Viral MHC class I inhibition evades CD8⁺ T-cell effector responses in vivo but not CD8⁺ T-cell priming

Abstract

Although viral MHC class I inhibition is considered a classic immune-evasion strategy, its in vivo role is largely unclear. Mutant cowpox virus lacking its MHC class I inhibitors is markedly attenuated during acute infection because of CD8⁺ T-cell–dependent control, but it was not known how CD8⁺ T-cell responses are affected. Interestingly, we found no major effect of MHC class I down-regulation on priming of functional cowpox virus-specific CD8⁺ T cells. Instead, we demonstrate that, during acute infection in vivo, MHC class I down-regulation prevents primed virus-specific CD8⁺ T cells from recognizing infected cells and exerting effector responses to control the infection.

CD8⁺ T cells play an essential role in the control of many virus infections (1). In naive hosts, there are low numbers of CD8⁺ T cells that express MHC class I-restricted T-cell receptors specific for particular viral peptides. During primary infection, virus-encoded peptides displayed by MHC class I molecules can be recognized by virus-specific CD8⁺ T cells, which then are primed to expand clonally in number via rapid cell division and gain effector functions. They then traffic to areas of virus replication and recognize virally infected cells by virtue of MHC class I molecules displaying viral peptides. CD8⁺ T-cell effector responses aid in viral clearance by direct killing of infected cells and secretion of proinflammatory cytokines (2). Depletion of CD8⁺ T cells often reveals enhanced viral infection as manifested by increased viral replication, morbidity, and lethality, highlighting the importance of CD8⁺ T cells in host immune control of viral infections.

Many viruses have evolved mechanisms to thwart MHC class I expression. In particular, herpesviruses encode multiple inhibitors that target diverse steps in the MHC class I biosynthesis and presentation pathway (3). These inhibitors can be grouped according to their functional effects on MHC class I biosynthesis, such as inhibitors of TAP (transporter associated with antigen processing), and retention of MHC class I molecules in the endoplasmic reticulum (ER), among others. Studies of these viral inhibitors have helped illuminate the mechanisms responsible for normal MHC class I biosynthesis, and it generally has been perceived that these viral mechanisms represent immune evasion from CD8⁺ T-cell control.

Despite progress in dissecting the mechanisms of viral inhibition of MHC class I expression in vitro, the role of endogenous viral MHC class I inhibition during acute infection has been difficult to demonstrate in vivo. This difficulty can be attributed in part to inhibition by human-specific viruses, such as CMV, that do not productively infect small animals that would allow detailed in vivo analysis. However, many studies have been performed with murine cytomegalovirus (MCMV). For example, MCMV-specific CD8⁺ T-cell lines preferentially killed cells infected with an MCMV mutant lacking all three MHC class I immune-evasion genes (m04, m06, and m152) but not cells infected with WT MCMV (4). However, when immunocompetent mice were infected with MCMV lacking m04, m06, and m152, there was essentially no effect on the CD8⁺ T-cell response, organ viral titers, or establishment of latency, with the exception of higher viral titers in the salivary glands of BALB/c mice at a relatively late phase of infection (5). Similarly, recent studies of rhesus CMV in rhesus macaques indicated that viral MHC class I inhibition did not affect primary infection (6). Finally, the murine γ-herpesvirus-68 MHC class I inhibitor (K3) allowed expansion of the number of latently infected cells but had no effect on viral clearance during the acute phase of infection (7). Surprisingly therefore, MHC class I inhibitors from herpesviruses have relatively little effect on acute infections in vivo, leaving some doubt as to the in vivo importance of MHC class I down-regulation by endogenous viral inhibitors.

Although most studies of viral MHC class I-evasion proteins have focused on herpesviruses, related studies of myxoma virus, a member of Leporipoxvirus genus of the Poxviridae family, indicated that the M153R protein is responsible for MHC class I down-regulation during infection (8). When the M153R ORF was disrupted, myxoma virus was much less virulent during acute infection in rabbits. However, the mechanism by which MHC class I down-regulation affects the immune response during acute myxoma virus infection is unclear, presumably because of the lack of available reagents for detailed analysis.

Recent studies on cowpox virus (CPXV) provide an opportunity to dissect the detailed mechanisms by which viral MHC class I inhibition affects acute infection in vivo (9). CPXV belongs to the Orthopoxvirus genus of the Poxviridae family, a genus that includes variola virus, vaccinia virus (VACV), monkeypox virus, and ectromelia virus. Although CPXV has a very wide host range and can cause zoonotic infections of humans and other animal species, its endemic reservoirs are wild rodents (10, 11). Because CPXV has coevolved with its rodent hosts, CPXV infection of rodents is appropriate for study of orthopoxvirus–host interactions. Two endogenous viral proteins, CPXV12 and CPXV203, act synergistically to down-regulate MHC class I molecules during CPXV infection (12–14). CPXV12 binds to the peptide-loading complex and inhibits loading of optimal peptides, whereas CPXV203 retains fully assembled MHC class I molecules in the ER by hijacking the KDEL receptor–ER retention pathway. A mutant CPXV lacking CPXV12 and CPXV203 (Δ12Δ203) can no longer down-regulate MHC class I molecules during infection, indicating that CPXV12 and CPXV203 are the only CPXV proteins responsible for MHC class I down-regulation. In contrast to mutant herpesvirus infection, Δ12Δ203 showed a striking reduction in lethality during primary infection (13). During acute intranasal (i.n.) infection, the reduced virulence of Δ12Δ203 was dependent on CD8⁺ T cells, because their depletion restored lethality to a level comparable to infection of untreated mice with WT CPXV. However, it remained to be determined what aspect of the CD8⁺ T-cell response was affected by endogenous viral MHC class I down-regulation during primary infection with CPXV.

