Suboptimal functional persistence limits the efficacy of adoptive T-cell therapies. CD28-based chimeric antigen receptors (CAR) impart potent effector function to T cells but with a limited lifespan. We show here that the genetic disruption of SUV39H1, which encodes a histone-3, lysine-9 methyl-transferase, enhances the early expansion, long-term persistence, and overall antitumor efficacy of human CAR T cells in leukemia and prostate cancer models. Persisting SUV39H1-edited CAR T cells demonstrate improved expansion and tumor rejection upon multiple rechallenges. Transcriptional and genome accessibility profiling of repeatedly challenged CAR T cells shows improved expression and accessibility of memory transcription factors in SUV39H1-edited CAR T cells. SUV39H1 editing also reduces expression of inhibitory receptors and limits exhaustion in CAR T cells that have undergone multiple rechallenges. Our findings thus demonstrate the potential of epigenetic programming of CAR T cells to balance their function and persistence for improved adoptive cell therapies.

Significance:

T cells engineered with CD28-based CARs possess robust effector function and antigen sensitivity but are hampered by limited persistence, which may result in tumor relapse. We report an epigenetic strategy involving disruption of the SUV39H1-mediated histone-silencing program that promotes the functional persistence of CD28-based CAR T cells.

See related article by López-Cobo et al., p. 120.

This article is featured in Selected Articles from This Issue, p. 5

Chimeric antigen receptors (CAR) are synthetic receptors that redirect T-cell antigen specificity and reprogram T-cell function (1). CAR therapy has changed the therapeutic landscape of several hematologic malignancies, including B lineage leukemias, non–Hodgkin lymphoma, and multiple myeloma (2, 3). Nonetheless, a significant fraction of patients will eventually relapse after initially responding to CAR therapy (4–7). Several mechanisms accounting for resistance or tumor relapse after CAR therapy have been identified. These include poor T-cell function or persistence (8, 9), low or absent target antigen expression (10, 11), inadequate tumor penetration and microenvironmental immune suppression, particularly in solid tumors (12–15). Insufficient persistence and dysfunction of CAR T cells have been associated with tumor relapse in several instances in both murine models (8, 16) and clinical trials (9, 17)

The CAR structure, its level of expression and signaling properties are critical determinants of function and persistence of engineered T cells (18–20). CD28-based CARs impart potent effector function and greater sensitivity to low antigen density relative to 4-1BB–based CARs (10, 11), but lesser persistence in preclinical models (21, 22) and in patients (20). Thus, enhancing the functional persistence of CD28-based CARs may improve clinical outcomes.

T cells, upon antigen exposure, undergo differentiation to memory and effector progeny (23). Effector T cells are endowed with potent cytolytic function but exhibit limited persistence (24, 25). Memory cells display lesser effector function but are endowed with greater functional persistence and may serve as a replenishing source for effector T cells (24–26). These transitions are accompanied by epigenomic changes that contribute to defining the transcriptional and functional profile of differentiating T cells (16, 27–29). The need to ensure locus accessibility to support gene expression and cell differentiation is well established (30) and provides a strong rationale for shaping the epigenetic landscape of engineered T cells to sustain their functional persistence. Disruption of regulators of DNA methylation, DNMT3a and TET2, in CAR T cells improves memory formation and has been suggested to enhance antitumor efficacy (31, 32). Modifications to histone tails provide another layer of control over locus accessibility (33).

Tri-methylation of H3 Lysine 9 (H3K9me3) marks facultative and constitutive heterochromatin, and thus contributes to maintaining cellular identity (34, 35). Three proteins prima­rily catalyze H3K9me3 – SUV39H1, SUV39H2, and SETDB1 (34). We focus here on SUV39H1, which has been reported to control Th2 effector lineage commitment in mice (36) and memory to effector transition in murine CD8+ T cells in an acute Listeria monocytogenes infection model (37). Thus, we hypothesized that disrupting SUV39H1 may improve functional persistence and antitumor efficacy in human CD28-based CAR T cells.

SUV39H1 Disruption Improves Antitumor Efficacy of CAR T Cells

To disrupt SUV39H1, we employed the CRISPR/Cas9 to edit T cells isolated from healthy donors and activated by CD3/28 beads (Fig. 1A). Activated T cells were then electroporated with Cas9 mRNA and a SUV39H1-targeting or scrambled guide RNA (gRNA; Fig. 1A). Multiple gRNAs targeting SUV39H1 were tested. The best candidate gRNA achieved an editing efficiency of approximately 80% as assessed by next-generation sequencing (NGS) of the site of edit (Fig. 1B and C). Loss of SUV39H1 was confirmed at a protein level by Western blot analysis (Fig. 1D). SUV39H1 disruption resulted in global loss of H3K9me3 levels within days (Fig. 1E and F).

Figure 1.

SUV39H1 disruption reduces global H3K9me3 levels in T cells. A, Schematics of H3k9me3 FACS assay. B and C, gRNA editing efficiency (B) and indel distribution - Red (unmodified) and Blue (modified) (C). D, Western blot analysis showing SUV39H1 disruption at protein level. E, Representative H3K9me3 flow cytometry data for each time point. F, Summary of H3K9me3 flow cytometry data from two replicates per donor (two donors). P values were determined by Mann–Whitney (F). P < 0.05 was considered statistically significant. P values are denoted: P > 0.05, NS, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001. KO, knockout; MFI, mean fluorescence intensity; WT, wild-type.

Figure 1.

SUV39H1 disruption reduces global H3K9me3 levels in T cells. A, Schematics of H3k9me3 FACS assay. B and C, gRNA editing efficiency (B) and indel distribution - Red (unmodified) and Blue (modified) (C). D, Western blot analysis showing SUV39H1 disruption at protein level. E, Representative H3K9me3 flow cytometry data for each time point. F, Summary of H3K9me3 flow cytometry data from two replicates per donor (two donors). P values were determined by Mann–Whitney (F). P < 0.05 was considered statistically significant. P values are denoted: P > 0.05, NS, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001. KO, knockout; MFI, mean fluorescence intensity; WT, wild-type.

Close modal

We next sought to assess the effect of SUV39H1 disruption on CAR T cells. Freshly isolated human T cells from healthy donors (Supplementary Fig. S1A) were electroporated with either scrambled gRNA or SUV39H1-targeting gRNA, followed by CAR transduction. We did not observe significant differences in T-cell phenotype after CAR transduction (Supplementary Fig. S1B). To assess the impact of SUV39H1 disruption on CAR T-cell antitumor efficacy, we treated NSG mice bearing the human B-cell acute lymphoblastic leukemia (B-ALL), NALM6, with a low dose of either SUV39H1-edited or unedited (scrambled gRNA) CAR T cells (Fig. 2A). SUV39H1 editing enhanced the antitumor efficacy of 1928z CAR T cells, with 9 of 10 NALM6-bearing mice treated with SUV39H1-edited 1928z CAR T cells [hereafter referred to as SUV knockout (KO) 1928z] surviving over the duration of observation (90 days) as compared with 1 of 12 mice treated with unedited 1928z CAR T cells [hereafter referred to as wild-type (WT) 1928z; Fig. 2B, left]. NALM6 were engineered to express GFP-luciferase to monitor tumor burden in mice (38). We noted very similar primary tumor clearance kinetics (first 10 days post CAR T-cell injection) in WT 1928z and SUV KO 1928z CAR T cells (Fig. 2B, middle and right). However, mice treated with WT 1928z CAR T cells relapsed after initial tumor clearance in contrast to mice treated with SUV KO 1928z CAR T cells, which maintained durable tumor control (Fig. 2B, middle and right). SUV39H1 disruption either in a nonfunctional CAR (19z1-delta) or a CAR targeting an irrelevant antigen prostate-specific membrane antigen (PSMA) (PSMA28z; ref. 39) did not result in any improvement of their antitumor efficacy (Supplementary Fig. S2A and S2B).

Figure 2.

SUV39H1 disruption enhances the antitumor efficacy of CAR T cells. A, Schematics of CAR T-cell generation protocol and murine NALM6 xenograft model. B, Mice survival (left) and tumor radiance (right) under 1928z CAR T-cell treatment dose: 2 × 105, n = 5 for untreated, n = 12 for WT (scrambled gRNA), and n = 10 for SUV39H1-edited. Survival trends were confirmed in another donor. C and D, CAR T-cell quantification (n = 5, each dot represents a mouse) in the bone marrow at day 10 (C) and day 17 (D). E, Representative CAR T-cell IL7Rα (left) and CD27 (right) flow cytometry plots at day 10 (top) and day 17 (bottom). F and G, Summary data for IL7Rα and CD27 flow cytometry plot replicates at day 10 (F) and day 17 (G). n = 5 for both WT and SUV39H1-edited groups at day 10. n = 4 for both WT and SUV39H1-edited groups at day 17. P values were determined by log-rank Mantel–Cox test (B) and Mann–Whitney (C, D, F, and G). P < 0.05 was considered statistically significant. P values are denoted: P > 0.05, NS, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001. The mouse illustration in A was generated using Servier Medical Art, CC BY 3.0. MFI, mean fluorescence intensity.

