Fig 2: EOMES induces epigenetic changes in enhancer regions. a MA plot of mean ATAC-seq counts per peaks showing the differentially open regions (DOR) of CD8SP WT (blue) and EomesTg (red) cells with the indicated number of regions. Histograms indicate the number of opening or closing regions in EomesTg in comparison to WT cells at promoters and enhancers. b Cumulative distribution plot generated by BETA algorithm showing the predicted activating/repressive functions of DOR in CD8SP WT and EomesTg with the indicated P-values determined by the Kolmogorov–Smirnov test. c Scatter plots display differentially active H3K27ac peaks at promoters (left) and enhancers (right) in CD8SP WT (blue) and EomesTg (red) cells. d CiiiDER analysis for putative transcription factors motifs from DOR of CD8SP EomesTg and WT at enhancers. Transcription factors are coloured according to their gene coverage P-value and whether they are over- (red) or under- (blue) represented. The size of each point is also proportional to log10 P-value. e Normalized coverage plot of histone modifications (H3K4me1, H3K4me3 and H3K27ac) and chromatin accessibility (ATAC-seq) centred on EOMES-binding sites at promoters and enhancers annotated to genes from Fig. 5e. f Representative ChIP-seq tracks of H3K4me1, H3K27ac, ATAC-seq, EOMES, and RUNX3 from the indicated population at the enhancers of Cxcr3 and Il2rb loci highlighted in grey. ChIP-seq was performed on three independent IPs (n = 3 mice per sample). ATAC-seq was performed on two independent samples (n = 2 mice per sample) from each group

Fig 3: TIM cells display classical features of conventional memory cells. a Flow cytometry of CD3+CD8SP thymocytes from wild-type (WT) Balb/c mice. Histograms represent the expression of the indicated protein in EOMEShi (red) and EOMESlo (gray) cells or the fluorescence minus one (FMO) controls (empty). Median fluorescence intensity (MFI) or the proportion of positive cells are displayed. b SPADE of flow cytometry data gated on CD3+CD8SP thymocytes from WT Balb/c mice. Circles represent cell nodes, colors indicate expression levels for the indicated marker and size is related to the number of cells within a node. Annotations indicate the identified innate memory (TIM) subsets and the bulk of naïve cells. Trees displaying expression levels of other markers are shown in Supplementary Fig. 1. c SPADE trees showing cell frequency of CD3+CD8SP thymocytes from the indicated strain (b, c, 4 WT Balb/c mice were used for tree construction. In c, one representative mouse out of four is shown for each group). d SPADE trees showing cell frequency in each node for naïve and TIM cells sorted from WT Balb/c mice (gating, Supplementary Fig. 2). EOMES expression (horizontal bars indicate median ± interquartile range) in sorted naïve and TIM cells are shown. Each point represents an individual sample. Statistics were calculated using Mann–Whitney test. *P < 0.05. e Volcano plot of RNA-seq data from naïve versus TIM cells shows the adjusted P-value (–log10) versus fold change (log2) (up in TIM, red; up in naïve, green). The numbers of differentially expressed genes are indicated. RNA-seq was performed in triplicates (each sample was generated from a pool of at least seven mice). f BubbleGUM gene set enrichment analysis (GSEA) map of datasets from naïve (N), conventional effector (TE), and memory (TM) cells or from naïve and TIM CD8SP thymocytes. Gene sets (with the indicated number of genes) were established by analysis of their expression patterns between N, TE, and TM, from available gene sets or from naïve and TIM-specific genes as defined in e. The panel summarizes the normalized enrichment score (NES) and false discovery rate (FDR) parameters

Fig 4: EOMES expression is sufficient to drive TIM transcriptional program. a Representative flow cytometry plots of total thymocytes from WT and EomesTg mice showing DN, DP, CD4SP, and CD8SP proportions. Histograms indicate the frequency of each population and their EOMES expression (MFI). b Expression of the indicated markers in WT and EomesTg CD3+CD8SP thymocytes and in EOMESlo and EOMEShi CD3+CD8SP thymocytes from Balb/c mice, shown as a reference. SPADE trees indicate cell frequency of CD3+CD8SP thymocytes from WT and EomesTg mice using the tree constructed from WT Balb/c CD3+CD8SP thymocytes as shown in Fig. 1. c IFN? expression measured by flow cytometry in CD3+CD8SP thymocytes from WT and EomesTg mice cultured with IL-12+IL-18 for 16 h or with phorbol-myristate-acetate and ionomycin (P/I) for 4 h. For P/I, EOMESlo and EOMEShi CD3+CD8SP thymocytes from Balb/c mice are shown. d Volcano plot of RNA-seq data of CD3+CD8SP thymocytes from WT versus EomesTg mice show the adjusted P-value versus fold-change (up in EomesTg, red; up in WT, green). The numbers of differentially expressed genes are indicated. e GSEA plots of innate memory (TIM)-specific (left) or naïve (N)-specific (right) gene sets in CD3+CD8SP thymocytes from EomesTg mice. NES normalized enrichment score. f Expression heatmaps of RNA-seq data comparing mean fold-change (FC) of TIM and EomesTg CD3+CD8SP thymocytes in comparison to their respective naïve and WT controls, respectively. CPM counts per million. In a–c horizontal bars indicate median ± interquartile range and are representative of more than three experiments. Each point represents an individual mouse. Statistics were calculated using Mann–Whitney test. ns not significant, *P < 0.05; **P < 0.01; and ***P < 0.001

Fig 5: EOMES and RUNX3 interact within chromatin-associated complexes. a Selected EOMES-interacting proteins identified by RIME in activated CD8+ T cells. Total spectral counts for each replicate (anti-EOMES or control IgG) are shown. b Genomic distribution of EOMES (4306) and RUNX3 (6741)-binding sites in innate memory (TIM) cells. c Venn diagram illustrating the intersection between EOMES (in TIM) and RUNX3 (in naïve [N] and TIM cells) peaks at active promoters and enhancers. Density plots centered on common EOMES/RUNX3 or EOMES-specific peaks (±250 bp) represent the distribution of the best predicted sites of the JASPAR RUNX3 (MA0684.1) and EOMES (MA0800.1) motifs. d Representative EOMES, RUNX3, H3K4me1, H3K27Ac ChIP-seq, and ATAC-seq tracks at the Ly6c2 locus showing the co-localisation of EOMES and RUNX3 highlighted in grey (top). Sequence of the highlighted region and the location of EOMES and RUNX motifs with their position P-value are indicated (down). e Cumulative distribution plot generated by BETA algorithm showing the predicted activating/repressive functions of RUNX3 and EOMES with the indicated P-values determined by the Kolmogorov–Smirnov test. f Heatmap showing expression of genes from e that are predicted to be targets of EOMES. Selected genes are shown in the right margin. g Normalized coverage plot of histones modifications (H3K4me1, H3K4me3, and H3K27ac) and chromatin accessibility (ATAC-seq) centered on EOMES-binding sites at promoters and enhancers annotated to predicted genes from e. ChIP-seq was performed on three independent IPs (n = 7 mice per sample)