Viral inhibition of MHC class I could affect either of the two major MHC class I-dependent steps in CD8⁺ T-cell control of acute in vivo infection. Inhibition could perturb initial priming and the generation of virus-specific CD8⁺ T cells. A failure to prime the appropriate T-cell response then would result in inadequate numbers of T cells for effective control. Alternatively, viral MHC inhibition may primarily affect CD8⁺ T-cell responses at the effector level by preventing recognition of virally infected cells, blocking functional virus-specific T cells from executing appropriate effector responses to control infection. Thus, it remained to be determined if CPXV-mediated MHC class I inhibition affected CD8⁺ T-cell priming or effector responses, or both.

Herein we report that, surprisingly, endogenous CPXV down-regulation of MHC class I molecules does not inhibit the priming of functional CPXV-specific CD8⁺ T cells during acute infection. Instead our data clearly demonstrate that down-regulation of MHC class I molecules potently inhibits the CD8⁺ T cells at the effector response level during primary virus infection by preventing recognition of virally infected cells in vivo.

Results

Clearance of the CPXV Δ12Δ203 Mutant and Reduced Morbidity Are CD8⁺ T-Cell Dependent.

Deletion of ORFs 12 and 203 (Δ12Δ203) from the CPXV genome greatly reduced CPXV lethality during i.n. infection of C57BL/6 (B6) mice that was dependent on CD8⁺ T-cell control (13). To determine if there was reduced morbidity and enhanced viral clearance with Δ12Δ203 CPXV infections, we infected B6 mice with WT or Δ12Δ203 CPXV. Weight loss was observed in mice infected with both viruses after day 6 postinfection (PI), although mice infected with Δ12Δ203 but not with WT CPXV began to regain their weight at day 8 PI (Fig. S1A). When mice were depleted of CD8⁺ T cells, mice infected with Δ12Δ203 CPXV showed prolonged weight loss similar to that seen in WT CPXV-infected mice treated with the isotype control antibody (Fig. S1B). Interestingly, depletion of CD8⁺ T cells during WT CPXV infection did not affect weight loss; this result is consistent with previously reported data that CD8 depletion does not enhance WT CPXV lethality (13).

The reduction in lethality and morbidity of Δ12Δ203 CPXV was associated with enhanced viral clearance from the lung. Although mice infected with WT or Δ12Δ203 CPXV showed similar lung titers on day 6 PI, Δ12Δ203-infected mice cleared virus completely by day 13 PI (Fig. S1C). In contrast, WT CPXV titers appeared to plateau on day 6 PI and persisted at the same levels even on day 13 PI. Like WT CPXV, the double-revertant virus r12r203 also was not cleared by day 13 PI (Fig. S2A). Together with previous studies showing that r12r203 is as lethal as WT CPXV during i.n. infection (13), these studies indicate that the findings with Δ12Δ203 CPXV are not caused by other changes in the Δ12Δ203 viral genome. All anti-CD8–depleted mice were unable to clear Δ12Δ203 CPXV, whereas Δ12Δ203 CPXV was cleared in mice treated with the isotype control antibody (Fig. S1D). In contrast, anti-CD8 depletion had no effect on WT CPXV lung titers on day 13 PI. Similar results were also found in CD8α-KO mice (Fig. S3). Taken together, these results demonstrate that the reduced morbidity and enhanced viral clearance of Δ12Δ203 CPXV, along with lethality (13), are dependent on CD8⁺ T-cell control, indicating that MHC class I down-regulation by endogenous WT CPXV inhibitors effectively neutralizes CD8⁺ T cells.

MHC Class I Down-Regulation Does Not Cause a Major Defect in CD8⁺ T-Cell Priming During WT CPXV Infection.

To determine if MHC class I down-regulation led to attenuated CD8⁺ T-cell priming during WT CPXV infection, we assessed if the subsequent expansion of virus-specific CD8⁺ T cells was impaired. We followed virus-specific CD8⁺ T cells by staining with H2K^b tetramers loaded with TSYKFESV peptide (Fig. 1A). This antigenic epitope is the immunodominant peptide during VACV infection of B6 mice, and its sequence is completely conserved in CPXV and ectromelia viruses (15). Moreover, when cells from CPXV-infected B6 mice were stimulated ex vivo, the number of CD8⁺ T cells producing IFN-γ in response to TSYKFESV peptide accounted for the vast majority of virus-specific T cells, nearly comparable to the number of virus-specific T cells stimulated by virus (Δ12Δ203)-infected target cells (Fig. 2 B and C). Therefore TSYKFESV also is an immunodominant epitope during CPXV infection of B6 mice. H2K^b-TSYKFESV tetramer staining showed similar numbers of virus-specific CD8⁺ T cells in the spleens, lungs, and draining mediastinal lymph nodes of mice infected with WT CPXV or with Δ12Δ203 CPXV at various times after infection (Fig. 1A ). Therefore, priming was not greatly affected by MHC class I down-regulation on WT CPXV-infected cells.