Figure 2.

SUV39H1 disruption enhances the antitumor efficacy of CAR T cells. A, Schematics of CAR T-cell generation protocol and murine NALM6 xenograft model. B, Mice survival (left) and tumor radiance (right) under 1928z CAR T-cell treatment dose: 2 × 105, n = 5 for untreated, n = 12 for WT (scrambled gRNA), and n = 10 for SUV39H1-edited. Survival trends were confirmed in another donor. C and D, CAR T-cell quantification (n = 5, each dot represents a mouse) in the bone marrow at day 10 (C) and day 17 (D). E, Representative CAR T-cell IL7Rα (left) and CD27 (right) flow cytometry plots at day 10 (top) and day 17 (bottom). F and G, Summary data for IL7Rα and CD27 flow cytometry plot replicates at day 10 (F) and day 17 (G). n = 5 for both WT and SUV39H1-edited groups at day 10. n = 4 for both WT and SUV39H1-edited groups at day 17. P values were determined by log-rank Mantel–Cox test (B) and Mann–Whitney (C, D, F, and G). P < 0.05 was considered statistically significant. P values are denoted: P > 0.05, NS, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001. The mouse illustration in A was generated using Servier Medical Art, CC BY 3.0. MFI, mean fluorescence intensity.

Close modal

SUV39H1-Edited CAR T Cells Have Improved Early Expansion

We measured CAR T-cell numbers in the bone marrow (primary site of disease) early after tumor clearance (day 10 post CAR T-cell infusion) and a week later (day 17 postinfusion). SUV39H1 editing enhanced day 10 CAR T-cell numbers (Fig. 2C) but not day 17 CAR T-cell numbers (Fig. 2D). We observed increased expression of IL7Rα (a marker of memory T-cell subset) in SUV KO CAR T cells at day 10 (Fig. 2E and F, left). At day 17, we also observed increased expression of IL7Rα (mean fluorescence intensity but not cell fraction; Fig. 2E and G, left). CD27 expression did not differ between WT and SUV KO 1928z CAR T cells at day 10 (Fig. 2E and F, right) but was reduced by day 17 in SUV KO 1928z CAR T cells (Fig. 2E and G, right), which was not surprising given the reemergence of tumor in WT 1928z CAR T cells at that time. PD-1 expression was reduced in SUV KO 1928z CAR T cells at day 10 (mean fluorescence intensity but not cell fraction) but not LAG3 and TIM3 [Supplementary Fig. S3A (top) and S3B]. The expression of all three inhibitory receptors was not significantly different between WT and SUV KO 1928z CAR T cells at day 17 [Supplementary Fig. S3A (bottom) and S3C]. SUV39H1 disruption thus boosts early CAR T-cell expansion, augmenting IL7R expression while decreasing that of CD27 and PD-1 in leukemia-bearing mice purposefully treated with a low CAR T-cell dose.

SUV39H1 gRNA Has Low Off-Target Activity, and Off-Target Editing Is Not Enriched Over Time

To determine whether an off-target effect of the gRNA may have contributed to enhancing CAR T-cell function, we determined editing efficiency at the SUV39H1 locus and at the top 10 predicted off-target (OT) sites before CAR stimulation and after 4 rounds of CAR stimulation by NGS (Supplementary Fig. S4A). The top 10 predicted off-target sites included two genomic sites with two mismatches to the SUV39H1 gRNA and eight genomic sites with three mismatches (Supplementary Fig. S4B). There were no genomic sites with only 1 mismatch to the gRNA. All editing efficiencies were computed through deep sequencing of the PCR product encompassing the editing site. The false discovery rate (FDR), determined through amplifying an unrelated genomic locus of T cells electroporated with Cas9 alone, was 0.06% (Supplementary Fig. S4C). All OT sites were edited at near FDR frequency, except for OT-9, which was edited at a frequency of 1% (Supplementary Fig. S4C). SUV39H1 editing before CAR stimulation (day 7) was 72.7% and increased to 83.7% after 4 rounds of CAR stimulation (Supplementary Fig. S4C). No enrichment was observed in any of the 10 OT sites (including OT-9) after 4 rounds of CAR stimulation (Supplementary Fig. S4C). These results suggest that the increased expansion observed upon treating CAR T cells with SUV39H1-targeting gRNA is unlikely to be due to an OT activity of gRNA.

Effect of SUV39H1 Disruption on CAR T-Cell Efficacy Extends to Other CAR Designs

To assess whether SUV39H1 disruption could enhance the efficacy of other CAR designs in a different model, we used a prostate cancer model (Supplementary Fig. S5A) in which three different CAR designs (PSMA28z, PSMA28z-1xx, PSMABBz) targeting the PSMA antigen were tested (39, 40). SUV39H1 disruption enhanced the antitumor efficacy of all 3 CAR designs (Supplementary Fig. S5B–S5D). These results show the potential of SUV39H1 disruption to serve as a general strategy to improve CAR T-cell function.

SUV39H1 Editing Enhances Proliferation, Limits Effector Differentiation, and Sustains Clonal Diversity

We performed sequential single-cell transcriptional analysis on SUV KO 1928z CAR T cells and WT 1928z CAR T cells to characterize the impact of SUV39H1 disruption on 1928z CAR T cells over time (Fig. 3A). CAR T cells were analyzed at three time points: preinfusion (day 0), 9, and 16 days postinfusion (Fig. 3A). Day 0 provided a baseline gene expression profile for both unedited and SUV39H1-edited groups. Day 9 corresponded to the peak of primary antitumor response and day 16 to contraction following the primary response. Across both genotypes and at all three time points, we identified 15 subpopulations (Fig. 3B and C).

Figure 3.

Single-cell transcriptional profiling of CAR T cells. A, Design of single-cell transcriptional profiling. Single-cell RNA sequencing was performed at three time points: Preinfusion (day 0), day 9, and day 16. Bone marrow was pooled together from five mice for each condition. CAR T cells were then sorted by flow cytometry. B, Uniform Manifold Approximation and Projection (UMAP) for all three time points (top) and CD4s and CD8s (bottom). C, Marker expression for Seurat clusters. D, GSEA analysis showing enrichment in proliferation-associated pathways in SUV KO 1928z CAR T cells and effector function–associated pathways in WT 1928z CAR T cells. E, Fraction of cycling cells at day 9. F, Gini index (inversely correlated with T-cell receptor diversity) over time [preinfusion (day 0), day 9 and day 16 in mice] of WT and SUV KO 1928z CAR T cells. The mouse illustration in A was generated using Servier Medical Art, CC BY 3.0. NES, normalized enrichment scale.

Figure 3.

Single-cell transcriptional profiling of CAR T cells. A, Design of single-cell transcriptional profiling. Single-cell RNA sequencing was performed at three time points: Preinfusion (day 0), day 9, and day 16. Bone marrow was pooled together from five mice for each condition. CAR T cells were then sorted by flow cytometry. B, Uniform Manifold Approximation and Projection (UMAP) for all three time points (top) and CD4s and CD8s (bottom). C, Marker expression for Seurat clusters. D, GSEA analysis showing enrichment in proliferation-associated pathways in SUV KO 1928z CAR T cells and effector function–associated pathways in WT 1928z CAR T cells. E, Fraction of cycling cells at day 9. F, Gini index (inversely correlated with T-cell receptor diversity) over time [preinfusion (day 0), day 9 and day 16 in mice] of WT and SUV KO 1928z CAR T cells. The mouse illustration in A was generated using Servier Medical Art, CC BY 3.0. NES, normalized enrichment scale.

Close modal

Gene-set enrichment analysis (GSEA) of differentially expressed genes between genotypes suggested higher proliferation and reduced effector function in SUV39H1-edited 1928z CAR T cells (Fig. 3D). At day 9, we observe an enrichment in proliferation pathways in SUV39H1-edited 1928z CAR T cells which was associated with increased relative abundance of both CD4 and CD8 cycling cells (Fig. 3E). Differential gene expression analysis revealed increased expression of genes associated with cytoskeletal regulation (ACTB, CAPG, PFN1) at day 9 in the SUV39H1-edited group (Supplementary Fig. S6A and S6B). WT 1928z CAR T cells on the other hand showed elevated expression of effector function associated genes both at day 9 and day 16 (GZMA, GZMK, KLRC1, KLRG1, PTGDR; Supplementary Fig. S6A–S6D). Interestingly, memory-associated transcription factors and receptors (KLF2, LEF1, TCF7, SELL) were elevated in SUV39H1-edited 1928z CAR T cells by day 16 but not at day 9 (Supplementary Fig. S6C and S6D). Precursor exhausted T cells (Tpex) express memory transcription factors such as TCF7, retaining some proliferative and effector function (41). However, at day 16 we did not find this subset in an appreciable fraction in both WT and SUVKO 1928z CAR T cells. Memory T cells have an improved ability to proliferate and persist as compared with effector T cells (24, 25). We hypothesized that continued expression of memory factors in SUV KO 1928z CAR T cells would promote greater clonal diversity over time. We therefore assessed the diversity of T-cell receptor (TCR)vβ sequences from single-cell sequencing data at the three time points. As expected, the overall diversity of TCRvβ sequences decreased from day 0 to day 16 (Fig. 3F). However, while clonal diversity was similar between WT and SUV KO 1928z CAR T cells at the earliest time points, the two groups diverged by day 16, with both CD4 and CD8 T cells exhibiting greater diversity in the SUV39H1-edited group (Fig. 3F). Single-cell transcriptional profiling thus suggested that SUV39H1 disruption in human CAR T cells promoted proliferative and memory programs, restrained effector programs, and maintained a broader clonal repertoire.

