Norovirus Immune Evasion


Noroviruses (NVs) are major global pathogens that cause an estimated 267 million infections and 200,000 deaths each year (Debbink et al., 2012). Vaccine efforts have shown promise, but complete protection against homologous challenge has not been achieved (Atmar et al., 2011). Moreover, volunteer and epidemiologic studies have generated conflicting data regarding the durability and protective capacity of NV immunity, while recent reports of persistent NV shedding by immunocompetent individuals have challenged the traditional view of NVs as acute and self-limited pathogens (Karst et al., 2014). These data suggest the existence of viral mechanisms of immune evasion, however the immunological determinants of NV clearance versus persistence remain poorly understood. Most studies have used antibodies as markers of pre-existing NV immunity and have shown that, depending on the setting, such responses are neither necessary nor sufficient for protection (Atmar et al., 2011). In contrast, cellular immunity, particularly at mucosal sites, remains largely unexplored.

We and others have used the mouse model of NV infection (MNV) to study MNV clearance versus persistence (Tomov et al., 2013). Early studies showed that all aspects of adaptive immunity are required for clearance of the non-persisting strain MNV-CW3 (Karst et al., 2014). On the other hand, a related strain, MNV-CR6, causes chronic infection even in immunocompetent mice. Recent studies have shown that MNV-CR6 persistence depends on the enteric microbiota and can be prevented by pre-treating mice with antibiotics (Baldridge et al., 2015, Jones et al., 2014). It remains to be defined whether these findings reflect direct bacterial-viral interactions (Kuss et al., 2011) or commensal modulation of the host antiviral immune response (Abt et al., 2012). In addition, persistent MNV-CR6 infection can be cleared by exogenous interferon-lambda (Nice et al., 2015), however this phenomenon is mediated by innate immunity and does not afford protection against subsequent re-challenge with the same strain. Thus, adaptive immune responses will be necessary to achieve durable immunity against MNV.

A key challenge to understanding MNV persistence is our limited knowledge of the host-pathogen interactions responsible for immune failure during chronic infection. Myeloid cells, B cells, and epithelial cells can support mouse and/or human NV replication in cell culture (Ettayebi et al., 2016, Jones et al., 2014), but the relevance of these cell types to persistence in vivo is unclear, and the precise cellular identity and anatomical location of the viral reservoir remain unknown. The identification of CD300lf as an MNV cellular receptor is a major step toward addressing this issue (Orchard et al., 2016). However, it is unclear whether CD300lf is sufficient to explain viral replication during established chronic infection as CD300lf expression is largely restricted to dendritic cells (DCs) (Gasiorowski et al., 2013) but persisting MNV-CR6 replicates in nonhematopoietic cells in vivo (Nice et al., 2015).

In previous studies, we demonstrated that the non-persisting strain MNV-CW3 induces robust virus-specific CD8+ T cell responses in the intestine (Tomov et al., 2013, Osborne et al., 2014). In contrast, infection with the persisting strain MNV-CR6 is associated with substantially fewer and less-functional virus-specific CD8+ T cells, suggesting that suboptimal T cell responses might contribute to viral persistence (Tomov et al., 2013). However, as the sequence of the immunodominant P1519 epitope differs between these two MNV strains, it was unclear whether the weak CD8+ T cell response to MNV-CR6 was due to intrinsic CD8+ T cell dysfunction or suboptimal epitope binding. In the current study, we have addressed this issue by engineering acute and chronic MNV strains that share the same immunodominant CD8+ T cell epitope. Using these strains, we demonstrate that improving the magnitude of the primary CD8+ T cell response did not prevent viral persistence. Moreover, virus-specific CD8+ T cells from chronic MNV infection developed a distinct transcriptional and phenotypic signature compared to memory CD8+ T cells generated during acutely-resolved infection. These cells showed strong similarity to inflationary effector CD8+ T cells responding to mouse cytomegalovirus (MCMV) infection. Consistent with these transcriptional features, virus-specific CD8+ T cells from chronic MNV infection remained responsive to antigen upon re-exposure, indicating that they retained functionality. MNV-specific memory CD8+ T cells mediated initial protection from challenge with a persisting MNV strain but in most cases this protection was short-lived. Analysis of early events following challenge of immunized mice revealed a marked deficiency in the ability of MNV-specific CD8+ T cells to respond to the chronic strain of MNV. Rather, during chronic infection, MNV-specific CD8+ T cells were largely ignorant of ongoing viral replication in vivo. Moreover, MNV-specific CD8+ T cells also failed to respond in vitro when co-cultured with intestinal cells from chronically infected mice unless the intestinal cells were first lysed to release antigen. Collectively our findings show that MNV persistence was associated with a unique differentiation state of virus-specific CD8+ T cells. While such cells could, in some settings, confer protection against MNV, T cell ignorance emerged early during chronic infection, likely due to the establishment of an immunoprivileged enteric niche that supported long-term viral replication. These findings further provide an explanation for the emergence of chronic NV infections and might help explain heterogeneous responses in humans.


