Environmental stress drives clearance of a persistent enteric virus in mice

Abstract

Persistent viral infections are associated with long-term health issues and prolonged transmission. How external perturbations after initial exposure affect the duration of infection is unclear. We discovered that murine astrovirus, an enteric RNA virus, persists indefinitely when mice remain unperturbed but is cleared rapidly after cage change. Besides eliminating the external viral reservoir, cage change also induced interferon-stimulated genes in the intestinal epithelium necessary for viral clearance. We further identified that displacing infected animals initially caused a temporary period of immune suppression through the stress hormone corticosterone, which was followed by an immune rebound characterized by activation of CD8 T cells responsible for downstream epithelial antiviral responses. Our findings show how viral persistence can be disrupted by preventing re-exposure and activating immunity upon stress recovery, indicating that external factors can be manipulated to shorten the duration of a viral infection.

Introduction

The time course of infection for many viruses is characterized by a brief period of peak viral burden followed by resolution within days. However, development of more sensitive techniques has enabled detection of viral products weeks and even months after the infection for many viruses previously considered non-persistent^1–13. Prolonged infection may contribute to long-term complications, such as post-acute sequalae of SARS-CoV-2 (long COVID) or increased opportunities for viral mutation and transmission^13–18.

Evasion of CD8 T cells is a hallmark of chronic infections¹⁹. While retroviruses and certain DNA viruses display periods of latency, non-integrating RNA viruses typically exhibit continuous replication and production of antigens. During persistent lymphocytic choriomeningitis virus (LCMV) infection of mice, unremitting stimulation of CD8 T cells leads to exhaustion and reduced activity^20,21. T cells specific for hepatitis B virus (HBV) and HCV in humans display exhaustion^19,22,23. CD8 T cells remain functional in other instances. In the gut for example, CD8 T cells fail to recognize immune-privileged tuft cells infected by persistent strains of murine norovirus (MNV)^24–26. CD4 T cell help partially prevents exhaustion during herpes simplex virus infection (HSV) and cytomegalovirus (CMV) causes a phenomenon called memory T-cell inflation distinct from exhaustion^27,28. Therapeutic blockade of PD-1 and innate immune activation through IFNλ administration overcome these failures to clear LCMV and MNV, respectively^29,30. Thus, identifying determinants of persistence upstream of CD8 T cells may reveal opportunities for non-pharmaceutical interventions.

Astroviruses are non-enveloped positive strand RNA viruses that infect a wide range of animal species³¹. In humans, astroviruses are a common cause of an acute, self-limiting gastroenteritis, and more than 91% of children have experienced at least one infection by age five³². Murine astroviruses were shown to also spread fecal-orally and infect goblet cells in the intestine similar to human astroviruses^35,36. Murine astrovirus-1 (MuAstV) is endemic in many mouse facilities where it persists asymptomatically in immunocompromised mice^34,37,38. In contrast, it has been reported that the virus is cleared within a couple weeks in immunocompetent wild-type (WT) mice^37,38. In this study, we found that lab MuAstV is only cleared because of cage change and can persistently infect WT mice if they remain in the same cage throughout the experiment. We show that cage change mediates MuAstV clearance by preventing reinfection while simultaneously inducing a short inflammatory burst through a stress-induced antiviral response by CD8 T cells and the intestinal epithelium. These findings demonstrate that environmental conditions determine the duration of a viral infection.

Results

Persistent MuAstV infection in the absence of cage change

Germ-free (GF) mice require less frequent exchange of cages to remove dirty bedding. While investigating the role of microbiota during MuAstV infection in WT mice, we observed that timing of virus decline was variable between experiments and coincided with cage change (CC). To test whether CC was responsible for MuAstV clearance, we compared viral shedding in the stool in WT mice ±CC at 14 days post inoculation (dpi). Virus levels peaked at 7dpi reaching ~10¹⁰ genome copies followed by an initial decline to ~10⁶ genome copies. Viral RNA (vRNA) dropped below the limit of detection by 21dpi in mice +CC, while mice –CC continued to shed virus at ~10⁶ genomes (nearly 2 months) (Fig. 1a). Similar to GF conditions, specific pathogen free (SPF) mice also cleared MuAstV by 28dpi following CC at 14dpi and remained infected –CC (Fig. 1b; Extended Data Fig. 1a). Virus levels declined more rapidly in SPF mice than GF mice (Fig. 1c-d). Tonic interferon signals, which are absent in GF mice, may contribute to this difference^39–41. Thus, +CC mediates viral clearance independently of the microbiome, and we used both GF and SPF mice to investigate the mechanism (as indicated in figure legends).

