Chronic viral infection alters PD-1 locus subnuclear localization in cytotoxic CD8+ T cells

Abstract

During chronic infection, virus-specific CD8⁺ cytotoxic T lymphocytes (CTLs) progressively lose their ability to mount effective antiviral responses. This "exhaustion" is coupled to persistent upregulation of inhibitory receptor programmed death-1 (PD-1) (Pdcd1)-key in suppressing antiviral CTL responses. Here, we investigate allelic Pdcd1 subnuclear localization and transcription during acute and chronic lymphocytic choriomeningitis virus (LCMV) infection in mice. Pdcd1 alleles dissociate from transcriptionally repressive chromatin domains (lamin B) in virus-specific exhausted CTLs but not in naive or effector CTLs. Relative to naive CTLs, nuclear positioning and Pdcd1-lamina dissociation in exhausted CTLs reflect loss of Pdcd1 promoter methylation and greater PD-1 upregulation, although a direct correlation is not observed in effector cells, 8 days post-infection. Genetic deletion of B lymphocyte-induced maturation protein 1 (Blimp-1) enhances Pdcd1-lamina dissociation in effector CTLs, suggesting that Blimp-1 contributes to maintaining Pdcd1 localization to repressive lamina. Our results identify mechanisms governing Pdcd1 subnuclear localization and the broader role of chromatin dynamics in T cell exhaustion.

Keywords: Blimp-1; CD8(+) T cell; CP: Immunology; L-selectin; PD-1; chronic viral infection; cytotoxic; exhaustion; methylation; subnuclear localization; transcription.

Graphical abstract

Figure 1

Exhausted antigen-specific CD8⁺ T cells lose biallelic Pdcd1 association to repressive nuclear lamina (A) Representative confocal microscopy images show DNA-immunoFISH scoring examples of Pdcd1 (green) and Sell (red) allelic association to lamin B (magenta) in LCMV-specific mouse effector CD8⁺ T lymphocytes (CTLs) as shown in (B). Left, middle, and right panels denote conditions of no association (none), monoallelic, or biallelic association to lamin B, respectively (specific focal planes shown). The far right panel displays an overlay example of Sell and Pdcd1 loci in one single cell. Scale bars represent 1 μm. (B) LCMV-specific CTLs from Arm (blue) or Cl13 (red) were obtained at denoted days post-infection (dpi). Top graphs: the frequencies of cells with Pdcd1 (left) or Sell (right) monoallelic or biallelic locus-specific association to lamin B (DNA-immunoFISH) are shown (% cells). Bottom graphs: Pdcd1 (left) or Sell (right) biallelic fold association to lamina was calculated relative to the remaining conditions (none + monoallelic). (C) Negative controls for experiments presented in (A) and (B) are shown. Top graphs: the frequency of cells with Cd4 (left) or Cd8 (right) monoallelic or biallelic locus-specific association to lamin B (DNA-immunoFISH) (% cells) is shown. Bottom graphs: Cd4 (left) or Cd8 (right) biallelic fold association to lamin B was calculated as in (B). The far right panel displays a representative overlay confocal microscopy image of Cd4 and Cd8 loci in a single cell. (A–C) Independent experiments were reproduced ≥ n = 2–3 times; n = 3 mice per condition, per experiment. FISH ≥100 cells. p values: ns, not significant; ^∗, significant; ^∗∗, very significant; ^∗∗∗, highly significant (see STAR Methods). Significance was calculated across all groups (Figures S1 and S2; Tables S1, S2, and S3). Graphs, p values combine 2–3 independent, representative experiments. (C) Differences in p, not significant. Errors bars = values ± SD.

Figure 2

Loss of biallelic Pdcd1-lamina association in exhausted antigen-specific CD8⁺ T cells is consistent with increased PD-1 mRNA and protein, as well as DNA demethylation (A–C) LCMV-specific CD8⁺ T lymphocytes (CTLs) from Arm- (blue) or Cl13-infected mice (red) were obtained at denoted days post-infection (dpi) (constituting the same samples and experiments shown in Figure 1). Representative and independent experiments were reproduced ≥ n = 3; n = 3 mice per condition, per experiment. (A) The graphs denote relative mRNA expression of Pdcd1 (left) or Sell (right) in CTLs. Error bars = values ± SD. (B) Results from genomic DNA bisulfite sequencing of Pdcd1 and Sell promoter regions are shown. Lines: individual sequenced clones. Filled and open circles: methylated and nonmethylated cytosines, respectively. (C) Shown are FACS histograms of PD-1, CD62L, CD44, KLRG1, CD25, CD27, CD127, and 2B4 protein expression from samples in (A) and (B) (Figures 1, S1, and S2).

Figure 3

Chronic LCMV-mediated exhaustion is associated with increased Blimp-1 transcription factor binding to Pdcd1 in CD8⁺ T cells LCMV-specific CD8⁺ T lymphocytes (CTLs) were obtained from Arm- (blue) or Cl13-infected mice (red) or wild-type (WT) C57BL/6 mice. CTLs from infected mice were collected at 8 dpi, as well as at 28 dpi for Cl13 mice. (A) ChIP analysis shows Blimp-1 binding to site 2 of the Pdcd1 promoter and a control site in CTLs. Results are presented as percentage of input DNA. 8- and 28-dpi experiments were performed separately. 28-dpi samples combined cells from 5 mice each. Data were averaged from 3 independent experiments. p values: ns, not significant; ∗, significant; ∗∗, very significant; ∗∗∗, highly significant (see STAR Methods). (B) The graph denotes the quantitation of intact Prdm1 locus by real-time PCR on DNA from Cl13 Blimp-1 conditional KO (Prdm1^−/−) and WT control CTLs used in Figures 4, S1C, and S3. (A) and (B) Error bars = values ± SD.

Figure 4

Blimp-1 modulates Pdcd1 subnuclear localization to lamina in LCMV-specific effector CD8⁺ T cells LCMV-specific effector CD8⁺ T lymphocytes (CTLs) from Arm- (blue) or Cl13-infected (red) Blimp-1 conditional KO (Prdm1^−/−) and wild-type (WT) mice were obtained at 8 days post-infection (dpi). Biallelic Pdcd1 association to lamin B was scored via DNA-immunoFISH. Samples corresponding to Figures 4A and 4B are shown. (A) Representative confocal microscopy images depict DNA-immunoFISH examples of Pdcd1 (green) and Ifng (negative control, red) allelic association to lamin B (magenta) in WT and Prdm1^−/− LCMV-specific mouse effector CTLs described in (B). Scale bars represent 1 μm. (B) Top graphs: the frequencies of cells with Pdcd1 (left) or Ifng (right) no association (none), monoallelic, or biallelic locus-specific association to lamin B (DNA-immunoFISH) are shown (% cells). Bottom graphs: Pdcd1 (left) or Ifng (right) biallelic fold association to lamina was calculated relative to the remaining conditions (remainder = none + monoallelic). (C) Relative mRNA expression of Pdcd1 (left) or Sell (right) from samples in (A) and (B) is shown. (D) Shown are FACS histograms of PD-1, CD62L, and CD127 protein expression from samples in (A) and (B). Blue, Prdm1^−/−; red, WT. Independent experiments were reproduced ≥ n = 2–3 times; n = 3 mice per condition, per experiment. FISH ≥100 cells. p values: ns, not significant; ^∗, significant; ^∗∗, very significant; ^∗∗∗, highly significant (see STAR Methods). Significance was calculated across all groups (Tables S4, S5, S6, and S7). Graphs, p values combine 2–3 independent, representative experiments. Errors bars = values ± SD. (See also Figures S1C, S3, and S4).