▼Single Amino Acid Determines the Magnitude and Function of MNV-Specific CD8+ T Cells

We previously mapped a conserved immunodominant epitope (P1519) that accounts for ∼80% of the total CD8+ T cell response against MNV (Figure S1A; Tomov et al., 2013). However, P1519 differs at position 7 between strains CW3 (Tyr) and CR6 (Phe), preventing direct comparison of epitope-specific CD8+ T cell responses. To address this issue, we changed position 7 in P1519 from Tyr to Phe (Y→F) or Phe to Tyr (F→Y) in MNV-CW3 and MNV-CR6, respectively, generating recombinant strains CR6F→Y and CW3Y→F (Figure 1A). These reverse engineered viruses grew with normal kinetics in the mouse macrophage-like RAW-264.7 cell line indicating that the changes in P1519 did not affect viral fitness in vitro (Figure 1B).

At day 8 p.i., mice infected with MNV-CW3Y→F had significantly fewer P1519 Tetramer-positive (Tet+) CD8+ T cells in the intestine compared to MNV-CW3WT-infected mice (Figure 1C). Conversely, infection with MNV-CR6F→Y resulted in a substantial boost in virus-specific CD8+ T cell responses with numbers of Tet+CD8+ T cells in the intestine approaching those from MNV-CW3WT infection (Figures 1C and 1D). Similar changes were observed in the spleen where P1519-specific CD8+ T cells accounted for ∼50%–75% of the MNV-induced CD44hiCD62Llo CD8+ T cells for MNV-CW3WT and MNV-CR6F→Y, compared to 20%–30% for MNV-CR6WT and MNV-CW3Y→F (Figure 1E). Moreover, the difference in the Tet+CD8+ T cell response between MNV-CR6WT and MNV-CR6F→Y persisted long-term (Figure 1F). Tet+CD8+ T cells responding to MNV-CR6F→Y produced more cytokine and were more polyfunctional (i.e., co-produced cytokines more efficiently) compared to Tet+CD8+ T cells from MNV-CR6WT-infected mice (Figures 1G and 1H; Figure S1B). Conversely, MNV-CW3Y→F induced Tet+CD8+ T cells with less polyfunctionality compared to MNV-CW3WT (Figure S1B). Moreover, the changes in P1519 did not inadvertently create a CD4+ T cell epitope (Figure S1C) or affect the overall antibody response (Figure S1D). Thus, position 7 of P1519 determines both the magnitude and quality of the CD8+ T cell response without affecting other aspects of adaptive immunity against MNV.

▼MNV Persistence Versus Clearance Is Independent of the CD8+ T Cell Response

To determine the effect of P1519-specific CD8+ T cells on MNV persistence, we measured viral RNA in the stool and intestine following infection with the parental and epitope-swap strains. MNV-CW3WT and MNV-CW3Y→F were efficiently cleared by ∼day 10 p.i. (Figures 1I and 1J). By contrast, MNV-CR6F→Y established chronic infection indistinguishably from the parental CR6WT strain (Figures 1I and 1J). Moreover, Tet+CD8+ T cells responding to persisting MNV-CR6F→Y maintained high expression of granzyme B and retained their cytotoxic potential at day 30 p.i. (Figures 1K and 1L), suggesting that exhaustion did not contribute to MNV persistence. Changes in P1519 did not impact CD4+ or regulatory T cell (Treg cell) responses (Figures S2A–S2C). Thus, MNV-CW3WT and CR6F→Y elicit similar adaptive immune responses and can be used to directly compare CD8+ T cells against the same epitope in acute versus chronic infection. Moreover, epitope reversion did not explain MNV-CR6F→Y persistence as the F→Y mutation was maintained even after 6 months of chronic infection (Figure 1M). Together these observations indicated that an augmented primary CD8+ T cell response did not exert immune pressure on MNV-CR6F→Y and was not sufficient to prevent persistence.

▼Distinct Transcriptional Profiles of Tet+CD8+ T Cells in Acute Versus Chronic MNV Infection

We compared the transcriptional profiles of intestinal Tet+CD8+ T cells from mice infected with MNV-CW3WT or MNV-CR6F→Y where the same epitope was being targeted. At day 30 p.i., 423 genes were differentially expressed by at least 2-fold (ANOVA p value < 0.05). Among genes upregulated during MNV-CR6F→Y infection were some with known roles in CD8+ T cell differentiation such as Itga2, Il12rb2, Cxcl10, and Fas, and others encoding the inhibitory molecules 2B4, Tim-3, and KLRG1 (Figure 2A). Genes that were downregulated during MNV-CR6F→Y infection included Il2, Sipr1, Tcf7, Lef1, Btla, Cxcr4, Slamf6, Ly6c2, Cd40lg, Klrk1, and Pdcd1 (Figure 2A).