Fig. 1: MuAstV establishes a persistent infection in the absence of cage change.

a, MuAstV genome levels in stool of germ-free (GF) B6 mice over time +/− cage change (CC) at 14 days post inoculation (dpi) detected by RT-qPCR. Mice were inoculated with 1E10 viral genomes via oral gavage. n= 3 mice. b, MuAstV genome levels in GF and specific-pathogen-free (SPF) mice +/−CC. n = 5–6 mice. c-d, MuAstV genome levels in stool from GF (c) and SPF (d) mice measured daily after CC. n = 4–5 mice. e, Ct values for MuAstV negative strand (-vRNA) RT-qPCR and matching DNA gel of negative strand RT-PCR in different gut sections of mock controls and MuAstV infected mice +/−CC normalized per 100ng total RNA. D - Duodenum, J - Jejunum, I - Ileum, C - Colon. Data pooled from GF and SPF mice. n = 4–10 mice. f, MuAstV genome levels by RT-qPCR detected in cell populations sorted from the ileum at 14dpi of SPF mice (no CC). Immune: CD45+, Epithelial (-GC): EpCAM+CLCA1-, Goblet Cells: EpCAM+CLCA1+. n = 4 mice. Each dot represents data from one animal in panels (b-f). All data are represented as mean ±SEM from at least two independent experiments. LOD = limit of detection. ns = not significant, *p≤0.05, **p≤0.01, ****p≤0.0001 by 2way ANOVA and Holm-Šídák’s comparison.

Next, we analyzed gut tissue for presence of negative strand vRNA (–vRNA), a replication intermediate of positive strand RNA viruses³¹. Samples from 7dpi served as positive control while mock-infected mice served as negative control (Extended Data Fig. 1b). At 21dpi, –vRNA was still detectable in the ileum and colon of mice –CC, while +CC at 14dpi resulted in levels comparable to mock-infected mice (Fig. 1e). Only the ileum and colon of mice –CC had bands for –vRNA after RT-PCR. Consistent with studies showing that MuAstV replicates in goblet cells at peak of infection^35,36, we found that MuAstV genomes were enriched in sorted goblet cells (EpCAM+CLCA1+) compared to immune cells (CD45+) or epithelial cells without goblet cells (EpCAM+CLCA1-) (Fig. 1f; Extended Data Fig. 1c). These results indicate that MuAstV persistence represents ongoing viral replication in goblet cells and are sensitive to CC-mediated clearance.

Continued exposure contributes to persistent infection

To determine whether input quantity affects persistence, we inoculated mice with a dilution series of MuAstV and quantified viral shedding –CC. Input of 1E7 to 1E10 viral genomes in peak of infection at 7dpi was followed by persistence. Not all mice became infected when inoculated with 1E6 genomes, but those that were infected displayed continuous shedding (Fig. 2a; Extended Data Fig. 1d). This demonstrates that, if infection occurs, mice are persistently infected. Regarding CC timing, we found that CC at 10dpi led to rapid viral clearance while CC 3 and 7dpi did not (Fig. 2b). Thus, CC was effective once viral burden decreased and leveled out, potentially coinciding with timing of adaptive immunity. CC at 28 or 42dpi still resulted in reduction of vRNA levels below the limit of detection (Fig. 2c). CC was also effective in mice inoculated with MuAsV isolated from persistently infected mice (Fig. 2d). This indicates that, even if given time to evolve, MuAstV remains susceptible to clearance following CC. Subsequent experiments were performed with 14dpi as standard time of CC.

Fig. 2: Continued exposure contributes to persistent infection.

a, time course of MuAstV levels in stool detected by RT-qPCR from mice inoculated with 10-fold dilutions of the viral inoculum (1E6 – 1E10 genome copies) in the absence of CC. n = 4–6 mice. b, MuAstV levels over time after CC on 3, 7 or 10dpi. n = 4–7 mice. c, MuAstV levels before and after CC at either 28 or 42dpi. n = 4 mice. d, MuAstV levels before and after CC in primary infection or after passage of the virus. n = 4 mice. e, MuAstV levels before and after cage ‘trade’ at 14dpi. Groups: mV = mock mice transferred into cages previously containing virus-infected mice, Vm = infected mice into cages previously containing mock mice. VV = infected mice into cages previously containing a separate cohort of virus-infected mice. dpt = day post trade. n = 4 mice. f, MuAstV levels in mice with repeated oral gavage of virus or PBS at 0h, 6h, 12h and 24h after CC. n = 4 mice. g, MuAstV levels in mice before and after sifting the bedding. n = 4 mice. Data from GF and SPF mice were pooled in panels (a) and (e). Only SPF mice were used for all others. Each dot represents data from one animal in all panels. All data are represented as mean ±SEM from at least two independent experiments. ns = not significant, ***p≤0.001, ****p≤0.0001 by 2way ANOVA and Holm-Šídák’s comparison.

We tested whether continued viral exposure contributes to persistence by swapping cages between infected and mock-infected mice (Fig. 2e). Mock-infected mice placed into cages that previously harbored virus-infected mice (m→V) became MuAstV-positive, indicating infectious virus present in the cage. Infected mice placed into cages previously containing mock-infected mice (V→m) cleared the virus. Infected mice placed into cages previously containing virus-infected mice (V→V) remained MuAstV-positive, indicating that presence of virus in the environment contributes to persistence. Repetitive administration of MuAstV delayed but did not prevent clearance of virus following CC (Fig. 2f), suggesting additional components of CC, such as stress, affect persistence⁴⁵. Consistent with this, mice kept in the same cage remained infected after we removed stool pellets (Fig. 2g; Extended Data Fig. 1e). Together, these data indicate that continued exposure contributes to MuAstV persistence, but other factors associated with CC are necessary for viral clearance.