SUV39H1 Disruption Restrains Cytokine Secretion and Improves Metabolic Fitness of CAR T Cells

Transcriptional profiling suggested enrichment in cytotoxicity and cytokine signaling in WT CAR T cells compared with SUV KO CAR T cells. We measured cytokine secretion 24 hours after CAR T-cell and target cell coculture (Fig. 4A) and observed that SUV39H1 disruption moderately attenuates cytokine (IL2, TNF, IFNγ) and granzyme B (GZMB) secretion upon activation (Fig. 4B).

Figure 4.

SUV39H1 disruption attenuates cytokine and granzyme B secretion upon repeated CAR stimulation. A, Schematics of repeated CAR stimulation assay. Cytokines were measured in the media 24 hours after coculture with target cells at indicated time points. B, Cytokine quantification for unedited and SUV39H1-edited 1928z CAR T cells. IL2 was below detection limit at day 21 while TNF was below assay detection limit at both, day 14 and day 21. Data is represented as mean ± SD, n = 5. The trends were confirmed in another donor. P values were determined by Mann–Whitney Test. P < 0.05 was considered statistically significant. P values are denoted: P > 0.05, NS, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 4.

SUV39H1 disruption attenuates cytokine and granzyme B secretion upon repeated CAR stimulation. A, Schematics of repeated CAR stimulation assay. Cytokines were measured in the media 24 hours after coculture with target cells at indicated time points. B, Cytokine quantification for unedited and SUV39H1-edited 1928z CAR T cells. IL2 was below detection limit at day 21 while TNF was below assay detection limit at both, day 14 and day 21. Data is represented as mean ± SD, n = 5. The trends were confirmed in another donor. P values were determined by Mann–Whitney Test. P < 0.05 was considered statistically significant. P values are denoted: P > 0.05, NS, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Close modal

Strong effector differentiation in T cells eventually culminates in a dysfunctional state associated with poor metabolic fitness (24, 25). Measuring mitochondrial capacity in CAR T cells over multiple rounds of stimulation (Fig. 5A), we did not find significant differences after two rounds of CAR stimulation between unedited and SUV39H1-edited CAR T cells (Fig. 5B, left). However, after the 3rd and 4th round of CAR stimulation, SUV KO 1928z CAR T cells demonstrated greater mitochondrial function, suggestive of improved cellular fitness (Fig. 5B, middle and right). Glycolytic rates, on the other hand, were not significantly different between WT and SUV KO 1928z CAR T cells (Fig. 5C). Altogether, these observations suggest that SUV39H1 disruption restrains cytokine secretion and promotes mitochondrial fitness of CAR T cells.

Figure 5.

SUV39H1 disruption improves metabolic fitness of CAR T cells under conditions of repeated stimulation. A, Schematics of repeated CAR stimulation assay. Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were measured at indicated time points. OCR (B) and ECAR (C) rates of scrambled gRNA treated and SUV39H1-edited 1928z CAR T cells were assessed at the indicated time points. Data are represented as mean ± SD, n = 4 or 5. B and C, The trends were confirmed in three different donors. P values were calculated by unpaired t test (B). P < 0.05 was considered statistically significant. P values are denoted: P > 0.05, NS, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Figure 5.

SUV39H1 disruption improves metabolic fitness of CAR T cells under conditions of repeated stimulation. A, Schematics of repeated CAR stimulation assay. Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were measured at indicated time points. OCR (B) and ECAR (C) rates of scrambled gRNA treated and SUV39H1-edited 1928z CAR T cells were assessed at the indicated time points. Data are represented as mean ± SD, n = 4 or 5. B and C, The trends were confirmed in three different donors. P values were calculated by unpaired t test (B). P < 0.05 was considered statistically significant. P values are denoted: P > 0.05, NS, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Close modal

SUV39H1 Disruption Sustains Cytolytic Function of 1928z CAR T Cells

The improved metabolic fitness and moderately reduced cytokine secretion observed in SUV39H1-edited CAR T cells prompted us to assess their cytolytic function under chronic activation over time (Supplementary Fig. S7A). As was noted in vivo, we observed enhanced expansion of 1928z CAR T cells over time upon SUV39H1-disruption (Supplementary Fig. S7B). Interestingly, in an assay to assess cytolytic activity of WT and SUVKO 1928z CAR T cells, we noted improved cytolytic activity of SUVKO 1928z at day 21, but not day 7 and 14 (Supplementary Fig. S7C).

SUV39H1-Edited 1928z CAR T Cells Protect Against Multiple Tumor Rechallenges

To address whether SUV39H1 editing translates to better long-term T-cell persistence, we treated NALM6-bearing mice with a low but curative CAR T-cell dose (4e5 CAR T cells/mouse; Supplementary Fig. S8A). Because tumor burden was effectively controlled (Supplementary Fig. S8B), we could assess long-term persistence in absence of chronic antigenic stimulation from residual tumor. Indeed, we observe improved persistence of both CD4 and CD8 CAR T cells upon SUV39H1 disruption (Supplementary Fig. S8C).

To test whether improved long-term persistence upon SUV39H1 disruption translates to improved function under chronic activation, we modified the NALM6 model to include repeated tumor rechallenges starting soon after primary tumor clearance (Fig. 6A). SUV39H1 disruption enhanced the ability of CAR T cells to overcome multiple tumor rechallenges (Fig. 6B and C). SUV39H1 disruption also improved CAR T-cell numbers in the bone marrow and spleen (Fig. 6D and E). Flow cytometry analyses showed no significant differences in CD27 expression (Fig. 6F and G, left). However, expression of inhibitory receptors was reduced in SUV39H1-edited 1928z CAR T cells (Fig. 6F and G, right). Collectively, these observations indicate that SUV39H1 disruption in 1928z CAR T cells enhanced their functional persistence.

Figure 6.

SUV39H1 disruption enhances the ability of CAR T cells to reject tumor upon rechallenge. A, Design of the rechallenge study. Mice were treated with a CAR T cell dose of 2 × 105. Tumor rechallenge was done with 2e6 NALM6. B, Tumor radiance over time. Tumor rechallenge is indicated by arrowheads. n = 10 for both groups. C, Comparing tumor radiance between unedited and SUV39H1-edited 1928z CAR T cells at day 51 (last tumor radiance measurement time point when all unedited 1928z CAR T cells are alive). D and E, CAR T-cell quantification in bone marrow (D) and spleen (E) after five rounds of rechallenge. F and G, Representative CAR T-cell CD27, PD1, LAG3, TIM3 flow cytometry plots at day 70 (F) and summary data for CD27 (left) and inhibitory receptor expression (right) at day 70 (G). Data are represented as mean ± SD (G). P values were determined by Mann–Whitney Test (CE and G). P values were corrected for multiple comparisons in G by BKY method. P values are denoted: P > 0.05, NS, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. The mouse illustration in A was generated using Servier Medical Art, CC BY 3.0.

Figure 6.

SUV39H1 disruption enhances the ability of CAR T cells to reject tumor upon rechallenge. A, Design of the rechallenge study. Mice were treated with a CAR T cell dose of 2 × 105. Tumor rechallenge was done with 2e6 NALM6. B, Tumor radiance over time. Tumor rechallenge is indicated by arrowheads. n = 10 for both groups. C, Comparing tumor radiance between unedited and SUV39H1-edited 1928z CAR T cells at day 51 (last tumor radiance measurement time point when all unedited 1928z CAR T cells are alive). D and E, CAR T-cell quantification in bone marrow (D) and spleen (E) after five rounds of rechallenge. F and G, Representative CAR T-cell CD27, PD1, LAG3, TIM3 flow cytometry plots at day 70 (F) and summary data for CD27 (left) and inhibitory receptor expression (right) at day 70 (G). Data are represented as mean ± SD (G). P values were determined by Mann–Whitney Test (CE and G). P values were corrected for multiple comparisons in G by BKY method. P values are denoted: P > 0.05, NS, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. The mouse illustration in A was generated using Servier Medical Art, CC BY 3.0.

Close modal

SUV39H1 Disruption Enhances Polyfunctional CAR T-cell Persistence

To better understand the effects of SUV39H1 disruption in CAR T cells, we modified the repeated rechallenge model to include simultaneous transcriptional and chromatin accessibility profiling after three consecutive rechallenges (Fig. 7A). Transcriptional analysis revealed increased expression of memory-associated transcription factors and receptors such as TCF7, LEF1, CCR7, and IL7R in SUV KO 1928z CAR T cells compared with WT 1928z CAR T cells (Fig. 7B). WT 1928z CAR T cells, on the other hand, expressed increased levels of effector/terminal-effector state associated transcription factors such as TBX21, EOMES, and PRDM1 (Fig. 7B). Increased expression of various inhibitory receptors such as PDCD1, LAG3, HAVCR2, CTLA4, CD38, KLRG1, and TIGIT was observed in WT 1928z CAR T cells (Fig. 7B). GSEA revealed increased enrichment of human T-cell exhaustion associated genes in WT 1928z CAR T cells (Fig. 7C).