Gene set enrichment analysis (GSEA) using gene ontology (GO) terms indicated that Tet+CD8+ T cells from chronically-infected mice were mitotically and metabolically active and upregulated genes involved in cell division, aerobic respiration, and lipid metabolism (Figure 2B). On the other hand, Tet+ memory CD8+ cells from resolved MNV-CW3WT infection upregulated genes associated with adaptive immune responses, T cell receptor (TCR) signaling, and cytokine production (Figure 2C). Thus, despite responding to the same epitope, CD8+ T cells induced during MNV-CW3WT versus MNV-CR6F→Y infection developed distinct metabolic and proliferative transcriptional programs.

Consistent with previous analyses (Tomov et al., 2013), expression of Pdcd1 encoding the inhibitory receptor PD-1, was lower in Tet+CD8+ T cells from chronic compared to resolved MNV infection. Another gene associated with T cell exhaustion, Btla, showed a similar pattern of relative expression (Figure 2A). On the other hand, genes for the inhibitory receptors 2B4 (Cd244) and Tim3 (Havcr2) were upregulated in Tet+CD8+ T cells from MNV-CR6F→Y infection, as was KLRG-1 (Figure 2A). Moreover, protein expression correlated with transcriptional differences for a number of differentially transcribed genes (Figure 2D). In sum, these data indicate an unusual differentiation state of MNV-specific CD8+ T cells compared to typical exhausted CD8+ T cells defined in studies of chronic LCMV and HIV infection (Wherry and Kurachi, 2015).

▼Distinct Tissue Resident CD8+ T Cell Subsets Exist in Acute Versus Chronic MNV Infection

We have previously noted differences in the expression of several integrins on Tet+CD8+ T cells from acute versus chronic MNV infection (Tomov et al., 2013). Gene-expression analysis confirmed 1.97-fold higher expression of Itgae (encoding CD103) in Tet+CD8+ T cells from MNV-CR6F→Y infection (ANOVA p value < 0.05). As CD103 is a marker of tissue residence (Carbone et al., 2013), we conducted additional analyses for expression of tissue residence memory (Trm) cell genes and proteins. At day 30 post MNV-CW3WT infection, a significant fraction of Tet+CD8+ T cells from the lamina propria (LP) and intraepithelial (IEL) compartments remained CD103lo (Figure 3A). By contrast, most Tet+CD8+ T cells from the small intestine and colon of mice infected with MNV-CR6F→Y were CD103hi (Figures 3A and 3B) consistent with previous observations for MNV-CR6WT (Tomov et al., 2013). The ratio of CD103hi to CD103lo Tet+ CD8+ T cells remained constant during MNV-CW3WT infection, but shifted to a predominantly CD103hi phenotype with MNV-CR6F→Y (Figure 3C). In contrast to CD103, CD11a, and CD49d were more highly expressed on Tet+CD8+ T cells from MNV-CW3WT compared to MNV-CR6F→Y infection (Figure 3D), consistent with differences between the WT strains (Tomov et al., 2013). Therefore, the distinct pattern of α-integrin expression by Tet+CD8+ T cells responding to MNV-CR6F→Y is attributable to chronic infection rather than the F→Y change in P1519.

Upregulation of CD103 is not an absolute requirement for tissue residence and CD103lo Trm cells are critical for control of some enteric pathogens (Bergsbaken and Bevan, 2015). Alternatively, the CD103lo Tet+CD8+ T cells from MNV-CW3WT-infected mice might be recirculating T cells rather than true Trm cells (Carbone et al., 2013). To distinguish between these possibilities, we used intravascular staining for CD8α (Beura et al., 2015), which distinguished between anatomic compartments with different access to the circulation (Figure 3E, CD8α+ versus CD8α− cells). In the intestine, 99% of CD8+ LPL and IEL were inaccessible to the i.v. administered antibody indicating that they were not from blood contamination (Figure 3E, green gates). Furthermore, these gut-associated lymphocytes contained both CD103lo and CD103hi Tet+ cells, suggesting that both CD103-defined subsets were Trm cells (Figure 3F).

To further investigate MNV-specific CD8+ T cell tissue residence, we applied GSEA using a Trm cell signature (Mackay et al., 2013). As expected, Trm cell genes were enriched in intestinal relative to splenic Tet+CD8+ T cells at all time points for both viruses (Figure S3). Comparison of enteric Tet+CD8+ T cells from MNV-CW3WT versus MNV-CR6F→Y showed a more prominent Trm cell program induced early during acutely-cleared infection (Figure 3G). By day 15, this enrichment bias was no longer evident (Figure 3H), while by day 30 p.i., the Trm cell signature was more highly enriched in Tet+CD8+ T cells from MNV-CR6F→Y infection (Figure 3I). Although at day 8 p.i. Trm cell signature enrichment was driven by 12 core genes, only 5 genes contributed at day 30 p.i. (Figure 3J). Therefore Tet+CD8+ T cells from either infection upregulated Trm cell-associated genes but in the case of MNV-CR6F→Y induction of the Trm cell program appeared to be delayed.