CD8 T cells are necessary for CC-mediated MuAstV clearance

A prior study showed that mice that clear MuAstV cannot be immediately reinfected ³⁷. We also found that mice +CC were not reinfected when rechallenged, and infected mice −CC remained persistently infected after a second inoculation (Fig. 3a). We examined humoral responses to the immunodominant spike domain by developing an ELISA using a 3D structure prediction of our specific strain^46–49 (Extended Data Fig. 2a). Levels of virus-specific IgA in ileum and IgG in serum were unaffected by CC (Fig. 3b; Extended Data Fig. 2b). Virus-specific antibodies emerged sometime between 7dpi and 14dpi (Fig. 3b). Additionally, CC did not enhance the development of neutralizing antibodies as virus incubated with serum from mock-infected, MuAstV-CC, and MuAstV+CC mice remained infectious (Fig. 3c). Together, these experiments do not support an effect of CC on antiviral antibody activity.

Fig. 3: CD8 T cells are necessary for cage change-mediated MuAstV clearance.

a, MuAstV genome levels over time +/−CC at 14dpi and reinfection with the same dose of MuAstV 28dpi (virus). n = 4–7 mice. b, Area under the curve (AUC) for gut anti-spike IgA levels in the ileum and time course of plasma anti-spike IgG from mice with MuAstV +/−CC at 14dpi measured by ELISA. n = 4–11 mice. c, MuAstV levels in stool of mice infected with virus inoculum incubated with serum collected at 28dpi from mock treated mice or MuAstV infected mice +/−CC. n = 4 mice. d-e, heatmaps of average fold changes in indicated groups compared to mock-CC for immune population frequencies (d) and frequency of cytokine production (e) in the lamina propria leukocytes (LPL) of the small intestine (SI) of GF mice as measured by flow cytometry. n = 4–5 mice. f, percentage of CD8 T cells of CD45+ cells in SI LPL. n = 4–5 mice. g, time course for the percentage of CD8 T cells out of CD45+ cells in the SI LPL of infected mice +/−CC at 14dpi. n = 2 mice for 7dpi, 4–5 mice for other data points. h, number of IFNγ+ CD8 T cells in SI LPL. n = 4–5 mice. i-j, MuAstV levels over time in Rag1^−/− (i) and muMT−/− (j) as measured by RT-qPCR. n = 4–6 mice. k, MuAstV levels over time in Cd8^−/− mice +/− CC at 14dpi and comparison of % of infected mice remaining in WT and Cd8^−/− SPF mice +/−CC. n = 6–7 mice. l, MuAstV levels over time and % of infected mice after CC of Cd8−/− mice treated with control or anti-CD4 antibody. n = 6–7 mice. Data from GF and SPF mice were pooled in panel (b), only GF mice were for panels (d-h) and only SPF mice for panels (i-l). Each dot represents data from one animal in panels (a-c) and (f-l). All data are represented as mean ±SEM from at least two independent experiments. ns = not significant, *p≤0.05, **p≤0.01, ****p≤0.0001 by 2way ANOVA and Holm-Šídák’s comparison for (a-c) and (f-l), two-sided Student’s T-test for (d-e), and Mantel-Cox analysis for survival curves (k-l). For panel (d-e) ‘+’ denotes fold change (FC) with p≤0.05.

To more broadly examine immune responses to CC during MuAstV infection, we analyzed lamina propria leukocytes (LPL) from the small intestine and colon by flow cytometry (Extended Data Fig. 2c). We used GF mice to avoid interfering immune reactions to other microbes⁵⁰. When compared with mock, MuAstV-CC mice showed changes of several immune cell populations including reduced B cells and upregulation of IL-22 producing cells (Fig. 3d-e; Extended Data Fig. 3a-b), as previously reported⁵⁰. MuAstV+CC mice were largely similar, except displayed an additional increase in CD8 T cells and IFNγ+ CD8 T cells at 28dpi (Fig. 3d-h; Extended Data Fig. 3c-d). We observed that MuAstV infection led to an increase of CD8 T cells at 7dpi that returned to baseline level in MuAstV-CC but remained above pre-infection levels in MuAstV+CC(Fig. 3g).

We analyzed immunocompromised mice to examine contributions of adaptive immunity. Many of these mouse lines, from vendors or bred in-house, were already infected with MuAstV and had to be re-established as virus-naïve (see methods). Previous reports showed Rag1^−/− mice lacking mature T and B cells are persistently infected with MuAstV³⁴. We found that after reaching its peak, viral burden remained high in Rag1^−/− mice without declining to the lower levels observed during persistence in WT mice and was unresponsive to CC (Fig. 3i). Similar observations were made in MuMT^−/− mice lacking mature B cells (Fig. 3j), consistent with a recent study reporting a role of IgA for MuAstV colonization⁵¹. Thus, B cells have a critical role that supersedes effects of CC. In contrast, in Cd8^−/− mice MuAstV levels peaked at 7dpi and then decreased to persistence levels similar to WT mice (Fig. 3k). MuAstV clearance was substantially delayed in Cd8^−/− mice +CC. As lack of CD8 T cells did not prevent MuAstV clearance entirely, we tested the role of CD4 T cells. Antibody-mediated depletion of CD4 T cells did not prevent MuAstV clearance in WT mice +CC, but completely abrogated CC-dependent virus clearance in Cd8^−/− mice (Fig. 4l; Extended Data Fig. 3e). These results suggest that the initial decrease of MuAstV levels from peak is mediated by mature B cells, leading to the establishment of a lower persistence level. Subsequently, a response dominated by CD8 T cells, with CD4 T cells as backup, mediates viral clearance after CC.

