Maintenance of PD‐1 on brain‐resident memory CD8 T cells is antigen independent

Abstract

Infection of the central nervous system (CNS) by murine polyomavirus (MuPyV), a persistent natural mouse pathogen, establishes brain‐resident memory CD8 T cells (bT_RM) that uniformly and chronically express programmed cell death protein 1 (PD‐1) irrespective of the expression of α_E integrin CD103, a T_RM cell marker. In contrast, memory antiviral CD8 T cells in the spleen are PD‐1⁻, despite viral loads being similar in both the brain and spleen during persistent infection. Repetitive antigen engagement is central to sustained PD‐1 expression by T cells in chronic viral infections; however, recent evidence indicates that expression of inhibitory receptors, including PD‐1, is part of the T_RM differentiation program. Here we asked whether PD‐1 expression by CD8 bT_RM cells during persistent MuPyV encephalitis is antigen dependent. By transferring MuPyV‐specific CD8 bT_RM cells into the brains of naive mice and mice infected with cognate epitope‐sufficient and ‐deficient MuPyVs, we demonstrate that antigen and inflammation are dispensable for PD‐1 maintenance. In vitro and direct ex vivo analyses indicate that CD103⁻ MuPyV‐specific CD8 bT_RM retain functional competence. We further show that the Pdcd‐1 promoter of anti‐MuPyV bT_RM cells is epigenetically fixed in a demethylated state in the brain. In contrast, the PD‐1 promoter of splenic antiviral memory CD8 T cells undergoes remethylation after being demethylated during acute infection. These data show that PD‐1 expression is an intrinsic property of brain T_RM cells in a persistent CNS viral infection.

Programmed cell death protein 1 (PD‐1) expression has been proposed to constitute a facet of the resident memory CD8 T cells (T_RM) differentiation program to prevent inadvertent deployment of poised mRNAs for effector molecules. ¹ In chronic lymphocytic choriomeningitis virus (LCMV) infection, T‐cell receptor (TCR) signaling upregulates PD‐1 expression at the effector stage of the splenic CD8 T cell response, with sustained PD‐1 driving differentiation of exhausted T (T_EX) cells to prevent immunopathology. ^{2 , 3} The state of PD‐1 expression and its dependence on antigen by tissue T_RM during persistent viral infection remains to be defined. For example, CD8 brain T_RM (bT_RM) cells from mice with acutely resolved vesicular stomatitis virus (VSV) encephalitis express PD‐1 transcripts but not PD‐1 receptors, whereas bT_RMs from mice persistently infected with mouse cytomegalovirus are PD‐1⁺. ^{4 , 5 , 6} This discrepancy in PD‐1 expression by bT_RM cells raised the question whether antigen and/or inflammation is involved in maintenance of PD‐1 expression by bT_RM cells during central nervous system (CNS) infection. Tissue‐intrinsic factors are also dominant determinants of the dependence on antigen for CD8 T_RM cell generation and/or maintenance. Antigen is required for T_RM cell formation and CD103 upregulation in the brain and dorsal root ganglion ^{5 , 7 , 8} but not in the skin, small intestine, female reproductive tract and salivary glands. ^{7 , 9 , 10 , 11 , 12} The role of antigen in maintenance of the expression of PD‐1 and CD103 by CD8 T_RM cells in the brain remains to be determined.

The PD‐1 promoter of virus‐specific CD8 T cells undergoes dynamic epigenetic reprogramming during development of memory T cells and T_EX cells. ¹³ In acutely resolved LCMV Armstrong infection, virus clearance was associated with remethylation of the Pdcd‐1 promoter and loss of PD‐1 expression; however, in the high‐level chronic LCMV clone 13 infection model, the Pdcd‐1 promoter remained unmethylated in T_EX cells even after virus levels fell below detection. ^{13 , 14} Notably, these epigenetic analyses were only performed on splenic LCMV‐specific CD8 T cells in an infection where PD‐1 is expressed by antiviral CD8 T cells in all nonlymphoid organs. ¹⁵ This led us to investigate the epigenetic programming of bT_RM cells during persistent viral encephalitis.

