Epigenetic training licenses naïve CD8+ T cell metabolic fitness and function Authors: Mengjie Lv, Jiyu Ding, Haizhen Long, Qian Chai, Lili Xu, Caiwei Jin, Wei Liang, Liwei Zhang, Bangmin Zhang, Zhihan Lv, Jicheng Zhao, Guohong Li, Mingzhao Zhu Published: 2026 ABSTRACT Naïve T cells maintain quiescence yet must respond rapidly to antigens, but how they prime this capacity is unclear. We identify histone variant H2A.Z as a key regulator of an epigenetic training program that licenses quiescent naïve CD8+ T cells for future activation. H2A.Z deficiency disrupts naïve T cell homeostasis and effector responses. Multiomics reveals that H2A.Z is selectively deposited at oxidative phosphorylation (OXPHOS) gene promoters in quiescent naïve CD8+ T cells, priming the chromatin for rapid transcriptional induction. This training is developmentally instructed by tonic interleukin-7 (IL-7) signaling and regulated by transcription factor GABPα. Age-related decline in IL-7 signaling reduces H2A.Z occupancy and impairs T cell activation, while IL-7 supplementation or enforced H2A.Z expression rescues this defect. H2A.Z overexpression also enhances chimeric antigen receptor T cell stemness and antitumor efficacy. Our work defines an IL-7R–GABPα–H2A.Z–OXPHOS axis that epigenetically establishes metabolic and functional fitness in quiescent T cells, offering insights for immunotherapy targeting ageing and tumors. INTRODUCTION Naïve T cells are poised at a critical balance between quiescence and activation, enabling both long-term persistence and rapid immune response. In young, healthy individuals, these cells remain in a quiescent state, marked by small cell size, limited proliferation, basal metabolic activity, and minimal production of effector molecules. However, upon recognition of cognate antigen via T cell receptor (TCR) and costimulatory signals, naïve T cells swiftly exit this quiescent state and undergo profound transcriptional and metabolic reprogramming. These changes support their activation, differentiation, proliferation, and acquisition of effector functions (1, 2). Thus, naïve T cells face a dual imperative: preserving a quiescent, low-energy state to support longevity, while maintaining the molecular preparedness required for immediate activation and response. Numerous studies have focused on naïve T cell responses following antigen encounter (1–5). However, whether and how naïve T cells, despite their quiescent state prior to this event, establish a ready capacity for rapid activation remains poorly understood. Epigenetic mechanisms, including DNA methylation, histone modifications, and higher-order chromatin structure, play a pivotal role in shaping adaptive T cell immunity. They dynamically regulate transcriptional programs that control lineage commitment, long-term memory formation, and the development of T cell exhaustion (6–11). Therapeutically, epigenetic modulation has emerged as a promising approach to reverse exhaustion signatures and enhance the persistence and functionality of chimeric antigen receptor T cells (CAR T cells) in cancer immunotherapy (12–14). While these mechanisms are well characterized in differentiating and effector T cells, it remains unclear whether and how epigenetic landscapes preconfigure naïve T cells, enabling their transition from quiescence to rapid activation upon initial antigen encounter. The histone variant H2A.Z is highly conserved across evolution and is incorporated into chromatin in a replication-independent manner, where it notably influences chromatin architecture, DNA replication, and transcriptional regulation. Our recent work has shown that H2A.Z facilitates the licensing and activation of early replication origins, thereby promoting efficient DNA replication and cell division (15). Furthermore, H2A.Z is enriched at regulatory regions such as promoters and enhancers, where it helps prime genes for rapid transcriptional activation (16–19). Despite these broad roles, the function of H2A.Z in naïve T cells remains poorly understood. It is therefore compelling to investigate whether H2A.Z acts as a critical molecular bridge that helps maintain the balance between quiescence and rapid activation in these cells. In this study, we identify H2A.Z as a central mediator in a metabolic-epigenetic training program in naïve CD8+ T cells. We show that H2A.Z orchestrates this training by poising oxidative phosphorylation (OXPHOS) genes for rapid transcription, a process directed by interleukin-7 (IL-7) signaling and GABPα. The failure of this program in ageing and its successful artificial execution in CAR T cells demonstrate that epigenetic training is critical for T cell immunity, establishing a new paradigm for how functional capacity is acquired during quiescence. RESULTS H2A.Z is essential for naïve T cell homeostasis and activation To study the role of H2A.Z in T cells, we generated CD4creH2A.Zfl/fl conditional knockout (cKO) mice. Western blot analysis confirmed that H2A.Z was efficiently knocked out at the protein level in peripheral T cells and that effective knockout began during thymic development in CD4 and CD8 single-positive (SP) thymocytes (fig. S1A). To detect the role of H2A.Z in T cell development and homeostasis, we analyzed the populations of T cell subsets in wild-type (WT) and H2A.Z cKO mice. Thymocyte development was grossly normal, with comparable CD4SP, CD8SP, and mature (CD69−CD62L+) thymocyte subsets observed in H2A.Z cKO mice (fig. S1, B to D). However, in the periphery, both the frequency and absolute number of CD8+ and CD4+ T cells in the spleen were significantly reduced in H2A.Z cKO mice (fig. S1E). Furthermore, the proportions of naïve, effector, and central memory-like CD44hi subsets among both CD8+ and CD4+ T cells were all markedly reduced (Fig. 1A and fig. S1F). The reduction in peripheral naïve T cells in H2A.Z cKO mice is unlikely due to impaired thymic egress, as the percentage of pre-egress mature CD69−CD62L+ SP thymocytes is normal (fig. S1, C and D) (8, 20). It is also unlikely to be caused by impaired survival, based on 7-AAD and annexin V staining (fig. S1, G and H). Instead, this reduction likely results from impaired homeostatic proliferation, as we previously reported (15). H2A.Z is required for naïve T cell homeostasis and response. (A) The percentages of naïve (Tn), effector (Teff), and memory (Tcm) CD8+ T cell subsets in the spleens of H2A.Z cKO mice and their littermate controls were measured by flow cytometry, and the cell numbers for the indicated populations were calculated (n = 4 mice per group). (B and C) Sorted naïve CD8+ T cells were stimulated with anti-CD3 and anti-CD28 antibodies for 24 hours, and the frequencies of CD69+ and CD25+ cells were assessed by flow cytometry (n = 4 biologically independent samples). (D and E) Sorted naïve CD8+ T cells were stimulated with anti-CD3 and anti-CD28 antibodies for 48 hours. Cytokine production was then measured by flow cytometry (n = 3 biologically independent samples). (F and G) The percentages of IFN-γ–producing T cells in the spleens of H2A.Z cKO mice and their littermate controls were determined 7 days after infection with LM-OVA. Splenocytes were restimulated with LLO190–210 (for CD4+ T cells) or OVA257–264 (for CD8+ T cells) peptide in the presence of BFA for 4 hours before analysis (n = 3 mice per group). (H) Tumor growth curves in H2A.Z cKO mice (red) and WT controls (blue) following inoculation with B16-OVA tumor cells (n = 8 mice per group). (I) The percentage of CD8+ T cells in the draining LNs and tumors of WT/H2A.Z cKO mixed bone marrow chimeric mice inoculated with B16 tumor cells (n = 3 mice per group). Data representative of two independent experiments. Data are means ± SEM. Statistical analysis was performed using two-tailed unpaired Student’s t test [(A) to (G) and( I)] or two-way analysis of variance (ANOVA) with Tukey’s multiple comparisons test (H). Given the notable reduction in effector/memory T cell subsets in H2A.Z cKO mice, we wondered whether H2A.Z may also regulate naïve T cell activation, in addition to proliferation. To directly test this, we used an in vitro system using naïve CD8+ and CD4+ T cells stimulated with agonistic antibodies against the TCR (anti-CD3) and the costimulatory molecule CD28 (anti-CD28). After 24 hours of stimulation, H2A.Z cKO CD8+ T cells showed significantly impaired up-regulation of the early activation markers CD69 and CD25 compared to WT cells (Fig. 1, B and C). The production of effector cytokines interferon-γ (IFN-γ) and IL-2 was also reduced at 48 hours in cKO CD8+ T cells (Fig. 1, D and E). Similar defects were observed in CD4+ T cells (fig. S1, I to L). To