Targeting p53 and histone methyltransferases restores exhausted CD8+ T cells in HCV infection

Abstract

Hepatitis C virus infection (HCV) represents a unique model to characterize, from early to late stages of infection, the T cell differentiation process leading to exhaustion of human CD8+ T cells. Here we show that in early HCV infection, exhaustion-committed virus-specific CD8+ T cells display a marked upregulation of transcription associated with impaired glycolytic and mitochondrial functions, that are linked to enhanced ataxia-telangiectasia mutated (ATM) and p53 signaling. After evolution to chronic infection, exhaustion of HCV-specific T cell responses is instead characterized by a broad gene downregulation associated with a wide metabolic and anti-viral function impairment, which can be rescued by histone methyltransferase inhibitors. These results have implications not only for treatment of HCV-positive patients not responding to last-generation antivirals, but also for other chronic pathologies associated with T cell dysfunction, including cancer.

Fig. 1. Gene-expression profiling of virus-specific CD8+ T cells in HCV infection.

a Principal-component analysis (PCA) of 4766 differentially expressed genes (DEGs) identified by ANOVA (q-value ≤ 0.05) in HCV-specific CD8+ T cells from patients with acute (n = 13), chronic (n = 7) and resolved (n = 4) HCV infections, as well as FLU-specific CD8+ T cells from healthy controls (n = 5). Data were normalized with the quantile method and filtered for probes detected in at least two-third of replicates for each condition. b Hierarchical-clustering representation of the 4766 DEGs. Data were median-normalized before clustering and expressed as single patient profiling. In red upregulated and in blue downregulated genes. c Transcriptome profiles of HCV-specific CD8+ T cells from chronically evolving and self-limited acute patients were compared by topological analysis at two different time-points (time of diagnosis/T1 and several months later/T2). Twenty-nine and 277 pathways were significantly dysregulated (Benjamini-Hochberg corrected q-value ≤ 0.05) at the T1/early and T2/late time-points, respectively. Venn-diagram distribution of pathways identified as dysregulated by comparative topological analysis of acute chronically-evolving vs. self-limited and chronic vs. resolved patients. The 15 pathways found to be significantly dysregulated in both comparisons, but with a largely predominant trend toward upregulation (red) at T1/early and the opposite trend (blue) at T2/late, are listed in the bottom panel; these include genes related to TCR signaling, DNA damage response and metabolism at the T1 comparison (each column shows the enrichment in upregulated or downregulated genes in each pathway derived from the calculation of the median gene expression fold change in the comparison of chronically evolving vs. self-limited–T1/early–and of late chronic vs. spontaneously resolved patients–T2/late). d List of the 14 T1/early-specific dysregulated pathways, half of which are upregulated (red), while the remaining half is downregulated (blue). e Heat-map of differentially expressed genes derived from GSEA (Molecular Signature Database, C2 canonical pathways and C5 gene ontology sets) at T1/early time-point, related to cell cycle, DNA damage/DNA repair, cell signaling, mitochondrion and metabolism. Upregulated genes in red; downregulated genes in blue.

Fig. 2. Glucose metabolism is impaired in HCV-specific CD8+ T cells from chronically evolving acute patients.

a Representative examples of virus-specific CD8+ T cells stained with HLA-A2+ dextramers ex vivo after overnight anti-CD3/anti-CD28 stimulation. Glucose uptake (b), measured by the incorporation of the glucose analog 2-NBDG (MFI), and Glut1 expression levels (c) in virus-specific CD8+ T cells from T1/early HCV patients and healthy controls stimulated as in a. Data are presented as median fluorescence intensity (MFI) values; median values are indicated by horizontal lines. Different numbers of patients (represented by individual dots) were tested in each assay depending on dextramer-positive cell frequencies. b–c Differences between multiple groups were evaluated with the non-parametric Kruskal-Wallis test; p-values were corrected for pairwise multiple comparisons with the Dunn’s test. d metabolic flux profiling of purified CD8+ T cells from 6 T1/early chronically-evolving (acute) patients. Cells were stimulated overnight with either HCV-NS3 (red) or control (FLU-specific, CMV-specific and EBV-specific) peptides (blue), or were not stimulated (green). The extracellular acidification rate (ECAR) was measured in real-time ex vivo before (basal level) and after oligomycin treatment in order to determine the maximum glycolytic capacity (MGC) and glycolytic reserve (difference between MGC and baseline ECAR) (see Methods section for details on Seahorse analysis). e Metabolic flux profiling of purified CD8+ T cells from T1/early self-limited (acute) patients (n = 4) stimulated overnight as in d. ECAR, maximum glycolytic capacity and glycolytic reserve were measured as in d. In d and e, ECAR values are given as the mean ± SD in the left-side and are presented as box-and-whisker plots (with median and 5–95 percentile) in the right-side. d–e Statistical analysis was performed with the Friedman test to compare different stimuli; p-values have been corrected for pair-wise multiple comparisons with the Conover’s test.

