Histone methyltransferase DOT1L maintains cell state and restricts cytotoxic potential of CD8 T cells

Abstract

The histone methyltransferase DOT1L is emerging as a central epigenetic regulator in immune cells. Loss of DOT1L during development of CD8 T cells in vivo leads to gain of memory characteristics but has also been reported to compromise CD8 T cell viability and antitumor reactivity. Here, we determined the cell-intrinsic role of DOT1L in mature mouse CD8 T cells. After conditional deletion of Dot1L in vitro, CD8 T cells retained in vivo proliferative capacity and antitumor reactivity. Moreover, Dot1L knockout CD8 T cells showed increased antigen-specific cytotoxicity toward tumor cells in vitro. Mechanistically, loss of DOT1L resulted in an altered cell state with loss of T cell and gain of innate-like features. These transcriptional changes were mediated by loss of DOT1L methyltransferase activity in a dose-dependent manner. Our findings show that in mature CD8 T cells, ablation of DOT1L activity is well tolerated and reprograms them to gain innate-like memory cell characteristics and enhance intrinsic cytotoxic capacity.

Fig. 1.. Dot1L deletion in mature CD8 T cells leads to replication-dependent loss of H3K79me2 without compromising viability.

(A) Experimental setup: Mature CD8 T cells were harvested from spleens of mice expressing tamoxifen-inducible Cre recombinase (CreERT2) and the OT-I transgenic TCR and carrying either WT (wt) or floxed (fl) Dot1L alleles. CD8 T cells were activated (anti-CD3/CD28), treated with 4-OHT to induce Cre, and expanded in the presence of cytokines as indicated. Created in BioRender. Van leeuwen, F. (2026) https://BioRender.com/rn56fdi. (B) Expansion of n = 4 Dot1L^wt/wt (WT) and n = 3 (biological replicates) Dot1L^∆/∆ (KO) CD8 T cells. (C) Flow cytometry analysis of KO-induced loss of H3K79me2 over time; histograms with n = 4 (WT) and 3 (KO) biological replicates. (D and E) Flow cytometry analysis of viability (D) and late apoptosis (E); CD8 T cells were gated on annexin V (AnnV)⁻/4′,6-diamidino-2-phenylindole (DAPI)⁻ for live and AnnV⁻/DAPI⁺ for late apoptotic/necrotic cells (WT n = 4 and KO n = 3 biological replicates). (F) Flow cytometry dot plots of differentiation status of WT and KO CD8 T cells at day 0 and day 8 based on CD44 and CD62L expression; n = 1 representative biological replicate plotted. (G) Quantification of CD8 T cell subsets as in (F) [mean of n = 3 biological replicates ± SD; two-way analysis of variance (ANOVA) and Šídák’s multiple comparisons test]. (H to K) Histograms of surface expression of CD3 (H) and CD8 (I) and nuclear staining for EOMES (J) and T-bet (K) based on flow cytometry (WT n = 4 and KO n = 3 biological replicates).

Fig. 2.. Adoptively transferred Dot1L KO CD8 TCR-T cells reduce tumor outgrowth in vivo.

(A) Experimental setup: CD45.1⁺ recipient animals were injected with B16F10-OVA melanoma cells, irradiated, and treated with HBSS (vehicle), or adoptively transferred (ACT) CD45.2⁺ WT or KO OT-I CD8 T cells. CD8 T cells were expanded as in Fig. 1A. Created in BioRender. Van leeuwen, F. (2026) https://BioRender.com/rn56fdi. sc, subcutaneous. (B to D) Flow cytometry quantification for H3K79me2 (B), OT-I (C), and CD8 surface expression (D) of donor CD8 T cells, before adoptive cell transfer (mean of n = 4 biological replicates ± SD, unpaired t test). (E) Cumulative tumor growth curve (n = 7 biological replicates). Significance was determined by cumulative Gaussian curve fitting and best-fit analyses. (F) Percentage donor WT and KO CD8 T cells in blood at indicated time points (mean of n = 7 biological replicates ± SD). (G) Percentage donor CD8 T cells in spleen and tumor at endpoint (mean of n = 7 biological replicates ± SD). (H) Differentiation state of donor CD8 T cells at tumor endpoint based on CD44 and CD62L (mean of n = 7 biological replicates ± SD). (I) Percentage CD49d⁺ donor CD8 T cells at tumor endpoint, grouped by CD62L expression (n = 3 biological replicates WT, n = 6 biological replicates KO, mean ± SD; samples with <150 CD45.2⁺ lymphocytes were excluded). For (F) to (I), significance was determined by two-way ANOVA and Šídák’s multiple comparisons tests. (J) Experimental setup: CD8 T cells from Dot1L^wt/wt and Dot1L^fl/fl mice were injected with tamoxifen in vivo, isolated, transferred to recipient mice, vaccinated with OVA peptide, and exposed to unpulsed and OVA-pulsed splenocytes. (K to L) Percentage OVA-pulsed target cell killing (K) and OT-I CD8 T cells (L) in spleen at t = 3 hours. (M and N) Percentage killing (M) and OT-I CD8 T cells (N) in spleen at T = 6 hours. (K) to (L) and (M) and (N) are two independent experiments (mean of n = 9 biological replicates ± SD, significance determined by unpaired t tests). MFI, median fluorescence intensity.