Murine polyomavirus (MuPyV) is a natural mouse pathogen that establishes a low‐level persistent infection. CNS infection with MuPyV yields a stable population of virus‐specific bT_RM cells. ¹⁶ Here we show that, during persistent MuPyV infection, PD‐1 is expressed by bT_RM cells but not splenic memory anti‐MuPyV CD8 T cells, despite virus loads being similar in both organs, suggesting dissociation between the viral load and PD‐1 expression. We further show that maintenance of PD‐1 expression by bT_RM cells is independent of cognate viral antigen and inflammation. As seen for splenic virus‐specific CD8 T cells in chronic LCMV infection, the pdcd‐1 promoter of bT_RM cells from MuPyV‐infected mice remains demethylated. However, the PD‐1 locus in splenic anti‐MuPyV CD8 T cells undergoes partial remethylation. Collectively, these findings indicate that PD‐1 expression is part of the developmental program of bT_RM cells to a persistent CNS viral infection.

Results and discussion

MuPyV‐specific bT_RM cells express PD‐1 during persistent infection

Naive B6 mice received a ‘physiological’ number (200 cells per mouse) of Thy1.1‐congenic TCR‐I cells and were inoculated intracerebrally (i.c.) the next day with MuPyV.LT206 virus. At day 9 postinfection (p.i.), the magnitude of the endogenous LT206‐specific CD8 response in the brain, kidney and spleen was similar to that of the donor TCR‐I cells (Supplementary Figure S1). Both the endogenous D^bLT206‐specific CD8 T cells and the TCR‐I cells in acutely infected mice expressed PD‐1 and CD69, with those in the brain having the highest level of expression (Figure 1a); notably, virus levels in these organs were similar at day 9 and also at day 30 p.i. (Figure 1b). In addition to CD69, CD103 (α_E integrin) is often used identify T_RM cells. ¹⁷ At day 9 p.i., D^bLT206‐specific CD8 T cells in both lymphoid and nonlymphoid organs did not express CD103; however, by day 30 p.i., a fraction of CD103⁺ endogenous and donor CD8 T cells were seen in the brain but not in the kidney or spleen. Despite a decline in virus levels by day 30 p.i. (Figure 1b), PD‐1 expression was sustained on both endogenous D^bLT206‐specific CD8 T cells and TCR‐I cells in the brain and kidney but not those in the spleen (Figure 1a), as was CD69 (Figure 1a). Analysis of PD‐1 transcript levels in TCR‐I cells from the spleen at days 8 and 30 p.i. showed that splenic memory TCR‐I cells expressed lower PD‐1 mRNA levels than those from the spleens of acutely infected mice, albeit this difference was not statistically significant (Figure 1c). Discordance between PD‐1 mRNA levels and surface expression by CD8 T_RM cells was similarly observed in the VSV acute encephalitis infection model. ⁴ To determine whether PD‐1 functionally inhibited MuPyV‐specific CD8 T cells, we assayed intracellular interferon (IFN)‐γ production by TCR‐I cells after stimulation by LT206 peptide‐pulsed PD‐L1⁺ DC2.4 cells in the presence of a PD‐L1 blocking monoclonal antibody (Ab). As shown in Figure 1d, PD‐L1 blockade resulted in a significant increase in frequency of IFN‐γ expressing TCR‐I cells from brains, kidney and spleen of mice at day 9 p.i. At day 30 p.i., PD‐1 blockade increased the frequency of IFN‐γ⁺ cells only of brain‐resident TCR‐I cells, which expressed the highest levels of PD‐1 among T cells from all the organs examined.

Figure 1.

Expression and function of PD‐1 on MuPyV‐specific CD8 T cells. (a) Surface expression of PD‐1, CD69 and CD103 by D^bLT206‐specific endogenous and Thy1.1 TCR‐I CD8 T cells from the indicated organs at days 9 and 30 p.i. (left panels), and frequency of PD‐1^hi cells at day 30 p.i. on endogenous D^bLT206‐specific CD8 T cells and TCR‐I cells (right panel). (b) Real‐time PCR assay of MuPyV genome copy numbers from total genomic DNA isolated from the brain, spleen and kidney at days 9 and 30 p.i. One‐way analysis of variance, Kruskal–Willis test with Dunn's multiple comparisons test was used to determine statistical significance. (c) Real‐time PCR analysis for the expression of PD‐1 on cDNA prepared from FACS‐purified GFP⁺ TCR‐I cells from spleens of naive, day 8‐ and day 30‐infected mice. (d) Mononuclear cells from the brain, kidney and spleen of TCR‐I‐recipient mice at days 9 and 30 p.i. were stimulated with LT206 peptide‐pulsed DC2.4 cells in the presence of control rat IgG () or PD‐L1 blocking Ab (). A nonparametric paired Student's t‐test with Wilcoxon matched‐pairs signed rank test was used to determine statistical significance between control rat IgG and PD‐L1 blocking Ab‐treated samples for each organ at a single time point. Each data point represents each mouse in that group. Data are from two independent experiments with 4–5 mice each group. Values are mean±s.d.; ∗∗P0.01, not significant (ns) P>0.05.