Fig. 3. Mitochondrial metabolism is impaired in HCV-specific CD8+ T cells from chronically evolving acute patients.

a Percentage of mitochondrial depolarized virus-specific CD8+ T cells, detected with HLA-A2+ dextramers ex vivo after overnight anti-CD3/anti-CD28 stimulation, by staining with the mitochondrial membrane potential (MMP) sensitive dye JC-1 (see Methods section for details). Dextramer-positive virus-specific depolarized cells were quantified by subtracting the percentage of FL1high/FL2low cells (JC-1 staining) detected in the unstimulated samples from the percentage of the corresponding cellular subsets detected in the stimulated samples, as previously reported. b Mitochondrial superoxide levels determined ex vivo as in a with the MitoSOX Red dye. c Cytoplasmic reactive oxygen species (ROS) determined ex vivo, as in a, with the superoxide-specific dye DHE and the intracellular H₂O₂ specific dye H2DCFDA are shown on the left and on the right, respectively. a–c Data are presented as median fluorescence intensity (MFI) values; median values are indicated by horizontal lines. Different numbers of patients (represented by individual dots) were tested in each assay depending on dextramer-positive cell frequencies. a–c Differences between multiple groups were evaluated with the non-parametric Kruskal-Wallis test; p-values were corrected for pairwise multiple comparisons with the Dunn’s test. d Oxygen consumption rate (OCR) data determined on the same samples (n = 6 T1/early chronically evolving acute patients) utilized for ECAR analysis (see Fig. 2d) before (basal level) and after addition of the mitochondrial stressors oligomycin, FCCP and rotenone/antimycin A, which were used to calculate ATP production, maximal respiration capacity, spare respiratory capacity, coupling efficiency and proton leak, as indicated. e, OCR, ATP production, maximal respiration capacity and spare respiratory capacity were determined on the same samples (n = 4 T1/early self-limited acute patients) utilized for ECAR analysis in Fig. 2e, and were calculated as in d. In d and e, OCR values are given as the mean ± SD in the left-side and are presented as box-and-whisker plots (with median and 5–95 percentile) in the right-side. d–e Statistical analysis was performed with the Friedman test to compare different stimuli; p-values have been corrected for pair-wise multiple comparisons with the Conover’s test.

Fig. 4. ATM and p53 pathways are activated in T1/early HCV-specific CD8+ T cells.

a Interaction network of genes involved in at least five of the eight dysregulated processes identified by GSEA (outlined in Supplementary Fig. 2). The network was generated using STRING v. 10.5. Node colors refer to enriched pathways associated with the proteins represented in the network. Line thickness indicates the degree of confidence prediction of the interactions. b Intracellular staining for total p53 of dextramer positive virus-specific CD8+ T cells from patients in the acute phase of HCV infection (chronically evolving T1/early n = 19 and HCV self-limited T1/early n = 14) or from healthy controls (n = 14), performed after overnight stimulation with anti-CD3/anti-CD28 (ex vivo staining). c Intracellular staining for phospho-p53 (Ser15) of dextramer positive virus-specific CD8+ T cells from PBMCs derived from patients in the acute phase of HCV infection or from healthy controls was performed with no stimulation and after overnight anti-CD3/anti-CD28 stimuli (left and middle panels, respectively). The plot on the right represents the ratio between anti-CD3/anti-CD28 stimulated and unstimulated virus-specific CD8+ T cells. d Intracellular staining for phospho-ATM (Ser1981) of dextramer positive virus-specific CD8+ T cells as in c. e Phospho-p38 (Thr180) intracellular staining of dextramer positive virus-specific CD8+ T cells as in c. Data in panels from b to e are presented as median fluorescence intensity (MFI), with median values indicated by horizontal lines. Different numbers of patients (represented by individual dots) were tested in each assay depending on dextramer-positive cell frequencies. Representative overlay histograms are shown next to each plot in panels from b to e. All data were analyzed with the Kolmogorov-Smirnov test. Differences between multiple groups were evaluated with the nonparametric Kruskal-Wallis test; p-values were corrected for pairwise multiple comparisons with the Dunn’s test (JASP software).

