Programmed Death-1 Impairs Secondary Effector Lung CD8⁺ T Cells during Respiratory Virus Reinfection

The respiratory viruses human metapneumovirus (HMPV) and respiratory syncytial virus (RSV) are important causes of acute lower respiratory infection (LRI), which results in significant morbidity and mortality, especially in infants, the elderly, and the immunocompromised (1–8). Despite the need to protect these populations against serious LRI, no licensed vaccines or therapeutics exist for these viruses. The majority of LRI beyond infancy are actually reinfections because nearly all individuals experience primary infection during early childhood (9). HMPV reinfection in children causes illness at a rate that equals primary infection (10) and can occur with both genetically heterologous and homologous viruses (11). Despite a high frequency of infection and minimal antigenic drift for both HMPV and RSV, protective immunity is poorly established because individuals can be repeatedly reinfected throughout life (12–14). High anti-HMPV Ab titer in serum is insufficient to prevent reinfection in adults (15). Respiratory virus reinfections cause important clinical disease, but the mechanisms governing susceptibility to recurrent viral LRI are poorly understood. Although much attention has been placed on humoral immunity, the above evidence argues that Abs are not always associated with protection. Indeed, in animal models, both arms of the adaptive immune system contribute (16–18).

The anti-HMPV T_CD8 response (19), like that against RSV (20) and influenza virus (19, 21), is functionally impaired in the respiratory tract. Virus-specific lung T_CD8 do not optimally respond to stimulation by releasing lytic granules or producing antiviral cytokines such as IFN-γ. We recently demonstrated that during primary HMPV and influenza virus infections the inhibitory receptor programmed death-1 (PD-1) significantly contributes to this impairment by repressing T_CD8 effector functions (19). Blockade of PD-1 signaling restored lung T_CD8 functions and enhanced viral control. Prior to this, PD-1 had mainly been associated with T cell exhaustion during chronic infection and cancer, where prolonged TCR stimulation by persistent viral or tumor Ags maintains PD-1 expression (22). PD-1 ligand (PD-L) binding to PD-1 antagonizes TCR signaling by blocking PI3K/Akt activation, leading to decreased protein synthesis, cytokine production, proliferation, and survival of effector T cells (23). PD-1 pathway blockade has recently proven effective at treating refractory malignancies (24, 25). Thus, therapeutic targeting of this pathway has significant clinical potential.

A poorly explored aspect of PD-1 biology is its contribution to reinfection. We recently demonstrated that secondary effector T_CD8 in wild-type (WT) mice rapidly re-expressed PD-1 and were highly impaired (19). Secondary T_CD8 also become exhausted during chronic infection in mice (26). In addition to PD-1, several other inhibitory receptors have been identified that contribute to functional T_CD8 impairment or exhaustion in a variety of settings and therefore may also contribute to impairment during reinfection (27). Exhausted T_CD8 express numerous inhibitory receptors, including TIM-3 (28), LAG-3 (29), 2B4 (26), and others (30). TIM-3 (31) and LAG-3 (32) also negatively regulate the T_CD8 response to acute viral LRI. The role these receptors play in causing lung T_CD8 functional impairment during respiratory virus reinfection has not been explored.

We hypothesized that inhibitory receptor signaling contributes to respiratory virus reinfections by causing lung T_CD8 impairment during memory immune responses. We sought to elucidate PD-1’s contribution to HMPV reinfection and to determine whether PD-1 limits the effectiveness of potential vaccination strategies directed at the cellular immune response. To study secondary effector T_CD8 impairment, we first established a model of HMPV reinfection using B cell–deficient mice because rodents are normally protected against respiratory virus reinfection by neutralizing Abs (17, 19, 33) and viral Ag directly modulates PD-1 expression through TCR signaling (19, 27, 34). B cell–deficient mice were susceptible to HMPV reinfection but were unable to better control early viral replication as compared with primary-infected mice. Secondary effector lung T_CD8 expressed the inhibitory receptors PD-1, LAG-3, 2B4, and TIM-3, with the former three expressed more highly than during primary infection. However, in vitro experiments demonstrated that PD-1 was the dominant functional inhibitory receptor early during reinfection. Therapeutic blockade of PD-1 signaling during HMPV reinfection restored secondary lung T_CD8 effector functions (i.e., degranulation and cytokine production) and reduced viral burdens in the lung. Furthermore, PD-1 limited the effectiveness of a peptide vaccination strategy against HMPV and also impaired secondary lung T_CD8 during heterotypic influenza virus challenge. These results indicate the PD-1 contributes to lung T_CD8 impairment during reinfection and prevents memory T_CD8 from controlling viral replication. Given the severity and frequency of reinfections by respiratory viruses and the lack of effective vaccines, these studies have important implications for future vaccines and therapeutic interventions.