Fig. 1.

CD8⁺ T-cell priming is not impaired during WT CPXV infection. (A) Tetramer analysis during infection. Eight-week-old female B6 mice were infected i.n. with 5,000 pfu of the indicated viruses. On days 4, 5, 6, and 8 PI, the number of CPXV-specific CD8⁺ T cells in the spleen, lung, and mediastinal lymph node was determined by H2K^b-TSYKFESV tetramer staining, as indicated. Data are representative of multiple individual time point experiments and are shown as mean ± SEM. n = 4 or 5 mice. (B) Tetramer analysis in Batf3-deficient mice. Female and male 8- to 11-wk-old Batf3-KO or B6 mice were infected with 5,000 pfu of the indicated viruses. On days 6 and 8 PI the number of CPXV-specific CD8⁺ T cells in the spleen was determined by H2K^b-TSYKFESV tetramer staining as indicated. Day 6 data are the combined results from two independent experiments. Day 8 data are combined results from three independent experiments. (C) CD8⁺ T-cell IFN-γ production. On day 8 splenocytes from the mice infected in B were stimulated with Δ12Δ203 CPXV-infected DC2.4 cells. The total number of IFN-γ–producing CD8⁺ T cells is shown. Data are the combined results of three independent experiments. The data were analyzed with an unpaired Student t test, P values between comparison groups are shown. Bars represent the mean for each sample group. Numbers followed by an “x” in B and C are the average fold reduction in the CD8⁺ T-cell response to the indicated virus in Batf3-KO mice.

Fig. 2.

CD8⁺ T cells generated during WT CPXV infection are functional. (A) Stimulation with Δ12Δ203 CPXV-infected DC2.4 cells. Female and male 9- to 11-wk-old B6 mice were infected i.n. with 5,000 pfu of WT or Δ12Δ203 CPXV. Mice were killed on day 8 PI, and single-cell lung suspensions were stimulated with DC2.4 cells infected with Δ12Δ203 CPXV. Then CD8⁺ T cells were stained for intracellular IFN-γ and TNF-α. (Upper) The total numbers of IFN-γ–producing CD8⁺ T cells are shown, representing the combined results from five independent experiments. (Lower) Percentages of cells producing both IFN-γ and TNF-α are shown, representing the combined results from three independent experiments. The bars represent the mean of each group. (B) Flow cytometry after stimulation with peptide-pulsed or infected DC2.4 cells. Single-cell lung suspensions were prepared as in A and were stimulated with DC2.4 cells that were pulsed with TSYKFESV peptide (Upper) or with DC2.4 cells infected with Δ12Δ203 CPXV (Lower) and were stained for intracellular IFN-γ. Representative dot plots are shown; the numbers represent the percentage of IFN-γ–producing CD8⁺ T cells. (C) Quantitation of stimulated cells and amount of IFN–γ produced by MFI. Cells were prepared and stimulated as in B. (Upper) Total number of IFN-γ⁺ CD8⁺ T cells. (Lower) MFI of IFN-γ staining. Data are the combined results of two independent experiments and are shown as mean ± SEM. n = 7 or 9 mice. (D) In vivo cytotoxicity assay. On day 8 PI mice mock infected or infected with indicated viruses were injected i.v. with CD45.1⁺ splenocytes that were labeled with 5 nM or 500 nM CFDA. The 500-nM–labeled cells were pulsed with TSYKFESV peptide before injection. Spleens were harvested 4 h after injection, and the number of CFDA-labeled cells was determined by flow cytometry. Histograms show the percentage of CD45.1⁺-, CFDA-high–, and CFDA-low–labeled cell populations at time of harvest from a representative experiment. (E) Target killing in combined experiments. The percentage of target cell (TSYKFESV peptide-pulsed cells) killing is shown. See Materials and Methods for the equation. Bars represent the mean of each sample group. Data shown are combined results from two independent experiments.

One potential explanation for the minimal effect of CPXV down-regulation of MHC class I on virus-specific CD8⁺ T-cell priming is that priming of CPXV-specific CD8⁺ T cells is dependent on dendritic cells (DCs) that are involved in cross-presentation. To examine this possibility, we infected Batf3-KO mice, because they lack the CD8α⁺ and developmentally related peripheral CD103⁺ DC subsets and are defective in cross-presentation (16, 17). H2K^b-TSYKFESV tetramer staining was reduced significantly in the spleens of Batf3-KO mice infected with either WT or Δ12Δ203 CPXV on day 6 and day 8 PI (Fig. 1B). When splenocytes from infected mice were stimulated with Δ12Δ203-infected target cells, the total number of IFN-γ–producing CD8⁺ T cells in the spleens of WT- or Δ12Δ203-infected Batf3-KO mice was greatly reduced also (Fig. 1C), indicating that the overall CD8⁺ T-cell response reflects the results seen with H2K^b-TSYKFESV tetramer staining (Fig. 1B). Viral lung titers were similar in B6 and Batf3-KO mice on day 6, indicating that the results could not be explained easily by changes in overall viral (antigenic) load in Batf3-KO mice (Fig. S4). Therefore, these results suggest CD8α⁺ (or CD103⁺) DCs are involved in priming of CD8⁺ T-cell responses to CPXV, regardless of MHC class I down-regulation.

