Cooperation of Tim-3 and PD-1 in CD8 T-cell exhaustion during chronic viral infection

Tim-3 Expression Defines a Subpopulation of PD-1⁺ Exhausted CD8 T Cells During Chronic LCMV Infection.

We examined expression of Tim-3 on LCMV-specific effector, memory, and exhausted CD8 T cells during acute or chronic infection using the Armstrong (Arm) or clone-13 (CL-13) strains of LCMV (14). First, we analyzed Tim-3 expression on LCMV-specific CD8 (GP33⁺CD8) T cells in lymphoid and nonlymphoid organs at various time points after infection. During acute infection, the percentage of GP33⁺CD8 T cells expressing Tim-3 (Tim3⁺GP33⁺) was significantly increased in spleen, lung, and liver at d 8 after infection (Fig. 1A and Fig. S1A). However, this frequency rapidly decreased to approximately 10% in all tissues at day 30 after Arm infection (Fig. 1B and Fig. S1A). On the contrary, during chronic infection, the frequency of Tim3⁺GP33⁺ T cells was markedly elevated at day 8 in all tissues, which was sustained at approximately 70% to 80% for at least 90 d (Fig. 1A and Fig. S1A). Although the frequency of Tim3⁺GP33⁺ T cells was similarly increased at day 8 in acute and chronic infection, we observed significantly stronger mean fluorescence intensity (MFI) of Tim-3 expression during chronic versus acute viral infections at day 8 (Fig. S1B). Moreover, this high intensity of Tim-3 expression on virus-specific CD8 T cells was maintained during chronic viral infection. These results indicate that Tim-3 is transiently up-regulated at intermediate levels on activated T cells, and is quickly down-regulated in acute infection, but continues to be expressed at much higher levels during chronic infection.

Because it was not yet clear whether Tim-3 and PD-1 are expressed by distinct or overlapping populations of CD8 T cells during chronic infection (12, 13), we next analyzed the coexpression of Tim-3 and PD-1. During acute viral infection, only at day 8 after infection was a small portion of GP33-specific CD8 T cells coexpressing Tim-3 and PD-1 (Tim3⁺PD1⁺) detectable (Fig. 1C). In contrast, chronic viral infection resulted in much higher frequency of Tim3⁺PD1⁺ cells, which persisted even at day 90 (Fig. 1C). We could also detect another dominant population expressing only PD-1 but not Tim-3 (Tim3⁻PD⁺) during chronic infection (Fig. 1D). Thus, exhausted CD8 T cells found during chronic LCMV infection could be divided into Tim3⁺PD1⁺ and Tim3⁻PD1⁺ subsets. Notably, the distribution of Tim3⁺PD1⁺ or Tim3⁻PD⁺ was slightly different depending on the tissues analyzed, and a small percentage of CD8 T cells detectable in the lung expressed only Tim-3 but not PD-1 (Tim3⁺PD1⁻; Fig. 1D), suggesting differential phenotypic characteristics of exhausted CD8 T cells in varying anatomical sites. The presence of this Tim3⁺PD1⁺ cell could also be defined by looking at total CD8 T cells (Fig. S1C).

Coexpression of Tim-3 and PD-1 Correlates with More Severe Exhaustion of Virus-Specific CD8 T Cells During Chronic Virus Infection.

We identified two dominant populations of virus-specific CD8 T cells expressing PD-1 alone (Tim3⁻PD1⁺) or coexpressing Tim-3 (Tim3⁺PD1⁺) during chronic LCMV infection. We further assessed the phenotypic profile of virus-specific Tim3⁻PD1⁺ and Tim3⁺PD1⁺ CD8 T cells by using a panel of markers, including other inhibitory and differentiation markers. The Tim3⁺PD1⁺ virus-specific CD8 T cells displayed higher levels of other inhibitory receptors such as LAG3 and 2B4, along with lower levels of CD62L and CD127, than the Tim3⁻PD1⁺ population (Fig. S2). Considering the phenotypic signature of CD8 T-cell exhaustion (8), this suggests that the virus-specific Tim3⁺PD1⁺ T cells may represent a more severely exhausted subpopulation.