Figure 7.

SUV39H1 disruption allows continued expression of memory transcription factors (TF) and limits expression of terminal effector transcription factors and inhibitory receptors. A, Design of the rechallenge study. Mice were treated with 2 × 105 1928z CAR T cells. Tumor rechallenge was done with 2e6 NALM6. Bone marrow was pooled together from 5 mice for each condition. CAR T cells were then sorted by flow cytometry. B, RNA sequencing (RNA-seq) heat map of memory, effector, and inhibitory receptor genes. C, GSEA showing enrichment of human T-cell exhaustion signature in unedited 1928z CAR T cells. D, Mean average plot (ATAC-seq). Red dots are peaks with Padj < 0.1. E, ATAC-seq heat map, most significant peak associated with memory, effector, and inhibitory receptor genes is highlighted. F, Motif enrichment analysis identifies TCF1/LEF1 motif in genes downregulated in SUV39H1-edited1928z CAR T cells. G, Graphical model summarizing the results. The mouse illustration in A was generated using Servier Medical Art, CC BY 3.0. Graphical model in G used elements generated in BioRender.

Figure 7.

SUV39H1 disruption allows continued expression of memory transcription factors (TF) and limits expression of terminal effector transcription factors and inhibitory receptors. A, Design of the rechallenge study. Mice were treated with 2 × 105 1928z CAR T cells. Tumor rechallenge was done with 2e6 NALM6. Bone marrow was pooled together from 5 mice for each condition. CAR T cells were then sorted by flow cytometry. B, RNA sequencing (RNA-seq) heat map of memory, effector, and inhibitory receptor genes. C, GSEA showing enrichment of human T-cell exhaustion signature in unedited 1928z CAR T cells. D, Mean average plot (ATAC-seq). Red dots are peaks with Padj < 0.1. E, ATAC-seq heat map, most significant peak associated with memory, effector, and inhibitory receptor genes is highlighted. F, Motif enrichment analysis identifies TCF1/LEF1 motif in genes downregulated in SUV39H1-edited1928z CAR T cells. G, Graphical model summarizing the results. The mouse illustration in A was generated using Servier Medical Art, CC BY 3.0. Graphical model in G used elements generated in BioRender.

Close modal

Chromatin accessibility profiling through assay for transposase-accessible chromatin using sequencing (ATAC-seq) showed increased global accessibility in SUV KO 1928z CAR T cells compared with WT 1928z CAR T cells, which is consistent with the role of SUV39H1 in mediating methylation at H3K9 and maintaining heterochromatin structure (Fig. 7D). Interestingly, while global chromatin accessibility was higher in SUV KO 1928z CAR T cells, we observed reduced chromatin accessibility of several inhibitory receptors and effector/terminal-effector state associated transcription factors in SUV KO 1928z CAR T cells (Fig. 7E). Motif analysis of genes that were downregulated in SUV KO 1928z relative to WT 1928z CAR T cells showed an enrichment of TCF7 (Protein TCF1)/LEF1 motif (Fig. 7F). To orthogonally validate whether SUV39H1 disruption sustains TCF1 and LEF1 expression in 1928z CAR T cells, we devised an in vitro experiment where TCF1 and LEF1 expression was measured in 1928z CAR T cells that were repeatedly exposed to antigen (Supplementary Fig. S9A). We observe that TCF1 and LEF1 expression drops dramatically from day 14 to day 21 under chronic activation for both WT and SUVKO 1928z CAR T cells (Supplementary Fig. S9B–S9E). However, SUV39H1-edited 1928z CAR T cells maintain higher levels of TCF1 and LEF1 by day 21 as compared with WT 1928z CAR T cells (Supplementary Fig. S9D and S9E). TCF1 and LEF1 are homologous proteins that have both transcriptional activating and inhibitory domains and remodel chromatin during T-cell development (42). Their ability to repress transcriptional factors that mediate effector differentiation (ID2, PRDM1; ref. 42) and inhibitory receptors (PDCD1, LAG3, CTLA4; ref. 42) are consistent with the function and phenotype we observed in SUV KO 1928z CAR T cells (Fig. 7G). Altogether, our findings show that, upon repeated exposure to tumor, SUV39H1-edited 1928z CAR T cells will maintain accessibility to loci known to promote T-cell memory and function. SUV39H1 editing in 1928z CAR T cells also reduces loci accessibility and expression of effector transcription factors and inhibitory receptors.

We show here that disrupting SUV39H1 enhances the antitumor efficacy of CAR T cells. SUV39H1 editing improves early CAR T-cell expansion and long-term persistence. Persisting SUV39H1-edited CAR T cells maintain superior ability to reject tumors upon rechallenge, whereupon they showed delayed expression of inhibitory receptors.

CD28-based CARs are one of two foundational CAR designs (43), alongside 4-1BB-based CARs (44), that have been used in the clinic for a range of hematologic malignancies with great success (4, 45, 46). CD28-based CAR T cells have potent antigen sensitivity (10, 11) but are limited by their rapid effector differentiation (21, 22) and relatively short persistence (17, 20, 45). As T cells differentiate, memory T-cell programs are epigenetically silenced through histone modification (37, 47) and DNA methylation (31, 48). A greater memory T-cell fraction in preinfusion CAR T-cell products has been shown to improve antitumor responses in preclinical models and patients (49, 50).

The initial tumor control achieved by WT and SUV39H1-edited CAR T cells was indistinguishable. Mice treated with SUV39H1-edited CAR T cells; however, show improved durable tumor remission and enhanced tumor rejection upon rechallenge. Single-cell transcriptional analysis reveals improved expression of memory transcription factors and clonal diversity post primary tumor clearance, consistent with improved CAR T-cell fitness. SUV39H1-edited CAR T cells show continued expression of several memory transcription factors such as KLF2, LEF1, and TCF7. SUV39H1-editing also limits effector cytokine secretion in CAR T cells. Early metabolic features (mitochondrial capacity and glycolysis) are indistinguishable in unedited and SUV39H1-edited CAR T cells. Over time, however, SUV39H1-edited CAR T cells show improved mitochondrial capacity while maintaining similar glycolytic rate as unedited CAR T cells.

SUV39H1 editing enhances early and late CAR T-cell numbers. Several studies have shown that SUV39H1 and other histone methyl transferases interact with E2F transcription factors through their association with retinoblastoma (RB1) during cell cycle that leads to deposition of H3K9me3 marks on E2F target genes such as CyclinE and CyclinA2 resulting in their silencing (51, 52). As a result, CyclinE and CyclinA2 activity are elevated in SUV39H1-null fibroblasts (52). Whether similar pathways are involved in CAR T cells remains to be determined.

Single-cell transcriptional analyses show that only a few CAR T-cell subpopulations display an enhanced cell-cycle signature in the SUV39H1-edited group compared with the unedited group. Transcriptional and genome accessibility analysis on CAR T cells that have undergone multiple rounds of rechallenge show continued expression of memory-associated genes and concurrent suppression of effector genes and inhibitory receptors in SUV39H1-edited CAR T cells. This expression profile is accompanied by improved accessibility of memory genes and reduced accessibility of effector genes in SUV39H1-edited CAR T cells. Motif analysis identifies TCF7/LEF1 among the top transcription factors mediating suppression of effector genes and inhibitory receptors in SUV39H1-edited CAR T cells. These results are in line with studies in murine T cells that have found that TCF7/LEF1 can remodel chromatin and repress transcriptional factors that mediate effector differentiation (ID2, PRDM1) and inhibitory receptors (PDCD1, LAG3, CTLA4; ref. 42). More corroborative evidence is found from GSEA analysis that shows enrichment of a gene signature associated with nonresponders to checkpoint blockade in unedited CAR T cells (53). TCF7 was identified in this signature as a marker in human T cells that respond to checkpoint blockade in melanoma (53). Altogether, these observations suggest that the continued expression of memory genes in SUV39H1-edited CAR T cells limits the expression of inhibitory receptors and extends their function. However, whether the improved accessibility at memory genes in SUV39H1-edited CAR T cells is a direct effect of reduced H3K9me3 deposition or a secondary effect is unclear as we were unable to conduct H3K9me3 chromatin immunoprecipitation sequencing (ChIP-seq) studies on CAR T cells from mice that have undergone repeated tumor rechallenge due to technical limitations.