Functional CPXV-Specific CD8⁺ T Cells Are Generated During WT CPXV Infection.

To assess the effector functions of CPXV-specific CD8⁺ T cells generated after infection of WT mice, we performed cytokine production and cytotoxicity assays. These are relevant functional studies, because we found that the enhanced survival during Δ12Δ203 CPXV infection was largely dependent on the CD8⁺ T-cell effector molecules IFN-γ and perforin: Mice deficient in these molecules no longer can control Δ12Δ203 CPXV (Fig. S5). On day 8 PI, similar numbers of CD8⁺ T cells isolated from the lungs (one of the major sites of infection) of WT- or Δ12Δ203 CPXV-infected mice produced IFN-γ after stimulation with Δ12Δ203-infected DC2.4 cells (Fig. 2A). Similar percentages of these cells also produced TNF-α after stimulation. A high percentage of CD8⁺ T cells produced IFN-γ after stimulation with TSYKFESV peptide-pulsed DC2.4 cells, accounting for the vast majority of virus-specific T cells as assessed by Δ12Δ203-infected DC2.4 stimulation (Fig. 2 B and C). The mean fluorescence intensity (MFI) of IFN-γ staining did not show a statistically significant difference when CD8⁺ T cells from WT- and Δ12Δ203 CPXV-infected lungs were compared, and the MFIs were similar regardless of stimulation with peptide-pulsed or Δ12Δ203-infected DC2.4 cells (Fig. 2C). Overall these results indicate that similar numbers of functional cytokine-producing, virus-specific CD8⁺ T cells are initially recruited to the lung after infection with WT and Δ12Δ203 CPXV.

To gauge the capacity of CPXV-specific CD8⁺ T cells to kill target cells, we injected mice with CD45.1⁺ splenocytes that were unpulsed or pulsed with TSYKFESV peptide, as distinguished by low or high carboxyfluorescein diacetate (CFDA) labeling, respectively (Fig. 2D). At 4 h post injection, more than 95% of peptide-pulsed target cells were eliminated in spleens of all WT- or Δ12Δ203 CXPV-infected mice examined (Fig. 2 D and E). These data further indicate that WT CXPV infection induces priming of virus-specific T cells that are functional in cytotoxicity as well as cytokine production and that these functional virus-specific CD8⁺ T cells are comparable to those generated after infection with Δ12Δ203 CXPV.

CD8⁺ T Cells Produce More IFN-γ in the Lungs of Δ12Δ203 CXPV-Infected Mice than in the Lungs of WT CPXV-Infected Mice.

Our results thus far indicate that MHC class I down-regulation during WT CPXV infection does not result in a large defect in the priming of functional CPXV-specific CD8⁺ T cells. On the other hand, when we assessed IFN-γ levels by ELISA in the lung, a site of viral replication, there were much higher levels of IFN-γ in Δ12Δ203 CPXV-infected mice than in WT CPXV-infected mice (Fig. 3A). The largest amount of IFN-γ was found in Δ12Δ203 CPXV-infected lungs on day 6 PI, the time when high numbers of CPXV-specific CD8⁺ T cells began accumulating in the lung (Fig. 1). High amounts of IFN-γ were not detected in the lungs of r12r203-infected mice on day 6 PI (Fig. S2B). Anti-CD8 depletion largely eliminated the difference between WT- and Δ12Δ203 CPXV-infected mice (Fig. 3B), despite similar viral titers at this time (Fig. 3C). These results indicate that most of the IFN-γ in the lungs of Δ12Δ203 CPXV-infected mice is produced by CD8⁺ T cells, suggesting that CPXV-induced MHC class I down-regulation directly affects CD8⁺ T-cell effector responses at a major site of virus replication.

Fig. 3.

CD8⁺ T cells produce more IFN-γ in the lungs of Δ12Δ203 CPXV-infected mice than in the lungs of WT CPXV-infected mice. (A) IFN-γ levels in lung homogenates during infection. Nine-week-old female B6 mice were infected i.n. with 5,000 pfu of WT or Δ12Δ203 CPXV as indicated. Lung homogenates were generated and used in IFN-γ ELISA on the indicated day PI. Data are representative of two independent experiments and are shown as means ± SEM. n = 3 or 4 mice. (B) IFN-γ production is CD8⁺ T-cell dependent. Nine-week-old female B6 mice were infected i.n. with 10,000 pfu of WT or Δ12Δ203 CPXV as indicated. Mice were treated 1 d before infection and on day 4 PI with an anti-CD8β-depleting antibody or control antibody. Lungs were harvested on day 6 PI, and homogenates were used in IFN-γ ELISA. Data are shown as mean ± SEM. (C) Viral lung titers were determined in the mice shown in B. Bars represent the mean of each sample group. (D) Direct ex vivo IFN-γ intracellular staining. Nine-week-old female B6 mice were mock infected or infected with 10,000 pfu of WT or Δ12Δ203 CPXV. Lungs were harvested on day 6 PI, and single-cell suspensions were intracellularly stained directly ex vivo without additional stimulation. Dot plots from two single representative mice from each infection group are shown. Numbers represent the percentage of IFNγ⁺CD8⁺ T cells in the sample. (E) Total number of IFN-γ⁺CD8⁺ T cells per lung. Analysis was performed as in D. Data are the combined results from two independent experiments and are shown as mean ± SEM. n = 4 mock, n = 9 WT CPXV, and n = 10 Δ12Δ203-infected mice. CPXV. The data were analyzed with an unpaired Student t test; P values between comparison groups are shown.