Maintenance of PD‐1 expression by brain T_RM cells is antigen and inflammation independent

We found that PD‐1^hi expression by MuPyV‐specific b_TRMs cells was a stable phenotype and dissociated from virus infection levels (Figure 1b); that is, splenic antiviral memory CD8 T cells were PD‐1⁻ but spleen and brain virus loads were similar. In contrast, PD‐1 expression by memory, TCR transgenic P14 CD8 T cells from LCMV clone13‐infected mice declines following transfer to LCMV‐immune mice or mice infected with a epitope‐null mutant LCMV. ¹⁴ To directly ask whether cognate antigen and/or virus‐associated inflammation was required for PD‐1 expression by brain T_RM cells, we sorted GFP⁺ TCR‐I cells from the brains of B6 mice infected 30 days earlier with MuPyV.LT206 and reintroduced i.c. in B6 at day 30 p.i. with MuPyV.LT206 (Group1: infection, cognate Ag), MuPyV.A2 virus (Group 2: infection, no cognate Ag) or into uninfected mice (Group 3) (Figure 2a). Seven days posttransfer, CD8 T cells from the brains were co‐stained with D^bLT206 or D^bLT359 tetramers (Figure 2b). As shown in Figure 2c, PD‐1 expression was similar among donor TCR‐I T_RM cells in the brains of mice regardless of expression of cognate epitope and/or virus‐associated inflammation. CD69 and CD103 expression by the TCR‐I cells in each recipient cohort was also similar and unchanged from the original donor TCR‐I cells in Group 1 (Figure 2c). High expression of PD‐1, CD103 and CD69 by the donor TCR‐I T_RM cells in naive recipients was maintained up to day 25 posttransfer (Supplementary Figure S1B). The possibility of MuPyV.LT206 virus carryover with the donor TCR‐I cells is unlikely given the absence of detectable MuPyV DNA by quantitative PCR in the brains of uninfected recipients (Supplementary Figure S2A) and by the absence of D^bLT206 tetramer binding the endogenous CD8 T cells either in the brain (Figure 2b) or in the spleen (Supplementary Figure S2B) of uninfected recipient mice. Naive TCR‐I cells transferred into mice infected by MuPyV.A2 failed to proliferate and did not upregulate CD44, indicating that maintenance of PD‐1 expression by TCR‐I T_RM cells in MuPyV.A2‐infected recipients could not be attributed to cross‐recognition of the D^bLT359 epitope (Supplementary Figure S2C). These data suggest that TCR stimulation and the inflammatory environment of persistent MuPyV CNS infection are dispensable for maintenance of PD‐1 expression by antiviral brain T_RM cells.

Figure 2.

Analysis of PD‐1, CD103 and CD69 expression on brain T_RM CD8 cells in the absence of cognate epitope and/or inflammation. (a) Experimental setup. (b) Representative dot plots for endogenous D^bLT206‐ or D^bLT359‐specific and transgenic GFP⁺ TCR‐I cells in respective groups of mice. (c) Expression of PD‐1, CD69 and CD103 on endogenous D^bLT206‐ or D^bLT359‐specific and GFP⁺ TCR‐I cells in groups of mice as indicated, represented as gMFI of the population. One‐way analysis of variance, Kruskal–Willis test with Dunn's multiple comparisons test was used to determine statistical significance. (d) Representative histograms showing the expression of CD103 D^bLT206‐ or D^bLT359‐specific and GFP⁺ TCR‐I cells. ‘LT206 in LT206’ indicates CD8 T cells specific to D^bLT206 epitope in Group 1, ‘LT359 in A2’ indicates CD8 T cells specific for D^bLT359 epitope in Group 2, ‘TCR‐I=>LT206/A2/Uninf’ indicates the transferred GFP⁺ TCR‐I cells in Groups 1, 2 and 3, respectively. Data are cumulative from two independent experiments, with 2–3 mice each group. Values are mean±s.d.; ∗P<0.05. (e) Number of GFP⁺ TCR‐I donor cells recovered from the brains of recipient mice. Data are representative from one of the two experiments performed where 8000 GFP⁺ TCR‐I cells were transferred i.c. into the recipient brains.