Fig. 3.. Deletion of Dot1L increases the cytotoxic activity of CD8 T cells toward tumor cells in vitro.

(A) Experimental setup: Expanded WT and KO CD8 T cells were cocultured with B16F10-OVA (target) or B16F10-MOCK (control) cells to determine target cell killing and CD8 phenotypes. (B) Quantification of live cell imaging of mCherry⁺ B16F10-OVA target cell outgrowth during coculture with different ratios of OT-I CD8 T cells to target cells {mean of biological replicates [n = 3 (KO) or n = 4 (WT) ± SD]}. The no-CD8 curves displayed in KO and WT are identical. (C) Quantification of area under the curve (AUC) at T = 48 hours of (B), relative to no CD8 [mean ± SD of biological replicates, n = 4 (WT) or n = 3 (KO)]. (D to G) Quantification of flow cytometry surface expression of CD3 (D), CD8 (E), and %CD5⁺ (F) and surface expression of CD69 (G) of CD8 T cells before coculture (0 hours) and after 6-hour coculture with B16F10-OVA target cells. (H) Quantification of T-bet and EOMES populations by flow cytometry after 6-hour coculture. For (D) to (H), values shown are mean ± SD of n = 3 biological replicates. (I and J) Quantification of apoptotic state of OT-I CD8 T cells by flow cytometry, before coculture (I) and after 48 hours with B16F10-OVA (J), mean ± SD [n = 3 (KO) and n = 4 (WT) biological replicates]. For (D) to (J), significance was determined by matched two-way ANOVA and Šídák’s multiple comparisons tests.

Fig. 4.. Loss of DOT1L leads to up-regulation of key effector proteins in CD8 T cells.

(A) Transcriptome analysis of expanded KO and WT CD8 T cells. Mean average (MA) plot shows differentially expressed genes between KO and WT CD8 T cells on day 8 (mean of n = 3 biological replicates). Positive log₂ fold change (FC) refers to an increase in KO. (B) GSEA-based GO enrichment for TCR signaling. (C) GSEA-based curated enriched gene sets. In (B) and (C), normalized enrichment scores (NES), adjusted P values (Padj), and FDR values are displayed. (D) Transcriptome heatmaps displaying log₂ FC in KO versus WT for selection of significantly altered (Padj) surface and cytokine signaling molecules and transcription factors (n = 3 biological replicates). (E) Proteomic volcano plot of differentially expressed proteins between KO and WT CD8 T cells on day 8 [n = 4 (KO) and n = 3 (WT) biological replicates], dotted lines represent cutoff values [log₂FC > 0.6 or < −0.6 and adjusted P value (Padj) <0.05, Student’s t test]. Highlighted are up-regulated cytotoxic granules and down-regulated TCR signaling proteins. (F) Overlay of proteome on MA plot of transcriptome. Up-regulated (yellow) and down-regulated (blue) proteins for KO CD8 T cells are colored; Prf1, Gzma, Gzmb, and Gzmc are highlighted. (G) Box plot showing the transcriptional trend of three sets of genes that were differentially down-regulated, up-regulated, or unchanged in KO compared to WT proteome analysis. Wilcoxon pairwise statistical analysis P values are plotted. (H and I) Heatmaps of selected proteins (H) [n = 4 (KO) and n = 3 (WT) biological replicates] and genes (I) involved in cytotoxic granules and TCR signaling (n = 3, biological replicates). Asterisks represent adjusted P value (Padj) significance.

Fig. 5.. Loss of catalytic activity of DOT1L increases the killing potential of CD8 T cells.

(A) Experimental setup: Mature OT-I CD8 T cells were treated with DOT1Li (10 μM SGC-0946) or DMSO (control) from day 0 onward. Created in BioRender. Van leeuwen, F. (2026) https://BioRender.com/rn56fdi. (B) Expansion of DOT1L inhibited and control CD8 T cells (n = 4 biological replicates). (C) Flow cytometry–based quantification of H3K79me2 of DMSO- and SGC-0946–treated CD8 T cells (mean ± SD, n = 4 paired biological replicates, significance determined by paired t tests). LCK-Cre Dot1L KO CD8 T cells (deletion in early T cell lineage) were used as a negative control (n = 1). (D) MA plot showing up-regulated (red) and down-regulated (blue) genes between SGC-0946– and DMSO-treated CD8 T cells [n = 4 biological replicates, based on adjusted P values (Padj)]. (E) Correlation of differential gene expression for SGC-0946 versus DMSO and KO versus WT (n = 3 biological replicates). The correlation coefficient and significance were determined by Pearson’s product-moment correlation. (F) Volcano plot showing up-regulated (orange) and down-regulated (blue) proteins between SGC-0946– and DMSO-treated T cells [n = 4 biological replicates, significance based on log₂FC > 0.6 or < −0.6 and adjusted P value (Padj) < 0.05 (Student’s t test)]. (G) Box plot showing the mRNA changes of down-regulated (blue), up-regulated (yellow), or unchanged (gray) proteins in SCG compared to DMSO treatment. (H) Heatmaps of selected proteins involved in cytotoxic granules and TCR signaling. Asterisks represent significance based on adjusted P value (Padj). (I) Quantification of live cell imaging cytotoxic killing assay. Coculture of DMSO- or SGC-0946–treated CD8 T cells with B16F10-OVA target cells. SD represents variation between independent cocultures (n = 4 biological replicates) with expanded DMSO- and SGC-0946–treated CD8 T cells. (J) Quantification of the area under the curve of (I), relative to no CD8 (mean ± SD, n = 4 biological replicates, paired two-way ANOVA and Šídák’s multiple comparisons tests).