Next, we used a MNV-CW3WT and MNV-CR6F→Y coinfection model. We reasoned that if the timing of Trm cell differentiation was important for the outcome of infection, either MNV-CW3WT would induce early Trm cells leading to control of the co-infecting MNV-CR6F→Y or co-infection with MNV-CR6F→Y would suppress Trm cell differentiation. Coinfected mice developed chronic infection with indistinguishable kinetics from mice infected with MNV-CR6F→Y alone indicating that MNV-CW3WT failed to induce a protective CD8+ T cell response (Figure 3K and 3L). Moreover, at day 30 both persistently infected mice from the MNV-CR6F→Y alone and the co-infected group had higher expression of CD103 on Tet+CD8+ cells compared to resolved MNV-CW3WT infection (Figure 3M). Thus, although CD8+ T cells are necessary for clearance of MNV-CW3WT (Chachu et al., 2008), these cells failed to prevent viral persistence in the setting of coinfection. Moreover, the persisting MNV strain determined the long-term phenotype of Tet+CD8+ T cells. Thus, a distinct viral mechanism of immune evasion, rather than intrinsic CD8+ T cell dysfunction enables MNV-CR6F→Y persistence.

▼Persistent MNV Infection Induces a Unique Transcriptional Program in Tet+CD8+ T Cells

Next, we defined transcriptional signatures of CD8+ T cells responding to persisting or resolved MNV infection using differentially expressed genes at day 30 p.i. (Figure 2A). We tested for enrichment of these MNV signatures in the transcriptional profiles of CD8+ T cells from acute versus chronic lymphocytic choriomeningitis virus (LCMV) infection (Doering et al., 2012). Indeed, GSEA revealed enrichment for the MNV-CR6F→Y signature in CD8+ T cells from chronic LCMV infection (Figure S4A). Conversely, the MNV-CW3WT signature was enriched in functional memory CD8+ T cells from resolved LCMV infection (Figure S4B). The transcriptional similarity between CD8+ T cells responding to chronic MNV and LCMV infections was driven, in part, by genes encoding exhaustion markers Tim3 and 2B4, integrin CD49b, and cytokine CXCL10 (Figure S4A). However, genes for other key markers of exhaustion, such as Pdcd1, were not part of the MNV-CR6F→Y CD8+ T cell signature and did not contribute to core enrichment in Figure S4A. To investigate this question further, we used published immunologic signatures to perform comprehensive GSEA of Tet+CD8+ T from day 30 MNV-CR6F→Y infection. There was significant enrichment of several exhaustion signatures derived from the LCMV model (Figure 4A) consistent with Figure S4A (Doering et al., 2012, West et al., 2011). However, the transcriptional profile of Tet+CD8+ T was also enriched for effector signatures (Figure 4B), defined as genes upregulated on day 8 versus 30 of LCMV or vaccinia virus infection. As effector and exhausted CD8+ T cells share features of activation (Doering et al., 2012, Singer et al., 2016), we used leading edge analysis to determine whether the same core genes contributed to enrichment for both types of signatures during MNV-CR6F→Y infection. Of the 489 genes driving enrichment for the signatures in Figures 4A and 4B, only 38 contributed to enrichment of both an exhaustion and an effector signature (Figure 4C). Moreover, only Litaf, encoding lipopolysaccharide (LPS)-induced tumor necrosis factor (TNF), contributed to more than one of the exhaustion and effector signatures used in the comparison. Thus, Tet+CD8+ T cells from chronic MNV-CR6F→Y infection had features of both exhausted and effector CD8+ T cells and these characteristics were largely driven by non-overlapping sets of genes, possibly reflecting recent activation and/or heterogeneity in the populations analyzed. By contrast, the transcriptome of Tet+CD8+ T cells from resolved MNV-CW3WT infection enriched for signatures of memory and/or quiescence, consistent with viral clearance (Figure S4C).

Given the preserved effector features of Tet+CD8+ T cells from chronic MNV infection, we considered the murine cytomegalovirus (MCMV) model of viral persistence that is characterized by latency and reactivation. These viral dynamics give rise to inflationary virus-specific CD8+ T cells that maintain an effector phenotype and do not become exhausted despite repeated antigenic stimulation (Snyder, 2011). GSEA showed enrichment of an inflationary effector CD8+ T cell signature (Quinn et al., 2015) in the transcriptional profile of CD8+ T cells responding to persistent MNV-CR6F→Y infection (Figures 4D and S4D), suggesting long-term maintenance of a functional effector CD8+ T cell (Teff cell) response. Indeed, the normalized enrichment scores (NES) for effector signatures in Figures 4B and 4D were higher than the NES for exhaustion signatures in Figure 4A (Figure 4E). Moreover, of the 45 genes that contributed to core enrichment of the MCMV Teff cell signature (Figure S4D), 18 (40%) also contributed to enrichment of other effector signatures (Figure 4F) while none contributed to enrichment of exhaustion signatures. Therefore, the transcriptional similarities between CD8+ T cells responding to chronic MNV and latent MCMV infection can be largely attributed to effector genes.

