IL-6/STAT3 Signaling Axis Enhances and Prolongs Pdcd1 Expression in Murine CD8 T Cells

A robust adaptive immune response requires contributions from CD8 cytotoxic T cells, which assist in viral pathogen and cancer cell clearance through direct killing of infected or cancerous cells (1–3). Although the cytotoxic functions of CD8 T cells are pivotal to pathogen clearance, curtailing their function is crucial to preventing overactivity or autoimmune responses (4, 5). One such brake on aberrant CD8 T cell functions is the well-documented upregulation of the surface protein Programmed Cell Death Protein-1 (PD-1), encoded by Pdcd1, which occurs after TCR stimulation (6, 7). Engagement of PD-1 by its ligand results in a multitude of outcomes antagonistic to CD8 T cell activation, including blocking of costimulatory signals through CD28, decreased cytokine production, and initiating cell-cycle arrest (8–13). The net outcome of CD8 T cells experiencing continued PD-1 signal transduction is a T cell state often referred to as exhaustion (14). As such, the PD-1 pathway has proven to be an effective target for cancer immunotherapies where blocking PD-1/PD-1 ligand interactions reinvigorates CD8 T cells to combat tumors (15–18).

Several studies have identified numerous cis- and trans-regulatory elements that control Pdcd1 expression (19–21). In response to TCR stimulation, NFATC1 binds to an element upstream of the transcription start site referred to as Conserved Region-C (CR-C) (19–22). Induction of the Pdcd1 locus is marked by changes in histone posttranslational modifications at key cis-regulatory elements, as well as changes in DNA methylation, with the known activities of these modifications each contributing to the regulation of the locus (20, 23, 24). Indeed, TCR stimulation of murine CD8 T cells is marked by loss of DNA methylation at CpGs, increased H3K27 acetylation (ac) and H3K9ac at CR-B (another cis-regulatory element close to the promoter) and CR-C, and enhanced formation of long-range interactions between distal regulatory regions and the Pdcd1 promoter (19, 23, 25, 26). In contrast, the transcription factor B Lymphocyte Maturation Protein 1 (BLIMP-1), encoded by Prdm1, has been shown to repress Pdcd1 expression through the recruitment of the chromatin remodeling enzyme Lysine-Specific Demethylase 1a. Lysine-Specific Demethylase 1a reduces Pdcd1 expression through the decommissioning of enhancer and promoter elements in the Pdcd1 gene (26, 27).

Cytokines also influence the expression of Pdcd1 in CD8 T cells. IL-6 and IL-12 signaling through STAT3 and STAT4, respectively, have been shown to enhance TCR-mediated induction of murine Pdcd1 (19, 28, 29). Signaling through these cytokines leads to JAK-dependent phosphorylation of their respective STAT factors, initiating STAT dimerization and translocation into the nucleus, where they can augment gene expression (30–32). At the Pdcd1 locus, STAT3 has been shown to associate directly with key cis-regulatory elements, including the two enhancer elements located at −3.7 and +17.1 kb from the transcription start site (19). Although such data begin to describe an IL-6–specific means of Pdcd1 regulation, questions remain regarding the exact, CD8 T cell–specific, mechanisms through which IL-6 promotes Pdcd1 expression.

This study focused on expanding the understanding of IL-6/STAT3–dependent regulation of Pdcd1 in murine CD8 T cells. Treatment of primary CD8 T cells with IL-6 led to enhanced and prolonged expression of Pdcd1, Stat3, and Prdm1. Furthermore, IL-6 signaling drove STAT3 phosphorylation/activation and subsequent binding to key Pdcd1 regulatory elements. STAT3 binding prevented BLIMP-1 from inducing a repressive epigenetic state while simultaneously promoting increased histone acetylation. Analysis of mice lacking STAT3 in activated CD8 T cells revealed that IL-6 enhancement of Pdcd1 expression was dependent on STAT3 activities. Infection of STAT3-deficient mice with lymphocytic choriomeningitis virus (LCMV) clone 13 revealed a role of STAT3 in the induction of PD-1 expression in vivo. LCMV-derived CD8 T cells were enriched for STAT3 at several elements within the Pdcd1 locus. Collectively, these findings expand on IL-6–dependent regulation of Pdcd1 expression, detailing a requirement for STAT3 in optimal PD-1 expression in vivo.