To examine the effector response more directly, we assessed IFN-γ production by intracellular cytokine staining of CD8⁺ T cells directly ex vivo without additional stimulation (Fig. 3 D and E). There were 3.8 times more CD8⁺ T cells producing IFN-γ in the lungs of Δ12Δ203 CPXV-infected mice than in the lungs of WT CPXV-infected mice. Note that at this time there were similar viral titers and numbers of virus-specific CD8⁺ T cells by tetramer staining in mice infected with either virus (Figs. 1 and 3C). These results suggest that, even though functional WT CPXV-specific CD8⁺ T cells are generated and recruited during infection, they are impaired at the site of infection and cannot execute effector responses.

CPXV-Primed CD8⁺ T Cells Preferentially Reduce the Severity of Δ12Δ203 CPXV Lesions.

To examine further the ability of CPXV-specific CD8⁺ T cells to recognize WT- or Δ12Δ203 CPXV-infected cells in vivo, we adapted the herpesvirus flank infection method (18). When we scratched adjacent areas of skin in the presence of WT or Δ12Δ203 CPXV, a single lesion developed at each site of skin disruption and virus inoculation after ∼3–4 d and remained localized without spreading to other areas of intact skin (Fig. 4C). In these mice both WT and Δ12Δ203 CPXV should contribute to CD8⁺ T-cell priming. WT and Δ12Δ203 CPXV lesions initially looked similar and had comparable viral titers on day 6 PI (Fig. 4 B and C), but by day 7 PI the Δ12Δ203 CPXV lesions began to show signs of scab formation and healing. By day 9 PI, 90% of Δ12Δ203 CPXV lesions, but none of the WT CPXV lesions, had lost their scab (Fig. 4 A and C). At this time, in 9 of 10 mice viral titers either were much lower in Δ12Δ203 CPXV lesions than in WT CPXV lesions or were undetectable (Fig. 4B), recapitulating the clinical sign after VACV inoculation in humans that spontaneous scab separation indicates a loss of lesion infectivity (19). The enhanced healing of Δ12Δ203 lesions was CD8⁺ T-cell dependent, because administration of anti-CD8–depleting antibody clearly delayed healing and viral clearance (Fig. 4 D, E, and F). r12r203 lesions did not show enhanced healing or viral clearance and were comparable to WT CPXV-induced lesions (Fig. S2 D, E, and F). These data suggest that, despite priming by WT and Δ12Δ203 CPXV, virus-specific CD8⁺ T cells were able to control Δ12Δ203 but not WT CPXV lesions.

Fig. 4.

CPXV-primed CD8⁺ T cells display less control of WT CPXV lesions. (A) Scab loss. Nine-week-old female B6 mice were infected on two adjacent areas of skin with 10-uL droplets containing 20,000 pfu of WT or Δ12Δ203 CPXV. Lesions were monitored daily; scab loss for each lesion is shown. n = 20 mice until day 9, then n = 10 mice. Data are representative of four independent experiments. (B) Viral titers. Skin lesions were collected separately on day 6, 9, and 13 PI, and viral titers were determined by plaque assay. The only Δ12Δ203 lesion with a scab still remaining on day 9 and the two WT lesions with scabs remaining on day 13 are denoted by arrows. Data were analyzed with an unpaired Student t test; P value between comparison groups is shown. (C) Photographs of skin lesions in two representative mice from the experiment described in A and B on days 6 and 9 PI (D and E) Mice were treated with an anti-CD8β-depleting antibody or a control antibody 2 d before infection as described in A. (D) Percentage of scab loss on day 9. (E) Lesion titers on day 9. Data in D and F are representative of two independent experiments; n = 10 mice. (F) Pictures of two representative mice from each treatment group.