As shown in Figure 2d, donor TCR‐I bT_RM cells were predominantly CD103⁺ irrespective of the presence of antigen or inflammation in the recipient brains, suggesting preferential survival of the CD103⁺ population. Although fewer donor TCR‐I bT_RM cells were recovered from the brains of naive and MuPyV.A2‐infected recipients than from MuPyV.LT206‐infected mice at day 7 posttransfer (Figure 2e), these CD8 bT_RM cells were found to persist in the brains of naive recipients for 25 days (Supplementary Figure S1). To ask whether PD‐1 expression was an intrinsic property of MuPyV‐specific bT_RM cells, TCR‐I cells were isolated from the brains of mice at day 30 p.i. and transferred intravenously to either naive or MuPyV.LT206 i.c. infected Thy1.2 congenic recipients. No donor cells, however, were detected at day 7 posttransfer in either cohort (data not shown), a finding mirroring that of Wakim et al. ⁵ using the VSV acute encephalitis mouse model and supporting the concept that CNS‐specific factors are required for maintenance of bT_RM cells.

To ask whether the brain environment per se induced PD‐1, memory GFP⁺ TCR‐I cells from the spleen were transferred i.c. into naive or acutely i.c. infected (MuPyV.LT206) mice. We found that the donor splenic memory TCR‐I cells (CD69⁻ PD‐1⁻, Figure 1a) 7 days posttransfer i.c. into naive mice or mice given MuPyV.LT206 virus i.c. 8 days earlier upregulated PD‐1 and CD69 expression but not that of CD103 (Supplementary Figure S1C). Although splenic donor memory cells upregulated PD‐1 expression in both naive and acutely infected recipients, it merits noting that PD‐1 expression is lower in the naive than in the infected mice, consistent with a role for antigen in driving PD‐1 expression. Taken together, these data confirm the ability of the brain environment to upregulate PD‐1 and CD69 on infiltrating MuPyV‐specific CD8 T cells.

CD103⁻ MuPyV‐specific CD8 bT_RM cells are functionally competent

CD8 T_RM cell populations defined by differential CD103 expression have been reported to vary in proliferative potential, transcriptome, antigen dependence and tissue distribution. ^{4 , 5 , 18 , 19} CD103 expression by donor TCR‐I cells, however, was higher than on the endogenous D^bLT359 or D^bLT206 CD8 T cells (Figures 2c and d), a difference that may be linked to the higher survival potential of CD103⁺ cells in the brain. ⁵

We thus asked whether MuPyV‐specific CD103⁻ CD8 T cells infiltrating the brain express phenotypic and functional characteristics of T‐cell exhaustion. Both CD103⁺ and CD103⁻ D^bLT359‐specific CD8 T cells were maintained in the absence of circulating CD8 cells (Figure 3a), suggesting that each of these populations is comprised of T_RM cells. CD103⁺ and CD103⁻ D^bLT359‐specific CD8 T cells also expressed similar levels of PD‐1 and Tim3 (Figure 3b). Of note, the CD103⁻ subset expressed higher TCF‐1 levels and had higher proportion of CXCR5^hi TCF‐1^hi cells compared with the CD103⁺ cells. Elevated TCF‐1 expression as well as the presence of a CXCR5^hi TCF‐1^hi population by memory CD8 T cells in chronic infection models have been reported to exhibit improved functional capability. ^{20 , 21} BLIMP‐1 and Eomesodermin (EOMES) T‐box transcription factors are highly expressed by exhausted CD8 T cells; ^{22 , 23} however, CD103⁻ D^bLT359‐specific CD8 T cells were found to express lower BLIMP‐1 and higher EOMES levels than the CD103⁺ cells. These populations also showed similar Ki67 expression levels, indicating their comparable rates of proliferation during persistent MuPyV CNS infection (Figure 3c). Taken together, the pattern of expression of these molecules indicates that CD103⁻ cells lack a phenotype associated with T‐cell exhaustion. Moreover, CD103⁻ and CD103⁺ LT359 peptide‐stimulated CD8 T cells from the brains of persistent MuPyV‐infected mice exhibited similar frequencies of intracellular IFN‐γ production (Figure 3d). To assess IFN‐γ production directly ex vivo by CD103⁻ and CD103⁺ MuPyV‐specific CD8 T cells in the brains of persistently infected mice, IFN‐γ EYFP reporter mice were inoculated with MuPyV.A2 virus i.c. As shown in Figure 3e, a higher fraction of CD103⁺ than CD103⁻ D^bLT359‐specific CD8 T cells were EYFP⁺, a pattern paralleled by the expression of IRF4, a transcription factor upregulated by TCR stimulation. ²⁴ The discrepancy between similar frequencies of IFN‐γ‐producing CD103⁺ and CD103⁻ bTRM cells following LT359 peptide stimulation vs lower EYFP⁺ and IRF4⁺ by the CD103‐ population may reflect regional differences between the CD103⁺ and CD103⁻ cells in engaging viral antigen‐expressing cells and/or differences in TCR affinity. ¹⁶