Murine EL4 cells were cultured in RPMI 1640 containing 5% FBS (Sigma-Aldrich), 5% bovine calf serum (HyClone), 100 U/ml penicillin/streptomycin, 1.0 mM sodium pyruvate, 10 mM HEPES, and 4.5 mg/ml glucose. Primary murine CD8 T cells were isolated from spleens of C57BL/6 mice using a negative selection method with the CD8a⁺ T Cell Isolation kit II (Miltenyi Biotec) according to the manufacturer’s protocol. Newly isolated primary CD8 T cells were cultured in the same media as EL4 cells. Anti-CD3/CD28 beads (Invitrogen) were directly added to the media at a bead-to-cell ratio of 1:1 to activate the CD8 T cells. Cells were treated with 20 ng/ml IL-6 (Miltenyi Biotec) or 20 ng/ml IL-10 (Miltenyi Biotec).

C57BL/6 wild-type (WT), B6.129S1 (Stat3^fl/fl), and B6;FVB-Tg1Jcb/J (GzmbCre) mice were purchased from Jackson Laboratories. STAT3 conditional knockout (KO; cKO) mice were generated by breeding Stat3^fl/fl mice with GzmbCre mice to delete STAT3 in activated T cells as previously described (26, 33). These are referred to as KO mice in this article. For some experiments, P14 mice were used (generously provided by Dr. R. Ahmed’s laboratory, Emory University). P14 mice contain a TCR transgene for LCMV antigenic peptide GP33. For these experiments, 10,000 P14 splenocytes were adoptively transferred into C57BL/6 Thy1.2 mice via tail-vein injection. Transferred cells were isolated from spleens of Thy1.2 hosts by positive magnetic selection kit on Thy1.1 (Miltenyi Biotec). Purity of isolated cells was confirmed by flow cytometry for Thy1.1 expression. In other experiments, WT and STAT3 KO mice were used as indicated. For all experiments, genotypes were confirmed by PCR. Mice used in this study were maintained and manipulated in compliance with the protocols approved by Emory University Institutional Animal Care and Use Committee.

Chromatin immunoprecipitation (ChIP) assays were performed as previously described (26). In brief, except where noted, 1–4 × 10⁷ cells were cross-linked in 1% formaldehyde for 15 min and then subjected to sonication to shear chromatin. A total of 10 µg of chromatin was used for each immunoprecipitation. Chromatin was incubated with the indicated Abs at 4°C overnight, then magnetic protein A or protein G beads (Invitrogen) were added to the sample and incubated at 4°C for 2 h to isolate the chromatin–Ab complexes. Abs used for immunoprecipitation are listed in Supplemental Table I. Nonimmune rabbit IgG was used as a nonspecific control for all the rabbit Abs. The mouse mAb anti-hemagglutinin (12CA5), which recognizes a short peptide from the influenza hemagglutinin protein, was used as a nonspecific control for the mouse monoclonal anti-NFATC1 Ab. The immunoprecipitated chromatin was eluted and incubated at 65°C overnight to reverse the protein–DNA cross-link; then chromatin DNA was purified and quantified by real-time PCR using a standard curve from sonicated murine genomic DNA. For H3K27ac ChIP on non-P14, LCMV-sorted CD8 T cells, 2–5 × 10⁵ cells were used per immunoprecipitation. Because of the low cell input, protein G beads were prebound with H3K27ac Ab overnight before introducing sheared chromatin for an additional 24-h incubation. Each ChIP assay was performed with chromatin purified from at least three independent experiments and represented relative to total input.

RNA was isolated from primary murine CD8 T cells using the RNeasy Mini Kit (74106; Qiagen) according to the manufacturer’s instructions. Template DNA was digested with DNase for 30 min at 37°C. cDNA was generated using the Superscript II reverse transcriptase (18064-14; Life Technologies). At least three independent RNA preparations were used for real-time PCR analysis using site-specific primers for Pdcd1, Prdm1, and Stat3. All values were normalized to 18s rRNA. Primer sequences are found in Supplemental Table I.

