt	Gold Biotechnology	Cat#115144–35–9
Tazemetostat (EPZ6438)	Cayman Chemicals	Cat#16174
Valemetostat (DS-3201)	Selleck Chemicals	Cat#S8926
CTS™ Dynabeads™ CD3/CD28	Thermo Fischer Scientific, Gibco	Cat#40203D
TrueCut Cas9 Protein v2	Invitrogen	Cat#A36499
Flow-Count Fluorospheres	Beckman Coulter	Cat#7547053
Nucleofector Solution Set P3 For Primary Cells	Lonza	Cat#PBP3–02250
Human IL-7	Miltenyi Biotech	Cat#170–076– 111
Human IL-15	Miltenyi Biotech	Cat#170–076–114
RIPA buffer	Research Products International	Cat#R26200–250.0
Protease inhibitor cocktail	Cell Signaling Technology	Cat#5871S
Intercept® (PBS) Blocking Buffer	LI-COR Biosciences	Cat# 927–70001
Discovery Inhibitor	Roche	Cat#760–4840
S Block	Roche	Cat#760–4212
EDTA-based preconditioning solution	Discovery CC1, Roche	Cat#950– 500
Critical commercial assays
Pierce Rapid Gold BCA Protein Assay Kit	Thermo Scientific	Cat#A55860
ZymoPURE II Plasmid Maxiprep Kit	Zymo Research	Cat#D4202
Pan T Cell Isolation Kit, human	Miltenyi Biotech	Cat#130–096–535
Quick-RNA Microprep Kit	Zymo Research	Cat#R1050
DNeasy Blood Kit Tissue Kit	QIAGEN	Cat#69504
P3 Primary Cell 4D-Nucleofector X Kit L	Lonza	Cat#V4XP-3024
ELISA MAX Standard Set Human IFN-γ	BioLegend	Cat#431801
ELISA MAX Standard Set Human IL-2	BioLegend	Cat#431801
ELISA MAX Standard Set Human TNF-α	BioLegend	Cat#430201
NEBNext® Ultra™ II Directional RNA Library Prep Kit for Illumina®	New England BioLabs	Cat#E7760L
ATAC-Seq Kit	Active Motif	Cat#53150
DNA Clean and Concentrator	Zymo Research	Cat#D4033
DISCOVERY Purple kit (RUO)	Ventana	Cat#760–229
DISCOVERY Teal HRP Kit (RUO)	Ventana	Cat#760–247
DISCOVERY Yellow Kit (RUO)	Ventana	Cat#760–239
CellTrace CFSE Cell Proliferation Kit	Invitrogen	Cat#50–591–407
CellTrace Far-Red	Invitrogen	Cat#C34564; RRID: N/A
Deposited data
Raw sequencing data	This paper	GSE265799, GSE267074, GSE266249, GSE285897
Experimental models: Cell lines
Human: SU-DH-L-4	ATCC	RRID: CVCL_0539
Human: OCI-Ly18	DSMZ	CVCL_1880
Human: Karpas-422	ATCC	RRID: CVCL_1325
Human: Toledo	ATCC	RRID: CVCL_3611
Human: KG-1	ATCC	RRID: CVCL_0374
Human: THP-1	ATCC	RRID: CVCL_0006
Human: SK-OV-3	ATCC	RRID: CVCL_0532
Human: PC3	ATCC	RRID: CVCL_E2RM
Human: A-673	ATCC	RRID: CVCL_0080
Human: RPMI-8226	ATCC	RRID: CVCL_0014
Human: HEK-293T	ATCC	RRID: CVCL_0063
Experimental models: Organisms/strains
Mouse: NOD-SCID-γc−/− (NSG)	Jackson Laboratory	Cat# JAX:005557; RRID: IMSR_JAX:005557
Oligonucleotides
Primers for Figure S5A Forward: GACTCTTGGCTTTAACGCATTCC Reverse: GTCCTTGACCATTCCAATAGC	N/A	N/A
gRNA for EZH2-KO (targeting exon 14): CAAGGGCACGAACTGTCACA	Liscovitch-Brauer et al.,71	N/A
shRNA for EZH2-KD mature antisense: TGAAGCTAAGGCAGCTGTTT	Porazzi et al.,22	N/A
Recombinant DNA
antiCD19-CART: pTRPE.CD19.41BB.CD3-ζ	Kalos et al.,72	N/A
antiBCMA-CART: pELPS.BCMA.41BB.CD3-ζ	Bu et al.,73	N/A
antiCD33-CART: pTRPE.CD33.41BB.CD3-ζ	Kenderian et al.,74	N/A
antiHER2-CART: pTRPE.4D5.41BB.CD3-ζ	Liu et al.,75	N/A
TCR-LOXHD1 (HLA-A*02:01)	Deng et al.,53	N/A
antiCD22-m971 short: pTRPE.CD22.41BB.CD3-ζ	Singh et al.,76	N/A
antiCD79b: pTRPE-CD79bPola	our lab	N/A
Software and algorithms
ImageJ	https://imagej.net/ij/	N/A
GraphPad Prism (version 9.2.0 or higher)	GraphPad Software Inc.	N/A
Lumicks Z-Movi Software	LUMICKS	N/A
FlowJo (version 10.8.0)	FloJo, LLC	N/A
Incucyte ZOOM software	Sartorius	N/A
IncuCyte SX5 Live-Cell Analysis System	Sartorius	N/A
SnapGene v5.2.3 SnapGene	SnapGene v5.2.3	N/A
Visiopharm Integrator System (VIS) software (Version 2023.09)	Visiopharm	N/A
Trimmomatic (version 0.36)	KBase Predictive Biology	https://github.com/usadellab/Trimmomatic/releases
STAR (version 2.6.0c)	Dobin et al.,77	https://github.com/alexdobin/STAR/releases/tag/2.6.0c
featureCounts (version 1.6.1)	Liao et al.,78	https://github.com/torkian/subread-1.6.1
DESeq2 (version 1.22.2)	Love et al.,79	https://bioconductor.org/packages/release/bioc/html/DESeq2.html
R Studio (version 1.2.5042, 4.0 or higher)	Posit	N/A
Gene Set Enrichment Analysis (GSEA) Mac App (version 4.1.0)	Broad Institute	http://software.broadinstitute.org/gsea/index.jsp
Seurat	Hao et al.,80	https://satijalab.org/seurat/

