Antigen-driven EGR2 expression is required for exhausted CD8+ T cell stability and maintenance

Abstract

Chronic stimulation of CD8⁺ T cells triggers exhaustion, a distinct differentiation state with diminished effector function. Exhausted cells exist in multiple differentiation states, from stem-like progenitors that are the key mediators of the response to checkpoint blockade, through to terminally exhausted cells. Due to its clinical relevance, there is substantial interest in defining the pathways that control differentiation and maintenance of these subsets. Here, we show that chronic antigen induces the anergy-associated transcription factor EGR2 selectively within progenitor exhausted cells in both chronic LCMV and tumours. EGR2 enables terminal exhaustion and stabilizes the exhausted transcriptional state by both direct EGR2-dependent control of key exhaustion-associated genes, and indirect maintenance of the exhausted epigenetic state. We show that EGR2 is a regulator of exhaustion that epigenetically and transcriptionally maintains the differentiation competency of progenitor exhausted cells.

Fig. 1. EGR2 expression is elevated within TCF1⁺ progenitor exhausted cells.

a EGR2 protein levels assessed by flow cytometry within splenic H2-D^bGP_33–41 tetramer-stained CD8⁺ T cells from B6 mice infected with LCMV-Arm or Cl13 for the indicated times. Representative histogram (left) shows data from day 8 p.i., with naive T cell staining included from polyclonal CD44^loCD8⁺ T cells from uninfected B6 mice. Right graph shows EGR2 expression relative to naive levels (as defined above). n = 11–15 mice per group per time point from 2–3 independent experiments. p-values were calculated using a two-tailed unpaired T test (days 5, 15, 20) or two-tailed Mann–Whitney test (days 8, 30). For all indicated significant differences p < 0.0001. b EGR2 protein levels assessed as in a within TCF1⁺ and TCF1^- tetramer^-stained CD8⁺ T cells from B6 mice infected with LCMV-Cl13 for the indicated times. Representative plot (left) shows data from day 20 p.i. Right graph shows EGR2 expression relative to naive levels (as defined above). n = 5–6 mice per group per time point from two independent experiments. p-values were calculated using a two-tailed unpaired T test. For indicated significant differences, exact p values (left to right) = 0.0060, 0.0141, 0.0035, 0.0022, 0.0024. c Egr2-GFP reporter expression within polyclonal CD44^loCD8⁺ T cells, or Slamf6⁺, Slamf6^-CD101^- and Slamf6^-CD101⁺ splenic H2^-D^bGP_33-41 tetramer-stained CD8⁺ T cells from chronic LCMV infected Egr2-GFP reporter mice at day 20 p.i. Representative histogram (top) and pooled data (bottom) are shown with n = 6 mice from two independent experiments. p-values were calculated using a one-way ANOVA with a Tukey’s post-test. For indicated significant differences, exact p values (left to right) = <0.0001, 0.0002, 0.0002, 0.0043. d, e Egr2-GFP mice were injected s.c. with 2 × 10⁵ B16-OVA cells, and at day 14 post-injection TILs were isolated from tumours for analysis. GFP levels in PD-1^hi vs PD-1^lo total CD45⁺CD8⁺ TILs (d), and Slamf6⁺ vs Slamf6^- CD45⁺CD8⁺PD-1^hi TILs (e) are shown. Representative plots/histograms are on the left, and pooled data are on the right showing MFI relative to naive. n = 9 mice per group from two independent experiments. p-values were calculated using a two-tailed Mann–Whitney test, exact p values =  <0.0001 (d), 0.0028 (e). Error bars depict SEM, *p < 0.05, **p < 0.01, ***p < 0.001.

Fig. 2. EGR2 expression is maintained by chronic antigen.

a Experiment outline for b, c. 2 × 10³ CD45.1⁺ P14 cells were transferred into B6 mice subsequently infected with LCMV-Arm or LCMV-Cl13. On day 8 p.i., splenocytes were isolated and the equivalent of 5 × 10⁵ P14 cells were transferred into LCMV-Cl13 or LCMV-Arm infection-matched B6 mice. Seven days later (day 15 p.i.), EGR2 protein levels were assessed within the transferred P14 cells. b, c EGR2 levels within cells transferred from LCMV-Arm (b) or LCMV-Cl13 (c) donors into Arm or Cl13 infected recipients. Left histograms show representative plots, with naive CD44^loPD-1^lo cells from Cl13 infected mice also shown in grey. Right graphs show EGR2 MFI relative to the above naive population. n = 13 (top) and 15 (bottom) (b), and n = 10 (top) and 9 (bottom) (c), mice per group from 2–3 independent experiments. d Experiment outline for e. Experiment was conducted as in a except that P14 cells derived from Cl13 infected donors were transferred into recipients infected with wild-type Cl13 or GP V35A mutant Cl13. e EGR2 levels within cells transferred from Cl13 donors into Cl13 or V35A recipients. Plots and data representation is as in b, c. n = 15 (top) and 18 (bottom) mice per group from three independent experiments. p-values were calculated using a two-tailed unpaired T test (c) or two-tailed Mann–Whitney test (b, e). For all indicated significant differences p < 0.0001. ***p < 0.001.