Cells were lysed on ice for 30 min in radioimmunoprecipitation buffer (150 mmol/l NaCl, 50 mmol/l Tris [pH 7.4], 1% Nonidet P-40, and 0.5% Na-deoxycholate) with freshly added protease and phosphatase inhibitors and spun down at 4°C to remove debris. Protein concentrations were determined using Bradford protein assay (Bio-Rad). Cell lysates were resolved on 10% SDS-PAGE and transferred to polyvinylidene difluoride membranes. Abs used for blotting were listed in Supplemental Table I. Protein band signals were detected with the ECL detection kit (ThermoFisher).

Mice were infected with 2 × 10⁶ PFUs of LCMV clone 13 i.v. as previously described to induce high PD-1 expression by CD8 T cells (27). All viral stocks were graciously provided by Dr. R. Ahmed (Emory University). WT and STAT3 cKO mice were infected in a blinded fashion as to ensure equal virus administration across genotype. LCMV clone 13 viral titers were measured as described previously (34). Mice were bled 8 d postinfection, plasma collected, RNA isolated, and cDNA generated as described earlier. Quantitative PCR was performed using LCMV-specific primers (Supplemental Table I). Titers were calculated using a standard dilution series of LCMV stock.

Primary murine CD8 T cells were isolated from spleens of mice using a negative selection method with the CD8a⁺ T Cell Isolation kit II (Miltenyi Biotec) according to the manufacturer’s protocol. Cells were resuspended in 100 µl of FACS buffer (PBS, 2 mM EDTA, and 1% BSA), stained with fluorophore-conjugated Abs for 30 min at room temperature, washed twice with 1 ml FACS, and resuspended in 300 µl FACS buffer. The fluorophore Abs used are detailed in Supplemental Table I. For ex vivo CD8 T cell experiments PD1-PerCp-Cy5.5 (RMPI-30), IL6Ra-allophycocyanin (D7715A7), IL-10R-PE (1B1.3a), Rat IgG2b,k-PerCp-Cy5.5 (RTK4530), rat IgG2b,k-allophycocyanin (RTK4530), and rat IgG1,k-PE (RTK2071) Abs were used. For LCMV experiments CD8-FITC (2.43), CD62lg-PerCp-Cy5.5 (MEL-14), CD44-AF700 (IM7), PD1-PE (RMPI-30), CD11b-allophycocyanin-Cy7 (M1/70), B220-APCC7 (RA3-6B2), F4/80-allophycocyanin-Cy7 (BM8), Thy1.1-Pacific Blue (OX-7), and a fixable viability dye (Ghost Dye-v510) were used. For indicated experiments, GP33-allophycocyanin (H-2D9b) LCMV tetramer staining was conducted before introducing extracellular epitope Abs. The visual gating strategy for all analyses is depicted in Supplemental Fig. 1: lymphocyte population determined by forward scatter-area and side scatter-area, single cells on forward scatter-area and forward scatter-height, viable cells on Ghost Dye v510, removal of non–T cell lineages using a dump gate on CD11b⁻B220⁻F4/80⁻, and finally gating on CD8-expressing cells. Flow cytometry was conducted with a LSRFortessa (BD Biosciences) or sorted with a FACSAria II (BD Biosciences) using BD FACSDiva software v8.0 (BD Biosciences). All data were analyzed, and figures were generated using FlowJo v10.6.2.