Fig. 6.. Loss of H3K79me2 is required to induce CD8 T cell phenotypes and NK-like differentiation.

(A) Overview experimental setup: Rcm2;Dot1L^wt/wt (WT) and Dot1L^fl/fl (KO) mice were treated with tamoxifen (75 mg/kg) for 3 consecutive days to genetically ablate Dot1L, after which mice were euthanized, and splenic OT-I CD8 T cells were enriched. Created in BioRender. Van leeuwen, F. (2026) https://BioRender.com/rn56fdi. (B) Flow cytometry histogram of H3K79me2 of WT and KO mice at day 0 (mean ± SD, n = 3 biological replicates). H3K79me2 of enriched splenic CD8 T cells of individual mice displayed. (C) Quantification of live cell imaging of cocultures with different ratios of CD8 to target cells (B16F10-OVA) (n = 3 biological replicates). CD8s were added to the coculture immediately following treatment and enrichment as described in (A). (D) Experimental setup of spike-in normalized H3K79me2 ChIP-seq workflow. Representative tracks (n = 1 biological replicate) are shown for the Lck and Hdac1 genes. Created in BioRender. Van leeuwen, F. (2026) https://BioRender.com/rn56fdi. (E) Tornado plot ranking genes based on the H3K79me2 signal (normalized reads per million; RPM) up to +2 kb from the TSS. Data shown are aggregates of IPs from CD8 T cells treated with DMSO (0.1%), EPZ-5676 (10 μM), or SGC-0946 (10 μM), after 8 days of expansion (n = 2 biological replicates). (F) Scatterplot of H3K79me2 TSS peak average signal (n = 2 biological replicates) for DMSO (x axis) and EPZ-5676– or SGC-0946–treated cells (y axis). (G) Correlation of differential gene expression between EPZ-5676 and SGC-0946 treatment, relative to control (DMSO) (n = 4 biological replicates). The displayed correlation coefficient and significance were determined by Pearson’s product-moment correlation. (H) MA plots showing differentially expressed genes for EPZ-5676 (left) and SGC-0946 (right) compared to DMSO, overlaid with the H3K79me2 signal in DMSO near the transcriptional start site (TSS; 0 to +2 kb, color gradient). Expression was transcript length normalized by calculating fragments per kilobase of transcript per million mapped reads (FPKM).

Fig. 7.. Dot1L-KO and DOT1Li CD8 T cells reversibly gain innate- and memory-like features.

(A) Heatmap displaying peak H3K79me2 TSS peak signal in DMSO (n = 2 biological replicates, green gradient) and transcriptional and proteomic changes at selected genes in Dot1L-KO versus WT and SGC-0946– versus DMSO-treated CD8 T cells (n = 4 biological replicates, red-blue gradient). All data shown are significant based on adjusted P values (Padj). (B) Flow cytometry histogram of IL18r1 for WT, KO, DMSO-, and SGC-treated CD8 T cells after 8 days of expansion. Data shown for (n = 2) representative biological replicates. (C) Quantification of (B) for n = 4 biological replicates. (D and E) Quantification of KIT (D) and NK1.1 (E) surface expression based on flow cytometry (n = 4) after 8 days of expansion. For (C) to (E), significance was determined by one-way ANOVA with Tukey’s multiple comparisons test. (F) Quantification of %IFN-γ⁺ by flow cytometry (n = 4 biological replicates) after a 5-hour stimulation with IL-2, IL-12/18, or IL-2 + PMA/ionomycin (abbreviated to PMA/iono). (G to I) Quantification (n = 4 biological replicates) of CD69 (G), %viability (based on ZombieNIR⁻) (H), and %PD-1⁺ (I) after 24 hours of stimulation. For (F) to (I), significance was determined by two-way ANOVA with Šídák’s multiple comparisons test. (J) Graphical summary highlighting key transcriptional and phenotypical changes in Dot1L-KO and DOT1Li CD8 T cells and their dependence on the loss of H3K79me. Created in BioRender. Van leeuwen, F. (2026) https://BioRender.com/rn56fdi.