Fig. 3. EGR2 maintains the exhausted CD8⁺ T cell response.

a–f WT or cKO mice were depleted of CD4⁺ T cells then infected with LCMV-Cl13 and the response tracked over time. a Inhibitory receptor MFI within H2-D^bGP_33-41 tetramer-stained CD8⁺ T cells over time (right; normalized to WT mean at each time-point) with representative day 20 p.i. histograms (left). p-values were calculated using a two-tailed Mann–Whitney test (2B4 day 20, CD160 day 90) or a two-tailed unpaired T test (all others). For indicated significant differences, exact p values (left to right) for PD-1= <0.0001, <0.0001, 0.0388, TIM3= <0.0001, 0.0005, 0.0044, 2B4= <0.0001, LAG3 = 0.0247, CD160 = 0.0004. From left to right in the graphs, WT n = 12, 10, 7, 13, and cKO n = 12, 10, 7, 8, from two independent experiments. b Summary of data in a expressed as the proportion of cells expressing 0–5 of the listed inhibitory receptors at day 20 p.i. p-values were calculated using a two-tailed unpaired T test comparing WT to cKO values in each quadrant. For indicated significant differences, exact p values = 0.0028 (red), 0.0029 (orange), 0.0212 (blue). c–d H2-D^bGP_33–41 tetramer-stained CD8⁺ T cells were sorted at day 20 p.i. and subjected to RNAseq analysis (n = 2 per genotype). c Volcano plot denoting differentially expressed genes in cKO versus WT cells. d GSEA analysis in cKO vs WT cells of genes selectively up or down within progenitor vs terminally exhausted virus-specific CD8⁺ T cells isolated from chronic LCMV infection. Plots show enrichment of progenitor (left; red) or terminal exhaustion (right; blue) gene signatures. Enrichment score p-values were calculated by the GSEA program using an empirical phenotype-based permutation test procedure. e Representative (top; day 20 p.i.) and pooled (bottom) proportions of tetramer+ cells that were TIM3^-TCF1⁺ over the course of infection from the experiment in a. p-values were calculated using a two-tailed Mann–Whitney test (day 8, 40) or a two-tailed unpaired T test (all others). For indicated significant differences, exact p values (left to right)= <0.0001, 0.0003, 0.0041, 0.0087. f Numbers of TCF1⁺ and TCF1^- cells from the experiment in e. p-values were calculated using a two-tailed Mann–Whitney test (TCF1^- day 20, 40) or a two-tailed unpaired T test (all others). For indicated significant differences, exact p values (left to right) for TCF1⁺ = 0.0101, for TCF1^- = 0.0156, 0.0015, 0.0059, 0.0317. g %Slamf6⁺TCF1⁺ cells of PD-1^hiCD8⁺ TILs in WT vs CD8 cKO mice given B16-OVA tumour cells 14 days prior. Representative (left) and pooled data (right) are shown. n = 7 (WT) and 8 (cKO) mice per group from two independent experiments. p-value was calculated using a two-tailed unpaired T test, exact p value = <0.0001. Error bars depict SEM, *p < 0.05, **p < 0.01, ***p < 0.001.

Fig. 4. EGR2 regulates the exhausted CD8⁺ T cell response to checkpoint blockade.

WT or cKO mice were depleted of CD4⁺ T cells then infected with LCMV-Cl13, and from days 28-42 p.i. mice were treated with anti-PD-L1 or isotype control antibody every 3 days prior to analysis at day 42 p.i. a The sum of GP_33–41 and GP_276–286 specific CD8⁺IFNγ⁺ T cell numbers (error bars show error of the summed numbers). For indicated significant differences, exact p values (left to right) = 0.0001, <0.0001, 0.0004. b %TIM3^-TCF1⁺ cells within H2-D^bGP_33–41 tetramer-stained CD8⁺ T cells. For indicated significant differences, exact p values (left to right)= <0.0001, <0.0001, 0.0049, 0.0136. c, d Representative (c) and pooled (d) %TNFα⁺ and %TNFα⁺IL-2⁺ cells within the GP_276–286 peptide-specific CD8⁺IFNγ⁺ T cells. n = 7–9 mice per group from three independent experiments. For indicated significant differences, exact p values (left to right) for left graph = 0.0003, 0.0102, 0.0048, for right graph = 0.0347, 0.0385. All p-values in this figure were calculated using a one-way ANOVA with a Tukey’s post-test. Error bars depict SEM, *p < 0.05, **p < 0.01, ***p < 0.001.