We next directly compared the functional properties of purified population of Tim3⁻PD1⁺ and Tim3⁺PD1⁺ CD8 T cells. As the majority of antigen-specific CD8 T cells in the spleen are Tim3⁻PD1⁺ or Tim3⁺PD1⁺ during chronic infection (Fig. 1 and Fig. S1), we focused on investigating the functionality of these two subsets. To investigate the proliferation capacity, equal numbers of GP33-specific Tim3⁻PD1⁺ or Tim3⁺PD1⁺ cells were labeled with carboxyfluorescein succinimidyl ester (CFSE) and stimulated with GP33-41 peptide or LCMV peptide pool. When cell proliferation was assessed by CFSE dilution, some division of Tim3⁻PD1⁺ cells was observed, whereas Tim3⁺PD1⁺ cells hardly showed any proliferation after peptide stimulation (Fig. 2A). These results show that the proliferative dysfunction is greater in the Tim3⁺PD1⁺ T cell population.

We then examined cytokine secretion by Tim3⁻PD1⁺ or Tim3⁺PD1⁺ cells in response to LCMV peptide stimulation. Consistent with our previous report (14), we confirmed the functional exhaustion of CD8 T cells from chronically infected mice characterized by a highly reduced ability to secret IFN-γ, TNF-α, and IL-2 compared with CD8 T cells from acutely infected mice (Fig. S3A). When we further accessed the dysfunction of Tim3⁻PD1⁺ or Tim3⁺PD1⁺ CD8 T cells in chronic infection, the frequency of IFN-γ–secreting cells was similar between the two populations, but the intensity of IFN-γ secretion (i.e., MFI) was approximately 2-fold higher in the Tim3⁻PD1⁺ subset compared with the Tim3⁺PD1⁺ subset (Fig. S3B). Furthermore, TNF-α and IL-2 secretion from CD8 T cells in response to stimulation was observed predominantly from the Tim3⁻PD1⁺ population, with minimal cytokine production observed in the Tim3⁺PD1⁺ population (Fig. S3B). To more quantitatively determine the functionality of these subsets, we examined cytokine secretion on a “per-cell” basis by calculating the frequency of epitope-specific CD8 T cells in each subset. The Tim3⁻PD1⁺ population showed a much higher percentage of both GP33- and GP276-specific CD8 T cell producing TNF-α or IL-2 than the Tim3⁺PD1⁺ population (Fig. 2B). These results therefore indicate that functional exhaustion of CD8 T cells during chronic infection is associated with coexpression of Tim-3 and PD-1, and that Tim3⁺PD1⁺ cells exhibit a deeper exhaustion.

Tim3⁺PD1⁺ Exhausted CD8 T Cell Produces the Inhibitory Cytokine IL-10.

A novel population of HIV-specific CD8 T cells has been shown to express IL-10 during HIV-1 infection (15, 16), suggesting that this specific population may have an important regulatory role in immune dysfunction in chronic infections. However, no phenotypic marker has been defined to identify the IL-10–positive CD8 T cells. In this regard, we examined whether Tim3⁻PD1⁺ or Tim3⁺PD1⁺ cells secrete IL-10 in response to stimulation with LCMV peptides. In contrast to effector cytokines (IFN-γ, TNF-α, and IL-2), IL-10 production was significantly observed only by the Tim3⁺PD1⁺ but not the Tim3⁻PD1⁺ populations (Fig. 2C and Fig. S3C). These results indicate that Tim-3 coexpression with PD-1 may provide an immunophenotypic marker for determining the IL-10–producing CD8 T cells as well as severely exhausted CD8 T cells.

Simultaneous in Vivo Blockade of Tim-3 and PD-1 Pathways Synergistically Restores CD8 T Cell Function and Enhances Viral Control.