To test if WT CPXV-primed CD8⁺ T cells could preferentially control Δ12Δ203 CPXV lesions, we performed adoptive transfer experiments with CD8⁺ T cells from mice infected with WT CPXV or control MCMV. On day 7 PI, we enriched CD8⁺ T cells by negative selection for i.v. transfer into Rag1-deficient mice. One day posttransfer, their skin was scratched with WT and Δ12Δ203 CPXV (Fig. 5A). Mice that received CD8⁺ T cells from CPXV-infected mice showed diminished Δ12Δ203 CPXV lesions but not diminished WT CPXV lesions, whereas WT and Δ12Δ203 CPXV lesions developed similarly in mice that received CD8⁺ T cells from MCMV-infected mice. These differences were noticeable beginning on day 5 after skin infection. The diminished Δ12Δ203 CPXV lesions were dependent on the transferred CD8⁺ T cells, because CD8 depletion at 16 h after transfer resulted in inability to control Δ12Δ203 CPXV lesions (Fig. 5B). Taken together, these data indicate that CPXV-specific CD8⁺ T cells generated during WT CPXV infection are able to control CPXV lesions, but only in the absence of MHC class I down-regulation.

Fig. 5.

WT CPXV-primed CD8⁺ T cells can control Δ12Δ203 but not WT CPXV lesions. Adoptive transfer experiments. (A) Nine-week-old female B6 mice were infected i.n. with 10,000 pfu WT CPXV or were infected i.p. with 100,000 pfu of MCMV. Spleens were harvested 7 d postinfection and were enriched by negative selection for CD8⁺ T cells, which then were transferred into naive Rag-1–deficient mice. At 24 h after transfer, mice were infected on two adjacent areas of skin with 10-uL droplets containing 20,000 pfu of WT or Δ12Δ203 CPXV. Lesions were monitored daily. Representative photographs of mice that received MCMV- or CPXV-immune CD8⁺ T cells are shown on days 5 and 9 PI. Data are representative of two independent experiments. n = 3 mice. (B) Negatively enriched CD8⁺ T cells from mice infected with CPXV as described in A were transferred into Rag-1–deficient mice, and 16 h later mice were treated with an anti-CD8β-depleting or control antibody. At 24 h after transfer, mice were infected as described in A. Photographs of two representative mice per treatment group on day 9 PI are shown. Data are representative of two independent experiments. n = 5 or 6 mice.

Discussion

Despite the widespread impression that virus-induced MHC class I down-regulation is an immune-evasion strategy, its in vivo effects have been poorly understood. This lack of understanding has stemmed largely from in vivo studies in which viruses lacking these inhibitors do not demonstrate the expected enhanced virulence. The capacity of CPXV to regulate MHC class I expression in a manner that clearly affects in vivo lethality during acute infection allowed us to examine this issue in detail. We found that reduced morbidity and enhanced Δ12Δ203 CPXV clearance correlated with the reduction in Δ12Δ203 CPXV lethality and was dependent on CD8⁺ T cells. Surprisingly, however, MHC class I down-regulation did not greatly impair priming and subsequent expansion of functional, CPXV-specific CD8⁺ T cells. Although these virus-specific CD8⁺ T cells generated during WT CPXV infection were functionally competent to produce cytokines and generate cytotoxicity and were recruited to a site of viral infection, they were inhibited in executing effector responses. These data indicate that the enhanced virulence associated with viral MHC class I down-regulation is caused by an inability of the primed virus-specific CD8⁺ T cells to recognize infected cells and exert effector responses to control the infection.

Effector CD8⁺ T cells can kill target cells after recognizing as few as three MHC class I molecules presenting agonist peptide (20). Therefore the finding that MHC class I down-regulation during CPXV infection is able to block CD8⁺ T cells at the effector stage is striking. Although additional studies are required to determine how memory T-cell responses are affected by MHC inhibition, our data are relevant to vaccine design and analysis because virus-specific CD8⁺ T-cell responses might be generated adequately by vaccination, but these virus-specific T cells might be compromised during subsequent natural infection with viruses capable of MHC inhibition.

Although CD8⁺ T cells generated during WT CPXV infection appear to be functional, CD8⁺ T-cell–depletion experiments in B6 mice showed that the elimination of these cells has no additional impact on mortality (13), morbidity, or viral clearance in WT CPXV-infected B6 mice. By themselves, these findings could lead to the interpretation that CD8⁺ T cells have no role in the control of WT CPXV infections. On the other hand, MHC class I down-regulation by WT CPXV clearly affected virulence, because deletion of the two CPXV ORFs responsible for down-regulation of MHC class I had a profound effect in vivo, allowing CD8⁺ T cells to control Δ12Δ203 infection. These apparently contrasting results now are best explained by the essentially complete neutralization of the CD8⁺ T-cell effector responses resulting from the capacity of WT CPXV to down-regulate MHC class I. These results also raise a caution about interpreting related studies that may be interpreted prematurely and erroneously as meaning that an immune component has no role in antiviral defense because of already inherent neutralization by unappreciated viral-evasion strategies.

Our results indicated that virus-specific CD8⁺ T-cell priming is not greatly affected by MHC class I down-regulation on infected cells. This observation is in contrast with prior results with VACV bearing the US11 or US2 MHC class I inhibitors from human CMV. Expression of either of these inhibitors caused a reduction in CD8⁺ T-cell priming (21, 22). However, this reduction in priming was dependent on the route of infection used (22), and these studies were compromised by the ectopic expression of human herpesvirus MHC inhibitors into VACV, unlike the studies here which involve viral inhibitors naturally present in CPXV. Intravital microscopy experiments of VACV-infected mice also have revealed that infected and uninfected DCs interact with CD8⁺ T cells during priming (23, 24). These studies and others (25, 26) imply that both direct presentation by infected cells and cross-presentation by uninfected DCs can be used for CD8⁺ T-cell priming during VACV infection of mice; however, it should be noted that VACV has been propagated extensively, and the natural endemic host of VACV is unknown (27), potentially obscuring the interpretation of host–pathogen studies.