Figure 3.

Functional and phenotypic characterization of CD103⁺ and CD103⁻ bT_RM cells. (a) Frequency and number of CD103⁺ and CD103⁻ D^bLT359 tetramer‐specific CD8 T cells from the brains of mice at day 30 p.i. after i.c. inoculation with MuPyV.A2 and treated weekly with anti‐CD8α or control rat IgG from day 10 to day 30 p.i. (b and c) Surface expression of PD‐1, Tim3 and CXCR5 and intracellular expression of TCF‐1, BLIMP1, EOMES and Ki67 on CD103⁺ and CD103⁻ D^bLT359 tetramer‐specific CD8 T cells from the brains of mice day 30 p.i. after i.c. inoculation with MuPyV.A2. (d) Frequency of CD103⁺ and CD103⁻ cells Thy1.1 TCR‐I cells and IFN‐γ‐producing cells upon in vitro LT359 peptide stimulation from the brains of mice at day 30 p.i. after i.c. inoculation with MuPyV.LT206 virus. (e) Direct ex vivo intracellular staining for EYFP and IRF4 on D^bLT359 tetramer‐specific CD8 T cells from the brains of mice at day 30 p.i. after i.c. inoculation with MuPyV.A2. Data are from two independent experiments with 4–5 mice per group. A nonparametric paired Student's t‐test with Mann–Whitney test was used to determine statistical significance. Values are mean±s.d.; ∗P0.05, ∗∗P0.01, not significant (ns) P>0.05.

Although the CD103 integrin tethers T_RM cells to E‐Cadherin and is important for their preferential localization at epithelial tissues, CD103⁻ CD8 T_RM cells exist in particular organs, including the brain. ^{5 , 25} CD103 expression by CD8 T_RM cells may also be influenced by the nature of the viral infection, as lung CD8 T_RM cells to influenza virus and respiratory syncytial virus express CD103, whereas those specific for human cytomegalovirus are CD103⁻. ²⁶

Demethylation of the PD‐1 promoter by MuPyV‐specific bT_RM cells

Using the mouse model of chronic LCMV infection, it has been previously demonstrated that the Pdcd1 promoter is extensively demethylated at early stages of the immune response. This locus was remethylated in the acutely resolved LCMV Armstrong infection but remained demethylated in chronic LCMV clone13 infection even after host control of infection, a result tying PD‐1 to the exhaustion state of differentiation. ^{13 , 14} Stability of the demethylated Pdcd1 promoter was further demonstrated in human EBV, CMV and HIV‐specific CD8 T cells. ^{13 , 27} It is yet to be confirmed whether the epigenetic changes observed at peripheral T_EX cells holds true for bT_RM cell population as well. The marked difference in PD‐1 expression by memory MuPyV‐specific CD8 T cells in the spleen and brain led us to ask whether cells in these organs differed in level and stability of Pdcd1 promoter demethylation. Pdcd1 methylation status was analyzed on Thy1.1 TCR‐I CD8 cells sorted from the brains and spleens of acute (day 8 p.i.) and persistent (day 30 p.i.) infected recipient mice (Figure 4a). As expected, the Pdcd1 promoter in naive splenic TCR‐I cells was almost entirely methylated (Figures 4b and e). At day 8 p.i., TCR‐I cells in both the spleen and brain exhibited extensive Pdcd1 promoter demethylation (Figures 4c and e). By day 30 p.i., however, TCR‐I cells in the spleen showed increased methylation of the Pdcd1 promoter, while brain TCR‐I cells retained a demethylated Pdcd1 promoter (Figures 4d and f). This intermediate level of methylation in TCR‐I cells from day 30 infected mice was intermediate between cells from a naive or day 8‐infected mice. This observation is consistent with the lower trending PD‐1 transcript levels seen at day 30 p.i. (Figure 1c). The PD‐1^hi phenotype of the brain TCR‐I T_RM cells further suggests that a minimum threshold of Pdcd1 promoter demethylation facilitates upregulation of PD‐1 gene expression. Thus these data show that the Pdcd1 promoter of CD8 bT_RM cells remains more extensively demethylated than splenic memory CD8 T cells despite having similar persistent MuPyV infection levels.