C57BL/6 (B6) were purchased from The Jackson Laboratory. μMT mice were provided by Dr. M. Boothby (Vanderbilt University). PD-1^−/− mice were obtained with permission from Dr. T. Honjo (Kyoto University, Kyoto, Japan). All animals were bred and maintained in specific pathogen-free conditions in accordance with the Vanderbilt Institutional Animal Care and Use Committee. Age- and gender-matched animals were used in all experiments. HMPV (pathogenic clinical strain TN/94-49, genotype A2) was grown and titered in LLC-MK2 cells as described previously (33). Influenza virus strains A/34/PR/8 (PR8; H1N1; American Type Culture Collection) and HK/x31 (x31; H3N2; provided by Dr. J. McCullers and P. Thomas, St. Jude Children’s Research Hospital, Memphis, TN) were grown in MDCK cells and titered on LLC-MK2 cells. Mice were anesthetized with ketamine-xylazine and infected intranasally (i.n.) with 1 × 10⁶ PFU HMPV. Viral titers in infected mouse lungs and nasal turbinates were measured by plaque titration as described previously (33). For influenza virus challenge experiments, mice were primed i.p. with 2 × 10⁵ PFU PR8 and challenged i.n. with 5 × 10² PFU of ×31 at least 15 wk later.

Lung lymphocytes were tetramer-stained or restimulated for intracellular cytokine staining (ICS) in parallel as described previously (19). Lung cells were stained for the inhibitory receptors PD-1 (clone RMP1-30), TIM-3 (clone RMT3-23), LAG-3 (clone C9B7W), and 2B4 (clone m2B4 (B6)458.1) or with appropriate isotype control Abs (all from BioLegend). Flow cytometric data were collected using an LSRII or Fortessa (BD Biosciences) and analyzed with FlowJo software (Tree Star). The Boolean gating function in FlowJo was used to assess inhibitory receptor coexpression, and patterns were visualized using the SPICE program (National Institute of Allergy and Infectious Diseases, Bethesda, MD).

ELISPOT assays were performed as previously described (35) with slight modifications. A total of 5 × 10⁴ lung cells were added to triplicate wells. Peptides were then added (10 μM final concentration), followed by inhibitory receptor blocking Abs (10 μg/ml final concentration). The following blocking Abs were used: isotype control (clone LTF-2; BioXCell), anti–PD-L1 (clone 10F.9G2; BioXCell), anti–PD-L2 (clone TY-25; BioXCell), anti–PD-1 (clone J43; BioXCell), anti-TIM-3 (clone RMT3-23; BioXCell), anti–LAG-3 (clone C9B7W; BioXCell), anti-2B4 (clone m2B4 (B6)458.1; BioLegend), and anti-CD48 (clone HM48-1; BioXCell). Plates were incubated at 37°C for 42–48 h, developed, and then counted using an ImmunoSpot Micro Analyzer (Cellular Technology). The average number of spots in wells stimulated with an irrelevant peptide was subtracted from each experimental value, which was then expressed as spot-forming cells per 10⁶ lymphocytes.

Mice were injected i.p. on days −1, 1, 3, and 5 postinfection (p.i.) with 200 μg rat isotype control Ab (clone LTF-2; BioXCell) or 200 μg of both rat anti-mouse PD-L1 (clone 10F.9G2; BioXCell) plus rat anti-mouse PD-L2 (clone TY-25; BioXCell) to block PD-1 signaling.