Phenotypic features of inflationary CD8+ T cells include upregulation of granzyme and CX3CR1 and downregulation of the co-stimulatory molecules CD27, CD28, and GITR (Snyder, 2011, van de Berg et al., 2012, Welten et al., 2013). Indeed, at day 30 of MNV-CR6F→Y compared to MNV-CW3WT infection Tet+CD8+ T cells had higher expression of granzyme B and lower expression of CD28 and GITR proteins (Figure S4E). Consistent with their inflationary differentiation pattern, Tet+CD8+ T cells during chronic MNV-CR6F→Y infection persisted long-term and comprised a significant fraction of intestinal CD8+ T cells at day 60 p.i. (Figure 4G). Thus, Tet+CD8+ T cells from chronic MNV-CR6F→Y infection share transcriptional and phenotypic features with inflationary Teff CD8+ T cells from latent MCMV infection.

▼Tet+CD8+ T Cells from Chronic MNV Infection Respond to Antigen In Vivo

To define in vivo functionality, we isolated Tet+CD8+ T cells from chronic MNV-CR6F→Y infection, labeled them with cell trace violet (CTV), and adoptively transferred them into congenic recipients. These recipients were either naive, acutely infected with MNV-CW3WT (day 4 p.i.) or chronically infected with MNV-CR6F→Y (day 60 p.i.) (Figures 5A–5C). As expected, 5 days post transfer, donor Tet+CD8+ T cells had not divided in naive hosts (Figure 5A). However, in mice acutely infected with MNV-CW3WT, donor Tet+CD8+ T cells divided extensively, expanded in number, and homed to the spleen, mesenteric lymph nodes (MLNs), and intestine (Figure 5B). Donor Tet+CD8+ T cells were particularly abundant in the intestine where they accounted for up to 80% of all donor-derived CD8+ T cells (Figure 5D). Thus, MNV-specific CD8+ T cells isolated even after 60 days of chronic infection were still capable of mounting robust responses to viral antigen in vivo during acute infection.

However, when the same Tet+CD8+ T cells were adoptively transferred to recipients that were chronically infected with MNV-CR6F→Y, these cells did not expand in number and were difficult to detect 5 days post transfer (Figure 5C). In a few chronically infected recipients, donor Tet+CD8+ T cells could be found in the intestine but these cells had not divided and accounted for less than 10% of all donor CD8+ T cells (Figure 5D), suggesting they had not encountered antigen. Consistent with this conclusion, viral loads in these mice remained unchanged after the adoptive transfer (data not shown).

The failure of adoptively transferred Tet+CD8+ T cells to respond to MNV-CR6F→Y suggested that these cells were ignorant of persisting antigen. Alternatively, poor donor responses might be due to competition from pre-existing host CD8+ T cells (Quinn et al., 2015). Therefore, we repeated the adoptive transfer experiment in Figures 5A–5C using Thy1.1 (recipients) and Thy1.2 (donors) as congenic markers, and used an anti-Thy1.1 antibody to selectively deplete host T cells from recipients prior to the adoptive transfer (Figures S5A and S5B). Five days post transfer, donor Thy1.2+CD8+ T cells were detected in all tissues examined but had not proliferated significantly in most mice (Figure S5C). Depletion of host T cells did enhance overall donor cell survival, especially in MLN, however this effect was modest and observed in both naive and chronically-infected mice, suggesting homeostatic proliferation rather than an antigen-specific response (Figure S5C). Indeed, most CTV-low donor CD8+ T cells were not Tet+ in either naive or CR6F→Y-infected mice (Figure 5E–5F). Adoptively-transferred CD8+ T cells were most abundant in the MLN of 2 out of 5 persistently infected Thy1.1-depleted recipients (Figure S5C, blue circles in MLN plot), however even in these mice Tet+ cells accounted for less than 0.5% of total donor cells (Figure 5H). Similarly, Tet+ cells accounted for at most 11% of donor CD8 T cells in the IEL and this was seen in just 1 out of 5 recipients (Figure 5H). Thus, even when depletion of host T cells gave an overall survival advantage, Tet+ cells failed to respond to chronic MNV-CR6F→Y infection.

Consistent with Figure 5B, when Thy1.2+CD8+ T cells were adoptively transferred into mice acutely infected with MNV-CW3WT, Tet+ donor cells proliferated robustly even without host T cell depletion (Figure 5G). By day 5 post transfer, virtually all Tet+ donor cells had diluted CTV and accounted for nearly half of donor cells in the gut (Figure 5H). Compared to Figure 5B, donor Tet+ cell expansion was less robust (red bars in Figure 5H versus 5D), consistent with the different timing of these adoptive transfers (recipient mice day 4 p.i. in Figure 5B versus day 1 p.i. in Figure 5H) and the fact that MNV-CW3WT viral titers peak at ∼day 4 (data not shown). Collectively, these data demonstrate that despite having the capacity to respond to antigen in vivo, virus-specific CD8+ T cells fail to do so during chronic MNV-CR6F→Y infection and this defect is not due to T cell competition. Therefore, Tet+CD8+ T cells in persistently infected mice might be limited by an inability to efficiently detect viral antigen.