Our results show a correlation between coexpression of Tim-3 with PD-1 and deeper CD8 T cell exhaustion. However, it is not clear whether the Tim-3 and PD-1 pathways have overlapping functions or contribute independently to T cell dysfunction during chronic viral infection. To address this issue, we blocked the Tim-3 and PD-1 pathways with Tim-3-Ig fusion protein (Tim3Ig) and blocking antibody to PD-L1 (αPDL1), either alone or in combination during chronic LCMV infection. We first measured the frequency of GP33-specific CD8 T cells in the blood before and after treatment (Fig. 3). Tim-3 blockade alone (Tim3Ig) induced almost no change in GP33-specific CD8 (GP33⁺CD8) T-cell population. PD-1 blockade alone (i.e., αPDL1) resulted in a 5.3-fold increase of GP33⁺CD8 T-cell population (P = 0.0013), and dual blockade of Tim-3 and PD-1 (Tim3Ig + αPDL1) led to the 6.8-fold increase of GP33⁺CD8 T cell pool (P = 0.001). We then analyzed virus-specific CD8 T cells responses in lymphoid and nonlymphoid tissues. Treatment with Tim3Ig alone showed only marginal effects on the absolute number of GP33⁺CD8 T cells, whereas blocking the PD-1 pathway significantly augmented the GP33⁺CD8 T cell response compared with isotype control (Fig. 3). More interestingly, the simultaneous administration of Tim3Ig and αPDL1 further increased the number of GP33⁺CD8 T cells even when compared with αPDL1 alone (Fig. 3). It is worth noting that, when treatments started at 200 d after infection, dual blockade still increased the quantity of virus-specific CD8 T cells, but αPDL1 alone was only minimally effective (Fig. S4).

To determine the effect of dual blockade of Tim-3 and PD-1 pathways on the function of virus-specific CD8 T cells during chronic viral infection, we measured the ability of LCMV epitope-specific CD8 T cells to produce the cytokine IFN-γ and to degranulate by monitoring CD107a/b expression. The treatment with Tim3Ig alone showed only a modest increase of CD107a/b expression and IFN-γ production, whereas treatment with αPDL1 significantly increased the percentage of functional CD8 T cells responding to multiple LCMV epitopes (Fig. 4A and Fig. S5A). Importantly, dual blockade with Tim3Ig and αPDL1 showed a much greater increase in the degranulation and IFN-γ–producing CD8 T cell population, particularly to GP33-41, GP276-286, NP396-404 peptides, and a pooled peptide mixture, compared with either treatment alone or isotype control (Fig. 4A and Fig. S5A). Furthermore, only dual blockade elevated the percentage of polyfunctional CD8 T cells coproducing IFN-γ and TNF-α (Fig. 4B). Next, to test the effect of the blockade treatments on the proliferation of virus-specific CD8 T cells, we analyzed the frequency of Ki67-positive GP33⁺CD8 T cells. Tim-3 blockade alone slightly increased the frequency of Ki67-positive GP33⁺CD8 T cells compared with isotype control (Fig. 4C and Fig. S5B). PD-1 blockade alone showed a significant increase in the Ki67-positive population, and dual blockade further increased the frequency of Ki67-positive GP33⁺CD8 T cells compared with αPDL1 alone (Fig. 4C and Fig. S5B). Taken together, these results indicate that dual blockade of Tim-3 and PD-1 pathways substantially enhanced virus-specific CD8 T cell responses in quality as well as quantity during chronic viral infection.