exclude the potential effect of lymphopenia on naïve T cell activation, mixed bone marrow chimeric mice were generated (fig. S2A). In these chimeric mice, the proportions of CD8+ and CD4+ T cell subsets derived from cKO mice were also significantly reduced (fig. S2, B to E), confirming the intrinsic role of H2A.Z in T cell homeostasis. In addition, a decreased percentage of Ki-67+ cells were observed in H2A.Z-deficient naïve CD8+ T cells, but not in naïve CD4+ T cells (fig. S2, F and G). Nonetheless, both CD8+ and CD4+ T cells lacking H2A.Z derived from mixed bone marrow chimeric mice displayed reduced activation capacity upon TCR stimulation (fig. S2, H to J, for CD8+ T cells and fig. S2, K to M, for CD4+ T cells), while the defect was less severe in CD4+ T cells. These data demonstrate that H2A.Z plays an important intrinsic role in the early activation of naïve T cells, highlighting its essential role in initiating effective T cell immune responses. Lack of H2A.Z diminishes the T cell response in vivo To further investigate the role of H2A.Z in T cell response in vivo, we used Listeria monocytogenes infection model, which elicits a robust T cell response characterized by the induction of IFN-γ–producing CD4+ T helper 1 (TH1) cells and CD8+ effector T cells. To assess the intrinsic role of H2A.Z in this process, mixed bone marrow chimeric mice were infected with recombinant L. monocytogenes expressing chicken ovalbumin (LM-OVA). Seven days postinfection, splenocytes were isolated and restimulated with major histocompatibility complex (MHC) class II–restricted listeriolysin O peptide (LLO190–201) and MHC class I–restricted OVA peptide (OVA257–264). In stark contrast to the robust CD8+ and CD4+ T cell responses observed in the WT compartment, antigen-specific T cell response was nearly undetectable in the H2A.Z KO T cell compartment (Fig. 1, F and G), highlighting a critical role for H2A.Z in regulating the T cell response during bacterial infection. To evaluate the role of H2A.Z in antitumor T cell response, WT and H2A.Z cKO mice were inoculated with B16-OVA melanoma cells. H2A.Z cKO mice showed significantly accelerated tumor growth compared to WT controls (Fig. 1H), indicating a defective antitumor T cell response in the absence of H2A.Z. To determine the intrinsic role of H2A.Z, we inoculated mixed bone marrow chimeric mice with B16-OVA tumor cells and quantified tumor-infiltrating CD8+ cytotoxic T lymphocytes (CTLs) by flow cytometry at various time points postinoculation. Consistently, the H2A.Z KO T cell compartment showed a markedly reduced frequency of tumor-infiltrating CTLs compared to the WT compartment (Fig. 1I). These findings suggest that H2A.Z is intrinsically required in T cells for the effective generation of both anti-infection and antitumor immune responses. H2A.Z regulates mitochondrial OXPHOS for naïve CD8+ T cell activation H2A.Z is highly enriched at the promoter regions of a large subset of genes across cell types, playing a role in transcriptional regulation (21–23). To uncover the molecular mechanisms underlying impaired T cell activation during H2A.Z deficiency, we performed RNA sequencing (RNA-seq) analysis on naïve and 24-hour in vitro–activated CD8+ T cells from WT and H2A.Z-deficient mice. While WT and KO T cells share the majority of up-regulated (727) or down-regulated (584) genes upon activation [|log2(fold change)| ≥ 1, FDR ≤ 0.05], WT- or KO-specific changes of differentially expressed genes (DEGs) exist (Fig. 2A). Gene Ontology (GO) analysis of those DEGs revealed that genes uniquely and preferentially up-regulated during T cell activation in WT T cells, compared to those preferentially up-regulated in KO T cells, were enriched in pathways of mitochondrial function (Fig. 2B). GSEA analysis further indicated that upon stimulation, WT transcriptome is enriched with OXPHOS metabolic pathway (Fig. 2C), whereas in KO cells, this enrichment was significantly attenuated compared with WT cells (Fig. 2D). Furthermore, upon stimulation, the majority of OXPHOS signature genes (88 out of 96) were down-regulated in KO T cells compared to WT controls (Fig. 2E). For CD4+ T cells, although WT cells showed enrichment of mitochondrial function and OXPHOS pathway upon stimulation (fig. S3, A to C), GSEA analysis did not reveal a significant defect of OXPHOS pathway in KO cells compared to WT cells (fig. S3D). H2A.Z deficiency impairs the transcription of genes involved in OXPHOS during CD8+ T cell activation. (A) The numbers in the Venn diagram indicate the DEGs between activated and naïve states in WT and H2A.Z cKO samples (n = 3 biologically independent samples). (B) Heatmap colors indicate –log10-transformed P values from GO enrichment analysis. (C and D) GSEA plots showing enrichment of OXPHOS gene signatures in the comparisons of activated versus naïve CD8+ T cells (C) and H2A.Z cKO versus WT samples (D). (E) Heatmap of DEGs from the OXPHOS gene list. The comparison focuses on DEGs between WT and cKO samples in activated CD8+ T cells. Heatmap colors represent z-scored FPKM values, with rows corresponding to DEGs and columns representing samples. (F to H) Sorted naïve CD8+ T cells were stimulated with anti-CD3/CD28 for 24 hours. MFI of MitoGreen (F), MitoRed (G), and MitoSOX (H) in T cells was measured by flow cytometry (n = 3 biologically independent samples). (I and J) Sorted naïve CD8+ T cells were stimulated for 24 hours and then analyzed for metabolic function (n = 6 to 8 biologically independent samples), indicated by OCR parameters: basal respiration, maximal respiration, ATP production, and spare respiratory capacity. (K) Jurkat T cells with the indicated gene knockout were stimulated with anti-CD3/CD28 for 24 hours. The expression of Il2 was measured by qPCR (n = 3 biologically independent samples). (L and M) Sorted naïve CD8+ T cells were stimulated with anti-CD3/CD28 for 24 hours with or without ATP addition. The percentages of CD69+ (L) and CD25+ (M) CD8+ T cells were measured by flow cytometry (n = 3 biologically independent samples). Data representative of two independent experiments. Data are means ± SEM. Statistical analysis was performed using two-tailed unpaired Student’s t test [(F) to (J), (L), and (M)] or one-way ANOVA and multiple comparisons test (K). Mitochondrial OXPHOS has been shown to be essential for T cell activation via reactive oxygen species (ROS) and adenosine 5′-triphosphate (ATP) production (24, 25). Because a number of OXPHOS-related genes are dysregulated in H2A.Z KO CD8+ T cells, we tested whether OXPHOS is impaired in H2A.Z KO CD8+ T cells. To this end, activated CD8+ T cells were first stained with MitoTracker Green and Deep Red to measure the mitochondrial mass and mitochondrial membrane potential. Flow cytometry analysis revealed that both mitochondrial mass and membrane potential were reduced in H2A.Z-deficient T cells (Fig. 2, F and G). In addition, direct measurement of mitochondrial ROS (mitoROS) levels showed a significant reduction in activated H2A.Z KO T cells compared to WT T cells (Fig. 2H). Furthermore, Seahorse assay showed that oxygen consumption rate (OCR), ATP production, and spare respiratory capacity were all diminished in H2A.Z-deficient CD8+ T cells upon activation (Fig. 2, I and J). In contrast, Seahorse assay did not reveal significant defect of OXPHOS in H2A.Z-deficient CD4+ T cells (fig. S3, E and F), consistent with the transcriptome assay, suggesting a less important role of H2A.Z on OXPHOS regulation in CD4+ T cells. To further confirm the function of H2A.Z-targeting OXPHOS genes in CD8+ T cell activation, we chose NDUFC1 (the largest subunit of complex I), NDUFA8 (a subunit of complex I), and ATP5E (a subunit of complex V) and generated KO Jurkat cells by CRISPR-Cas9. These genes are critically required for IL-2 expression upon stimulation (Fig. 2K). To further test the contribution of reduced OXPHOS on defective T cell activation in H2A.Z-deficient T cells, ATP was added exogenously to T cell cultures. Flow cytometry analysis showed that T cell activation was largely rescued as measured by the expression of CD69 and CD25 in H2A.Z-deficeint CD8+ T cells (Fig. 2, L and M), suggesting that the defective OXPHOS is likely a major reason in H2A.Z-deficient CD8+ T cells accounting for its defective activation. H2A.Z establishes a poised epigenetic status on OXPHOS gene promoters in quiescent CD8+ T cells To investigate how H2A.Z regulates the expression of OXPHOS genes during T cell activation, we first detected the deposition pattern of H2A.Z in activated CD8+ T cells via chromatin immunoprecipitation sequencing (ChIP-seq) analysis. The data showed that H2A.Z is deposited