Mice were injected i.v.with a mixture of 200 μg D^b/F528 peptide, 50 μg anti-CD40 Ab (clone FGK4.5; BioXCell), and 50 μg polyinosinic:polycytidylic acid (InvivoGen). Peripheral blood was obtained to check immune responses in all vaccinated animals.

Data analysis was performed using Prism version 6.0 (GraphPad Software). Comparisons between tetramer staining and intracellular cytokine staining within the same animals were performed using a paired, two-tailed t test. Comparisons between two groups were performed using an unpaired, two-tailed Student t test. Multiple group comparisons were performed using a one-way ANOVA with a Bonferroni posttest. Error bars on each graph represent SEM unless otherwise noted.

Rodents are normally protected against reinfection with respiratory viruses because of the presence of neutralizing Abs and the fact that they are semipermissive hosts (33, 36, 37). We therefore used the B cell–deficient mouse strain μMT, which has been used to model reinfection with influenza virus and RSV (36, 38, 39), to test whether this strain was susceptible to reinfection with HMPV. HMPV demonstrated similar replication kinetics in the lungs and nasal turbinates during primary infection of μMT and WT mice (Fig. 1A), although there was a slight decrease in lung virus titers at day 7 in WT compared with μMT mice. Virus was cleared from the lungs by day 10 in both strains. Upon reinfection, WT mice were completely protected against viral replication, as expected (Fig. 1B). However, HMPV replicated to similarly high titers in reinfected μMT mice as in primary infection. Indeed, lung titers were indistinguishable between primary and secondary infection at day 5, the usual peak of viral replication (19). On day 7, lung titers were diminished in many μMT mice, with some undetectable, and by day 10, most had cleared the infection. μMT mice were also able to clear virus from the nasal epithelium during secondary infection because mice had lower titers at days 5 and 7 postchallenge compared with primary infection. Thus, μMT mice are susceptible to HMPV reinfection, although they clear virus more rapidly than primary infection.

Because the μMT mouse represents a new model of HMPV reinfection, we next tested whether the kinetics and functionality of T_CD8 during primary and secondary infection were similar to WT B6 mice. WT and μMT mice were either primarily infected with HMPV, or for secondary infection, mice were infected with HMPV, allowed to clear the infection, and then challenged with HMPV 16 wk later. Lung lymphocytes were collected on days 7, 10, and 14 after primary or secondary infection and enumerated by tetramer staining and intracellular cytokine staining for the two dominant HMPV epitopes D^b/F_528–536 (F528) and K^b/N_11–19 (N11) (19) (Fig. 2). During primary infection, the percentage of F528-specific T_CD8 was similar between μMT and WT mice; in both groups, a minority of the T_CD8 degranulated or produced IFN-γ, with the percent functional declining over time as described previously (19) (Fig. 2 A, 2B, top panels). Interestingly, the percentage of N11-specific T_CD8 increased from days 7 to 14 in both groups, although functionality also declined over time (Fig. 2 A, 2B, bottom panels). The results were similar for both populations of epitope-specific T_CD8 during secondary infection of μMT and WT mice (Fig. 2 C, 2D). Although the magnitude of the N11-specific T_CD8 response was higher in both groups during secondary infection as judged by tetramer staining, the percentage that were functional was even lower than during primary infection. Thus, the μMT mouse serves as an appropriate model of T_CD8 responses compared with WT B6 mice.

The ability of μMT mice to clear HMPV during secondary infection was presumably due to cellular immunity (i.e., CD4⁺ and CD8⁺ T cells). However, we were surprised to see that at days 5 and 7 after reinfection lung viral titers were equivalent to peak titers during primary infection. We therefore quantified inhibitory receptor expression during both primary and secondary HMPV infection of μMT mice to search for an explanation for the large degree of T_CD8 impairment observed during reinfection. Reinfected μMT mice generated a 3-fold greater F528-specific response compared with primary infected mice (Fig. 3A). Lung F528–specific T_CD8 expressed the inhibitory receptors PD-1, LAG-3, TIM-3, and 2B4 (Fig. 3B); however, PD-1, LAG-3, and 2B4 were more highly expressed by secondary T_CD8 as compared with primary (Fig. 3B, 3C).