▼Preexisting CD8+ T Cell Immunity Confers Partial Protection against Chronic MNV

Given the response of Tet+CD8+ T cells to MNV-CW3WT (Figure 5B), we tested the effectiveness of pre-existing immunity against MNV-CR6F→Y. Mice that had previously cleared MNV-CW3WT or MNV-CW3Y→F were challenged with MNV-CR6F→Y or MNV-CR6WT, and protection was measured by viral shedding and tissue titers. We carried out both “matched” and “mismatched” experiments in which the prime and challenge strains had the same or different P1519 epitopes. When taken together, there was no consistent pattern for priming strain related to control upon challenge (data not shown). Therefore, we combined data for MNV-CW3WT and MNV-CW3Y→F-immune mice and report all outcomes based on the chronic strain used in the challenge.

As expected, all non-immunized mice became chronically infected when challenged with MNV-CR6WT or MNV-CR6F→Y (Figures 6A and 6B , circles). Mice immunized with MNV-CW3WT or MNV-CW3Y→F and then challenged with MNV-CR6WT also developed chronic infection (Figures 6A and 6B, solid squares). By contrast, all immune mice challenged with MNV-CR6F→Y had decreased viral shedding for the first 48–96 hours p.i. (Figure 6A, open squares; time points in blue box), although in most cases replication of the challenge MNV-CR6F→Y virus eventually rebounded. However, ∼1/3 of animals achieved long-term viral control with undetectable titers in the stool for at least ∼30 days despite being co-housed with mice who remained infected and shedding virus (Figure 6A). At day 30 post challenge, these protected mice also had undetectable viral titers in the proximal colon (Figure 6B).

These data showed that early viral control following MNV-CR6F→Y challenge was universal, whereas long-term protection was sporadic (Figure 6C). Long-term viral control did not correlate with the number or functionality of Tet+CD8+ T cells (data not shown), and boosting MNV-CW3WT-immune mice with a recombinant vaccinia virus expressing the P1519Y epitope did not improve outcomes to MNV-CR6F→Y challenge (data not shown) despite significantly augmenting the P1519-specific CD8+ T cell response (see Figures S6A and S6B). Taken together, these observations suggested that long-term protection was likely due to events that occurred early after viral challenge and prevented establishment of a niche for MNV-CR6F→Y persistence. Consistent with this interpretation, in 4 out of 6 primed mice that had developed chronic MNV-CR6F→Y infection (see box in Figure 6B) the F→Y mutation was preserved at day 30 post challenge (Figure 6D). Similarly, MNV-CR6F→Y sequences from the fecal pellets of 4 other primed and chronically-infected mice also contained the F→Y mutation (data not shown). These findings further suggested that CD8+ T cells failed to exert immune pressure on MNV-CR6F→Y in the chronic phase of infection. In 2 of the primed and chronically infected mice, the P1519Y epitope had, in fact, mutated back to P1519F (Figure 6D). Moreover, this was a true reversion since the Phe codon in these mutant sequences was TTC, whereas it is TTT in the wild-type CR6 strain (compare to Figure 1M). We hypothesize that this sequence change represents an alternative pathway of viral escape early after challenge when CD8+ T cells exerted some degree of control over viral replication.

As partial protection against chronic MNV was observed for MNV-CR6F→Y but never for MNV-CR6WT or in unimmunized mice, we surmised that such protection was mediated by CD8+ T cells. To directly test this conclusion, we depleted a subset of immunized mice of CD8+ T cells (Figure S6C) and measured viral titers at early time points post MNV-CR6F→Y challenge. Consistent with our previous observations, immunized mice with intact CD8+ T cells controlled viral replication at early time points (Figure 6E, squares). By contrast, most mice treated with anti-CD8β had high viral titers in the proximal colon at days 4 and 6 post challenge (Figure 6G, triangles). Of note, we did not detect MNV in the liver or spleen of these CD8-depleted mice, consistent with the gut-specific tropism of MNV-CR6 (Nice et al., 2013). Thus, CD8+ T cells were necessary for early enteric control of MNV-CR6F→Y in immunized mice.

▼Inconsistent Detection of Antigen by Tet+CD8+ T Cells Early after MNV-CR6F→Y Challenge

Prime-challenge data indicated a narrow window for Tet+CD8+ T cells to clear MNV-CR6F→Y (Figure 6A), while the adoptive transfers suggested that these cells were ignorant of viral replication during chronic infection (Figure 5C). To investigate these dynamics further, we generated a recombinant vaccinia virus expressing the P1519Y epitope (rVV519Y) and used it to boost the P1519-specific CD8+ T cell response in MNV-CW3WT-immune mice (Figures S6A–S6B). We used this prime-boost strategy to generate large numbers of Tet+CD8+ T cells, labeled these cells with CTV, and adoptively transferred them to congenic recipients. Recipients mice were either naive or at different time points of MNV-CW3WT (day 1 or 7) or MNV-CR6F→Y (day 1, 7, or 30) infection.