differently across various gene loci, especially at the promoter regions (TSS) of OXPHOS genes compared to the whole genome (Fig. 3A). Moreover, the deposition levels of H2A.Z at the loci of OXPHOS genes are higher than those at the loci of glycolysis or effector genes (fig. S4, A and B). These data suggested that H2A.Z is selectively enriched at the OXPHOS gene locus in activated CD8+ T cells, consistent with the active transcription of these genes during T cell activation. H2A.Z establishes a poised epigenetic status on OXPHOS gene promoters in quiescent CD8+ T cells. (A) Line plots show H2A.Z binding profiles for gene sets including those involved in oxidative phosphorylation (OP) and all mouse genes (mm10), both upstream and downstream of the transcription start site (TSS) in activated CD8+ T cells. The y axis shows normalized H2A.Z binding levels, averaged per gene set. (B) Line plots and box plots display normalized signal values within 1 kb upstream and downstream of TSS regions for H3K27me3 and H3K27ac datasets in activated WT (blue) and H2A.Z cKO (red) CD8+ T cells. (C) Line plots show H2A.Z binding profiles for gene sets including those involved in OXPHOS and all mouse genes (mm10), both upstream and downstream of the TSS in naïve CD8+ T cells. The y axis shows normalized H2A.Z binding levels, averaged per gene set. (D) Line plots and box plots display normalized signal values within 1 kb upstream and downstream of TSS regions for H3K27me3 and H3K27ac datasets in naïve WT (blue) and H2A.Z cKO (red) CD8+ T cells. (E) Line and box plots display H3K4me3 signal changes in WT and H2A.Z cKO samples under naïve conditions, near TSSs (±1 kb). (F) Line plot displaying the log2(H3K4me3/H3K27me3) ratio within ±1 kb of TSSs for WT and H2A.Z cKO samples. (G and H) Line and box plots display Pol IIS5 signal changes in WT and H2A.Z cKO samples under naïve and activated conditions, near TSSs (±1 kb). All box plots were analyzed using the Wilcoxon rank-sum test, and P values are displayed. A differentially expressed OXPHOS gene set was filtered by RNA-seq dataset [WT versus H2A.Z cKO, for (B) and (D) to (H)]. To further define how H2A.Z might alter the epigenetic status of OXPHOS genes in activated T cells, CUT&Tag assay was performed for H3K27ac and H3K27me3 in WT and KO activated CD8+ T cells. The data demonstrate that H2A.Z depletion significantly alters the epigenetic landscape of OXPHOS promoters in activated CD8+ T cells, characterized by decreased levels of the active histone mark H3K27ac and a concomitant increase of the repressive mark H3K27me3 (Fig. 3B). These data further revealed that H2A.Z is essential for permissive chromatin landscape maintenance in activated T cells. To next test when this H2A.Z-mediated epigenetic status is established, naïve CD8+ T cells were measured. H2A.Z is also highly enriched at OXPHOS gene loci in naïve CD8+ T cells compared to the whole genome (Fig. 3C) or the loci of glycolysis and effector genes (fig. S4, A and B). In H2A.Z deficiency, H3K27ac levels were also decreased and H3K27me3 levels increased in naïve CD8+ T cells (Fig. 3D). Co-occupancy of H3K4me3-H3K27me3 defines a poised modification for gene expression. CUT&Tag assay of H3K4me3 showed impaired deposition in naïve CD8+ T cells in the absence of H2A.Z (Fig. 3E). The ratio of H3K4me3/H3K27me3 is also reduced (Fig. 3F). These data together suggest that H2A.Z establishes a poised epigenetic status on OXPHOS genes even at quiescent state of naïve T cells prior activation. To further test the H2A.Z-mediated epigenetic priming on gene transcription, CUT&Tag assay of Pol IIS5 was performed. While normal level of Pol IIS5 deposition was detected on OXPHOS gene promoters in H2A.Z-deficient naïve CD8+ T cells (Fig. 3G), upon stimulation, significantly less Pol IIS5 deposition was found in H2A.Z-deficient cells than in WT cells (Fig. 3H), suggesting the H2A.Z-mediated poised chromatin status is required for efficient gene transcription induction in T cells upon stimulation. GABPα recruits H2A.Z and regulates mitochondrial OXPHOS for naïve CD8+ T cell activation To elucidate how OXPHOS gene-specific H2A.Z deposition is formed, we first conducted stable isotope labeling with amino acids in cell culture (SILAC) for liquid chromatography–mass spectrometry (MS) to find potential factors interacting with H2A.Z. Jurkat-H2A.Z and Jurkat-H2A cells were cultured with heavy or light amino acids medium, respectively, and then collected for mononucleosome immunoprecipitation (IP) and mass spectrum analysis (Fig. 4, A and B). The top listed factors identified specifically binding with H2A.Z include SRCAP, a well-known chaperone factor involved in H2A.Z deposition, validating the efficacy of this assay. GABPα is a member of Ets family transcription factors and binds DNA via its conserved Ets domain. Deficiency of GABPα affects T cell homeostasis (26). CD4creGABPαfl/fl cKO mice highly phenocopied H2A.Z cKO mice, characterized by reduced T cell homeostasis (fig. S5, A to C), activation (fig. S5, D and E), mitochondrial function, and mitoROS production (fig. S5, F to H), as well as defective expression of OXPHOS genes in activated CD8+ T cells (fig. S5, I and J). These suggests a close functional relationship between GABPα and H2A.Z in the regulation of T cell homeostasis and activation. GABPα recruits H2A.Z deposition on OXPHOS gene promoters. (A) Experimental procedure of SILAC. Created in PowerPoint by M.L. (B) Ranking of log10 (heavy/light signal intensity), with gene names of proteins showing a 100-fold difference in heavy/light signal intensity listed. (C) Mononucleosome co-IP of H2A.Z and GABPα, with H2A serving as a control. (D) Expression of H2A.Z or GABPα in CD8+ T cells sorted from the spleens of WT and cKO mice, determined by Western blot. (E and F) Heatmaps display H2A.Z and GABPα binding densities in CD8+ T cells (n = 2 biologically independent samples), with regions ranked by H2A.Z binding levels. Color intensity corresponds to binding density. (G and H) Line and box plots display normalized signals near TSSs (±1 kb) for WT, H2A.Z cKO (G), and GABPα cKO (H), based on GABPα and H2A.Z datasets (n = 2 biologically independent samples). Blue, WT; red, H2A.Z cKO; orange, GABPα cKO. Significance was assessed using the Wilcoxon rank-sum test; P values are shown. A differentially expressed OXPHOS gene set was filtered by RNA-seq datasets (WT versus H2A.Z cKO, WT versus GABPα cKO) for (G) and (H). (I and J) Heatmaps show normalized signal densities of H2A.Z, GABPα, and H3K27ac in WT and H2A.Z1 KO (I) and GABPα KO (J) Jurkat cells (n = 2 biologically independent samples). Each row represents a ±5-kb genomic region centered on each dataset’s peak summit. Heatmaps were generated using datasets from Jurkat cell lines. (K) Electroporation of the plasmid (dCas9-GABPα-sgRNA) into GABPα KO Jurkat cells was performed, and dCas9-GABPα–induced H2A.Z enrichment at specific gene loci was detected by CUT&Tag-qPCR assay (n = 3 biologically independent samples). Statistical analysis was performed with a two-tailed unpaired Student’s t test. Data representative of two independent experiments. To further validate the interaction between GABPα and H2A.Z, we performed mononucleosome coimmunoprecipitation (co-IP), which confirmed their colocalization on nucleosomes (Fig. 4C). Western blot analysis showed that knockdown of either GABPα or H2A.Z did not affect the protein expression level of the other (Fig. 4D). Further ChIP-seq analysis revealed that 89% of GABPα binding sites in naïve CD8+ T cells overlapped with H2A.Z deposition sites, indicating extensive chromatin colocalization (Fig. 4, E and F). Beyond their physical colocalization, we investigated whether these two proteins influence each other’s deposition at OXPHOS gene loci. CUT&Tag analysis revealed that, upon H2A.Z deletion, GABPα enrichment at OXPHOS genes was even enhanced in naïve CD8+ T cells (Fig. 4G), while in GABPα-deficient naïve CD8+ T cells, H2A.Z enrichment at OXPHOS genes was significantly reduced (Fig. 4H). These findings were also recapitulated in Jurkat cell lines with GABPα or H2A.Z deficiency (Fig. 4, I and J). To further directly test the role of GABPα on H2A.Z recruitment on specific OXPHOS genes, overexpressed dCas9-GABPα was guided to specific genomic loci via single guide RNA (sgRNA) in GABPα-deficient Jurkat cells. The target genes, ATP5F1E, CYB5R3, and NDUFS8, were selected from the set of H2A.Z-regulated OXPHOS genes that are co-occupied by GABPα and H2A.Z and exhibit reduced H2A.Z deposition upon GABPα deletion. CUT&Tag–quantitative polymerase chain reaction (qPCR) was performed to determine the efficacy of GABPα-mediated