Coexpression of inhibitory receptors is important in T_CD8 exhaustion during chronic infection because the number of receptors expressed on the cell surface corresponds to the degree of functional impairment (29). During HMPV LRI, most F528-specific lung T_CD8 expressed two or more inhibitory receptors and a large fraction expressed three or more (Fig. 3D). A greater proportion of lung T_CD8 expressed three or four inhibitory receptors during secondary infection compared with primary infection (Fig. 3D). Although the expression pattern of these receptors was similar during primary and secondary infection, the percentage of F528-specific T_CD8 expressing all four receptors, or PD-1, TIM-3, and LAG-3 was increased during reinfection (Fig. 3E). In addition, the fraction of cells that expressed only PD-1 or no inhibitory receptors was decreased during secondary infection (Fig. 3E), confirming a skewing toward augmented inhibitory receptor expression during reinfection. These results indicate that secondary T_CD8 express numerous inhibitory receptors, with several of these receptors more frequently expressed than primary T_CD8 at the same time p.i.

Given the coexpression of numerous inhibitory receptors during reinfection, we next sought to determine their potential contribution to lung T_CD8 impairment. To do so, we isolated lung cells from reinfected μMT mice and added them to an IFN-γ–detecting ELISPOT assay. Cells were restimulated with F528 peptide and incubated with blocking mAbs against each inhibitory receptors or its ligand(s). F528 peptide restimulation of lung cells plus PD-L1 blockade resulted in significantly more IFN-γ–secreting cells than isotype-treated cells (i.e., more spots) (Fig. 4A), confirming our previous in vivo results during primary infection (19). Blockade of the PD-1 receptor itself also resulted in more spot-forming cells (Fig. 4B). Blocking PD-L2, TIM-3, LAG-3, 2B4, or CD48 (the ligand for 2B4) alone did not result in any significant changes. There was a trend toward decreased responsiveness during 2B4 blockade. Combined blockade of TIM-3, LAG-3, or 2B4 with PD-L1 did not result in greater responses than PD-L1 or PD-1 blockade alone. Anti–PD-L1 and anti–PD-1 treatment also resulted in larger spots (Fig. 4C), indicating a greater amount of IFN-γ secreted by each cell compared with isotype treatment. Again, blockade of other inhibitory receptors or their ligands failed to increase spot size. These results indicate that despite coexpression of numerous inhibitory receptors, PD-1 is the dominant mediator of lung T_CD8 impairment early during respiratory virus reinfection.

To directly test whether PD-1 signaling contributes to lung T_CD8 impairment during HMPV reinfection, we therapeutically blocked this pathway by injecting μMT mice undergoing secondary infection with blocking Abs directed against PD-1 ligands or with isotype control Ab. Naive μMT mice were also infected for comparison of T_CD8 responses and viral titers. Both the F528-specific and N11-specific T_CD8 responses were quantified. Primary-infected μMT mice possessed F528-specific T_CD8 that were mostly functional (Fig. 5A, 5B), as previously reported for early times p.i. (19). Secondary lung T_CD8 in isotype-treated, reinfected mice were more impaired than primary T_CD8 (Fig. 5A, 5B), indicating that elevated inhibitory receptor coexpression during reinfection (Fig. 3) is correlated with decreased functionality. Anti–PD-L treatment resulted in enhanced degranulation ability (i.e., CD107a mobilization) and greater IFN-γ production for N11-specifc T_CD8 compared with isotype-treated mice (Fig. 5A). Treatment also resulted in a greater percentage of functional F528- and N11-specific T_CD8 (Fig. 5B). TNF production was significantly enhanced for F528-specific T_CD8 and trended higher for N11-specific cells. The absolute numbers of T_CD8 were similar between isotype- and anti–PD-L–treated mice, although the number of functional N11-specific T_CD8 was higher in anti–PD-L–treated mice (Fig. 5C). PD-1 was more highly expressed during secondary infection, and PD-L blockade further increased PD-1 expression (Fig. 5D), as previously described for primary infection (19). Surprisingly, at day 6 p.i., isotype-treated, reinfected mice still exhibited lung viral titers indistinguishable from primary infected μMT mice (Fig. 5E). Anti–PD-L treatment reduced lung viral titers ∼3-fold in reinfected mice, suggesting that PD-1 signaling impairs the ability of secondary effector lung T_CD8 to effectively reduce viral replication. Despite the increased numbers of functional T_CD8, anti–PD-L was not associated with exacerbated lung histopathology (data not shown). These results indicate that PD-1 signaling impairs the T cell response during reinfection and may therefore contribute to respiratory virus reinfection.