As expected, no proliferation of donor Tet+CD8+ T cells was observed in naive recipients 5 days after adoptive transfer (Figure 7A). By contrast, donor Tet+CD8+ T cells divided extensively in all recipients at day 1 p.i. with MNV-CW3WT (Figure 7B). These, donor Tet+CD8+ T cells homed to the spleen and MLN, as well as to the intestinal mucosa and LP where they accumulated to the highest frequency (Figure 7G). Consistent with the viral clearance kinetics for MNV-CW3WT (see Figures 1I and 1J) there was little division or accumulation of donor Tet+CD8+ T cells in MNV-CW3WT-infected recipients at day 7 p.i., though occasional CTV dilution likely reflected presence of residual antigen (Figure 7C).

When Tet+CD8+ T cells were adoptively transferred to MNV-CR6F→Y-infected recipients at days 1 or 7 p.i., we observed heterogeneous responses (representative plots in Figures 7D and 7E). In some recipients, essentially no division of donor Tet+CD8+ T cells took place, while in others Tet+ donor cells divided extensively, although not to the same extent as with MNV-CW3WT (Figure 7B versus 7D and 7E). When donor Tet+ cells responded to MNV-CR6F→Y, they accumulated primarily in the intestine where P1519-specific cells accounted for at most 30% of total donor CD8+ T cells (compared to 80% for MNV-CW3WT) (Figure 7G). At day 5 post transfer, all MNV-CR6F→Y infected mice had detectable virus in the colon and there was no correlation between the extent of donor Tet+CD8+ T cell proliferation and intestinal viral titers in individual mice (data not shown). This was expected because the number of MNV-specific CD8+ T cells following adoptive transfer was low compared to intact MNV immune mice and these mice lacked preexisting CD4+ T cell and B cell immunity.

Despite ongoing viral replication (and consistent with Figure 5C), no significant division of donor CD8+Tet+ cells took place in recipients at day 30 post MNV-CR6F→Y infection and few donor cells were detected at this late time point (Figures 7F and 7G). Collectively, these data suggested that although MNV-specific CD8+ T cells could sense and respond to MNV-CR6F→Y early after infection, these responses were sporadic and less robust than for MNV-CW3WT. Moreover, the ability of Tet+CD8+ T cells to detect ongoing MNV-CR6F→Y replication waned over time and was nearly absent in established chronic infection.

▼Failure of Antigen Presentation by Chronically Infected Intestinal Cells

The preserved functionality and progressive inability of Tet+CD8+ T cells to detect ongoing viral replication suggested that MNV-CR6F→Y was inaccessible to CD8+ T cells during established chronic infection. Alternatively, MNV-CR6F→Y might directly antagonize immune cell function via virally-encoded proteins (Zhu et al., 2016, Zhu et al., 2013). To distinguish between these possibilities, we designed an in vitro experiment to “force” interactions between Tet+CD8+ T cells and cells infected with MNV-CR6F→Y. CD8+ T cells from the spleens of day 30 MNV-CR6F→Y-infected mice were labeled with CTV and incubated with single-cell intestinal suspensions from congenic mice that were either naive or also persistently infected with MNV-CR6F→Y. These “intestinal preps” contained multiple cell types including epithelial cells, fibroblasts, and immune subsets from the intraepithelial and LP compartments (data not shown). Although the cellular reservoir of persisting MNV-CR6F→Y is currently unknown, we reasoned that cells harboring virus were likely to be present in intestinal preps.

Intestinal preps from naive mice did not stimulate Tet+CD8+ T cell division (Figure 7H). Preps from persistently-infected mice also failed to induce an MNV-specific CD8+ T cell response (Figure 7I), consistent with in vivo observations (Figure 5C). To determine whether host or viral factors prevented Tet+CD8+ T cell responses, we lysed intestinal preps from chronically infected mice via freeze-thaw to release potentially inaccessible intracellular viral antigens and added splenocytes from naive mice to act as antigen-presenting cells (APC). In this case, Tet+CD8+ T cells responded by dilution of CTV consistent with antigen release from a cellular viral reservoir and presentation by splenic APC (Figure 7J). Similar results were obtained when bone marrow-derived DCs were used as APC (Figure S7). Moreover, proliferation of Tet+CD8+ T cells in these in vitro assays was antigen-specific, as it was not observed with lysed intestinal preps from naive mice (Figure S7A). These data suggested restricted availability and/or release of viral antigen from the intestinal cellular reservoir despite continued replication of MNV-CR6F→Y in vivo. Taken together, our in vivo and in vitro experiments support a mechanism of CD8+ T cell evasion by MNV based on persistence in an immune-privileged enteric niche.