General flow cytometry

Cells were resuspended in FACS staining buffer (PBS + 2% FBS) using the following antibodies: human CD3 (clone UCHT1, Beckman-Coulter), human CD45 (clone J33, Beckman-Coulter #IA07782), or mouse CD45 (clone 30-F11, Biolegend), human CCR7 (clone G043H7, Beckman-Coulter) human CD45RA (clone 2H4, Beckman-Coulter). CAR expression was detected using a PE-conjugated anti-G4S linker antibody (see above). To monitor cell proliferation, cells were stained with CFSE according to manufacturer’s instructions (Invitrogen). For quantification of absolute cell numbers (tumor or T cells) acquired during flow cytometry, FlowCount FluoroSpheres were used (BeckmanCoulter). Cell viability was established using ViaKrome 808 Fixable Viability Dye (Beckman Coulter). Unless otherwise specified, all gating strategies began in the following order: SSC-A vs. FSC-A (Lymphocytes), FSC-A vs. FSC-H (Singlets), FSC-A vs. Live/Dead (Viability). All data were acquired on Beckman Coulter CytoFLEX S Flow Cytometer. All data analysis was performed using FlowJo version 10.8.1 (FlowJo, LLC).

Enzyme-linked immunosorbent assay (ELISA)

Target cells were pre-incubated with DMSO or tazemetostat for 3 days, and then incubated with effector cells at 1:0.25 E:T ratios in standard media for 48 h. Supernatant was collected and stored at −80°C if not used immediately. Analyses were performed using ELISA MAX^™ standard set human IFN-γ (BioLegend, #430107), IL-2 (BioLegend, #431801) and TNFA (BioLegend, #430201) kits.

Retro orbital bleeding, peripheral blood staining for flow cytometry and analysis

Mice were anesthetized with isoflurane, and 100–200 μL of blood was collected via retro-orbital bleeding using heparinized capillary tubes. Blood was treated with ACK lysis buffer for 5–10 minutes at room temperature to lyse red blood cells, then centrifuged at 500 × g for 5 minutes. The pellet was washed with PBS and centrifuged again. For flow cytometry, cells were resuspended in 100 μL of flow buffer and stained with anti-mouse CD45 (FITC), anti-human CD45 (PE), and anti-human CD3 (APC) antibodies for 30 minutes at 4°C. After washing, samples were resuspended in 300 μL of buffer, and counting beads were added for quantification. Flow cytometry was used to distinguish murine and human CD45⁺ populations, and human CD3⁺ cells were identified within the human CD45⁺ population.