Vaccine strategies that only elicit humoral immune responses against the related virus RSV have thus far proven unsuccessful and potentially hazardous (40), highlighting the need for better understanding of the contribution of T cells to protective immunity against RSV and HMPV. Recently, a peptide vaccination strategy against RSV proved highly effective in mice when given close to the time of challenge infection; however, the efficacy waned when mice were challenged several weeks later (41). We therefore tested whether PD-1 was responsible for the decreased effectiveness of T_CD8-directed peptide vaccination. WT and PD-1^−/− mice were immunized i.v. with F528 TriVax (F528 peptide + anti-CD40 Ab + polyinosinic:polycytidylic acid), and the T_CD8 response was monitored in peripheral blood. PD-1^−/− mice were used to ensure PD-1 signaling did not occur throughout the entire experiment, from vaccine prime to the end of viral challenge. We found that WT mice generated more F528-specific T_CD8 than did PD-1^−/− mice, which was true in three independent experiments (Fig. 6A). PD-1 may temper some of the strong stimulatory signals received during priming that potentially cause activation-induced cell death (42). Five days post-HMPV challenge, we detected a greater overall F528-response in WT mice compared with PD-1^−/− mice as determined by tetramer staining (Fig. 6B), which corresponded to the greater magnitude of the initial immunization. There were similar percentages of degranulating or IFN-γ–producing F528-specific T_CD8 in both groups. However, calculation of the percentage of functional lung T_CD8 (which takes into account the different magnitude of tetramer response) revealed that F528-specific T_CD8 were more functional in PD-1^−/− mice (Fig. 6C). The absolute numbers of T_CD8 were similar between WT and PD-1^−/− mice (Fig. 6D).

To determine whether PD-1 affects the ability of the vaccine-elicited T_CD8 to control viral replication, we measured viral titers in vaccinated WT and PD-1^−/− mice as well as in unvaccinated or control vaccinated mice (Fig. 6E). We found that in all the groups examined, only PD-1^−/− mice receiving F528 TriVax had decreased viral titers at day 5 postchallenge. Despite the increased percentage of functional T_CD8, lung histopathology was comparable in the WT and PD-1^−/− mice (data not shown). These results suggest that PD-1 limits the effectiveness of antiviral T_CD8 elicited by peptide vaccination. Interestingly, F528 was not protective in WT mice, unlike an immunodominant RSV epitope tested using TriVax in BALB/c mice (41, 43).

To determine whether PD-1 impairs secondary effector T_CD8 responses during reinfection with other respiratory viruses, we primed mice with PR8 influenza (H1N1) virus and challenged with the heterologous x31 (H3N2) influenza strain. The percentage and absolute number of H2-D^b/NP_366–374 tetramer-specific T_CD8 was similar between WT and PD-1^−/− mice (Fig. 7 A, 7B). However, the percentage and number of functional T_CD8 was higher in PD-1^−/− mice (Fig. 7).