Fig. 5. EGR2 loss leads to altered phenotypes in exhausted cell subsets.

a–e H2-D^bGP_33-41 tetramer-stained CD8⁺ T cells were sorted from CD4-depleted, chronic LCMV infected WT or cKO mice at day 20 p.i., and subjected to multiplexed scRNAseq. a tSNE plot showing clusters identified within the pooled WT and cKO cells using published signatures. b Proportional distribution of the clusters identified in a within WT and cKO cells. c tSNE plot from a showing the location of WT (aqua) and cKO (orange) cells. d Volcano plots generated by comparing gene expression within cKO vs WT cells in each cluster from a. Dotted lines indicate p-value (0.05) and log₂ fold change (>0.8 or <−0.8) cut-off values. Red dots indicate genes that satisfy both p-value and fold change cut-offs, while blue dots indicate genes that satisfy the p-value but not the fold change cut-off. Genes up in cKO are on the right, while genes up in WT are on the left. e GSEA analysis of the comparisons from d looking at enrichment of genes selectively upregulated in exhaustion (“Exhaustion signature”; upper plots in red), or genes selectively up in effector cells at day 8 of acute vs chronic LCMV (“Effector signature”; lower plots in blue). f, g Differentially expressed markers identified by the scRNAseq analysis in d were confirmed in H2-D^bGP_33–41 tetramer-stained CD8⁺ T cells from CD4-depleted, chronic LCMV infected WT or cKO mice at day 20 p.i. Representative plots (f) and pooled data (g) are shown. For the markers in g, WT n = 20 (PD-1, EOMES), 10 (TIGIT, Ly6C, GzmB, FR4, IL-10Rα, CD160), 6 (Ly6A), and KO n = 22 (PD-1, EOMES), 12 (TIGIT, Ly6C, GzmB, FR4, IL-10Rα, CD160), 6 (Ly6A), mice per group from 2–5 independent experiments. p-values were calculated using a two-tailed Mann–Whitney test (TCF1^- Ly6C, TCF1^- Ly6A, TCF1⁺ GzmB, TCF1^- GzmB, TCF1⁺ CD160) or a two-tailed unpaired T test (all others). For indicated significant differences, exact p values (left to right graphs) for TCF1⁺= <0.0001, 0.0466, 0^.0014, 0.0015, 0.0056, 0.0070, <0.0001, <0.0001, for TCF1⁻= <0.0001, <0.0001, 0.0059, 0.0090, 0.0411, <0.0001, 0.0084. *p < 0.05, **p < 0.01, ***p < 0.001.

Fig. 6. EGR2 directly regulates key exhaustion genes and indirectly stabilises the exhausted epigenetic state.

a Z-score heatmap of differentially open regions (DORs; FDR < 0.01) identified by ATACseq between cKO (left) and WT (right) H2-D^bGP_33–41 tetramer-stained CD8⁺ T cells sorted from CD4-depleted, chronic LCMV infected WT or cKO mice at day 20 p.i. (n = 2 replicates per genotype). b, c CD8⁺ T cells were isolated from B6 mice at day 20 p.i. with LCMV-Cl13 and subjected to EGR2 ChIP-seq analysis. b Venn diagram summarizing the overlap between genes identified as EGR2 bound by ChIP-seq within CD8⁺ T cells isolated from LCMV-Cl13 infection (EGR2 bound) and genes differentially expressed within exhausted cKO vs WT cells from Fig. 3c (EGR2 regulated). c Representative profiles of EGR2 ChIP-seq peaks within genes repressed (left) or induced (right) by EGR2. Traces show EGR2 Ab signal and Total input signal. Red boxes (and arrows) denote significant peaks. d The top 50 WT or cKO-specific ATACseq peaks were analysed by HOMER to identify transcription factor motifs enriched above background, and the differential enrichment of these motifs was then measured between cKO vs WT peaks. Plot shows log₂(fold enrichment) (x-axis), and difference in enrichment p-value (y-axis), between cKO vs WT. Dotted lines mark cut-offs for p-value difference (≥3) and log₂(fold enrichment) (≥0.7 or ≤−0.7), while blue box (enriched in WT) and red box (enriched in KO) indicate enriched motifs that passed these cut-offs. e Z-score heatmap showing average AP-1 gene expression level within WT vs cKO cells in each exhausted cell cluster from the scRNAseq experiment in Fig. 5. f EGR2 binding to AP-1 genes. Red boxes (and arrows) denote significant peaks.