recruitment of H2A.Z. The results showed that dCas9-GABPα induced significant deposition of H2A.Z on the promoter of ATP5F1E, CYB5R3, and NDUFS8 (Fig. 4K). Collectively, these data demonstrate that GABPα is required for the proper localization and deposition of H2A.Z at the promoter regions of OXPHOS genes in naïve T cells, suggesting a hierarchical relationship in which GABPα acts upstream of H2A.Z to facilitate epigenetic priming for metabolic gene activation. IL-7 tonic signaling instructs H2A.Z deposition during T cell maturation for future activation Our findings indicate that H2A.Z-mediated epigenetic priming of OXPHOS genes has been established at naïve T cell stage. This raises an important question exactly when this is established. To answer this question, we isolated immature SP CD8+ thymocytes, mature SP CD8+ thymocytes, and peripheral naïve CD8+ T cells and determined H2A.Z deposition. Gradual increase of H2A.Z deposition on OXPHOS genes was observed (Fig. 5A), suggesting progressive epigenetic maturation during T cell development and homeostasis. Given the critical role of IL-7 tonic signal in both T cell development and homeostasis (27–29), we wondered whether IL-7 tonic signal may induce this epigenetic maturation. To test this, we generated CD4–IL-7Rα cKO mice and determined H2A.Z deposition in naïve CD8+ T cells. Significant decrease of H2A.Z deposition on OXPHOS genes was found in the absence of IL-7Rα (Fig. 5B). This was also confirmed in IL-7Rα KO naïve CD8+ T cells on a Bcl2 transgenic (tg) background (Fig. 5C), excluding the potential impact from lymphopenia. IL-7 tonic signaling instructs H2A.Z deposition for CD8+ T cell activation. (A) Line and box plots show normalized H2A.Z signals across the genome in imSP8 (CD62L−CD69+CD8+ thymocytes), mSP8 (CD62L+CD69−CD8+ thymocytes), and naïve CD8+ (CD62L+CD44loCD69−CD8+ splenocytes) cells (n = 2 biologically independent samples). (B) Line and box plots show normalized H2A.Z signal levels of OXPHOS genes around TSSs (±1 kb). Blue, WT; light purple, IL-7Rα cKO. n = 2 biologically independent samples. (C) Same as (B) but with Bcl2 overexpression in all samples (n = 2 biologically independent samples). (D to G) Sorted naïve CD8+ T cells were stimulated with anti-CD3/CD28. [(D) and (E)] The percentages of CD69+ and CD25+ T cells were measured by flow cytometry at 24 hours. (F) After 48 hours, IFN-γ production was measured. (G) MFI of MitoRed in T cells was measured after 24 hours of stimulation (n = 3 biologically independent samples). (H) Heatmap depicting DEGs from the OXPHOS gene list. The comparison focuses on DEGs between Bcl2-tg WT and Bcl2-tg IL-7Rα cKO naïve CD8+ T cells before and after activation (n = 2 to 3 biologically independent samples). (I) GSEA plots showing enrichment of OXPHOS gene signatures in activated Bcl2-tg IL-7Rα cKO versus Bcl2-tg WT CD8+ T cells. (J and K) CD8+ T cells isolated from Bcl2-tg WT and Bcl2-tg IL-7 KO mice were transduced with H2A.Z-overexpressing lentivirus for 72 hours. Intracellular IFN-γ production was subsequently analyzed (n = 4 biologically independent samples). Data representative of two independent experiments. Data are means ± SEM. Statistical analysis was performed using two-sided unpaired Wilcoxon rank-sum test [(A) to (C)], two-tailed unpaired Student’s t test [(D) to (G)], and one-way ANOVA Tukey’s multiple comparisons test (K). T cell activation was further examined for naïve CD8+ T cells from WT and IL-7Rα cKO mice on Bcl2-tg background. The results showed significantly reduced expression of activation makers (CD69 and CD25), cytokine production (IFN-γ) and mitochondrial membrane potential upon TCR stimulation (Fig. 5, D to G). RNA-seq analysis also confirmed the impaired expression of OXPHOS genes in IL-7Rα KO T cells upon stimulation (Fig. 5, H and I). To further test whether H2A.Z may rescue the activation defect of naïve CD8+ T cells from mice without IL-7 tonic signal, CD8+ naïve T cells isolated from WT or IL-7 KO mice on Bcl2-tg background were preactivated and transduced to force H2AZ overexpression. Following this rescue, the IFN-γ secretion capacity of the KO T cells was markedly enhanced, reaching levels nearly comparable to those of WT T cells (Fig. 5, J and K). Together, these data suggest that IL-7 tonic signaling instruct the deposition of H2A.Z on OXPHOS genes for metabolic fitness required for T cell activation; forced H2A.Z expression may overcome the defect resulting from lack of IL-7 tonic signaling. Loss of IL-7 tonic signaling in mid-aged naïve CD8+ T cells leads to reduced H2A.Z deposition and function It is well known that during ageing, CD8+ T cell responses to newly encountered pathogens and malignant cells decrease. This is considered mainly due to the reduced number of naïve CD8+ T cells, while the functional quality of naïve T cells remains less defined. A previous study reported decreased chromatin accessibility of IL-7Rα in aged memory CD8+ T cells (30). We wondered whether the expression of IL-7Rα or IL-7 signaling is also reduced in aged naïve CD8+ T cells, which might lead to reduced H2A.Z deposition. To interrogate this, we first analyzed published single-cell RNA sequencing (scRNA-seq) expression profiles from the PRJNA1181768 dataset to characterize the age-related trend of IL-7Rα expression in naïve CD8+ T cells (31). Specifically, we extracted the gene expression matrix encompassing different age groups (10 to 98 weeks) and focused on naïve CD8+ T cells to investigate IL-7Rα expression patterns. We found that IL-7Rα expression decreased progressively with ageing (Fig. 6A). When assessing IL-7Rα expression levels in mid-aged (40-week-old) naïve CD8+ T cells by flow cytometry, we observed a significant reduction compared to young (8-week-old) naïve CD8+ T cells (Fig. 6B). Meanwhile, IL-7 protein levels in peripheral lymph nodes (LNs) as determined by enzyme-linked immunosorbent assay (ELISA) were gradually reduced during mouse ageing, with a notable decline as early as 40 weeks of age (Fig. 6C). Consistent with the previous data, mid-aged naïve CD8+ T cells had significantly reduced H2A.Z deposition on OXPHOS genes compared to young T cells (Fig. 6D). In vitro stimulation assay also confirmed the activation defect for naïve CD8+ T cells from mid-aged mice, as determined by Ifng and Il2 mRNA expression levels (Fig. 6, E and F). Furthermore, the LM-OVA infection model and MC38-OVA tumor model were used to assess the function of mid-aged naïve CD8+ T cells in vivo. Mid-aged mice showed defective antigen-specific CD8+ T cell response upon LM-OVA infection compared to young mice (Fig. 6G). This defect is T cell intrinsic as confirmed using the naïve T cell cotransfer-infection model (Fig. 6, H and I). To assess the cell-intrinsic antitumor function of T cells, naïve OT-I CD8+ T cells isolated from young and mid-aged mice were adoptively transferred into recipient mice 1 day prior to subcutaneous MC38-OVA tumor inoculation. Naïve CD8+ T cells isolated from mid-aged mice showed significantly less antitumor effect than those from young mice (Fig. 6J). The percentages of IFN-γ+ OT-I T cells were also significantly reduced in mid-aged CD8+ donor T cells (Fig. 6K). These data indicated that during ageing, reduced tonic IL-7 signaling in naïve CD8+ T cells could lead to impaired H2A.Z deposition resulting in defective CD8+ T cell function. Loss of IL-7 tonic signaling leads to reduced H2A.Z deposition and T cell dysfunction. (A) IL-7Rα mRNA expression during ageing in naïve T cells. (B) IL-7Rα protein expression on murine naïve T cells (n = 3 mice per group). (C) IL-7 protein level in murine peripheral LNs (n = 3 mice per group). (D) H2A.Z CUT&Tag analysis of naïve CD8+ T cells (n = 2 biologically independent samples). (E and F) Ifng and Il2 mRNA expression in activated CD8+ T cells (n = 3 to 6 biologically independent samples). (G) IFN-γ production by splenic CD8+ T cells on day 7 after LM-OVA infection (n = 7 biologically independent samples). (H and I) IFN-γ production in splenic CD8+ donor T cells from 8- and 40-week mice on day 7 after LM-OVA infection (n = 5 biologically independent samples). (J) MC38-OVA tumors with or without naïve OT-I cell transfer (n = 9 to 10 mice per group). (K) IFN-γ production in tumor-infiltrating OT-I cells (n = 3 to 4 biologically independent samples). (L and M) Naïve CD8+ T cells from IL-7–pretreated mice were stimulated with anti-CD3/CD28. IFN-γ expression was detected at 48 hours (L), and MitoRed MFI was measured at 24 hours (M) (n = 4 biologically independent samples). (N) H2A.Z CUT&Tag analysis of naïve CD8+ T cells from IL-7–pretreated mice (n = 2 biologically independent