Fig. 4: Virus-specific CD8 T cells are not exhausted despite persistent infection.

a, percentage of IFNγ+ CD8 T cells in SI LPL of MuAstV infected mice after stimulation with DMSO (−), a small peptide pool of MuAstV capsid peptides, the immunodominant peptide AAIWNPIVV, or positive control cell stimulation cocktail (+) as measured by flow cytometry. n = 3–4 mice. b, representative flow plots of MuAstV Tetramer staining of CD8ab T cells in SI LPL for mock and MuAstV mice. Tet-, Tetramer negative; Tet+, Tetramer positive. c, percentage of Tet+ CD8ab T cells in different organs around 28dpi detected by flow cytometry. d, percentage of Tet+ CD8ab T cells in SI LPL over time. e, percentage of Tet- and Tet+ CD8ab T cells out of CD45+ immune cells in SI LPL over time. f-n, percentage of Tet- and Tet+ CD8ab T cells in SI LPL over time that are (f) CD62L+CD44-, (g) CD38+, (h) CD103+, (i) Ki67+, (j) T-bet+, (k) GzmB+, (l) IFNγ+, (m) TNFα+, and (n) IFNγ+TNFα+ (k-n after stimulation). o, representative histograms of PD-1 levels on Tet- and Tet+ populations at different times post infection. p-q, percentage of Tet- and Tet+ CD8ab T cells that are PD-1+ (p) or CD39+PD-1+ (q). n = 6–16 mice in (c-q). r, MuAstV levels as measured by RT-qPCR in Rag1−/− recipient mice before and after transfer of PBS or CD8 T cells isolated from the pooled LPL and IEL of infected WT donor mice. n = 2–3 mice. s, levels of CD8ab, Tet+ and CD39+PD-1+ cells in the SI LPL at input and after transfer into Rag1-/−. SPF mice were used for all experiments. Each dot represents data from one animal in panels (a), (c-n) and (p-s). All data are represented as mean ±SEM from at least two independent experiments. ns = not significant, *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001 by 2way ANOVA and Holm-Šídák’s comparison.

Virus-specific CD8 T cells are not exhausted

To generate a tetramer for identifying MuAstV-specific CD8 T cells, we created a peptide library containing the 50 top predicted MHC epitopes. This library stimulated IFNγ production in CD8 T cells from the small intestine LPL of infected but not mock mice (Fig. 4a; Extended Data Fig. 4a). Subdividing the library into pools and individual peptides identified a 9 amino-acid motif (AAIWNPIVV) predicted to bind MHC-I allele H-2-Db (Extended Data Fig. 4b-d). This peptide alone stimulated IFNγ production by CD8 T cells similarly to the full peptide pool (Fig. 4a).

A tetramer using this immunodominant peptide stained up to 30% of CD8 T cells in the small intestine LPL of infected mice (Tet+) (Fig. 4b; Extended Data Fig. 4e). At 28dpi without CC, ~15% of CD8 T cells were Tet+ in small intestine LPL and intraepithelial lymphocyte (IEL) fractions (Fig. 4c). Less Tet+ CD8 T cells were found in colon LPL, and almost none in mesenteric lymph nodes (MLN), blood, and spleen. In the small intestine LPL and IEL, Tet+ cells were detectable at 7dpi and remained at ~15% of all CD8 T cells (Fig. 4d; Extended Data Fig. 5a-b). Tet- and Tet+ CD8 T cells contracted sharply after 7dpi (Fig. 4e; Extended Data Fig. 5c-e). In contrast to Tet- cells that acquired markers of naïve CD8 T cells (CD62L+CD44-) at later time points, naïve Tet+ CD8 T cells remained low, and were positive for activation marker CD38 instead (Fig. 4f-g; Extended Data Fig. 5f-g). These results show that MuAstV elicits an antigen-specific CD8 T cell response.

Tissue-resident memory T cells in non-lymphoid tissues provide rapid local responses and are characterized by surface marker CD103 in the gut⁵². The initially high proportion of CD103+ CD8 T cells in LPL and IEL at resting conditions⁵³ decreased by 7dpi, consistent with influx of new cells, followed by an increase at 14dpi and 28dpi once residency was re-established (Fig. 4h; Extended Data Fig. 5h). The proportion of proliferating Ki67+ CD8 T cells spiked at 7dpi and decreased below mock levels by 14dpi (Fig. 4i; Extended Data Fig. 5i), matching CD8 T cell contraction. The CD8 T cells are likely effector cells as they express type 1 transcription factor T-bet (Fig. 4j; Extended Data Fig. 5j). The main functions of effector CD8 T cells are killing of target cells through granzymes and perforin, and cytokine production⁵⁴. Granzyme B (GzmB) was upregulated at 7dpi and then returned to baseline, while production of cytokines IFNγ and TNFα increased at later time points (Fig. 4k-n; Extended Data Fig. 5k-n)