Recent advances in our understanding of Pdcd1 regulation highlight the ability of extracellular cytokine signals, including IL-6, to modulate Pdcd1 expression (20). To determine the temporal kinetics of cytokine treatment on Pdcd1 expression, we conducted a time course on magnetically enriched primary murine CD8 T cells stimulated with anti-CD3/CD28 beads. Isolated CD8 T cells were treated with IL-6 or another STAT3 signaling cytokine, IL-10, for 5 d, and Pdcd1 transcript levels were measured (Fig. 1A). IL-6 treatment led to a significant increase in Pdcd1 mRNA expression over stimulation alone at each point in the time course. Previously it was shown that Pdcd1 mRNA decreases to baseline over time because of the induction of BLIMP-1 (26). Intriguingly, the addition of IL-6 to the cultures also increased Prdm1 (encodes BLIMP-1) transcript levels (Fig. 1A). Furthermore, IL-6 treatment enhanced Stat3 transcript levels above stimulation alone (Fig. 1B). In contrast, IL-10 treatment did not induce Pdcd1 mRNA expression. To assess IL-6 and IL-10 sensitivity of naive murine CD8 T cells, we measured surface expression of IL-6Rα and IL-10R (Fig. 1C, Supplemental Fig. 1A). Naive CD8 T cells express high levels of IL-6Ra, supporting the IL-6–specific influence on Pdcd1 and Prdm1 transcription. In contrast, IL-10R was not detected above background. Because the activities of STAT factors are largely dependent on phosphorylation-marked activation, the phosphorylation status of STAT3 at Y705 after IL-6 treatment was measured (Fig. 1D). Indeed, p-STAT3 was dependent on IL-6 and detected over the 5 d of culture. The observed IL-6–dependent p-STAT3 activation correlated with increased Pdcd1 transcript, as well as increased PD-1 surface expression (Fig. 1E). Notably, although Pdcd1 transcript increased rapidly after 1 d in the presence of IL-6, an IL-6–specific increase in PD-1 surface levels was not detected until 3 d after IL-6 administration. Taken together, these data implicate IL-6 as a potent and stable driver of Pdcd1, Prdm1, and STAT3 activation.

Building on a previous report indicating that STAT3 downstream of IL-6 directly associates with Pdcd1 regulatory regions, we performed a more comprehensive analysis of STAT3 occupation at the Pdcd1 locus (19). Specifically, STAT3 enrichment was analyzed at an expanded number of Pdcd1 regulatory elements, including CR-B, CR-C, −2.7 kb, −3.7 kb, and +17.1 kb (Fig. 2A). Such regions were selected through the analysis of previously published DNA hypersensitivity data at the Pdcd1 locus and the presence of STAT factor motifs within each region (19, 20, 27). To determine the stability of STAT3 binding at each of these regions, we determined enrichment at both day 1 and day 5 after IL-6 treatment in primary murine CD8 T cells by ChIP. Consistent with previous findings (19), STAT3 was found to associate with the Pdcd1 locus only in the presence of IL-6, occurring at −3.7, −2.7, and +17.1 (Fig. 2B). STAT3 remained significantly enriched at each of these elements 5 d after IL-6 treatment. CR-C, which binds NFATC1, did not bind STAT3, demonstrating cis-element specificity.

The TCR/NFATC1 signaling axis has been established as a potent inducer of Pdcd1 and is a required prerequisite for IL-6–mediated enhancement of Pdcd1 expression (19). To determine whether IL-6 augmented the TCR signaling pathway, we assessed occupancy of NFATC1 at CR-C in the presence of IL-6 (Fig. 2C). IL-6 treatment had no effect on NFATC1 binding to CR-C 1 d after TCR stimulation. In addition, IL-6 did not prolong the occupancy of NFATC1 at day 5, indicating a TCR/NFATC1-independent mechanism of IL-6–specific Pdcd1 enhancement. Interestingly, the observation that Pdcd1 expression is induced despite a similar increase in Prdm1 transcript on IL-6 treatment (Fig. 1A) suggests that IL-6 signaling may be capable of overcoming BLIMP-1–mediated transcriptional silencing. Surprisingly, irrespective of IL-6 treatment, BLIMP-1 was bound to its site within the Pdcd1 locus at day 5 (Fig. 2D). These data suggest that IL-6–dependent STAT3 enrichment at the Pdcd1 locus is dominant to the repressive actions of BLIMP-1.