samples). (O) Naïve CD8+ T cells were activated and transduced with H2A.Z-overexpressing lentivirus for 72 hours. Intracellular IFN-γ production was subsequently analyzed (n = 4 biologically independent samples). Data representative of two independent experiments. Data are means ± SEM. Statistical analysis was performed using Jonckheere-Terpstra trend test (A), two-tailed unpaired Student’s t test [(B), (E) to (I), and (K)], one-way ANOVA with Tukey’s multiple comparisons test [(C) and (L) to (O)], two-way ANOVA with Tukey’s multiple comparisons test (J), or two-sided unpaired Wilcoxon rank-sum test [(D) and (N)]. wk, weeks. To further investigate the effects of IL-7 and H2A.Z on the activation of mid-aged T cells, we treated mid-aged mice with IL-7 and assessed the activation of CD8+ T cells in vitro. IL-7 pretreatment (administered twice) rescued the impaired OXPHOS function and restored T cell activation (Fig. 6, L and M). Moreover, IL-7 administration ameliorated the impaired H2A.Z chromatin deposition (Fig. 6N). Furthermore, overexpression of H2A.Z via viral transduction markedly enhanced IFN-γ secretion capacity in mid-aged T cells (Fig. 6O). Collectively, these findings demonstrate that IL-7 supplementation or H2A.Z overexpression may restore impaired CD8+ T cell function in aged individuals. Forced expression of H2A.Z enhances CAR T cell antitumor efficacy Defective T cell activation is a typical feature of exhausted T cells in tumor microenvironment. The impaired metabolic fitness and function of H2A.Z-deficient CD8+ T cells prompt us to wonder its function in T cell exhaustion and immunotherapy. We first compared the expression level of H2A.Z in CD8+ T cells from the spleen, tumor-draining lymph nodes (TDLNs), and tumor-infiltrating lymphocytes (TILs). H2A.Z expression level was significantly reduced in the CD8+ TILs from tumors than that in the CD8+ T cells from the spleen and TDLNs (Fig. 7A). Further experiments revealed that H2A.Z expression gradually decreased in CD8+ T cells as they progressed toward exhaustion, with the PD-1+ TIM-3+ subset exhibiting lower levels of H2A.Z than the PD-1− TIM-3− subset in tumors (Fig. 7B), indicating a role of H2A.Z for inhibiting T cell exhaustion. To further apply this in tumor immunotherapy, we generated H2A.Z1, the major isoform of H2A.Z, overexpressing CLDN18.2-specific CAR T cells, and tested its function in MC38-CLDN18.2 tumor model (Fig. 7C). Significantly enhanced tumor-inhibition efficacy was found for H2A.Z1-overexpressing CAR T cells compared to control CAR T cells in the MC38-CLDN18.2 tumor model (Fig. 7D). This enhanced antitumor effect was consistently observed in the KPC tumor model as well (Fig. 7E). To more specifically examine the effect of H2A.Z1 overexpression on CAR T cells, control and H2A.Z1-overexpressing CLDN18.2-specific CAR T cells were compared. H2A.Z1 overexpression significantly enhanced OXPHOS in CAR T cells, as evidenced by increased mitochondrial membrane potential (Fig. 7F). Moreover, flow cytometry analysis of PD-1 and TIM-3 expressions on TILs demonstrated an increased percentage of PD-1− TIM-3− subset of CAR T cells in the H2A.Z1-overexpression group (Fig. 7G). Additionally, H2A.Z1-overexpressing CAR T cells isolated from both the tumor microenvironment and TDLNs exhibited elevated IFN-γ production (Fig. 7H). Thus, these data suggest that the diminished response of exhausted T cells may be, at least in part, attributable to impaired H2A.Z-mediated chromatin remodeling; overexpression of H2A.Z can reverse T cell exhaustion within the tumor microenvironment and promote the metabolic activity and cytokine secretion of CAR T cells, thereby enhancing the therapeutic efficacy of CAR T cells. H2A.Z overexpression in CAR T cells improves CAR T cell effector function. (A) MFI of H2A.Z in CD8+ T cells from spleen, TDLN, and MC38 tumors (n = 5 to 6 mice per group). (B) MFI of H2A.Z in PD-1− TIM-3− and PD-1+ TIM-3+ CD8+ TILs (n = 5 mice per group). (C) Experimental scheme of in vivo antitumor experiment under lymphodepletion conditions. Created in PowerPoint by M.L. (D) MC38-CLDN18.2 tumor growth following treatment with CLDN18.2 CAR T cells, CLDN18.2-H2A.Z1 CAR T cells, or control T cells (n = 6 mice per group). (E) KPC tumor growth following treatment with CLDN18.2 CAR T cells, CLDN18.2-H2A.Z1 CAR T cells, or control T cells (n = 6 mice per group). (F) MitoRed MFI was determined by flow cytometry on CLDN18.2 CAR T or CLDN18.2-H2A.Z1 CAR T cells (n = 3 to 4 biologically independent samples). (G and H) Expression levels of PD-1 and TIM-3 (G), and IFN-γ (H) were measured on CLDN18.2 CAR T or CLDN18.2-H2A.Z1 CAR T cells isolated from tumors or TDLNs (n = 4 to 5 mice per group). Data representative of two independent experiments. Data are means ± SEM. Statistical analysis was performed using one-way ANOVA with Tukey’s multiple comparisons test (A), two-tailed unpaired Student’s t test [(B) and (F) to (H)], or two-way ANOVA with Tukey’s multiple comparisons test [(D) and (E)]. d, day; ip, intraperitoneal; iv, intravenous. DISCUSSION Naïve T cells are maintained at a quiescent state before activation. While T cell activation has been extensively studied, whether and how the quality of quiescence influences subsequent activation and immune function is poorly understood. Here, we reveal that an IL-7R–GABPα–H2A.Z–OXPHOS axis acts as an epigenetic training program during T cell quiescence, priming cells for rapid effector function. Disruption of this axis during ageing abrogates such cellular preparedness and impairs immune responses, whereas supplementation with IL-7 or forced expression of H2A.Z rescues the defect. Furthermore, H2A.Z overexpression also inhibits CAR T cell exhaustion and enhances their antitumor efficacy. Our findings establish that epigenetic training during quiescence is indispensable for T cell immunity. We term this T cell trained immunity. A previously unrecognized instructing function of IL-7 was identified in this study. IL-7, mainly produced by lymphoid stromal cells, has been well documented for its essential role on T cell survival, proliferation, and metabolic fitness during thymocyte development and peripheral naïve T cell homeostasis by enhancing glucose and amino acid catabolism through AKT-mediated up-regulation of GLUT1 and support of OXPHOS (1, 32–34). Key regulators of IL-7R or downstream signaling pathway such as FOXO1, FOXP1, and BTG1/BTG2 and N6-methyladenosine (m6A) mRNA methylation also help sustain this quiescent state (35–39). This study identifies a previously unrecognized role for IL-7 tonic signaling in epigenetically training T cells for future activation. This is consistent with previous observation that naïve T cells from IL-7R–deficient mice were significantly less responsive than those from WT mice (28, 40, 41). Mechanistically, we revealed that IL-7 governs the epigenetic deposition of the H2A.Z onto OXPHOS genes, poising them for rapid transcription upon activation. This H2A.Z deposition increases progressively during thymocyte maturation and naïve T cell homeostasis, suggesting a qualitative “education” process that establishes T cell functional fitness. This represents a crucial addition to the known roles of lymphoid tissues, moving beyond clonal selection and numerical maintenance to include the epigenetic programming of metabolic readiness. Our findings provide a mechanistic explanation for the observed functional maturation of T cells during development and post–thymic homeostasis (42, 43). Ageing in humans is associated with decreased IL-7 levels in thymus and periphery (44, 45). While this decline is frequently considered a major cause of reduced naïve T cell numbers, its direct impact on T cell functional competence per se has been largely overlooked. In this study, we demonstrate that diminished IL-7 tonic signaling during ageing is associated with reduced H2A.Z occupancy on OXPHOS genes and consequent impairment of naïve T cell activation. More importantly, IL-7 supplementation enhanced H2A.Z deposition on OXPHOS genes and rescued the defective response of aged T cells, which may provide an alternative explanation for recent work on immune rejuvenation in the aged (46). Thus, our findings identify a reduced IL-7–H2A.Z–mediated epigenetic-metabolic program as a previously unrecognized hallmark of T cell functional ageing. This mechanism may offer a unifying explanation for the differential T cell responses observed among individuals across various physiological and pathological conditions. GABPα, an Ets transcription factor, was found in our study to be required for OXPHOS gene-specific H2A.Z deposition. First, GABPα deficiency