We next assessed if the heightened CD8 T-cell response induced by dual blockade of these inhibitory receptors also enhanced viral control. The viral load of mice treated with isotype control or Tim3Ig alone in the serum was not significantly altered after treatment (Fig. 5). In contrast, αPDL1 alone or dual blockade (Tim3Ig + αPDL1) treatments led to 4.2-fold and a 6.3-fold reductions, respectively, in virus titer of serum (Fig. 5). Similarly, treatment with Tim3Ig alone resulted in slightly lower viral loads in the spleen, liver, and lung, whereas treatment with αPDL1 alone showed significant reduction of viral load in all tissues compared with the isotype control group (Fig. 5). More impressively, dual blockade with Tim3Ig and αPDL1 led to further reduction in viral loads in all tissues compared with blockade with αPDL1 alone (Fig. 5).

Mice and Infections.

Six-week-old female C57BL/6 mice were purchased from Jackson Laboratory. LCMV strains were propagated, titered, and used as previously described (14). For acute infection, mice were intraperitoneally infected with 2 × 10⁵ pfu of LCMV Armstrong strain. For chronic infection, mice were infected i.v. with 2 × 10⁶ pfu of LCMV clone-13 strain, and were intraperitoneally injected with 500 μg anti-CD4 antibody (GK 1.5; BioExpress) on day −1 and day +1 relative to infection with LCMV clone-13 on d 0. All mice were used in accordance with National Institutes of Health and the Emory University Institutional Animal Care and use Committee guidelines.

Flow Cytometry and Intracellular Cytokine Staining.

Lymphocytes were isolated from tissue including spleen, liver, lung, and blood as previously described (3). The antibodies to CD8 (53-6.7), Ki67 (B56), IFN-γ (XMG 1.2), TNF-α (MP6-XT22), IL-2 (JES6-5H4), IL-10 (JES5-16E3), CD107a (1D4B), and CD107b (ABL093) were purchased from BD Bioscience. Anti-CD44 (IM7), anti–Tim-3 (215008), and anti–PD-1 (RMP 1–30) were obtained from eBioscience, R&D Systems, and BioLegend, respectively. MHC class I tetramers or multimers conjugated with APC and Qdot565 (Invitrogen) were generated and used as previously described (14, 25). Surface and intracellular cytokine staining was performed as described (14). To detect degranulation, splenocytes were stimulated with individual LCMV peptides or a pool of eight LCMV epitopes for 5 h in the presence of brefeldin, monensin, anti–CD107a-FITC, and anti–CD107b-FITC. Cells were then analyzed on an LSR II flow cytometer (BD Immunocytometry Systems). Data were analyzed with FlowJo v.8.8 (TreeStar). Dead cells were removed by gating on Live/Dead NEAR IR (Invitrogen).

In Vitro Proliferation Assay.

CD8 T cells were purified to more than 90% purity by using magnetic beads (Miltenyi Biotec). Tim3⁺PD1⁺ and Tim3⁻PD1⁺ CD8 T cells were isolated by FACS sorting. CFSE-labeled Tim3⁺PD1⁺ and Tim3⁻PD1⁺ CD8 T cells were cocultured with splenocytes from Thy1.1⁺ C57BL/6 mice in the presence of LCMV peptides for 3 d. To analyze proliferation, loss of CFSE in Thy1.2⁺-gated cells was determined by flow cytometry.

In Vivo Blockade and Virus Titration.

For blockade of PD-1 pathway, 200 μg of rat antimouse PD-L1 antibody (10F.9G2; prepared in house) or rat IgG2b isotype control were administered intraperitoneally every 3 d for 2 wk. For blockade of Tim-3 pathway, 100 μg of Tim-3-Ig fusion protein (prepared in house) were injected intraperitoneally every 2 d for 2 wk. The ability of Tim-3-Ig and anti–PD-L1 to block the Tim-3 and PD-1 pathways, respectively, has been previously demonstrated (3, 26). Titers of virus from serum or homogenized tissue sample were determined by plaque assay on Vero cells as previously described (27).

Statistical Analysis.

Data were analyzed using Prism 5.0 software (GraphPad). Experiments were repeated two or three times. The data presenting the differences between the groups were assessed using two-tailed unpaired Student t tests. P < 0.05 indicated that the value of the test sample was significantly different from that of the relevant controls.