BLIMP-1 is known to promote a repressive epigenetic landscape, in part through influencing the posttranslational histone modifications at key Pdcd1 regulatory regions (26, 27). Thus, one way that IL-6 may bypass BLIMP-1 silencing is through influencing the ability of BLIMP-1 to alter the composition of histone modifications at key Pdcd1 regulatory elements. To test this hypothesis, we assayed the effect of IL-6 on active (H3K9ac, H3K27ac), promoter (H3K4me3), and repressive (H3K9me3, H3K27me3) chromatin marks within Pdcd1 regulatory elements by ChIP (Fig. 3A). Because BLIMP-1’s association with the Pdcd1 locus is delayed, and not detected until day 5 (Fig. 2D), histone modifications were assayed at both day 1 and day 5. TCR stimulation induced active histone modifications (H3K9ac, H3K27ac, and H3K4me3, respectively) at day 1 at CR-B and CR-C (Fig. 3B). These modifications were lost at day 5, consistent with Pdcd1 mRNA expression presented earlier. In the presence of IL-6, the earlier active histone modifications were now detected at day 5 at −3.7, −2.7, CR-C, CR-B, and +17.1 and enhanced at CR-B and CR-C compared with stimulation alone. Conversely, and consistent with the binding of BLIMP-1, repressive modifications (H3K9me3 and H3K27me3) appear at day 5 but were diminished in their enrichment in cells cultured with IL-6 (Fig. 3C). Collectively, these data indicate that IL-6/STAT3 induces Pdcd1 expression, at least in part, by impeding activities attributed to BLIMP-1–dependent epigenetic silencing (26, 27, 35).

To establish whether STAT3 was essential for IL-6 induction of Pdcd1, we bred Stat3^fl/flGzmbCre cKO mice. This genotype results in Gzmb-dependent expression of Cre recombinase, resulting in a deletion of Stat3 within the activated CD8 T cell compartment (26, 33). Primary naive CD8 T cells were isolated from both WT and KO mice, stimulated with anti-CD3/CD28 beads with and without IL-6, and cultured for 4 d (Fig. 4A). As earlier, WT cells treated with IL-6 upregulated Pdcd1; however, STAT3-deficient KO cells failed to enhance and prolong Pdcd1 transcript levels relative to stimulation alone (Fig. 4B). As expected, Stat3 transcripts were drastically reduced in the KO mice after anti-CD3/CD28 beads stimulation. These data place STAT3 as a key mediator of IL-6–dependent Pdcd1 induction in vitro.

The LCMV clone 13 infection model has been shown to elicit high levels of PD-1 surface protein and Pdcd1 mRNA transcripts in CD8 T cells responding to the infection (7, 36–38). To determine whether STAT3 plays a role in governing CD8 immune responses in an in vivo system, we infected WT and KO mice with LCMV clone 13, and after 8 d CD8 T cells were magnetically enriched and phenotyped by flow cytometry (Fig. 5A, Supplemental Fig. 1B). Relative to uninfected controls, the infected WT and KO mice displayed elevated activated CD8 T cell frequencies, although there was no discernable difference in frequency of activated T cells between the two genotypes (Fig. 5B). However, significantly lower LCMV virus titers were detected in the peripheral blood of STAT3 KO mice relative to WT mice 8 d postinfection, suggesting a role for STAT3 in CD8-dependent viral clearance (Fig. 5C).

The observed enhanced viral clearance in the KO mice may be caused in part by impaired PD-1 expression by STAT3-deficient CD8 T cells, because upregulation of PD-1 on CD8 T cells in response to LCMV clone 13 infection is a well-known mechanism to disrupt CD8 T cell functionality (8–13). Intriguingly, LCMV-infected KO mice exhibited a modest yet consistently lower frequency of PD-1^Hi–activated CD8 T cells compared with WT controls (Fig. 5D, Supplemental Fig. 2). Notably, the loss of STAT3 in an in vivo context had a diminished effect on PD-1 expression relative to that observed in an in vitro cell culture environment.

As established in the in vitro system, STAT3 induces PD-1 expression in part through promoting active chromatin marks at the Pdcd1 locus. To determine whether the diminished PD-1 expression by STAT3 KO CD8 T cells was accompanied by altered histone modifications, we performed ChIP assays for H3K27ac on sorted activated CD8 T cells from WT and KO mice 8 d after LCMV infection (Fig. 5E, 5F). Relative to naive CD8 T cells isolated from uninfected mice, both WT and KO activated T cells displayed increased H3K27ac throughout the Pdcd1 locus. Correlating with decreased PD-1 expression, KO activated CD8 T cells displayed reduced enrichment for H3K27ac at the CR-B and +17.1 regulatory elements. Collectively, these data depict a role for STAT3 in promoting an epigenetic landscape conducive for heightened PD-1 expression in LCMV-activated CD8 T cell populations.