Figure 4.

Methylation of the Pdcd1 promoter from TCR‐I cells in uninfected and MuPyV acutely infected mice. (a) Experimental setup. (b–d) Bisulfite sequencing of nucleotides −778 to −465 upstream of the Pdcd1 promoter from genomic DNA of naive TCR‐I mice (b) and of Thy1.1 FACS‐sorted TCR‐I cells from the spleens and brains of MuPyV‐infected mice at day 8 p.i. (c) and day 30 p.i. (d). Quantification of methylation of nine CpGs sites (closed circles) in the Pdcd1 promoter, as described previously. ¹³ (d and e) Combined quantification of percentage of methylation of the Pdcd1 promoter for day 8 p.i. (d) and day 30 p.i. (e) (pooled cells from three to five mice, two independent experiments). A nonparametric paired Student's t‐test with Mann–Whitney test was used to determine statistical significance. Values are mean±s.d.; ∗P0.05, not significant (ns) P>0.05.

Recent studies document expression of PD‐1 by memory CD8 T cells in inflamed immune‐privileged organs, such as the eye and brain. ^{6 , 28} Expression of PD‐1 on brain CD8 T cells has been shown to limit axonal bystander damage in a mouse coronavirus CNS infection model and mediate protection from experimental autoimmune encephalomyelitis. ^{29 , 30 , 31 , 32 , 33} Our finding that PD‐1 blockade elevates effector function by anti‐MuPyV CD8 bT_RM cell is in line with the concept that PD‐1 expression by tissue‐resident T cells may serve to restrain immunopathology, which would prove especially deleterious in the infected CNS. The divergence in PD‐1 expression between brain and spleen MuPyV‐specific CD8 T cells during persistent infection, with bT_RM cells maintaining PD‐1 in the absence of cognate antigen and inflammation, supports the concept that the brain microenvironment favors PD‐1 expression.

Methods

Mice and viruses

C57BL/6NCr female mice were purchased from the National Cancer Institute (Frederick, MD, USA). B6.PL (Thy1.1) and C57BL/6‐Tg (UBC‐GFP) 30Scha/J mice (‘GFP mice’) and B6.129S4‐Ifng^tm3.1Lky/J, IFN‐γ reporter ‘IFN‐γ EYFP’ mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA). TCR‐I mice are transgenic for a TCR for the D^b‐restricted peptide corresponding to amino acids 206–215 of the SV40 Large T Ag (‘LT206’). ²² The D^b‐restricted LT359–368 epitope in MuPyV Large T Ag (strain A2; ‘MuPyV.A2’) was replaced by the D^bLT206 epitope (‘MuPyV.LT206’). ²² In all, 6.5 × 10⁴ plaque‐forming units per mouse of MuPyV.A2 or MuPyV.LT206 were injected i.c. ¹⁶ Mice were bred and housed in accordance with the guidelines of the IACUC of The Penn State College of Medicine.

Depletion of CD8 cells

CD8 Ab (clone YTS 169.4; BioXCell, West Lebanon, NH, USA) or rat IgG Ab (Jackson ImmunoResearch, West Grove, PA, USA) was injected via intraperitoneal route at 250 μg Ab per mouse once per week starting from day 10 p.i. to day 30 p.i. Efficiency of depletion of CD8⁺ cells in blood was 90% (data not shown).

Adoptive cell transfer

Naive CD8 T cell (CD44^lowCD62L^high) were purified from TCR‐I mice using a negative selection isolation kit (Miltenyi Biotec, Ausburn, CA, USA), and unstained or carboxyfluorescein succinimidyl ester‐labeled cells were injected intravenously per tail vain in 100 μl phosphate‐buffered saline or by i.c. injection in 30 μl phosphate‐buffered saline.