results in a significant reduction of H2A.Z deposition on OXPHOS genes. Second, gRNA-mediated dCas9-GABPα targeting enhances H2A.Z deposition at specific gene loci. In addition, GABPα-deficient T cells and H2A.Z-deficient T cells show similar homeostasis and activation phenotypes at the cellular level, as well as largely overlapping OXPHOS gene expression impairments during activation. These data suggest that GABPα works upstream of H2A.Z for epigenetic programming. GABPα plays important roles on lymphocyte development and function (26, 47–49). The role of GABPα in T cells was first established by Xue et al. (47), who identified it as essential for regulating IL-7Rα expression in double-negative thymocytes. However, in peripheral naïve T cells, GABPα deficiency does not appear to influence IL-7Rα expression (26), suggesting that its role in regulating naïve T cell fitness is independent of IL-7R signaling. Furthermore, using CD4-Cre cKO models, Luo et al. have demonstrated that GABPα is indispensable for peripheral T cell homeostasis and effector responses, despite being dispensable for thymocyte development (26). This phenotypic distinction is consistent with our current findings. Furthermore, transcriptional profiling of GABPα-deficient T cells revealed a notable enrichment of DEGs involved in mitochondrial biogenesis and function, a result that corroborates our own data. Our finding that GABPα is required for the targeted deposition of H2A.Z onto OXPHOS genes unveils a previously unrecognized epigenetic mechanism through which GABPα regulates transcriptional poising and governs T cell metabolic fitness. The critical role of the H2A.Z-mediated epigenetic-metabolic axis in governing T cell fitness prompted us to investigate its potential in T cell immunotherapy. In settings such as the tumor microenvironment or chronic infections, persistent antigen exposure drives T cell exhaustion—a dysfunctional state marked by impaired effector function and the sustained expression of inhibitory receptors like PD-1 (50–52). This state is increasingly understood to be stabilized by epigenetic mechanisms (51, 52). Notably, we found that CD8+ TILs exhibit significantly reduced H2A.Z expression compared to effector/memory CD8+ T cells from non–tumor-bearing mice, implicating H2A.Z loss in the exhaustion process. Notably, forced expression of H2A.Z in CAR T cells inhibits their exhaustion and resulted in superior antitumor efficacy correlated with increased OXPHOS and IFN-γ production in the tumor microenvironment, underscoring its therapeutic potential. H2A.Z seems to have a different effect on naïve CD4+ T cells than on CD8+ T cells. First, the percentage of Ki-67+ cells is reduced in H2A.Z-deficient naïve CD8+ T cells, but not in their CD4+ counterparts. While this reduction suggests a role for H2A.Z in homeostatic proliferation of naïve CD8+ T cells, the normal Ki-67 levels in naïve CD4+ T cells do not necessarily rule out a role for H2A.Z, as Ki-67 positivity simply indicates cell cycle entry (G1/S/G2/M) rather than cycling speed. The exact function of H2A.Z in naïve CD4+ T cell proliferation thus warrants further investigation. Second, while H2A.Z is also required for CD4+ naïve T cell activation, this requirement may not depend on OXPHOS priming. No significant defect in OXPHOS gene expression was observed in activated H2A.Z-deficient CD4+ T cells, correlating with comparable mitochondrial respiration as measured by Seahorse. How H2A.Z regulates CD4+ T cell activation and why its effects are specific to different T cell subsets remain unclear from the current study and need to be investigated in the future. In conclusion, our study reveals that the IL-7 signaling axis, via GABPα, orchestrates the deposition of the histone variant H2A.Z to establish a permissive chromatin architecture in naïve CD8+ T cells. This epigenetic training program is fundamental to their quiescent fitness and defines their subsequent activation potential. We further demonstrate that the decline of this pathway is linked to T cell functional impairment and may underpin immune senescence. Therapeutically, enforced H2A.Z expression enhances the antitumor efficacy of CAR T cells. Our work thus uncovers a critical epigenetic training mechanism that governs T cell responsiveness, identifying the targeting of the IL-7–H2A.Z regulatory axis as a promising therapeutic strategy to enhance T cell–based immunotherapies for cancer, chronic infections, and age-related immune decline. MATERIALS AND METHODS Mice H2A.Zf/f mice (RIKEN Center, RBRC05765) were obtained from G. Li from the Institute of Biophysics, Chinese Academy of Sciences. Cd4-Cre (022071), Ly5.1(002014), and Thy1.1(000983) transgenic mice were obtained from the Jackson Laboratory. Gabpa-flox (T008829), Il7ra-flox (T005464), and Il7 KO (T052425) mice were purchased from GemPharmatech (Nanjing, China). Bcl2 transgenic mice (037905) were obtained from L. Lu from Zhejiang University. Mice used in the experiments were 6 to 8 weeks old and sex-matched in each experiment. Bone marrow chimeras were generated by adoptively transferring single WT or H2A.Z cKO bone marrows or mixed WT (Thy1.1) and H2A.Z cKO (Thy1.2) bone marrows (1:1 mixed, 5 × 106 cells per mouse) intravenously into lethally irradiated Ly5.1 mice. Eight to 10 weeks after transplantation, the bone marrow chimeric mice were used. All mice are on the C57BL/6 background and were maintained under specific pathogen-free conditions with approval by the institutional committee of Institute of Biophysics, Chinese Academy of Sciences. All animal experimental procedures were performed with approval (SYXK2020035, ABSL-2-2021006) from the institutional committee of the Institute of Biophysics, Chinese Academy of Sciences. Cell lines Jurkat cells [American Type Culture Collection (ATCC)] were maintained in RPMI-1640 medium (Gibco) containing 10% fetal bovine serum (FBS; BI), 10 mM Hepes (Gibco), and 1% penicillin-streptomycin (Gibco). Phoenix-ECO cells (ATCC) were maintained in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% FBS, 10 mM Hepes, and 1% penicillin-streptomycin. All these cells were grown at 37°C in humidified atmosphere with 5% CO2. Flow cytometry analysis and antibodies Fluorescence labeled antibodies for CD4 (RM4-5), CD8 (53-6.7), CD45.1 (A20), CD45.2 (104), CD90.1 (HIS51), CD90.2 (53-2.1), CD3 (2C11), TCRβ (H57-597), CD69 (H1.2F3), CD25 (PC61), and their isotype controls were purchased from eBioscience; antibodies for IFN-γ (XMG1.2), CD62L (MEL-14), and CD44 (IM7) were purchased from BioLegend. For cell surface staining, 2 × 106 freshly isolated cells were incubated with 2.4G2 for 10 min at 4°C and then stained with a mixture of cell surface antibodies in fluorescence-activated cell sorting (FACS) buffer [phosphate-buffered saline (PBS) containing 2% FBS] for 30 min at 4°C. For the measurement of mitoROS and mitochondrial membrane potential, activated T cells were incubated with 2 μM MitoSOX (Thermo Fisher Scientific, M36008), 20 nM MitoTracker Deep Red (YEASEN, 40743ES50), and 20 nM MitoTracker Green (YEASEN, 40742ES50) for 15 min at 37°C in culture medium. Samples were collected on BD LSRFortessa and analyzed using FlowJo software (Tree Star Inc). Antibodies used for FACS are listed in table S1. Western blot Sorted T cells were lysed with nuclei lysis buffer (1% SDS, 10 mM Tris-HCl, pH 8.0, and 50 mM EDTA) and sonicated until the buffer is clear. Cell lysis (10 μg) was resolved by 12% SDS–polyacrylamide gel electrophoresis and blotted to polyvinylidene difluoride membrane (Millipore, IPVH00010). Antibodies for H2A.Z (Abcam, ab4174), H4 (Millipore, 05858), β-actin (Santa Cruz, sc-47778), and GABPα (Santa Cruz, sc-28312) were used. In vitro T cell activation Naïve CD8+ T cells (CD8+CD25−CD62L+CD44lo) and CD4+ T cells (CD4+CD25−CD62L+CD44lo) were isolated by FACS using a BD FACSAria II system (purity > 95%). Sorted naïve T cells were cultured in RPMI-1640 medium supplemented with 10% FBS, l-glutamine, penicillin-streptomycin, and 2-mercaptoethanol. For activation, 2 × 106 naïve T cells were seeded per well of a 24-well plate and stimulated with 2 μg/ml plate-bound anti-CD3 (BioLegend, 100340) and 2 μg/ml soluble anti-CD28 (BioLegend, 102116) for the indicated durations. qPCR For reverse transcription of RNA samples, Thermo Fisher Scientific Reverse Transcription Kit (catalog no. EN032) was used. The reverse transcription PCR reagent used was SYBR Green Mix from Takara Company, and a 96-well plate reaction was conducted using the ABI QuantStudio 6. Relative mRNA expression was calculated with the 2−ΔΔCT method. The internal reference gene selected is the housekeeping gene Actb (β-actin). ELISA Mouse periphery LNs were harvested and mechanically homogenized to obtain supernatants. The protein level of IL-7 in the supernatants was measured by ELISA Kit (Proteintech) according to kit instructions. L. monocytogenes infection Recombinant L. monocytogenes expressing truncated OVA protein (LM-OVA) was provided by H. Xu (National Vaccine and Serum Institute, Beijing, China). WT and H2A.Z cKO mixed bone marrow chimeric mice were infected intraperitoneally with LM-OVA (1 × 105 colony-forming units per mouse) and sacrificed on day 7 after infection. Splenocytes were restimulated with MHC-II restricted LLO190–201 peptide (NEKYAQAYPNVS, 10 μg/ml) for CD4+ T cell or MHC-I restricted OVA257–264 peptide (SIINFEKL, 10 μg/ml) for CD8+ T cell in the presence of brefeldin A solution (BioLegend, 420601) for 6 hours. The percentage of IFN-γ+ antigen specific T cell was analyzed. Tumor inoculation MC38-OVA, MC38-CLDN18.2, B16-OVA, and KPC tumor cells were cultured in DMEM supplement with 10% FBS, Hepes, and penicillin-streptomycin. Tumor cells (5 × 105) were injected subcutaneously at the right flank of mice, and the tumor growth was monitored. Tumor tissues were collected at indicated time points, and tumor infiltration T cells were isolated by collagenase VI (1 mg/ml; Sigma-Aldrich, C0130-5G) and deoxyribonuclease I (66.7 μg/ml; Roche, 11284932001) digestion in 2% FBS RPMI-1640 for 20 min at 37°C. The digested tumor cells were stained with indicated antibodies for flow cytometry analysis. RNA-seq analysis Total RNA was extracted from 1 × 106 naïve or activated T cells using TRIzol reagent (Invitrogen, 15596018). RNA-seq libraries were constructed with the NEBNext Ultra II RNA Library Prep Kit for Illumina (NEB, E7770L) and sequenced on an Illumina HiSeq 2000 platform. Raw reads were filtered, and adaptor sequences were cut using FASTX-Tools (http://hannonlab.cshl.edu/fastx_toolkit/) with the “Q30” threshold. Then, clean reads were first mapped to fruit fly genome using TopHat (v2.0.0) with default parameters (53). HTSeq tool (v0.11.1) was used to compute RNA fragment counts for all mouse genes with default parameters (54). Those genes with |log2(fold change)| ≥ 1 and FDR ≤ 0.05 were selected as DEGs. GSEA and GO enrichment analysis were done using clusterProfiler (v3.4, FDR ≤ 0.05) (55). Metabolism assays The OCRs and extracellular acidification rates of activated T cells were measured in XF media (nonbuffered RPMI-1640 containing 10 mM glucose, 2 mM l-glutamine, and 1 mM sodium pyruvate) under basal conditions and in response to 2 μM oligomycin (Abcam, ab141829), 1 μM fluorocarbonyl cyanide phenylhydrazone (Sigma-Aldrich, C2920), 100 nM rotenone (Sigma-Aldrich, R8875), and 1 μM antimycin A (Abcam, ab141904), with the XF-96 Extracellular Flux Analyzer (Seahorse Bioscience). CRISPR-Cas9 gene editing The CRISPR-Cas9 vector was purchased from Addgene (Addgene, 82416). Oligo sgRNA were annealed and constructed into Bsmb I site. The sgRNA sequences were used as following: NDUFC1: CGCCGAATGCCAAACCTGACTGG; NDUFA8: AGATCCGAGGCGGTGTTTAGAGG; ATP5E: GCAAATGCTGAGAAGACTTCTGG; GABPα: AACTGTACAGCTTAGTGTAC; and H2AFZ: GGCAAGCCGGCTCTCTGCGAGCGG. Jurkat cells were transfected with the sgRNA-Cas9-EGFP plasmid (250 V, 950 μF), and GFP+ single cells were sorted 48 hours after transfection. Knocked out clones were selected by sequencing the mutation site. For the activation of Jurkat cells, anti-human CD3 (10 μg/ml) and PMA (20 ng/ml) were added to the culture medium. Cells were harvested 24 hours after stimulation, and the expression of IL2 was measured by qPCR. Generation of dCas9-GABPα fusion construct To generate a dCas9-hGABPα fusion construct, human GABPα amplified by PCR from cDNA of human peripheral blood mononuclear cells (Milestone Biotechnologies, PB050C) was inserted at the C terminus of the dCas9 gene, by a short Gly4Ser linker. The sequence of dCas9 gene is contributed by plasmid SP-dCas9-VPR (D10A, H840A mutations in SpCas9). The sequence was integrated onto the lentiCRISPRv2 plasmid vector through homologous recombination, and under the control of the EF1α promoter. Oligo sgRNA designed to bind promoter regions of target genes (e.g., NDUFS8: AGTAACGTGTTACCACGTT; ATP5F1E: GGAAGATCAGACCACAGAAG; and CYB5R3: TAAGAGGCGGAATCTTCCCC) were annealed and constructed into the Bsmb I site. The final sgRNA-dCas9-hGABPα-EGFP plasmid was electroporated into GABPα KO Jurkat cells. After 72 hours of electroporation, GFP+ cells were sorted and collected for CUT&Tag-qPCR analysis. CUT&Tag and computational analysis CUT&Tag assay was performed using the Hyperactive Universal CUT&Tag Assay Kit for Illumina Pro (Vazyme, catalog no. TD904-02) following the manufacturer’s instructions. As previously described (19), a total of 1 × 105 freshly sorted naïve or activated CD8+ T cells were tagged with concanavalin A–coated magnetic beads and then incubated with 1 μg of primary antibody [rabbit polyclonal anti-histone H2A.Z, anti-H3K4me3, anti-H3K27me3, anti-H3K27ac, and anti-RNA polymerase II CTD repeat YSPTSPS (phospho S5)] overnight at 4°C with slow rotation. After incubation with 1 μg of secondary antibody [goat anti-rabbit immunoglobulin G (IgG) H&L] at room temperature for 1 hour, cells were incubated with 0.04 μM hyperactive pG-Tn5 transposon and then resuspended in tagmentation buffer for tagmentation. DNA was purified and amplified by 12 to 14 PCR cycles to generate the libraries. Libraries were sequenced on Illumina Nova-PE150, and 150-bp paired-end reads were generated. For CUT&Tag qPCR detection, 5 μl of stop buffer was added to extract DNA after tagmentation. Using the negative control (IgG) as a control, the 2−ΔΔCT values were calculated for each group to determine the relative enrichment fold: relative enrichment fold (treatment) = 2 − [ΔCT treatment (CT treatment – CT treatment DNA spike-in) − ΔCT IgG (CT IgG – CT IgG DNA spike-in)]; relative enrichment fold (control) = 2 – [ΔCT control (CT control – CT control DNA spike-in) – ΔCT IgG (CT IgG – CT IgG DNA spike-in)]. Antibodies used for CUT&Tag are listed in table S2. For data processing, raw sequencing data were trimmed and filtered using Trim Galore. Bowtie2 (v2.5.1) was used to align reads to the mouse reference genome mm10 with default parameters and then sorted using Samtools (v1.15). MACS2 (v2.2.7.1) was used to call peaks. bamCoverage (v3.5.1) was used to generate bigwig files from BAM files. The distribution of CUT&Tag peaks was annotated with the R package ChIPseeker. Native ChIP-seq and computational analysis ChIP refers to the ULI-NChIP method published by Brind’Amour et al. (56). The method used here follows our earlier report (19). Native ChIP-seq was performed to conduct a detailed analysis of H2A.Z deposition in naïve and activated CD8+ T cells (24 hours post–anti-CD3/CD28 stimulation). Antibodies used for native ChIP are listed in table S2. For ChIP-seq, Trim Galore software was used to perform quality control on the raw data, and sequencing fragments less than 20 bp in length was filtered out. The genome alignment software used is Bowtie2. After alignment, MarkDuplicates in Picard is used to remove redundant alignment results, followed by the use of Samtools v1.10 (57) to eliminate reads with MAPQ < 10. Read visualization is carried out using IGV software. It is necessary to use the bamCompare function to generate bigwig format input files, with RPKM selected as the standardization method. The Peak calculation software used is MACS2, with the threshold set at q < 0.05. The count number calculation of the Peak is implemented using the featureCounts function of the Subread (http://subread.sourceforge.net) software. During the calculation process, the Drosophila dm6 version is selected as the spike-in genome to calculate the scale factor between different mouse samples. Peak annotation is completed using the ChIPseeker package (58). The bigwig file was used as input, the computeMatrix function of deepTools software (59) was used to calculate the density value of the transcript in each region, and the ggplot2 package was used for drawing display. ATAC-seq and computational analysis Naïve and activated (8 × 104; 6 hours post–anti-CD3/CD28 stimulation) CD8+ T cells isolated from WT or cKO mice were subjected to ATAC-seq (assay for transposase-accessible chromatin using high-throughput sequencing) to assess chromatin accessibility. ATAC-seq was performed as described previously (19). The TruePrep DNA Library Prep Kit V2 for Illumina (Vazyme, catalog no. TD501-01) was used according to the manufacturer’s instructions to prepare the sequencing libraries. The Trim Galore software was used for quality control of