Members of the STAT family are known to have similar binding sequences, often playing redundant roles (39). The enrichment of various STAT factors was assayed in Ag-specific CD8 T cells generated in response to LCMV clone 13 infection to gain insight into the specificity and stability of STAT3 in driving Pdcd1 expression in vivo (Fig. 6A). Splenocytes from Thy1.1⁺ P14 mice, which have been genetically engineered to have TCR specific for LCMV clone13 (GP33), were adoptively transferred into Thy1.2⁺ WT mice to achieve robust numbers of Ag-specific CD8 T cells. Splenic CD8 T cells were isolated from the recipient mice 28 d after LCMV infection and subsequently enriched for Thy1.1. Flow cytometric analysis of enriched cells revealed a high frequency of Thy1.1 and GP33 Ag-specific cells (Supplemental Fig. 1C). Intriguingly, ChIP analysis of the isolated CD8⁺ Thy1.1⁺ cells, which can be recovered in high numbers, revealed STAT3 to be significantly enriched at Pdcd1 regulatory regions, whereas STAT1, STAT4, STAT5, and STAT6 were absent from the locus (Fig. 6B). Specifically, STAT3 was found to associate with the −3.7 and +17.1 kb regulatory elements, consistent with the in vitro findings (Fig. 2B). Thus, STAT3 binds the Pdcd1 locus in Ag-specific CD8 T cells of mice responding to LCMV.

This study places the IL-6/STAT3 signaling pathway as a regulator of Pdcd1 expression in murine CD8 T cells. Treatment with IL-6 resulted in activation of STAT3 and prolonged and enhanced expression of Pdcd1. Activated STAT3 associated with key Pdcd1 cis-regulatory elements, −3.7, −2.7, and +17.1, yet had no effect on BLIMP-1 association at its binding site within the locus. Moreover, IL-6/STAT3 was able to counter BLIMP-1–mediated epigenetic silencing of Pdcd1 expression, both by promoting/prolonging active histone modifications and preventing/reducing the application of repressive chromatin modifications. Furthermore, CD8 T cells lacking STAT3 displayed diminished Pdcd1 expression in vitro and surface PD-1 expression in an in vivo LCMV clone 13 infection model. This observed loss of PD-1 by STAT3 KO CD8 T cells in vivo was associated with reduced H3K27ac of key Pdcd1 regulatory elements. Importantly, STAT3 was the only STAT family member bound to Pdcd1 cis-regulatory elements in Ag-specific CD8 T cells after LCMV clone 13 infection. Together, this study provides mechanistic insight into the observed TCR/IL-6 enhancement of Pdcd1 expression, placing STAT3 as a key molecular regulator of this process.

It is intriguing that Pdcd1 transcripts were observed after 1 d of IL-6 treatment ex vivo, but protein expression did not appear until day 3, a time point in which untreated cells showed a decrease in surface PD-1 expression. This suggests that the system may be more highly regulated than through the traditional central dogma. Previously, STAT and NFAT factors were shown to bind and associate with the Pdcd1 locus downstream of cytokines and TCR, respectively, correlating with increased transcription (19, 21). However, the longevity of these transcription factor–chromatin associations and subsequent mechanisms of transcriptional induction remained unclear. In this study, we established that IL-6 signaling resulted in STAT3 enrichment at Pdcd1 enhancer elements (−3.7, −2.7, and +17.1) but had no effect on NFATC1 association at CR-C. Moreover, at an extended 5-d ex vivo time point, NFATC1 binding at CR-C was lost, an event previously shown to be due to BLIMP-1 binding (26), whereas STAT3 binding remained. This suggests a model wherein initial signals via TCR/NFATC1 result in activation of the locus at CR-C/CR-B (40) followed by a secondary event in the form of IL-6 to exacerbate expression through association of STAT3 to cis-regulatory elements. Indeed, stimulation of CD8 cells was shown to facilitate increased accessibility at the −3.7, −2.7, and +17.1 regions, paving the way for STAT3 and/or other factors to enhance expression (19). In concordance with the transcriptional events, active histone modifications were observed at Pdcd1 regulatory elements; importantly, in the presence of activated STAT3, the expected repressive modifications attributed to the binding of BLIMP-1 (26, 27, 35) were diminished or prevented from being deposited. Thus, the positive/active role of STAT3 is dominant to BLIMP-1 in controlling the epigenetic landscape of the region.