Previously, Lu and colleagues described that a drug (F5446) mediated inhibition of H3K9me3 results in increased effector cytokine secretion in murine CD8+ T cells (54). On the other hand, in transgenic mice expressing the ovalbumin-specific OT-1 TCR, loss of SUV39H1 in CD8+ T cells leads to improved memory formation with a block of effector differentiation upon rechallenge (37). As a result, SUV39H1-knockout mice rechallenged with L. monocytogenes display poor L. monocytogenes control as compared with wild-type mice. In contrast, human CAR T cells lacking SUV39H1 preserve potent effector function, possibly owing to the CAR signal overcoming their restriction imparted by SUV39H1 loss. Furthermore, in a murine melanoma model, SUV39H1 loss by itself failed to enhance the antitumor response of tumor-infiltrating lymphocytes without the addition of PD-1 blockade (55). These divergent outcomes between mouse T cells and human CAR T cells in the context of SUV39H1 disruption may be due to species/experimental-model differences, off-target effects of the drug (in the case of Lu and colleagues), and/or differences in strength and quality of signaling between CAR T cells and natural, TCR-dependent T cells. Thus, we find that 28z CARs preserve effector differentiation in SUV39H1-edited T cells to in contrast to the block in differentiation observed in the OT-1 TCR transgenic model.

Disruption of DNA methylation programs mediated by DNMT3a and TET2 in CAR T cells has been shown to improve their memory phenotype and antitumor efficacy (31, 32). Both these genes, however, are potent tumor suppressor (56, 57). While disruption of DNMT3a in mature T cells did not result in aberrant proliferation, TET2 disruption can lead to antigen-independent expansion and genomic instability in CAR T cells (32). SUV39H1 mutations, on the other hand, are not commonly found in cancer. Depending on the cell of origin, SUV39H1 disruption may limit tumor progression (58) or tumor migration (59). We did not observe abnormal expansions in mice treated with SUV39H1-edited CAR T cells over a 200-day follow-up period. The longer-term safety profile of SUV39H1-edited CAR T cells will nonetheless need to be carefully evaluated in future clinical trials.

In summary, we find that SUV39H1 editing enhances the antitumor efficacy of CAR T cells by improving their proliferation, persistence, and limiting the onset of exhaustion. These features are associated with continued expression of memory transcription factors, particularly TCF7 and LEF1. Our results indicate that disruption of SUV39H1, and potentially other histone methyl transferases, is a promising strategy for balancing the effector functions and persistence imparted by the CAR to human T cells.

Retroviral Vector Constructs and Retroviral Production

Plasmids encoding the retroviral vector were prepared using standard molecular biology techniques (60). Dsred or LNGFR (a truncated and mutated TNF-R family homolog; ref. 61) was used as a control molecule to ensure comparable CAR expression levels from different bicistronic vectors. Synthesis of 1928z (62), PSMA28z (39), PSMABBz (39) has been described before. PSMA28z-1xx was designed to target PSMA with a 1xx cytoplasmic domain described before (40). 19z1 delta is a nonfunctional CAR that targets CD19 with a truncated CD3z domain. VSV-G pseudotyped retroviral supernatants derived from transfected gpg29 HEK293 (H29) were used to construct stable retroviral-producing cell lines as previously described (63).

Generation of Human CAR T Cells

Buffy coats from anonymous healthy donors were purchased from the New York Blood Center (institutional review board-exempted) and peripheral blood was obtained from healthy volunteers. All blood samples were handled following the required ethical and safety procedures. Peripheral blood mononuclear cells were isolated by density-gradient centrifugation. T cells were then purified by using the Pan T Cell Isolation Kit (Miltenyi Biotec). T cells were then stimulated with CD3/CD28 T cell Activator Dynabeads (Invitrogen) at 1:1 ratio and cultured in RPMI +10% FBS, 5 ng/mL IL7, and 5 ng/mL IL15 (Miltenyi Biotec) for retroviral transduction experiments. The medium was supplemented every 2 days, and cells were plated at 106 cells per mL.

Mouse Systemic Tumor Model

We used 6- to 12-week-old NOD/SCID/IL-2Rγ–null mice (The Jackson Laboratory), under a protocol approved by the Memorial Sloan Kettering Cancer Center (MSKCC) Institutional Animal Care and Use Committee. All relevant animal use guidelines and ethical regulations were followed. NALM6 expressing firefly luciferase-GFP were described previously (62). Mice were inoculated with 0.5  ×  106 FFLuc-GFP NALM6 cells by tail vein injection, CAR T cells (electroporated with either scrambled or SUV39H1 gRNA) were then injected 4 days later at varying doses. In the rechallenge model, starting at day 17 post CAR T-cell infusion, 2  ×  106 NALM6 were injected via tail vein every 10 days (3–5 times in total). NALM6 produce very even tumor burdens, and no mice were excluded before treatment. For prostate cancer model, male mice were inoculated with 2e6 PC3-PSMA FFLuc-GFP (64) cells by tail vein injection. CAR T cells [electroporated with (TRAC+scrambled) gRNA or (TRAC+SUV39H1) gRNA] were injected 4 weeks later. gRNAs used for gene editing can be found in Supplementary Table S1. No randomization or blinding methods were used. Bioluminescence imaging was performed using the IVIS Imaging System (PerkinElmer) with the Living Image software (PerkinElmer) for the acquisition of imaging datasets.

Ex Vivo CAR T-cell Quantification

Tibia, femur, and spleen were isolated from mice. Both tibia and femur were crushed with a mortar to obtain a single-cell suspension. Spleen was mashed to obtain a single-cell suspension. This suspension was then treated with an ACK Lysing Buffer (Lonza, catalog no. BP10–548E) to lyse red blood cells. Between 10% and 20% of the lysate was used to quantify CAR T-cell numbers using absolute counting beads (C36950) on a flow cytometer.

In Vitro Rechallenge Model

3T3 cells were irradiated with 3,000 rads. They were plated at 150,000 cells/well (24-well plate) one day prior to CAR T cells (5e5 CAR T cells/well). No supplemental cytokines were added. This assay was used to determine various functional properties of CAR T cells—expansion, cytolytic capacity, cytokine secretion, metabolic fitness, memory transcription factor expression. The schema for each individual measurement is described in their respective figure. Antibodies used for flow cytometry analysis can be found in Supplementary Table S1.

Cytotoxicity Assays

The cytotoxicity of T cells transduced with a CAR was determined by a luciferase-based assay. NALM6-expressing FFLuc–GFP cells served as target cells. The effector and tumour cells were cocultured at indicated effector to target (E/T) ratio in the black-walled 96-well plates (in triplicates) with 1 × 105 target cells in a total volume of 100 μL per well. Target cells alone were planted at the same cell density to determine the maximal luciferase expression [relative light units (RLUmax)]. Eighteen hours later, 100 μL luciferase substrate (Bright-Glo, Promega) was directly added to each well. Emitted light was measured by a luminescence plate reader or the Xenogen IVIS Imaging System (Xenogen) with Living Image V4.4 software (Xenogen) for acquisition of imaging datasets. Lysis was determined as [1 − (RLUsample)/(RLUmax)] × 100.

RNA-seq 10X Genomics Library Construction and Sequencing

Preinfusion T cells (day 0), and flow-sorted CAR T cells from mouse bone marrow (day 9 and 16) were stained for 30 minutes at room temperature with a custom panel of 101 Total-Seq-C antibodies (BioLegend; ref. 65), and washed three times using the HT1000 laminar wash system (Curiox). Cells were then counted using the Cellaca MX High-throughput Automated Cell Counter as described in the manufacturer's protocol (Nexcelom) and loaded on the 10x Chromium Next GEM Chip K Kit. TCR CDR3 sequences were enriched using the human V(D)J T-cell enrichment. Libraries were prepared according to manufacturer's protocol (5′ V2 kit, 10x Genomics) and sequenced on a NovaSeq 6000 System using the S4 2 × 150 kit (Illumina).

Preprocessing of Single-cell RNA Sequencing, CITE-seq, and VDJ Data

The single-cell RNA sequencing (scRNA-seq) reads were aligned to the human transcriptome (GRCh38) and unique molecular identifier (UMI) counts quantified to generate a gene-barcode matrix using Cell Ranger pipeline (10X Genomics, version cellranger-5.0.1). Cellular indexing of transcriptomes and epitopes (CITE-seq) antibody expression matrices were generated using the Cell Ranger pipeline (10X Genomics, version cellranger-5.0.1). TCR reads were aligned to the GRCh38 reference genome and consensus TCR annotation was performed using Cell Ranger vdj pipeline (10X Genomics, version cellranger-5.0.1).

Cluster Analysis of All Immune Cells

The processed expression matrix was subjected to several preprocessing steps. Genes that were expressed in less than 10 cells were removed. As a quality control step, cells were filtered on the basis of the number of detected genes, number of detected UMIs, house-keeping gene expression, and percentage of mitochondrial gene expression. Surface proteins were normalized using the centered-log ratio (CLR) method. Variable genes were identified, and a dimensionality-reduced representation of the cells was created based on those variable genes. We corrected for batch effect by performing Negative Binomial regression per sequencing sample using scTransform with method glmGamPoi without using additional covariates. Shared nearest neighbors were computed and cells were then clustered using graph community clustering methods. Multi-omic data was further utilized to remove low-quality cells and previously undetected doublets (e.g., cells that express both CD8 and CD4 protein tags, cells that express both a high B cells signature and have a detected TCR). All cluster analyses were performed in Seurat version 4.0.6. Data are available at GEO (accession number GSE245187).