The inhibitory molecule associated with exhaustion PD-1 was high at 7dpi and was subsequently downregulated (Fig. 4o-p; Extended Data Fig. 5o). CD39, a marker of terminal exhaustion, was similar when comparing MuAstV and mock-infected mice at 28dpi (Fig. 4q). These data are consistent with CD8 T cells actively reacting to MuAstV at peak of infection followed by T cell ignorance, potentially similar to observations with MNV^24,25. To examine whether virus-specific cells are functional, we sorted intestinal CD8 T cells from persistently infected WT mice and adoptively transferred them into Rag1^−/− recipients. These cells successfully engrafted and reduced MuAstV levels despite absence of B and CD4 T cells (Fig. 4r-s; Extended Data Fig. 5p-q). Transferred CD8 T cells upregulated PD-1 and CD39, likely through overstimulation by antigen in these highly infected Rag1^−/− mice. Thus, functional virus-specific CD8 T cells are present before CC.

CC induces CD8 T cell activation and epithelial ISG response

Flow cytometric analyses of CD8 T cells in LPL and IEL of MuAstV-infected mice 1d +CC trended towards decreased naïve and increased effector capacity marked by GzmB and cytokine production (Extended Data Fig. 5r-s). To further examine cellular responses, we performed single cell RNA sequencing (scRNA-Seq) of small intestinal tissue from mock- and MuAstV-infected mice 1d +/−CC (Extended Data Fig. 6a). Analysis of immune cells (CD45+) yielded 15 distinct populations and showed increased cytotoxic CD8 T cells levels in MuAstV+CC mice (Fig. 5a-b; Extended Data Fig. 6b-c). Sub-clustering of T cells revealed that CD8 central memory (TCM) and effector memory (TEM) cells were highest in MuAstV+CC mice (Extended Data Fig. 6d-g). Activation marker CD69 was increased after CC specifically in CD8 TEM (Fig. 5c). Pathway analysis for all T cells or CD8 TEM revealed that CC, in mock and infected mice, resulted in upregulation of pathways involved in immunological synapses including RNA processing and organelle organization (Extended Data Fig. 7a-b)^55,56. Thus, CC induced CD8 T cell activation and effector functions in both mock and infected mice.

Fig. 5: Cage change activates CD8 T cells, leading to downstream epithelial ISG responses and virus clearance.

a, UMAP visualization of major immune cell clusters in scRNA-Seq data combining mock-CC, mock+CC, MuAstV-CC, and MuAstV+CC. Cells from 2 mice were pooled per condition. b, Proportion of immune cell clusters shown in A in each condition. c, Cd69 expression in indicated T cell clusters. d, UMAP visualization of major epithelial cell clusters in scRNA-Seq data. e, Pathway enrichment analysis of genes upregulated in both CC conditions showing the six most upregulated pathways in mock (top) and MuAstV (bottom) in the EpCAM+ cells with GeneRatio for each. f, heatmap for Z-scores of genes in the ‘response to virus’ pathway in EpCAM+ cells. g, Expression levels of individual ISGs in goblet cells. h, Heatmap showing changes in gene expression of ISGs and Tnfa over time in ileal epithelial layer measured by RT-qPCR. Values are normalized to 0h post CC. FC fold change. n = 3–10 mice. i, MuAstV levels from the mice used to generate (h) measured by RT-qPCR. n = 3–10 mice. j, normalized Ifit2 RNA levels in mock infected WT, MuAstV infected WT and MuAstV infected Cd8^−/− mice 24h +/−CC measured by RT-qPCR. n = 4–14 mice. k, MuAstV levels from Cd8^−/− mice in (j) measured by RT-qPCR. n = 5–10 mice. Square (□) in (j+k) indicates the same mouse that behaved as an outlier. l, MuAstV levels in Ctr and Stat1 ΔIEC mice before and after CC measured by RT-qPCR. Stat1 deletion was induced through tamoxifen feed for 7d before CC. n = 3 mice. SPF mice were used for all experiments. Each dot represents data from one animal in panels. (i-k) and the data are represented as mean ±SEM from at least two independent experiments. ns = not significant, *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001 by two-sided Student’s T-test compared to no CC for panels (h) and (j-k). For panel (h) ‘+’ denotes FC with p≤0.05.

Next, we examined how CD8 T cell activation might affect the EpCAM+ fraction (Fig. 5d; Extended Data Fig. 7c-d). Pathway analysis revealed that CC increased innate immune responses in the epithelium of both mock and infected mice, which included interferon-stimulated genes (ISGs) (Fig. 5e-f; Extended Data Fig. 8a). An ISG score (averaged level of 16 ISGs) showed highest ISG expression in MuAstV+CC mice (Extended Data Fig. 8b). Goblet cells, which harbor MuAstV (Figure 1f), also upregulated ISGs +CC and produced additional antiviral factors such as TNFα (Fig. 5g; Extended Data Fig. 8b). Examining the timing of ISG and Tnfa expression in the intestinal epithelium revealed a modest increase at 3h, followed by a sharp decrease by 12h and another increase by 24h post CC in MuAstV-infected mice. The increased gene expression at 24h was also observed in mock mice (Fig. 5h; Extended Data Fig. 9a-e). MuAstV levels were increased at 12h post CC followed by a rapid decline, consistent with the antiviral gene pattern (Fig. 5i). We hypothesized that CD8 T cells are upstream of this epithelial immune response. Most genes, with exception of Isg15, were not increased in Cd8^−/− mice after CC (Fig. 5j; Extended Data Fig. 9a-e). One of ten mice had high ISG induction and was the only one with reduction in vRNA at 24h after CC (Fig. 5k). CD4 T cells likely compensate for the absence of CD8 T cells according to our earlier findings (Fig. 3k-l). The impact of blocking or administering individual interferons, TNFα, and Perforin1 had minimal effects on virus clearance (Extended Data Fig. 9f-l). However, blocking ISG induction through inducible deletion in the intestinal epithelium of STAT1, a signaling protein downstream of all interferons, prevented CC-mediated MuAstV clearance (Fig. 5l). Together, these data demonstrate that CC results in CD8 T cells activation, leading to downstream ISG induction in the intestinal epithelium that mediates virus clearance.