Two challenging aspects of NV biology have recently been appreciated. First, NVs cause chronic infections in some immunocompetent individuals thereby creating a potential reservoir for viral evolution and spread (Karst et al., 2014). Second, NVs evade adaptive immune responses even in previously vaccinated individuals who have robust NV-specific antibody titers (Atmar et al., 2011). Understanding why adaptive immunity fails to prevent or clear NV infections can facilitate future vaccine and public health efforts.

CD8+ T cells are critical for control of many mucosal pathogens including NVs (Chachu et al., 2008). However, the impact of chronic enteric infection on Trm cell responses and differentiation remains poorly understood. As chronic viral infections are often facilitated by T cell dysfunction, defining the features of T cell failure could have implications for therapeutic interventions and preventative vaccines. Using the MNV model we discovered two key features of the CD8+ T cell response and protective immunity during chronic infection. First, MNV-specific CD8+ T cells retained functional responsiveness and did not display typical features of exhaustion. Indeed, MNV-specific CD8+ T cells maintained transcriptional features of activation resembling inflationary effector CD8+ T cells from herpesvirus infections. Second, there was a narrow window of opportunity for CD8+ T cells to exert protection against chronic MNV infection, leading to an apparent paradox of continued robust viral replication despite a largely functional antiviral CD8+ T cell response. Our findings point to a mechanism of largely immunologically undetectable persistence of an RNA virus in the intestinal tract.

Although pathogen persistence is not unique to MNV, our data suggest an unusual pattern of immune evasion. First, the immunodominant P1519Y epitope did not revert to the less immunogenic 519F variant following primary and most secondary infections, suggesting that antigenic escape is not a major pathway facilitating MNV persistence. Second, CD8+ T cell exhaustion or senescence did not develop in P1519Y-specific CD8+ T cells, as these cells remained functional even after 30 days of chronic infection. Third, overt tolerance to MNV-CR6F→Y also appears unlikely given the absence of Treg cell expansion and the robust response of P1519Y-specific CD8+ T cells when adoptively transferred to MNV-CW3WT-infected mice.

A fourth possibility is that Tet+CD8+ T cells fail to clear MNV-CR6F→Y due to limited access to viral antigen during chronic infection. True CD8+ T cell ignorance is thought to arise from insufficient anatomical access to antigen and/or lack of costimulatory signals (Woodham et al., 2016). However, in the case of established chronic MNV-CR6F→Y infection, the effector properties of P1519-specific CD8+ T cells and their transcriptional similarity to inflationary MCMV-specific responses suggest that these cells do, at least occasionally, encounter and respond to antigen in vivo, but such interactions might be transient and/or inefficient.

How does a virus persistently replicating at high levels induce a CD8+ T cell transcriptional program normally reserved for latent pathogens? MNV cellular tropism might be at the core of this question and while the identity of the in vivo cellular reservoir remains poorly understood, our data might be relevant for elucidating the pathogenesis of this chronic enteric infection. Although we demonstrate a role of CD8+ T cells in preventing chronic MNV infection, at least in some mice, our data also show that these cells have a narrow window of opportunity to prevent persistence. This observation suggests a possible “switch” in tropism associated with the transition to viral persistence. CD300lf+ DCs can support MNV-CR6 infection in vitro and might be necessary for at least the initial establishment of infection in vivo (Orchard et al., 2016). Thus, one possibility is that initial DC infection allows priming and early control of MNV-CR6F→Y infection by CD8+ T cells, but subsequently viral persistence is established in a cell type that is relatively inaccessible or unrecognizable by Tet+CD8+ T cells. The transcriptional program of MNV-specific CD8+ T cells from chronic infection does, however, implicate occasional re-stimulation with antigen in vivo perhaps due to low-level release from the viral reservoir. Moreover, the CD103 status of MNV-specific CD8+ T cells in acutely resolved versus chronic infection might be relevant. For example, data from the Yersinia model show that the physical proximity of CD103lo Trm cells to antigen-presenting cells is critical for successful pathogen clearance (Bergsbaken and Bevan, 2015). Whether the distinct patterns of CD103 expression on Tet+CD8+ T cells from resolved versus chronic MNV infection reflect differences in viral tropism and local immune environment, and whether modulating this axis could have antiviral effects will be interesting to interrogate in the future.

We have shown that a robust and functional CD8+ T cell response fails to prevent MNV persistence. The behavior of MNV-specific CD8+ T cells during chronic infection suggests a “blind spot” in their ability to perceive antigen. Whether such ignorance is the result of a maladaptive CD8+ T cell differentiation pattern or viral replication inside an immunoprivileged enteric niche remains to be determined and will have important implications for the design of future NV vaccines.

▼Author Contributions

V.T.T. and E.J.W. designed the study with input from H.W.V. and T.J.N. V.T.T., O.P., C.W.L., A.P., Y.S., and R.T. performed experiments. V.T.T., B.B., and S.M. analyzed the data. G.L.C., L.E., T.J.N., and H.W.V. contributed reagents. V.T.T. and E.J.W. wrote the paper. All authors edited the paper.