Cell Type Annotation

Cells were annotated as CD4 and CD8 based on cluster level of CD4, CD8A and CD8B RNA and protein expression. CD4 and CD8 T cells are easily separated using CITE expression on day 0, but the noise was higher at the other time points. Cells with intermediate expression of CD4 and CD8 protein expression were labeled as doublets. Therefore, we use day 0 data to train an SVM using sums of CD4 and CD8 RNA and protein expression where CD4_total = CD4 + cite-CD4, CD8_total = CD8A + CD8B + cite-CD8. The trained model was used to generate CD4 and CD8 labels for day 9 and day 16 cells.

Differential Gene Expression Analysis

Differential expression was performed with the FindAllMarkers functions in Seurat R package (version 4.0.6) using the Wilcoxon rank sum test. The single-cell expression is log normalized with the NormalizeData function before inputting to the DEG analysis. CD4 and CD8 T cells are analyzed separately. For time-point related DEGs, unedited and SUV39H1-edited cells are analyzed separately. For genotype-related DEGs, cells from day 0, 9, and 16 were analyzed separately.

Seahorse Assay

Seahorse 96-well plates were coated with 20 μL Poly-l-lysine (50 μg/mL) for 20 to 30 minutes at room temperature. Culture media for Mito stress test (DMEM with glutamax, pyruvate, and glucose, no FBS) and glycolysis stress test (DMEM with glutamax and pyruvate, no glucose, no FBS) was prepared. Cells were spun down and resuspended to a concentration of 5e6/mL with respective mediums. Discard the poly-l-lysine and wash the plate with 200 μL distilled water. Let the plate dry completely. Add 40 μL of cell suspension per well (n = 4–5 replicates). Spin the plate at 300 × g for 1 minute to settle the cells. Add 140 μL respective medium to each well. The assay was then run as per manufacturer's instructions.

Cytokine Secretion Assessment

CAR T (1.25e5) cells were cultured with irradiated 3T3-CD19 (0.375e5) cells in 96 well plate. Anti-CD25 antibody was added at a concentration of 1 μg/mL. Supernatant was collected 24 hours after coculture. Cytokines were measured using cytometric bead arrays (BD Biosciences) for human IL2, IFNγ, TNF, and Granzyme B as per the manufacturer's instructions.

Bulk Transcriptome Sequencing

After RiboGreen quantification and quality control by Agilent BioAnalyzer, 2 ng total RNA with RNA integrity numbers ranging from 7.3 to 9.7 underwent amplification using the SMART-Seq v4 Ultra Low Input RNA Kit (Clontech, catalog no. 63488), with 12 cycles of amplification. Subsequently, 10 ng of amplified cDNA was used to prepare libraries with the KAPA Hyper Prep Kit (Kapa Biosystems KK8504) using 12 cycles of PCR. Samples were barcoded and run on a HiSeq 4000 in a PE50 run, using the HiSeq 3000/4000 SBS Kit (Illumina). An average of 40 million paired reads were generated per sample and the percent of mRNA bases per sample ranged from 31% to 69%. Data is available at GEO (accession number GSE245187).

ATAC-Seq

Profiling of chromatin was performed by ATAC-seq as described by Buenrostro and colleagues (66). Briefly, 50,000 fresh T cells were washed in cold PBS and lysed. The transposition reaction containing TDE1 Tagment DNA Enzyme (Illumina, catalog no. 20034198) was incubated at 37°C for 30 minutes. The DNA was cleaned with the MinElute PCR Purification Kit (QIAGEN, catalog no. 28004) and material was amplified for five cycles using NEBNext High-Fidelity 2X PCR Master Mix (New England Biolabs, catalog no. M0541L). After evaluation by real-time PCR, 8 additional PCR cycles were done. The final product was cleaned by aMPure XP beads (Beckman Coulter, catalog no. A63882) at a 1× ratio, and size selection was performed at a 0.5× ratio. Libraries were sequenced on a HiSeq4000 in a PE50 run, using the HiSeq 3000/4000 SBS Kit (Illumina). An average of 108 million paired reads were generated per sample.

ATAC-Seq Data Analysis

Reads were trimmed for both quality (≤15) and Illumina adaptor sequences using Trim Galore v0.4.5 (https://github.com/FelixKrueger/TrimGalore) then aligned to human assembly hg38 using bowtie2 (67) with the default parameters. Duplicated reads that have the same start site and orientation were removed using the Picard tool MarkDuplicates (http://broadinstitute.github.io/picard). Accessible regions were called using MACS2 v2.1.2 (68) against standard input as the control (fold change >2 and P < 0.001). Regions overlapping with genomic “blacklisted” regions (http://mitra.stanford.edu/kundaje/akundaje/release/blacklists/hg38-human/hg38.blacklist.bed.gz) were removed. The peaks from all samples were then merged within 500 bp to create a full peak atlas. Raw read counts were tabulated over this peak atlas using featureCounts (69). All genome browser tracks were normalized to a sequencing depth of ten million mapped reads. Peaks were annotated using linear genomic distance, with a gene assigned to a peak if it was within 50 kb up- or downstream of the gene start or end, respectively. Raw read counts in the peak atlas were normalized in DESeq2 (70) prior differential and motif analyses. Significant differential regions were generally accepted with fold change greater than 1.5 and Padj value less than 0.1. Data is available at GEO (accession number GSE245187).

Data Availability

Data (bulk RNA-seq, bulk ATAC-seq, scRNA-seq) generated in this study can accessed at GEO (accession number GSE245187).

Statistical Analysis

All statistical analyses were performed using the Prism 9 (GraphPad) software. No statistical methods were used to predetermine sample size. Statistical comparisons between two groups were determined by two-tailed parametric or nonparametric (Mann–Whitney U test) t tests for unpaired data or by two-tailed paired Student t tests for matched samples. For in vivo experiments, the overall survival was depicted by a Kaplan–Meier curve and the log-rank test was used to compare survival differences between the groups. P values < 0.05 were considered to be statistically significant. The statistical test used for each figure is described in the corresponding figure legend. *, P < 0.05; **, P < 0.01; ***, P < 0.0001; ****, P < 0.00001.

N. Jain reports grants from Mnemo Therapeutics during the conduct of the study. Z. Zhao reports grants from Mnemo Therapeutics during the conduct of the study. T. Giavridis reports personal fees from Mnemo Therapeutics during the conduct of the study; in addition, T. Giavridis has a patent for WO2023126458A1 pending to Mnemo Therapeutics, is listed as a coinventor in other cell therapy–related patents that are unrelated to this work, and is an equity holder in Arsenal Biosciences, Inc. M. Sadelain reports grants from Mnemo Therapeutics during the conduct of the study, grants from Atara Biotherapeutics and Fate Therapeutics, and grants from Takeda Pharmaceuticals outside the submitted work. No other disclosures were reported.

N. Jain: Conceptualization, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. Z. Zhao: Conceptualization, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. R.P. Koche: Software, formal analysis, investigation, visualization. C. Antelope: Software, formal analysis, investigation, visualization. Y. Gozlan: Software, formal analysis, investigation, visualization. A. Montalbano: Formal analysis, investigation. D. Brocks: Software, formal analysis, investigation, visualization. M. Lopez: Software, formal analysis, investigation, visualization. A. Dobrin: Formal analysis, investigation. Y. Shi: Formal analysis, investigation. G. Gunset: Project administration. T. Giavridis: Conceptualization. M. Sadelain: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, visualization, methodology, writing–original draft, project administration, writing-review and editing.

We thank members of the Sadelain lab for helpful discussion and feedback. We thank the following SKI core facilities for their support: Flow Cytometry, Anti-tumor Assessment, Integrated Genomics Operation, and the Cell Therapy and Cell Engineering Facility (CTCEF). We thank the Immunai Platform team for support with single-cell multiomic profiling and analysis. This work was supported by Mnemo Therapeutics, the Pasteur-Weizmann/Servier award (to M. Sadelain), the Leopold Griffuel/Fondation ARC award (to M. Sadelain), and NCI MSKCC core grant P30 CA008748.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.

Note: Supplementary data for this article are available at Cancer Discovery Online (http://cancerdiscovery.aacrjournals.org/).