ATAC-seq raw data, filtering out sequencing fragments less than 20 bp in length. The genome comparison software used is Bowtie2 v2.5.1 (60). After comparison, MarkDuplicates in Picard was used to remove redundant comparison results. For quality control of BAM files after comparison, refer to the official documentation of ENCODE (www.encodeproject.org/). Before peak analysis, the BAM file needs to be shifted according to the occupancy of the Tn5 enzyme. This operation is completed by alignmentSieve in deepTools v3.4.3 (59). Read visualization is completed using IGV software (61). The bamCoverage function needs to be used to generate the bigwig format input file, and RPKM is selected as the standardization method. The Peak calculation software used is MACS2 v2.2.7.1 (62), with the threshold set at q < 0.05. Peak count calculation is implemented using the featureCounts function of Subread v1.6.2 (63) software. Peak annotation was completed using the ChIPseeker package. scRNA-seq data processing and analysis scRNA-seq data were sourced from the NCBI BioProject under accession number PRJNA1181768 and, from across eight ages (from 10 to 98 weeks), were filtered using Seurat package (v5.3.1). Cells were filtered and processed using the following thresholds: (i) minimum 200 and maximum 6000 unique genes detected per cell; (ii) minimum 1000 and maximum 40,000 UMI counts per cell; (iii) mitochondrial gene content < 10%; and (iv) minimum three cells expressing each gene. After normalization and principal components analysis clustering with the Leiden algorithm, batch correction was performed using Harmony. To isolate naïve T cells, we applied a gene expression–based filtering strategy on the integrated Seurat object. Cells were retained if they expressed canonical T cell markers (Cd4, Cd8a, or Cd8b1) and showed no detectable expression of Cd44 (defined as zero counts in the normalized feature matrix, reflecting low activation status). In addition, cells were required to coexpress a panel of genes associated with naïve phenotype and LN homing, including Dapl1, Igfbp4, Lef1, Ccr7, Vps37b, and Dusp10. This stringent criterion was used to enrich for quiescent, antigen-inexperienced T cells. Sample identifiers were parsed from cell barcodes, and the numeric week value was extracted and converted into an ordered factor to preserve temporal sequence. For visualization, normalized expression levels of Il7ra were plotted. Temporal trends from 10 to 98 weeks were visualized using line plots showing mean expression per age group. Mononucleosome IP assay and SILAC Mononucleosome IP assay was performed as previously described with some modifications (64). Cells were cross-linked with 1% formaldehyde for 10 min at room temperature. The fixation was quenched by adding glycine to a final concentration of 125 mM for 5 min. After that, nuclei were washed by PBS and resuspended in cold lysis buffer (50 mM Hepes, pH 7.5, 140 mM NaCl, 10% glycerol, 0.5% Nonidet P-40, 0.25% Triton X-100, and 1× protease inhibitors) at a density of 1.5 × 107 cells/ml for 10 min at 4°C. The nuclei were washed with wash buffer I (10 mM tris, pH 8.0, 200 mM NaCl, and 1× protease inhibitors) and resuspended in MNase buffer II (10 mM tris, pH 7.4, 15 mM KCl, 60 mM NaCl, and 1 mM CaCl2). Then, the nuclei were digested with 1.2 U/ml MNase (Sigma-Aldrich, N3755) for 30 min at 37°C to generate mononucleosomes, followed by sonication in buffer III (10 mM tris, pH 8.0, 1 mM EDTA, 0.5 mM EGTA, 0.5% N-lauroyl-sarcosine, and 1× protease inhibitors) to release the mononucleosomes. The soluble mononucleosomes were enriched with anti-Flag resins (Sigma-Aldrich, A2220). For mass spectrum, the Flag-IP sample was eluted directly with 2× SDS loading buffer containing 100 mM dithiothreitol followed by incubation at 100°C for 12 min. The sample was centrifuged, vortexed, and then incubated at 100°C for another 12 min. Then, the samples were resolved on Novex NuPAGE 12% Bis-Tris gel (Life Technologies). The gel bands for samples were excised according to the prestained protein markers and subjected to in-gel digestion. The peptides were analyzed using tandem MS. For SILAC labeling, Jurkat-H2A.Z cells were grown for three passages (∼8 doublings) in Arg/Lys-depleted RPMP-1640 medium supplemented with 10% (v/v) dialyzed FBS and heavy isotope-labeled amino acids: l-lysine-13C6,15N2 hydrochloride (Sigma-Aldrich, 608041-1G) and l-arginine-13C6,15N4 hydrochloride (Sigma-Aldrich, 608033-500MG). Jurkat-H2A cells were cultured in a medium with opposite light amino acid (12C6-Arg/Lys) additions. When Jurkat-H2A.Z/H2A cells grew to 80 to 90% confluency, cells were harvested with cell scrapers, washed with cold PBS for three times, and pelleted by centrifugation at 500g for 5 min at 4°C. The collected cells carried out the above operation for mononucleosome IP assay. After exchanging the heavy/light amino acids culture media for Jurkat-H2A.Z/H2A, cells were collected and analyzed as the biological replicate group samples. H2A.Z and IL-7 rescue T cells were isolated from the spleens of Bcl2tgIL-7KO or aged mice and corresponding control mice using a mouse T cell isolation kit. Purified T lymphocytes were activated in vitro with anti-CD3 and anti-CD28 antibodies (5 μg/ml each) for 24 hours. Cells were then transduced via spin infection (800g, 32°C, 2 hours) in the presence of polybrene (10 μg/ml), using retroviruses encoding H2AZ overexpression or empty control vectors. Posttransduction, cells were maintained in T cell culture medium supplemented with IL-2 (10 ng/ml) for 4 days, after which IFN-γ expression and other functional parameters were assessed. Eight-week-old and 40-week-old mice were administered two intraperitoneal injections of 20 μg of IL-7 on day 0 and day 2, respectively. Splenic CD8+ naïve T cells were harvested on day 7, and 2 × 105 CD8+ naïve T cells were stimulated with anti-CD3/CD28 for 24 or 48 hours. The MFI of MitoRed was assessed at 24 hours after stimulation, and the IFN-γ expression was measured at 48 hours after stimulation via flow cytometry. CAR structure design The second generation of CLDN18.2-CAR was constructed using an anti-CLDN18.2 single-chain variable fragment (hu8E5-2I), followed by the murine CD8α hinge and transmembrane region, the murine 4-1BB costimulatory domain, and the CD3ζ intracellular region, which was then inserted into the pMIGR1-IRES-EGFP retroviral vector (Addgene, 27490). Full-length mouse H2A.Z1 was amplified by PCR from cDNA. CLDN18.2-H2A.Z1-CAR construct was generated by fusing P2A self-cleaving peptide and the H2A.Z1 gene fragments into the CLDN18.2-CAR containing viral vector. Mouse T cell isolation and CAR T cell production Mouse T cells were isolated from the spleen of C57BL/6 mice using a mouse T cell isolation kit (STEMCELL Technologies, 19851) according to the manufacturer’s instructions. Purified mouse T cells were activated in vitro for 24 hours with anti-CD3 (10 μg/ml) and anti-CD28 (5 μg/ml) before retrovirus transduction. Retrovirus was produced by cotransfecting Phoenix-ECO cells (ATCC) with the pMIGR1-overexpressing plasmid and the helper plasmid pCL-Eco (Addgene, 12371) and was collected at 48 and 72 hours after transfection. Viral transduction was performed by spin-infection at 800g and 32°C for 2 hours with polybrene (10 μg/ml; Sigma-Aldrich, H9268). After transduction, cells were cultured in T cell medium with human IL-2 (10 ng/ml; PeproTech, 200-02-10) for 4 days. EGFP+ CAR T cells were sorted using a cell sorter (BDAria Fusion) before adoptive transfer into recipients. Adoptive T cell transfer for tumor therapy MC38-CLDN18.2 cells (5 × 105) or KPC cells (2 × 106) were injected subcutaneously into female C57BL/6 mice (7 to 8 weeks of age). At day 7 (MC38-CLDN18.2) or day 8 (KPC), tumor-bearing mice were intraperitoneally administered 150 mg/kg cyclophosphamide (CTX, Macklin, C849559). CLDN18.2-CAR T/CLDN18.2-H2A.Z1-CAR T/control T cells (8 × 105) were adoptively transferred intravenously into these recipients after 72 hours of cyclophosphamide injection. For OT-I adoptive transfer, naive OT-I CD8+ T cells were isolated from 8-week-old and 40-week-old OT-I mice. On day 0, 2 × 105 cells were adoptively transferred into 8-week-old female C57BL/6 recipient mice. On day 1, 8 × 105 MC38-OVA tumor cells were inoculated subcutaneously. Tumor volumes were calculated by the formula: length × width × width / 2. Statistical analysis Prism software (GraphPad Software) was used to prepare graphs and to perform statistical analysis. All results are shown as means ± SEM. Statistical analysis were performed as indicated in the figure legends. A P value of <0.05 was considered significant.