Because priming of the CPXV-specific response was not greatly affected by the endogenous capacity of CPXV to down-regulate MHC class I expression potently, we currently favor a model in which cross-presentation is the main pathway used during CPXV infection. Given that our studies now demonstrate that effector CD8⁺ T cells are blocked effectively in vitro and in vivo, direct priming also should be susceptible to CPXV-induced MHC class I down-regulation, and thus cross-presentation would be the means to allow priming of CD8⁺ T cells despite viral MHC class I down-regulation during acute WT CPXV infection. As is consistent with this hypothesis, our results show that WT- and Δ12Δ203 CPXV-specific CD8⁺ T-cell priming is reduced by up to fivefold in Batf3-deficient mice that lack CD8α⁺ and CD103⁺ cross-presenting DCs (16, 17). Although we favor cross-priming as the most likely explanation for these findings, an alternative hypothesis is that these DCs may be infected directly by CPXV. If so, CPXV 12 and 203 proteins may be unable to down-regulate MHC class I molecules efficiently in these directly infected DCs but remain able to down-regulate MHC class I molecules efficiently in other cell types in the lung and thus evade CD8⁺ effector responses but not direct priming. Although further studies will be required to examine these complex possibilities, detailed physiologically relevant analysis will be enabled by the distinctive, intrinsic capacity of WT CPXV to block MHC class I expression and markedly alter CD8⁺ T-cell responses in vivo in a relevant rodent species.

Although the potent down-regulation of MHC class I molecules allows CPXV to evade CD8⁺ T-cell effector responses, it also would leave infected cells open to attack by natural killer (NK) cells. This effect is a basic tenet of the missing-self hypothesis in which NK cells survey the ubiquitous expression of MHC class I molecules preventing NK cell killing of normal targets (28). In the absence of MHC class I, NK cells are able to attack. These effects now can be explained by MHC-specific inhibitory receptors on NK cells that provide negative signals that override NK cell activation receptors such as NKG2D that otherwise stimulate NK cells when NKG2D engages its stress-induced ligands (29, 30). Interestingly, CPXV expresses Orthopoxvirus MHC class I-like protein (OMCP) (31), a soluble, high-affinity antagonist of NKG2D. Moreover, other orthopoxviruses containing homologs of CPXV203 also encode OMCP molecules. Although further details should be forthcoming, these observations provide major support for the in vivo relevance of missing-self in the context of ongoing viral infections, because these viral processes affect the effector responses of both CD8⁺ T cells and NK cells rather than separable events such as T-cell priming versus NK effector responses.

In this regard, the previous inability to discern in vivo effects of MHC class I down-regulation by MCMV may be caused by the pleiotropic roles of MCMV inhibitors such as m152 on both MHC class I expression and NK cell responses (32, 33). In vivo studies also are complicated further by the differential capacity of multiple arms of the immune system, such as NK cells, CD4⁺ T cells, and CD8⁺ T cells, to compensate for each other in controlling different infections (34, 35). It also remains possible that other inherent differences between orthopoxviruses and herpesviruses, such as the ability of herpesviruses to establish latency, may account for differences in the role of inhibition of MHC class I-dependent responses to these major classes of dsDNA viruses.

The CPXV Brighton Red strain used in the current study originally was isolated in 1937 from the hand of a milker in the United Kingdom near the area where Edward Jenner conducted the initial vaccination studies with CPXV in 1798 (36). Although Jenner observed the protective effects of CPXV against variola virus, he surprisingly also noted several instances in which CPXV infection paradoxically was not protective against recurrent acute infections with CPXV itself (37). On the other hand, variola virus lacks homologs of the CPXV MHC class I inhibitors (9, 12–14), suggesting that primed, cross-reactive, virus-specific T cells should be unrestrained in conferring protection against variola virus. Thus, our current studies suggest that the capacity of WT CPXV to prime a CD8⁺ T-cell response but evade CD8⁺ T-cell effector responses may help provide a possible explanation for Jenner’s 200-year-old puzzling observation.

Methods

Cell Lines, Antibodies, Viruses, and Mice.

Detailed information on sources of cell lines, antibodies, viruses, and mice is available in SI Methods. Batf3^−/− mice crossed to the C57BL/6 background were kindly provided by Kenneth Murphy (Washington University in St. Louis, St. Louis, MO) (16). Mice were maintained under specific pathogen-free conditions and used at 8–11 wk of age. Mice were sex and age matched for each experiment.

Mouse Infection and Depletion of CD8⁺ T Cells.