Fig. 5. Blocking dysregulated intracellular signaling pathways can reverse early metabolic and functional CD8+ T cell defects.

a PBMC from T1/early chronically-evolving patients were stimulated overnight with HCV-NS3 peptides in the presence or absence of the ROS scavenger resveratrol^, (treated vs. untreated) and then stained with MitoSOX Red to assess mitochondrial superoxide content and with anti-phospho-ATM (Ser1981), phospho-p38 (Thr180), and phospho-p53 (Ser15) antibodies. Bars represent mean fold-change values + SEM derived from 6 patients. Representative overlay histograms are illustrated on the right. b PBMCs from T1/early chronically-evolving patients were stimulated for 40 h with HCV-NS3 peptides in the presence or absence of specific ATM (KU-55933), p53 (Pifithrin-α), AMPK (Dorsomorphin), and p38a (SB203580) inhibitors, followed by flow cytometry determination of GLUT-1 expression levels. Representative dot plots are illustrated on the right. Glucose uptake studied via incorporation of the glucose analog 2-NBDG (c) and PD-1 expression (d) have been measured as in b. e IFN-γ, TNF-α, and IL2 production by CD8+ T cells cultured as in b. Data are presented as the ratio between the percentage of cytokine positive CD8+ T cells detected in inhibitor-treated vs. untreated cultures (fold-change). f IFN-γ, TNF-α, IL2 single positive, as well as double-positive IFN-γ+/TNF-α+CD8+ T cells generated in short-term T cell lines upon 10-days stimulation with HCV-NS3 peptides in the presence or absence of the inhibitors specified in the legend to panel b. Data shown in all panels are presented as fold-change of treated vs. untreated CD8+ T cells. Horizontal lines in panels b to f represent median values; data were analyzed statistically with the Wilcoxon signed-rank test; NS = not significant.

Fig. 6. Epigenetic transcriptional repression in exhausted HCV specific CD8+ T cells from chronic patients.

a Six distinct functional groups of pathways enriched in upregulated genes (red) in T1/early and displaying the opposite trend (blue) in T2/late identified by GSEA (MSigDB, C2 canonical pathways and C5 Gene Ontology sets) in HCV-specific CD8+ T cells from chronically evolving patients. NES = normalized enrichment score; FDR = False Discovery Rate. In red, above the p-value columns, is shown the total number of genes significantly upregulated in T1/early and downregulated in T2/late in each group of pathways. b Heat-maps comparing the expression profiles of leading genes belonging to the DNA repair/damage response, metabolism and cell signaling pathways in chronic and resolved infections (see also the T2 sheets in Supplementary Data 2). c Mitochondrial (left panel, JC-1 staining) and proteasomal (right panel, ProteoStat staining) functions were assessed in dextramer-stained HCV-specific CD8+ T cells from T2/late chronic and T2/late self-limited HCV infection and in healthy controls following PBMC overnight stimulation with anti-CD3/CD28. MFI, Median Fluorescence Intensity. d Heat-map comparing the expression levels (average log₂ fold change) of epigenetic regulatory complexes (derived from the EpiFactors database as detailed in Methods section) in chronic vs. self-limited infection (T1/early), chronic vs. resolved infection (T2/late) and chronic (T2/late) vs. healthy controls. e Repressive H3K9me2 (upper panels) and permissive H3K9ac2 (lower panel) histone marks determined by flow cytometry in dextramer-stained HCV specific CD8+ T cells from T2/late chronic and T2/late resolved HCV patients or healthy controls. PBMC were stimulated overnight with anti-CD3/CD28 (ex vivo staining) or for 10 days with HLA-A2-restricted HCV-specific or FLU-specific peptides. Horizontal lines in panels c and e represent median values. Differences between multiple groups in panels c and e were evaluated with the non-parametric Kruskal-Wallis test; p-values were corrected for pairwise multiple comparisons with the Dunn’s test.