Glucocorticoid fluctuation contributes to virus clearance

The decrease of antiviral genes at 12h post CC was unexpected. Among pathways enriched after CC was response to peptide hormones (Extended Data Fig. 8a). Hormones released upon stress, such as glucocorticoids, have potent anti-inflammatory functions⁵⁷. Several genes affected by glucocorticoid signaling were responsive to CC (Fig. 6a). We examined the expression of Sgk1 as a none-immune glucocorticoid target gene and found that it peaked at 12h, followed by a sharp decline at 18h and returning to baseline by 24h (Fig. 6b). Similarly, serum levels of corticosterone, the main murine glucocorticoid, increased up to 12h and then dropped sharply (Fig. 6c). This pattern corresponded remarkably well with MuAstV levels and an inverse pattern of ISG and Tnfa expression (Fig. 6d). Glucocorticoid, ISG expression, and virus dynamics were independent of the circadian rhythm (Extended Data Fig. 10a-b). Cage trade induced ISG expression while removing stool did not (Extended Data Fig. 10c). Although glucocorticoids were detected in blood, we observed only partial upregulation of ISGs and Tnfa in blood and a decrease in spleen and MLN after CC (Extended Data Fig. 10d).

Fig. 6: Glucocorticoid fluctuation contributes to ISG expression and virus clearance.

a, heatmap for Z-scores of genes in the response to peptide hormone pathway that are related to glucocorticoid signaling in EpCAM+ cells in scRNA-Seq dataset. Genes are grouped based on whether prior studies have shown them to be upregulated (Up), downregulated (Down), or interact with the glucocorticoid receptor during hormone signaling (Signal). b, time course of Sgk1 expression after CC in samples from Fig. 5J. n = 3–8 mice. c, corticosterone levels in serum from mice from Fig. 5h as determined by ELISA. n = 3–10 mice. d, comparison of behavior of averaged corticosterone levels, ISG + Tnfa expression (as shown in Fig. 5h), and MuAstV levels after CC. e, percentage of mice with detectable MuAstV in their stool treated with either vehicle control (Ctr), glucocorticoid antagonist Mifepristone (Mif) or synthetic glucocorticoid Methylprednisolone acetate (MPA) +/−CC. n = 6–11 mice. f, Heatmap of normalized ISG and Tnfa expression in Mifepristone and Methylprednisolone acetate treated mice 24h post CC compared to control. n = 4–6 mice. g, model of cage change-induced MuAstV clearance. SPF mice were used for all experiments. Each dot represents data from one animal in panels (b-c) and (e-f) and the data are represented as mean ±SEM from at least two independent experiments. ns = not significant, *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001 by 2way ANOVA using Holm-Šídák’s comparison for panels (c) and (e), and two-sided Student’s T-test compared to Ctr for panel (f). For panel f ‘+’ denotes FC with p≤0.05.

In the gut, we hypothesized that the short-lived anti-inflammatory glucocorticoid signaling following CC led to an immune rebound reaction that mediated MuAstV clearance. Therefore, breaking the glucocorticoid fluctuation should interfere with CC-mediated viral clearance. The glucocorticoid-antagonist Mifepristone or synthetic glucocorticoid Methylprednisolone acetate inhibited MuAstV clearance after CC and reduced ISG expression (Fig. 6e-f; Extended Data Fig. 10e-j), placing glucocorticoids upstream of ISG induction.

Discussion

We discovered that an intervention as simple as changing cages determines the duration of a model viral infection in mice. Cage change causes a stress response characterized by temporary production of corticosterone, triggering transient immune suppression and increase in viral replication. This is followed by a rebound in CD8 T cells and downstream antiviral gene expression in the epithelium necessary for virus elimination (Fig. 6g). This antiviral response is likely not sufficient to prevent reinfection when viruses remain in the same environment, thus explaining why removing the external reservoir and breaking the chain of reinfection is necessary.

Persistent infections might be more common than appreciated for MuAstV. High prevalence of MuAstV in wild and pet store mice^58–60 may reflect prolonged infections and windows of transmission. Although free-living mammals do not experience cage changes, predators or natural disasters can force infected animals to migrate. In those cases, the removal of the external reservoir and stress-mediated strengthening of immune barriers could induce clearance of ongoing infections while protecting from new pathogens.