For i.n. infections, mice were anesthetized i.p with ketamine/xylazine. Mice were inoculated with 50 μL of virus diluted in minimum essential medium (MEM) (Mediatech). For skin infections, fur was removed 1 d before infection by trimming with clippers followed by treatment with Nair (Church and Dwight Company). A 10-μL droplet of MEM containing 20,000 pfu of virus was placed on the skin, and a 27-gauge needle was dragged gently through the droplet 20 times in the shape of an asterisk. For MCMV infections, 10⁵ pfu was diluted into 200 μL of PBS solution (Mediatech) and was injected i.p. For CD8-depletion studies, 0.5 mg of H35, a monoclonal antibody specific for CD8β, or SFR3-DR5 isotype control was administered i.p. every 4 d beginning 1 d before infection. Depletion efficiency was >95%, as determined by flow cytometry. Animal studies were approved by the Animal Studies Committee at Washington University in St. Louis.

Virus Isolation from Mice.

Tissue collected from mice was stored at −80 until processing. Lungs were weighed, and 1 mL of MEM was added to each sample. Lungs were disrupted using a MiniBeadbeater-8 (Biospec Products) for 2 min. Skin lesions were collected individually and placed into tubes containing 1 mL MEM. Skin lesions were disrupted using a MiniBeadbeater-8 for 2 min, sonicated for 1 min, frozen/thawed twice, and then sonicated again for 1 min. Supernatants were serially diluted in MEM and used in plaque assays.

Flow Cytometry.

H2K^b-TSYKFESV tetramers were produced in the laboratory or were obtained from the National Institutes of Health Tetramer Core Facility (Emory University, Atlanta, GA). Before lung tissue collection, mice were perfused with PBS. Lungs were rinsed in PBS and then were minced into small fragments. Fragments were digested with 22.4 μg/mL DNase I (type II; Sigma) and 0.7 mg/mL collagenase (Sigma) at 37 °C for 30–45 min. Digested lung fragments were passed through 70-μm cell strainers, and the red blood cells were lysed (0.15 M NH₄Cl, 10 mM KHCO₃, and 0.1 mM EDTA). Spleens and lymph nodes were passed through 70-μm strainers without digestion, and then red blood cells were lysed before staining. For intracellular staining of IFN-γ, cells were fixed and permeabilized with using Cytofix/Cytoperm solution (BD Biosciences). All flow cytometry experiments were analyzed with FlowJo software (TreeStar).

Cytokine Production Assays.

Single-cell lung suspensions were generated as described above, and 10⁶ viable lung cells resuspended in RPMI-1640 (Mediatech) containing 10% (vol/vol) FBS and 2-Mercaptoethanol were added to 96-well V-bottomed plates. Then 5 × 10⁵ DC2.4 cells that had been pulsed with 0.1 ug/mL TSYKFESV (New England Peptide) or had been infected for at least 4 h earlier with Δ12Δ203 CPXV (multiplicity of infection of 5) were added to lung cells and placed at 37 °C, 5% CO₂. One hour later GolgiPlug (BD Biosciences) was added to each well. Five hours later cells were stained on ice for CD8α and CD3 surface markers and then were fixed/permeabilized and stained overnight for IFN-γ and TNF-α.

In Vivo Cytotoxicity Assays.

CD45.1⁺ splenocytes were isolated from naive B6-LY5.2/Cr mice as described above and then were labeled with 500 nM or 5 nM CFDA (Molecular Probes). The cells labeled with 500-nM CFDA were pulsed with 1 uM TSYKFESV peptide for 1 h at 37 °C, 5% CO₂. Cells labeled with 500 nM and 5 nM were mixed at a 1:1 ratio, and then 10⁷ cells were injected i.v. into mice that had been mock-infected or infected 8 d before with WT or Δ12Δ203 CPXV. Mice were killed 4 h after injection, and splenocytes were processed as described above and stained for CD45.1⁺ donor cells. The percentage of target cell killing was calculated as [1 − (CFDA^hi/CFDA^lo)] × 100 (38).

IFN-γ Assays.

Lungs were harvested at the indicated days PI and were homogenized for 2 min in a MiniBeadbeater-8. Virus was inactivated from supernatants with 0.1% Nonidet P40 (Sigma) for 15 min. The amount of IFN-γ in the supernatant then was determined using Ready-SET-Go Mouse IFN-γ ELISA kits (eBioscience), following the manufacturer’s instructions. For direct ex vivo IFN-γ intracellular staining, mice were perfused with PBS, and lungs were harvested and rinsed in PBS. Lungs were minced immediately over ice and then were forced through 70-μm strainers in the presence of RPMI-1640 medium containing 10% FBS, 2-Mercaptoethanol, and GolgiPlug. Cells were kept on ice until all samples were processed. Cells were stained for CD8α and CD3 and then were fixed/permeabilized and stained for IFN-γ for flow cytometry as described above.

CD8⁺ T-Cell Adoptive Transfer Experiments.

Splenocytes were removed from female B6 mice that had been infected 7 d earlier with WT CPXV or MCMV. CD8⁺ T cells were enriched by negative selection using the CD8⁺ T-cell isolation kit II (Miltenyi Biotec) following the manufacturer’s instructions. CD8⁺ T cells were enriched to 72–91% purity. Then 3–5 ×10⁶ cells were i.v. transferred into B6.129S7-Rag1^tm1Mom/J mice. Approximately 24 h after transfer mice were infected via skin scratch as described above.