Fig. 7. HMT and p53 inhibitors improve anti-viral and metabolic functions of exhausted HCV-specific CD8+ T cells.

PBMC from chronically evolving (T1/early) or chronic (T2/late) HCV patients were stimulated for 40 h with HCV-NS3 peptides in the presence or absence of the EZH2 inhibitors GSK126 (GSK) and EPZ005687 (EPZ) (red dots), of the EHMT2/G9a inhibitors UNC0638 (UNC) and BIX01294 (BIX) (green dots), and of the p53 inhibitor pifithrin-alfa (p53) (blue dots). HCV-stimulated CD8+ T cells were then tested in flow cytometry for GLUT-1 levels (a), glucose uptake (b), PD-1 expression (c), IFN-γ (d), TNF-α (e), and IL2 (f) production. g–j PBMC from chronically evolving (T1/early) or chronic (T2/late) HCV patients were stimulated for 10 days with HCV-NS3 peptides in the presence or absence of the inhibitors specified in a and CD8+ T cells were then tested for IFN-γ, TNF-α, IL2, and IFN-γ plus TNF-α production as indicated. Data are presented as the ratio (fold-change) between positive CD8+ T cells detected in treated vs. untreated cultures from individual patients. Statistical analysis was performed with the Wilcoxon signed-rank test; horizontal lines represent median values. k Reduction of the repressive H3K9me2 histone mark was assessed by flow cytometry on CD8+ T cells from T2/late chronic HCV patients stimulated for 40 h or 10 days as in a. Data are presented as the ratio (fold-change) between MFI (Median fluorescence intensity) of H3K9me2 CD8+ T cells detected in treated vs. untreated cultures from individual patients; statistical analysis was performed with the Wilcoxon signed-rank test; columns and dots represent median values and single patients, respectively.

Fig. 8. Effect of DAA treatment and HMTs/p53 modulation on HCV-specific CD8+ T cells from chronic patients.

a Glucose uptake (measured by the incorporation of the glucose analog 2-NBDG), percentage of depolarized mitochondria (by staining with the mitochondrial membrane potential sensitive dye JC-1), PD-1 expression, proteasomal function (by ProteoStat staining) and repressive H3K9me2 mark were assessed in virus-specific, dextramer-stained CD8+ T cells from T2/late HCV patients (n = 9) at baseline and at EOT, from T2/late resolved patients and from healthy controls after overnight anti-CD3/anti-CD28 stimulation (see Methods section for details). All data were analyzed with the Kolmogorov-Smirnov test, followed by Wilcoxon matched-pairs signed rank test (paired for chronic patients at baseline vs. EOT). Conversely, differences between multiple patient groups were evaluated with the non-parametric Kruskal-Wallis test; p-values were corrected for pairwise multiple comparisons, with the Dunn’s test. b PBMC from chronic HCV patients (T2/late; n = 6) at baseline and at EOT were stimulated for 40 h with HCV-NS3 peptides in the presence or absence of the EZH2 inhibitors GSK126 (GSK) and EPZ005687 (EPZ) (red dots), of the G9a inhibitors UNC0638 (UNC) and BIX01294 (BIX) (green dots), and of the p53 inhibitor pifithrin-alfa (p53) (blue dots), followed by co-staining for IFN-γ, IL2, and TNFα (left). c PBMC from baseline and EOT of T2/late chronic HCV patients (n = 11) were stimulated for 10 days with HCV-NS3 peptides as in b. Data in b and c are presented as the ratio (fold-change) between cytokine producing CD8+ T cells detected in treated vs. untreated cultures from individual patients; statistical analysis was performed with the Wilcoxon signed-rank test; horizontal lines represent median values.