Our observations also have implications for mouse husbandry. Cage change during the daytime is standard practice but causes considerable stress by disrupting the normal resting period of mice, which are nocturnal⁴⁵. Despite representing such a drastic procedure, cage changes are rarely considered as an experimental variable in mouse studies. Careful documentation of husbandry practices may aid efforts to improve experimental reproducibility or reveal pathways that are sensitive to external perturbations.

A limitation of our study is that we used chemicals to modulate glucocorticoid signaling without delving into cell type and tissue-specific responses and have not ruled out other pathways that potentially intersect stress responses to cage change. Stress and glucocorticoids have established immunomodulatory effects and clinical applications, but the target cell types and signaling pathways are frequently not known because they are tissue- and context-specific⁶¹. Glucocorticoids can impact CD8 T cells by favoring memory over effector programming, which limits immunopathology during infections^62–64. Although typically considered anti-inflammatory, sustained glucocorticoid production due to psychological stress exacerbates TNFα-induced intestinal inflammation⁶⁵. Given the variety of external and internal factors that trigger stress responses, an important future direction is to define the exact relationship between corticosterone, CD8 T cells, and the epithelium.

Glucocorticoid administration during viral infections improves health outcomes by preventing immune-mediated damage⁶⁶. Our results suggest that persistence of viruses may be impacted by the timing of therapy. Additionally, psychological stress and re-exposure to viruses varies considerably between individuals and could explain why many viruses or their products can be detected in a subset of individuals after acute illness¹⁷. A better understanding of how external factors impact immune responses may reveal non-pharmaceutical intervention strategies or inform the use of existing drugs to control viral infections.

Methods

Ethics oversight

All procedures and methods were conducted in compliance with regulations at New York University and the University of Pennsylvania. All animal experiments were conducted according to the Public Health Service Policy on Humane Care and Use of Laboratory Animals, and were evaluated and approved by the Institutional Animal Care and Use Committee (IACUC) at at New York University (protocols #IA16–0087 and #IA16–00864) and the University of Pennsylvania (protocols #807356 and #807406).

Mice

C57BL/6J (WT, #000664), Cd8^−/− (#002665), muMT−/− (#002288), Ifngr−/− (#003288), Ifnar1^fl (# 028256), Prf1−/− (#002407), and Villin-cre (#004586) were originally purchased from The Jackson Laboratory and bred onsite or directly used for experiments. Ifnlr1^fl mice were provided by Skip Virgin at Washington University⁶⁷. Ifnar1^fl and Ifnlr1^fl were bred to Villin-cre mice to generate specific deletions in intestinal epithelial cells (Ifnar^ΔIEC and Ifnlr^ΔIEC, respectively). Rag1^−/− mice were obtained from the Gnotobiotic Mouse Core at the University of Pennsylvania to avoid MuAstV contamination risk. Stat1^fl Villin-cre^ERT2 (Stat1^ΔIEC) mice were provided by the Hunter lab at the University of Pennsylvania, which originally acquired Stat1^fl and Villin-cre^ERT2 from The Jackson Laboratory⁶⁸. Stat1 deletion was induced by using mouse chow containing 250mg/kg Tamoxifen (Inotiv, TD.130855) for 7 days. Mice were confirmed to be negative for MuAstV by RT-qPCR before experiments. For mice that were positive for MuAstV infection, the line underwent backcrossing through C57BL/6J or early cross-fostering to eliminate virus contamination.

Mice between 6–12 weeks of age were used for all experiments. Age- and sex-matched littermate controls were used whenever possible. Mice were housed in ventilated cages at room temperature with a 12h light-dark cycle and given food and water ad libitum. Mice of both sexes were used for all experiments besides the immune profiling in Figure 3 and the scRNA-Seq in Figure 5, for which only female mice were used. GF mice were bred in flexible-film isolators either in the New York University Grossman School of Medicine Gnotobiotics Animal Facility or the University of Pennsylvania Gnotobiotic Mouse Core. For experiments, GF mice were transferred to ISOcages (Tecniplast) with sterile food and water, and absence of bacteria was monitored throughout the experiment by qPCR for 16S.

Cage change involved transfer of mice into a clean cage with fresh food and water during the daytime, unless otherwise indicated (see Extended Data Fig. 10b in which morning versus night were compared). Cage trade involved transfer of mice into a cage that previously housed other mice of the same sex. Cage sifting was performed by transferring mice to a clean cage temporarily (up to 15 minutes) and removing as much feces as possible by sifting the cage contents with a Garden Potting Mix Sieve (Practicool) by combining the 2, 3 and 4mm mesh filters followed by manual removal of remaining stool pellets. For viral passage, infected mice were housed with uninfected mice at 14dpi. A pilot study was conducted for prolonged housing of mice without cage change. In short, ammonia levels, humidity and temperature were measured daily for 12 weeks. It was determined that housing of 2 mice in the same cage for up to 6 weeks was possible without increase of ammonia above 50 ppm. Food and water were replenished as needed.