Reinfections with respiratory viruses are extremely common among humans of all ages. For influenza virus, progressive evolution through RNA genome mutations leads to antigenic drift and immune escape (44). However, paramyxoviruses, including HMPV and RSV, do not exhibit the same degree of genetic evolution and remain antigenically stable over decades (11, 45, 46). Healthy adults who have experienced natural RSV infection have been experimentally challenged with the same RSV strain and productively reinfected repeatedly over months, even within 2 mo of the first infection (14). Thus, there is limited protection against reinfection with antigenically identical viruses. Many theories have been proposed to explain this phenomenon, including waning humoral Ab, insufficient IgA production at mucosal surfaces, and viral evasion of innate immunity (40). HMPV reinfection in children causes illness at an equal rate to primary infection (10), highlighting the importance of better understanding the mechanisms that contribute to this proclivity. Reinfection with HMPV occurs in healthy and immunocompromised humans, despite the presence of serum Ab (8, 15, 47). This may occur because of limited cross-protective immunity between different strains of HMPV or may indicate that Ab-mediated protection is not sufficient to prevent HMPV infection.

T cell immunity is important for respiratory virus clearance and resolution of infection. HMPV infections are more severe, and even fatal, in HIV-infected or immunocompromised patients (2, 7, 48, 49). The contribution of T cells to protection against HMPV in humans remains poorly defined. However, we previously discovered that T_CD8 are significantly impaired in the respiratory tract during HMPV infection of mice and this impairment was mediated by the inhibitory receptor PD-1 (19), suggesting that optimal T cell immunity may be compromised by inhibitory pathways present in the immune system. Furthermore, the PD-1 pathway is activated in humans with severe viral LRI (19), indicating a potential role for PD-1 signaling during human respiratory virus infections. A previous report using a mouse model of vaccinia virus infection found that primary and memory T_CD8 were increased in number and function in PD-1^−/− mice (50). However, that study used primary i.n. vaccinia virus infection but secondary i.p. infection and thus did not directly address reinfection of the respiratory tract. Thus, a key unanswered question is whether inhibitory receptor–mediated T cell impairment contributes to respiratory virus reinfection.

We found that lung T_CD8 highly express numerous inhibitory receptors, including PD-1, LAG-3, TIM-3, and 2B4. PD-1, LAG-3, and 2B4 were more highly expressed by secondary lung T_CD8. TIM-3 (31) and LAG-3 (32) have been shown to limit the magnitude of the T_CD8 response in models of LRI caused by influenza virus and Sendai virus, respectively. The NK cell receptor 2B4 can deliver inhibitory or costimulatory signals to T_CD8 depending on the isoform expressed (50), but the role of NK-expressed 2B4 during acute LRI has not been explored. All of these receptors are coexpressed with PD-1 and contribute to T cell exhaustion during chronic infection (26, 28, 29). The mechanisms regulating expression and signaling via these receptors and the role of PD-1 signaling in CD4⁺ T cells during respiratory viral infection is not known.

Interestingly, PD-1 is poised for rapid re-expression at the gene expression level because of demethylation of its promoter during primary infection (34). Such epigenetic alterations could explain the higher and more rapid expression of these receptors by secondary effector T_CD8. In addition, inhibitory receptor expression increases over time during primary infection (J.J. Erickson, unpublished observations), suggesting that the increased expression by secondary effector T_CD8 could be due to more rapid differentiation from memory precursors compared with naive T_CD8 during primary infection.

Despite coexpression of multiple inhibitory receptors, we found that only PD-1 impairs T_CD8 early during reinfection. Blockade of these receptors at later time points does result in increased IFN-γ production by lung T_CD8 (data not shown), indicating that they are indeed capable of delivering inhibitory signals. One possible explanation for why these receptors showed no functional role could be that their ligands are not expressed at a high enough level in the lung to mediate receptor interaction. We previously showed that PD-L1 is upregulated by infection and peaks at day 7 (19). The ligands for LAG-3, TIM-3, and 2B4 are MHC class II, galectin-9, and CD48, respectively. It may take a longer amount of time for each of these ligands to be sufficiently upregulated to cause inhibitory signaling than the PD-1:PD-L1 interaction. Alternatively, the downstream mediators of the signaling pathways used by TIM-3, LAG-3, and 2B4 may take additional time to be expressed and trafficked to the correct subcellular location. It is interesting to note that although each of these receptors can contribute to T cell dysfunction, PD-1 is capable of limiting all of the cellular functions associated with T cell exhaustion single-handedly. A recent study showed that increasing the level of surface-expressed PD-1 could directly reduce the ability of T cells to produce cytokines, proliferate, and be cytotoxic (51). Therefore, PD-1 may solely contribute to early impairment, but additional inhibitory receptors may be activated to maintain this impairment at later time points. This theory requires further investigation.