Cell-to-cell avidity assays

Avidity was measured using the Lumicks Z-Movi assay. Z-Movi-compatible acoustofluidic chips were coated with poly-L-lysine for 3 h prior to attaching a monolayer of CAR-T cells. CellTrace far-red labeled tumor cells were allowed to incubate on the monolayer of CAR-T cells for 10 min, and then a ramping acoustic force was applied. Cell-detachment was analyzed using ImageJ and R. Avidity experiments were conducted according to the manufacturer’s (LUMICKS^™) instructions.

Bulk RNA sequencing and analysis

Total RNA was extracted using Zymo Quick-RNA Microprep Kit (car#R1050). RNA sequencing was performed by the Novogene Sequencing Service (Novogene CSS America) on a NovaSeq 6000 Sequencing System (Illumina) as 100 base single-end reads using an NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (New England BioLabs; Cat# E7760L). Reads of the samples were trimmed for adapters and low-quality bases using Trimmomatic (version 0.36) before alignment with the hg19 reference genome and the annotated transcripts using Spliced Transcripts Alignment to a Reference (STAR; version 2.6.0c).⁷⁷ Gene expression quantification analysis was performed for all samples using STAR and featureCounts (version 1.6.1).⁷⁸ Differential expression analyses between samples were computed using DESeq2 (version 1.22.2)⁷⁹ within R Studio (version 1.2.5042). Pathway enrichment was performed using the Gene Set Enrichment Analysis (GSEA) Mac App (version 4.1.0) with the Hallmark gene set from the Molecular Signatures Database.^83–85 Visualization of differential gene expression was performed by volcano plot depicting log 2 fold change on the x-axis and False Discovery Rate (FDR) adjusted p value on the y-axis.

Single-cell RNA sequencing and analysis

For single-cell sequencing experiments, peripheral blood or single cell suspension from sub-cutaneous tumors were collected from mice and red blood cell lysis was performed. Samples were then sorted using anti-hCD3, anti-hCD5 antibodies to stain for human T cells and GFP (for tumor cells) using a BD FACSMelody^™ Cell Sorter. Cells were then tagged, washed twice with 0.04% BSA in PBS and loaded onto the 10X Chromium controller. Library preparation was performed using the v2 Chemistry kit (10X Genomics). After NGS and mapping, data were filtered in Seurat⁸⁰ with the following parameters prior to clustering: 200 < number of RNA features < 5500; percent mitochondrial genes < 10. In addition, the filters MS4A1<1 and CD2>1 were applied to remove any residual contaminating tumor cells. Clustering in both experiments was driven by a predefined list of canonical/classical T cell genes as described by Szabo et al.⁴⁹ Additional details on cell processing and handling are provided in Figure legends.

ATAC sequencing

ATAC-seq libraries were prepared using the Active Motif ATAC-Seq kit following the manufacturer’s protocol. Nuclei from 50,000 sorted SU-DHL-4 cells from all the experimental conditions were isolated and were subjected to the transposition reaction using Tn5 transposase for 30 minutes at 37°C with 1000rp mixing. The transposed DNA was purified using DNA Clean and Concentrator (Zymo) and fragments were barcoded with unique i5 and/or i7 indexed primers. Final libraries were double size selected using AM-Pure beads prior to sequencing. Sequencing was performed on an Illumina NextSeq 2000 instrument to produce 75-bp paired-end reads.

QUANTIFICATION AND STATISTICAL ANALYSIS

Sample size was not predetermined using statistical methods as there was no predetermined expected effect. No data were artificially removed from statistical analysis. All in vitro data presented are representative of at least two independent experiments except for ex vivo-based scRNA-seq experiments. Sequencing results were processed in R version 4.0 or higher. All comparisons between two groups were performed using unpaired Student’s t-test unless otherwise specified in the figure legends. All comparisons between three or more groups were performed using one-way ANOVA with post hoc Tukey tests, unless otherwise specified. All results are represented as mean ± SD unless otherwise noted. Survival data were analyzed using the log-rank (Mantel-Cox) test. Data analysis was performed using Prism v9.2.0 or higher (GraphPad software).

Supplementary Material

SUPPLEMENTAL INFORMATION

Supplemental information can be found online at https://doi.org/10.1016/j.ccell.2025.01.013.

Highlights.

• EZH1/2 inhibition enhances the efficacy of CAR-T and TCR-T cells in multiple cancer models

• EZH1/2 inhibition rewires cancer cells to a more immunogenic state

• Inhibition of tumor EZH1/EZH2 enhances CART activation, expansion, and infiltration