1.
Sadelain
M
,
Riviere
I
,
Riddell
S
.
Therapeutic T cell engineering
.
Nature
2017
;
545
:
423
31
.
2.
June
CH
,
Sadelain
M
.
Chimeric antigen receptor therapy
.
N Engl J Med
2018
;
379
:
64
73
.
3.
Salter
AI
,
Pont
MJ
,
Riddell
SR
.
Chimeric antigen receptor-modified T cells: CD19 and the road beyond
.
Blood
2018
;
131
:
2621
9
.
4.
Park
JH
,
Riviere
I
,
Gonen
M
,
Wang
X
,
Senechal
B
,
Curran
KJ
, et al
.
Long-term follow-up of CD19 CAR therapy in acute lymphoblastic leukemia
.
N Engl J Med
2018
;
378
:
449
59
.
5.
Maude
SL
,
Laetsch
TW
,
Buechner
J
,
Rives
S
,
Boyer
M
,
Bittencourt
H
, et al
.
Tisagenlecleucel in children and young adults with B-cell lymphoblastic leukemia
.
N Engl J Med
2018
;
378
:
439
48
.
6.
Cappell
KM
,
Sherry
RM
,
Yang
JC
,
Goff
SL
,
Vanasse
DA
,
McIntyre
L
, et al
.
Long-term follow-up of anti-CD19 chimeric antigen receptor t-cell therapy
.
J Clin Oncol
2020
;
38
:
3805
15
.
7.
Chong
EA
,
Ruella
M
,
Schuster
SJ
;
Lymphoma Program Investigators at the University of Pennsylvania
.
Five-year outcomes for refractory B-cell lymphomas with CAR T-cell therapy
.
N Engl J Med
2021
;
384
:
673
4
.
8.
Gade
TP
,
Hassen
W
,
Santos
E
,
Gunset
G
,
Saudemont
A
,
Gong
MC
, et al
.
Targeted elimination of prostate cancer by genetically directed human T lymphocytes
.
Cancer Res
2005
;
65
:
9080
8
.
9.
Turtle
CJ
,
Hanafi
LA
,
Berger
C
,
Gooley
TA
,
Cherian
S
,
Hudecek
M
, et al
.
CD19 CAR-T cells of defined CD4+:CD8+ composition in adult B cell all patients
.
J Clin Invest
2016
;
126
:
2123
38
.
10.
Hamieh
M
,
Dobrin
A
,
Cabriolu
A
,
van der Stegen
SJC
,
Giavridis
T
,
Mansilla-Soto
J
, et al
.
CAR T cell trogocytosis and cooperative killing regulate tumour antigen escape
.
Nature
2019
;
568
:
112
6
.
11.
Majzner
RG
,
Rietberg
SP
,
Sotillo
E
,
Dong
R
,
Vachharajani
VT
,
Labanieh
L
, et al
.
Tuning the antigen density requirement for CAR T-cell activity
.
Cancer Discov
2020
;
10
:
702
23
.
12.
Kakarla
S
,
Gottschalk
S
.
CAR T cells for solid tumors: armed and ready to go?
Cancer J
2014
;
20
:
151
5
.
13.
Globerson Levin
A
,
Riviere
I
,
Eshhar
Z
,
Sadelain
M
.
CAR T cells: building on the CD19 paradigm
.
Eur J Immunol
2021
;
51
:
2151
63
.
14.
Fuca
G
,
Reppel
L
,
Landoni
E
,
Savoldo
B
,
Dotti
G
.
Enhancing chimeric antigen receptor T-cell efficacy in solid tumors
.
Clin Cancer Res
2020
;
26
:
2444
51
.
15.
Hull
CM
,
Maher
J
.
Novel approaches to promote CAR T-cell function in solid tumors
.
Expert Opin Biol Ther
2019
;
19
:
789
99
.
16.
Weber
EW
,
Parker
KR
,
Sotillo
E
,
Lynn
RC
,
Anbunathan
H
,
Lattin
J
, et al
.
Transient rest restores functionality in exhausted CAR-T cells through epigenetic remodeling
.
Science
2021
;
372
:
eaba1786
.
17.
Maude
SL
,
Frey
N
,
Shaw
PA
,
Aplenc
R
,
Barrett
DM
,
Bunin
NJ
, et al
.
Chimeric antigen receptor T cells for sustained remissions in leukemia
.
N Engl J Med
2014
;
371
:
1507
17
.
18.
van der Stegen
SJ
,
Hamieh
M
,
Sadelain
M
.
The pharmacology of second-generation chimeric antigen receptors
.
Nat Rev Drug Discov
2015
;
14
:
499
509
.
19.
Eyquem
J
,
Mansilla-Soto
J
,
Giavridis
T
,
van der Stegen
SJ
,
Hamieh
M
,
Cunanan
KM
, et al
.
Targeting a CAR to the TRAC locus with CRISPR/Cas9 enhances tumour rejection
.
Nature
2017
;
543
:
113
7
.
20.
Cappell
KM
,
Kochenderfer
JN
.
A comparison of chimeric antigen receptors containing CD28 versus 4-1BB costimulatory domains
.
Nat Rev Clin Oncol
2021
;
18
:
715
27
.
21.
Zhao
Z
,
Condomines
M
,
van der Stegen
SJC
,
Perna
F
,
Kloss
CC
,
Gunset
G
, et al
.
Structural design of engineered costimulation determines tumor rejection kinetics and persistence of CAR T cells
.
Cancer Cell
2015
;
28
:
415
28
.
22.
Long
AH
,
Haso
WM
,
Shern
JF
,
Wanhainen
KM
,
Murgai
M
,
Ingaramo
M
, et al
.
4-1BB costimulation ameliorates T cell exhaustion induced by tonic signaling of chimeric antigen receptors
.
Nat Med
2015
;
21
:
581
90
.
23.
Kaech
SM
,
Cui
W
.
Transcriptional control of effector and memory CD8+ T cell differentiation
.
Nat Rev Immunol
2012
;
12
:
749
61
.
24.
Blank
CU
,
Haining
WN
,
Held
W
,
Hogan
PG
,
Kallies
A
,
Lugli
E
, et al
.
Defining ‘T cell exhaustion.’
Nat Rev Immunol
2019
;
19
:
665
74
.
25.
Wherry
EJ
.
T cell exhaustion
.
Nat Immunol
2011
;
12
:
492
9
.
26.
Kaech
SM
,
Wherry
EJ
,
Ahmed
R
.
Effector and memory T-cell differentiation: implications for vaccine development
.
Nat Rev Immunol
2002
;
2
:
251
62
.
27.
Philip
M
,
Fairchild
L
,
Sun
L
,
Horste
EL
,
Camara
S
,
Shakiba
M
, et al
.
Chromatin states define tumour-specific T cell dysfunction and reprogramming
.
Nature
2017
;
545
:
452
6
.
28.
Zebley
CC
,
Gottschalk
S
,
Youngblood
B
.
Rewriting history: epigenetic reprogramming of CD8(+) T cell differentiation to enhance immunotherapy
.
Trends Immunol
2020
;
41
:
665
75
.
29.
Pace
L
,
Amigorena
S
.
Epigenetics of T cell fate decision
.
Curr Opin Immunol
2020
;
63
:
43
50
.
30.
Allis
CD
,
Jenuwein
T
.
The molecular hallmarks of epigenetic control
.
Nat Rev Genet
2016
;
17
:
487
500
.
31.
Prinzing
B
,
Zebley
CC
,
Petersen
CT
,
Fan
Y
,
Anido
AA
,
Yi
Z
, et al
.
Deleting DNMT3A in CAR T cells prevents exhaustion and enhances antitumor activity
.
Sci Transl Med
2021
;
13
:
eabh0272
.
32.
Jain
N
,
Zhao
Z
,
Feucht
J
,
Koche
R
,
Iyer
A
,
Dobrin
A
, et al
.
TET2 guards against unchecked BATF3-induced CAR T cell expansion
.
Nature
2023
;
615
:
315
22
.
33.
Rea
S
,
Eisenhaber
F
,
O'Carroll
D
,
Strahl
BD
,
Sun
ZW
,
Schmid
M
, et al
.
Regulation of chromatin structure by site-specific histone H3 methyltransferases
.
Nature
2000
;
406
:
593
9
.
34.
Nicetto
D
,
Donahue
G
,
Jain
T
,
Peng
T
,
Sidoli
S
,
Sheng
L
, et al
.
H3K9me3-heterochromatin loss at protein-coding genes enables developmental lineage specification
.
Science
2019
;
363
:
294
7
.
35.
Ninova
M
,
Fejes Toth
K
,
Aravin
AA
.
The control of gene expression and cell identity by H3K9 trimethylation
.
Development
2019
;
146
:
dev181180
.
36.
Allan
RS
,
Zueva
E
,
Cammas
F
,
Schreiber
HA
,
Masson
V
,
Belz
GT
, et al
.
An epigenetic silencing pathway controlling T helper 2 cell lineage commitment
.
Nature
2012
;
487
:
249
53
.
37.
Pace
L
,
Goudot
C
,
Zueva
E
,
Gueguen
P
,
Burgdorf
N
,
Waterfall
JJ
, et al
.
The epigenetic control of stemness in CD8(+) T cell fate commitment
.
Science
2018
;
359
:
177
86
.
38.
Brentjens
RJ
,
Latouche
JB
,
Santos
E
,
Marti
F
,
Gong
MC
,
Lyddane
C
, et al
.
Eradication of systemic B-cell tumors by genetically targeted human T lymphocytes co-stimulated by CD80 and interleukin-15
.
Nat Med
2003
;
9
:
279
86
.
39.
Zhong
XS
,
Matsushita
M