Importantly, we show that blockade of PD-1 signaling significantly improves T_CD8 functionality and viral clearance in a mouse model of HMPV reinfection. This suggests that PD-1–mediated T_CD8 impairment directly contributes to respiratory virus reinfection. Use of the μMT mouse strain overcame the major barrier to studying HMPV reinfection in rodents: Ab-mediated virus neutralization. A limitation to this model is the complete lack of a B cell compartment, which could have unintended effects on the immune system. For instance, the microbiota of these mice are significantly different compared with WT mice (52), which could affect numerous other immune cell types. Care must be taken in extrapolating these results to humans who are not B cell deficient. B cells can also provide costimulation to T cells, which could alter the functionality and magnitude of the T cell response. However, by directly comparing primary infected and reinfected μMT mice, we demonstrated that despite a quantitatively greater secondary T_CD8 response, virus titers at day 5 and 6 postreinfection were indistinguishable from primary infection. Blocking PD-1 overcame this, allowing for better viral control. Furthermore, PD-1^−/− mice reinfected with heterosubtypic influenza virus exhibited greater T_CD8 functionality compared with WT mice.

In addition, PD-1 limited the effectiveness of a peptide vaccination formula. TriVax was previously shown to completely protect against RSV infection when mice were challenged soon after the time of vaccination (41). The T_CD8 response in that report was high in magnitude and mostly functional, suggesting that a large antiviral T_CD8 response that prevents initial viral infection also prevents T_CD8 impairment. This may be due to low levels of viral Ag following viral clearance. Indeed, rapid clearance of LCMV prevents functional impairment as well (53, 54). However, when mice were challenged with RSV several weeks after TriVax administration, the efficacy was greatly reduced. In contrast, we found that F528 TriVax was not protective in WT mice. However, when PD-1 signaling was removed, we found that HMPV-specific T_CD8 were more functional and viral replication was reduced. This suggests that the effect on viral titers might be even more pronounced if the PD-1^−/− mice generated a similar immune response to F528 TriVax. It is unclear why the WT mice responded with a higher number of epitope-specific cells. TriVax is a potent vaccination strategy that elicits a robust T_CD8 response (42). PD-1 may temper some of the strong stimulatory signals received during priming that potentially cause activation-induced cell death (43). We demonstrate that this may be because of impairment of secondary T_CD8 because viral control was only improved in the absence of PD-1 signaling.

The exact mechanisms allowing respiratory viruses to repeatedly reinfect individuals throughout life have been unclear. Taken together, our findings indicate that PD-1 inhibits secondary effector lung T_CD8 during respiratory virus reinfection. We provide data to suggest that rapid re-expression of multiple inhibitory receptors contributes to the functional impairment of secondary effector T_CD8 in the respiratory tract. In vitro studies only uncovered a role for PD-1 in impairing these cells; thus, it remains to be determined whether TIM-3, LAG-3, and 2B4 play a role in vivo. We focused on the dominant inhibitory receptor PD-1 and found that PD-1 potently inhibits T_CD8 effector functions during reinfection. These results highlight the importance of better understanding the role of PD-1 and other inhibitory receptors in modulating lung T_CD8 effector functions to design more effective vaccines and therapeutics against respiratory viruses. Further work is warranted to explore whether PD-1–mediated T_CD8 impairment contributes to respiratory virus reinfection in humans.

We thank Drs. T. Honjo and M. Boothby for providing mice used in these experiments. We also thank Dr. Jennifer Schuster for assistance with manuscript preparation.

J.V.W. serves on the Scientific Advisory Board of Quidel. All other authors have no financial conflicts of interest.