Insight into how STAT3 may circumvent BLIMP-1 activities is likely derived from its potential interactions with other transcriptional activators, such as ZIPK, Y14, STAP-2, and AIOLOS (41–44). Furthermore, and potentially of higher relevance to the observed Pdcd1 induction, STAT3 binding results in chromatin acetylation through its partnership with the histone acetyltransferase P300 and the BRG1 subunit of the SWI/SNF chromatin remodeling complex (45–47). Some studies have indicated that STAT3-dependent gene regulation requires direct recruitment of histone acetyltransferases (48, 49). As such, the detected increase in H3K9ac and H3K27ac at Pdcd1 cis-regulatory elements on IL-6 treatment in this study could be a result of P300 recruitment or stabilization of P300 to the locus by STAT3, as well as the recruitment of additional regulatory elements to enhance expression.

Although mice lacking STAT3 exhibited a lower frequency of PD1^Hi CD8 T cells in response to LCMV infection, the observed reduction was not as robust as seen in ex vivo cultures. In addition, we noted that the lack of STAT3 did not significantly alter the H3K27ac levels at −3.7 and −2.7 kb, despite the fact that we can observe STAT3 binding in ex vivo–derived CD8 T cells in response to IL-6 treatment. Although this may be caused by the in vivo LCMV infection model, or the fact that low cell numbers were available for the ChIP assay, other explanations are also possible. For example, physiologically, the extracellular cytokine milieu throughout the course of active infection is highly complex, consisting of a plethora of proinflammatory cytokines and other signaling molecules with the downstream transcriptional program of CD8 T cells in vivo being the culmination of all present signaling molecules. Thus, other factors may be able to compensate for loss of STAT3 to drive Pdcd1 expression. Notably, other members of the STAT factor family have been shown to bind similar motifs, specifically IFN-γ–activated sequences, with site-specific coenrichment of STAT factors having been previously demonstrated (29, 50–52). Although in the WT setting only STAT3 enrichment at Pdcd1 regulatory regions was observed in vivo, it is possible that, on loss of STAT3, other factors may compensate.

IL-6/STAT3 signaling has been shown to play a role in the pathogenesis of human and murine viral infections, including LCMV clone 13 infection (53–55). Recent studies have begun to connect elevated serum IL-6 to increased disease severity in severe acute respiratory syndrome coronavirus 2–infected individuals (56–58). Notably, coronavirus disease 2019 patients in the intensive care unit (ICU) exhibited elevated serum IL-6 levels relative to a non-ICU control group (59). Moreover, CD8 T cells from the ICU cohort exhibited an exhaustive phenotype characterized by elevated PD-1 expression (59). Intriguingly, therapeutic treatment blocking IL-6/IL-6Rα has been used in a subset of severe coronavirus disease 2019 patients with some success (60). This opens the possibility that perhaps IL-6 in these infections is augmenting PD-1 expression on virus-responding T cells, resulting in reduced activity of these cells.

Collectively, this study provides insight into the regulatory mechanisms and requirements for IL-6/STAT3–dependent Pdcd1 expression in CD8 T cells. Remarkably, initiation of the IL-6/STAT3 pathway functions to counteract BLIMP-1–driven formation of an epigenetically silenced chromatin state within the Pdcd1 locus. Ultimately, acquisition of insight into Pdcd1 regulatory pathways has proven clinical implications, effectively providing the basis for future therapies aimed at manipulating PD-1 expression.

We thank the members of the Boss and Scharer laboratories for scientific input and careful critiquing of the manuscript, and Royce Butler for orchestrating mouse care.

The authors have no financial conflicts of interest.