,
Plotkin
J
,
Riviere
I
,
Sadelain
M
.
Chimeric antigen receptors combining 4-1BB and CD28 signaling domains augment PI3kinase/AKT/Bcl-XL activation and CD8+ T cell-mediated tumor eradication
.
Mol Ther
2010
;
18
:
413
20
.
40.
Feucht
J
,
Sun
J
,
Eyquem
J
,
Ho
YJ
,
Zhao
Z
,
Leibold
J
, et al
.
Calibration of CAR activation potential directs alternative T cell fates and therapeutic potency
.
Nat Med
2019
;
25
:
82
8
.
41.
Kallies
A
,
Zehn
D
,
Utzschneider
DT
.
Precursor exhausted T cells: key to successful immunotherapy?
Nat Rev Immunol
2020
;
20
:
128
36
.
42.
Zhao
X
,
Shan
Q
,
Xue
HH
.
TCF1 in T cell immunity: a broadened frontier
.
Nat Rev Immunol
2022
;
22
:
147
57
.
43.
Maher
J
,
Brentjens
RJ
,
Gunset
G
,
Riviere
I
,
Sadelain
M
.
Human T-lymphocyte cytotoxicity and proliferation directed by a single chimeric TCRzeta /CD28 receptor
.
Nat Biotechnol
2002
;
20
:
70
5
.
44.
Imai
C
,
Mihara
K
,
Andreansky
M
,
Nicholson
IC
,
Pui
CH
,
Geiger
TL
, et al
.
Chimeric receptors with 4-1BB signaling capacity provoke potent cytotoxicity against acute lymphoblastic leukemia
.
Leukemia
2004
;
18
:
676
84
.
45.
Davila
ML
,
Riviere
I
,
Wang
X
,
Bartido
S
,
Park
J
,
Curran
K
, et al
.
Efficacy and toxicity management of 19-28z CAR T cell therapy in B cell acute lymphoblastic leukemia
.
Sci Transl Med
2014
;
6
:
224ra25
.
46.
Lee
DW
,
Kochenderfer
JN
,
Stetler-Stevenson
M
,
Cui
YK
,
Delbrook
C
,
Feldman
SA
, et al
.
T cells expressing CD19 chimeric antigen receptors for acute lymphoblastic leukaemia in children and young adults: a phase 1 dose-escalation trial
.
Lancet
2015
;
385
:
517
28
.
47.
Gray
SM
,
Amezquita
RA
,
Guan
T
,
Kleinstein
SH
,
Kaech
SM
.
Polycomb repressive complex 2-mediated chromatin repression guides effector CD8(+) T cell terminal differentiation and loss of multipotency
.
Immunity
2017
;
46
:
596
608
.
48.
Youngblood
B
,
Hale
JS
,
Kissick
HT
,
Ahn
E
,
Xu
X
,
Wieland
A
, et al
.
Effector CD8 T cells dedifferentiate into long-lived memory cells
.
Nature
2017
;
552
:
404
9
.
49.
Xu
Y
,
Zhang
M
,
Ramos
CA
,
Durett
A
,
Liu
E
,
Dakhova
O
, et al
.
Closely related T-memory stem cells correlate with in vivo expansion of CAR.CD19-T cells and are preserved by IL-7 and IL-15
.
Blood
2014
;
123
:
3750
9
.
50.
Monfrini
C
,
Stella
F
,
Aragona
V
,
Magni
M
,
Ljevar
S
,
Vella
C
, et al
.
Phenotypic composition of commercial anti-CD19 CAR T cells affects in vivo expansion and disease response in patients with large B-cell lymphoma
.
Clin Cancer Res
2022
;
28
:
3378
86
.
51.
Vandel
L
,
Nicolas
E
,
Vaute
O
,
Ferreira
R
,
Ait-Si-Ali
S
,
Trouche
D
.
Transcriptional repression by the retinoblastoma protein through the recruitment of a histone methyltransferase
.
Mol Cell Biol
2001
;
21
:
6484
94
.
52.
Nielsen
SJ
,
Schneider
R
,
Bauer
UM
,
Bannister
AJ
,
Morrison
A
,
O'Carroll
D
, et al
.
Rb targets histone H3 methylation and HP1 to promoters
.
Nature
2001
;
412
:
561
5
.
53.
Sade-Feldman
M
,
Yizhak
K
,
Bjorgaard
SL
,
Ray
JP
,
de Boer
CG
,
Jenkins
RW
, et al
.
Defining T cell states associated with response to checkpoint immunotherapy in melanoma
.
Cell
2018
;
175
:
998
1013
.
54.
Lu
C
,
Yang
D
,
Klement
JD
,
Oh
IK
,
Savage
NM
,
Waller
JL
, et al
.
SUV39H1 represses the expression of cytotoxic T-lymphocyte effector genes to promote colon tumor immune evasion
.
Cancer Immunol Res
2019
;
7
:
414
27
.
55.
Niborski
LL
,
Gueguen
P
,
Ye
M
,
Thiolat
A
,
Ramos
RN
,
Caudana
P
, et al
.
CD8+ T cell responsiveness to anti–PD-1 is epigenetically regulated by Suv39h1 in melanomas
.
Nat Commun
2022
;
13
:
3739
.
56.
Tefferi
A
,
Lim
KH
,
Levine
R
.
Mutation in TET2 in myeloid cancers
.
N Engl J Med
2009
;
361
:
1117
.
57.
Couronne
L
,
Bastard
C
,
Bernard
OA
.
TET2 and DNMT3A mutations in human T-cell lymphoma
.
N Engl J Med
2012
;
366
:
95
6
.
58.
Goyama
S
,
Nitta
E
,
Yoshino
T
,
Kako
S
,
Watanabe-Okochi
N
,
Shimabe
M
, et al
.
EVI-1 interacts with histone methyltransferases SUV39H1 and G9a for transcriptional repression and bone marrow immortalization
.
Leukemia
2010
;
24
:
81
8
.
59.
Rodrigues
C
,
Pattabiraman
C
,
Vijaykumar
A
,
Arora
R
,
Narayana
SM
,
Kumar
RV
, et al
.
A SUV39H1-low chromatin state characterises and promotes migratory properties of cervical cancer cells
.
Exp Cell Res
2019
;
378
:
206
16
.
60.
Riviere
I
,
Brose
K
,
Mulligan
RC
.
Effects of retroviral vector design on expression of human adenosine deaminase in murine bone marrow transplant recipients engrafted with genetically modified cells
.
Proc Natl Acad Sci U S A
1995
;
92
:
6733
7
.
61.
Gallardo
HF
,
Tan
C
,
Ory
D
,
Sadelain
M
.
Recombinant retroviruses pseudotyped with the vesicular stomatitis virus G glycoprotein mediate both stable gene transfer and pseudotransduction in human peripheral blood lymphocytes
.
Blood
1997
;
90
:
952
7
.
62.
Brentjens
RJ
,
Santos
E
,
Nikhamin
Y
,
Yeh
R
,
Matsushita
M
,
La Perle
K
, et al
.
Genetically targeted T cells eradicate systemic acute lymphoblastic leukemia xenografts
.
Clin Cancer Res
2007
;
13
:
5426
35
.
63.
Gong
MC
,
Latouche
JB
,
Krause
A
,
Heston
WD
,
Bander
NH
,
Sadelain
M
.
Cancer patient T cells genetically targeted to prostate-specific membrane antigen specifically lyse prostate cancer cells and release cytokines in response to prostate-specific membrane antigen
.
Neoplasia
1999
;
1
:
123
7
.
64.
Stephan
MT
,
Ponomarev
V
,
Brentjens
RJ
,
Chang
AH
,
Dobrenkov
KV
,
Heller
G
, et al
.
T cell-encoded CD80 and 4-1BBL induce auto- and transcostimulation, resulting in potent tumor rejection
.
Nat Med
2007
;
13
:
1440
9
.
65.
Stoeckius
M
,
Hafemeister
C
,
Stephenson
W
,
Houck-Loomis
B
,
Chattopadhyay
PK
,
Swerdlow
H
, et al
.
Simultaneous epitope and transcriptome measurement in single cells
.
Nat Methods
2017
;
14
:
865
8
.
66.
Buenrostro
JD
,
Giresi
PG
,
Zaba
LC
,
Chang
HY
,
Greenleaf
WJ
.
Transposition of native chromatin for fast and sensitive epigenomic profiling of open chromatin, DNA-binding proteins and nucleosome position
.
Nat Methods
2013
;
10
:
1213
8
.
67.
Langmead
B
,
Salzberg
SL
.
Fast gapped-read alignment with Bowtie 2
.
Nat Methods
2012
;
9
:
357
9
.
68.
Zhang
Y
,
Liu
T
,
Meyer
CA
,
Eeckhoute
J
,
Johnson
DS
,
Bernstein
BE
, et al
.
Model-based analysis of ChIP-Seq (MACS)
.
Genome Biol
2008
;
9
:
R137
.
69.
Liao
Y
,
Smyth
GK
,
Shi
W
.
featureCounts: an efficient general purpose program for assigning sequence reads to genomic features
.
Bioinformatics
2014
;
30
:
923
30
.
70.
Love
MI
,
Huber
W
,
Anders
S
.
Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2
.
Genome Biol
2014
;
15
:
550
.