Quantitation of MuPyV genomes

Quantitative TaqMan‐based PCR was performed using 10 ng template DNA purified from tissues, as described. ²²

Quantitative real‐time PCR for transcript analysis

Fluorescence‐activated cell sorter (FACS)‐purified GFP⁺ TCR‐I cells were lysed in TRIzol (Ambion, Carlsbad, CA, USA) and total RNA was isolated per the manufacturer's instruction. cDNA was prepared using random primers and RevertAid H Minus Reverse Transcriptase enzyme (ThermoFisher Scientific, Leesport, PA, USA). Taqman primer probe set for Pdcd1 and Tbp from IDT Technologies (Coralville, IA, USA) was used to perform quantitative PCR using the ABI StepOnePlus Real‐Time PCR System (ThermoFisher Scientific). Sequences of the primer probe sets are as follows: PD‐1, 5′‐ATTTGCTCCCTCTGACACTG‐3′; 5′‐GTACCCTGGTCATTCACTTGG‐3′; 5′‐/56‐FAM/TCCCTCACC/ZEN/TTCTACCCAGCCT/3IANKFQ/‐3′ and TBP: 5′‐TGTATCTACCGTGAATCTTGGC‐3′; 5′‐CCAGAACTGAAAATCAACGCAG‐3′; 5′‐/56‐FAM/ACTTGACCT/ZEN/AAAGACCATTGCACTTCGT/3IABkFQ/‐3′. Relative fold change over naive control was determined using the threshold cycle (2^{−ΔΔ Ct}) method.

Cell isolation, intracellular cytokine staining and flow cytometry

Mononuclear cells from brain and kidney were isolated from transcardially perfused mice by collagenase‐DNase digestion and percoll gradient centrifugation, then treated with Fixable Viability Dye (eBioscience, San Diego, CA, USA) and stained with APC‐D^bLT359 or BV421‐D^bLT206 tetramers (NIH Tetramer Core Facility, Atlanta, GA, USA) and antibodies, as described. ¹⁶ Samples were acquired on a BD LSRII or BD LSRFortessa (BD Bioscience, San Jose, CA, USA) and analyzed using the FlowJo software (FlowJo, LLC, Ashland, OR, USA). GFP⁺ cells stained with anti‐CD8α and 4,6‐diamidino‐2‐phenylindole were sorted using a BD FACSAria SORP Cell Sorter (BD Bioscience).

In vitro PD‐L1 blockade

DC2.4 cells were exposed to IFN‐γ to upregulate MHC I and PD‐L1 (data not shown) and then pulsed with 10 μM LT206 peptide. Cells were added in the presence of 50 μg ml⁻¹ control rat IgG or PD‐L1 blocking Ab (10F.9G2; BioXCell). After 5 h incubation with BD GolgiPlug (BD Bioscience), cells were stained with Fixable Viability Dye, anti‐CD8α and anti‐Thy1.1 and then stained for intracellular IFN‐γ.

Epigenetic analysis of PD‐1 promoter

Thy1.1 TCR‐I cells isolated from the spleens or brains of naive mice or from mice at days 8, 30 or 45 p.i. were FACS‐sorted to >95% purity. Genomic DNA was isolated from the purified cells and subjected to bisulfite treatment. Bisulfite‐modified DNA was PCR‐amplified using PD‐1 promoter‐specific primers and cloned, and sequences were analyzed using the BISMA software (Bremen, Germany), as described. ¹³

Statistical analysis

P‐values were determined by an unpaired Student t‐test with Mann–Whitney posttest or one‐way analysis of variance with Kruskal–Wallis posttest using the GraphPad Prism software (GraphPad Software, La Jolla, CA, USA). P‐values0.05 were considered significant.

ACKNOWLEDGEMENTS

This work was supported by the National Institutes of Health (R01 NS088367, R01 NS092662 and R01 AI102543 to AEL; R01 AI114442 to BAY; and T32 AI007610 and F31 NS083336 to ELF) and funds from the American Lebanese Syrian Associated Charities (ALSAC) to BAY. We thank J Bednarczyk of the Penn State College of Medicine Flow Cytometry Core Facility for technical assistance with cell sorting.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

Supporting information

Supplementary Material