Virus purification and infection

We used murine astrovirus-1 strain NYU1 for all experiments⁵⁰. Viral stock was generated as previously described⁵⁰. In short, GF C57BL/6J were inoculated and fecal pellets containing virus collected at peak of infection before onset of the humoral immune response. This was followed by homogenization in PBS, removal of particulates by repeated spins at 17000g and filtration of the supernatant through 0.22μm Millex-GP syringe-driven filter unit (Millipore Sigma) at least twice. Viral titers were determined by RNA extraction and RT-qPCR for the viral genome. Mice were inoculated with a standard dose of 1E10 genome copies/mouse of MuAstV diluted in PBS via oral gavage. For repeated gavage of mice with virus, stool pellets from 7dpi of the respective mice were used and equal to 1 pellet/mouse/time point was homogenized in PBS and administered by oral gavage. For the neutralization assay, 10μl of virus was incubated with 40μl of serum for 90min at RT. Virus was diluted 1:100 in PBS for inoculation of mice equaling the standard infectious dose used.

Viral quantification

Fresh stool pellets were collected from mice before infection and throughout the experiment and stored at −70°C until further processing. Pellets were weighed and homogenized in RLT buffer (RNeasy kit, Qiagen) with 1.0 mm Zirconia/Silica Beads (BioSpec Products) using a FastPrep-24 homogenizer (MP Biomedicals). Debris was removed using two spins of 2,000g and 10,000g, each 5 min at RT. RNA was isolated as described below. The following probes were used for quantifying viral genomes: MuAstV Fwd 5’-TAC ATC GAG CGG GTG GTC GC-3’ and Rev 5’-GTG TCA CTA ACG CGC ACC TTT TCA-3’. Absolute amounts were calculated by comparison to a linearized plasmid.

Negative strand RT-qPCR

The protocol was adapted from a similar approach used for MNV⁶⁹. RNA was isolated from 1cm of the respective gut sections from mice at 21dpi with and without cage change at 14dpi using the homogenization method mentioned above. 100ng of total RNA were used for reverse transcription using Superscript-IV enzyme (Invitrogen) using the following primers: MuAstV RTnegTag 5’-GTC CAA CGT GGT GAA GGA TTG TCG GGA TAA GGC AAG TAG TGC TCC CCT ATT C-3’. The RT primer contained a non-viral tag sequence (bolded) attached to the strand specific sequence. The RT reaction contained 5X buffer, 20mM DTT, 0.5mM dNTPs, 100nM RT-primer and 0.025μl of Superscript-IV enzyme with a total volume of 20μl. The RT reaction was carried out at 55°C for 15min followed by 80°C for 10min. cDNA was diluted 1:10 in H₂O and qPCR was performed using the following primers: MuAstV neg. strand Fwd 5’- GTC CAA CGT GGT GAA GGA TTG TCG-3’ and Rev 5’- GTA CAA GGA CAT CTT TGG CAT GTG GG-3’.

RNA isolation and quantification

RNA was isolated using the RNeasy kit including the DNase incubation step (Qiagen). cDNA was generated using High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems) according to the manufacturer’s protocol and stored at −20°C. Transcripts were quantified using either the LightCycler 480 SYBR Green I Master (Roche) on a Roche480II Lightcycler or the PowerTrack™ SYBR Green Master Mix for qPCR (Applied Biosystems) on a QuantStudio 5 Real-Time PCR System (Applied Biosystems). For list of primers see Supplemental Table 1.

MuAstV Spike purification and ELISA

3D structure of the MuAstV NYU-1 capsid protein was modeled using a public version of AlphaFold⁴⁶(https://colab.research.google.com/github/deepmind/alphafold/blob/main/notebooks/AlphaFold.ipynb). Alignment with processing sites reported for human Astrovirus cut sites were used to decide design ELISA epitope (amino acids 425–686)⁷⁰. A gene block covering the spike portion of NYU-1 strain was ordered from IDT and cloned into pGSTag plasmid (Addgene, 21877) through Gibson Assembly (NEB, E2611) using the following primer: Spike-GST Fwd GGT GGT GGT GGT GGA ATT CTA GAC ATG CCT GAA ACC GCA G and Rev G TCA GTC AGT CAC GTG AAT TAA GCT T CTA AGG CCC AGT GGG TGG TAT TTC. Construct was transformed into BL21 (DE3) cells (NEB C2527I) and expression induced with 0.1 mM isopropyl-β-d-thiogalactoside (IPTG; Sigma) for 4h. Cells were lysed (Buffer: 50 mM Tris-HCl pH 8, 10 mM NaCl, 2 mM MgCl2, 1 mM PMSF, Protease inhibitor cocktail, 0.05% NP-40, 0.5 mg/ml Lysozyme and 25 U/ml benzonase), spun for 30min at 4°C 4,000g and protein purified from supernatant using glutathione agarose beads (Pierce, 16100) and elution with reduced L-Glutathione (VWR, 76177–906). Protein purity was confirmed on a protein gel and concentration was determined using a Nanodrop, diluted in 25% glycerol and stored at −20°C.

For the MuAstV ELISA, 96 well plates (USA Scientific, 5667–5074) were coated with 2μg/ml MuAstV spike overnight at 4°C. Wells were washed and blocked in block buffer (PBS, 2%BSA, 0.05% Tween) for 2h at room temperature. After washing, serum samples isolated by cheek bleed or cardiac puncture were diluted in block buffer, added to the plate and incubated for 2h at room temperature. After washing goat anti-mouse IgG or IgA HRP antibody (Invitrogen, 31430) was added for 1h at room temperature. Plate was washed, ELISA developed using TMB (eBioscience, 00–4201-56), stopped using H₂SO₄ and read at 450nm using a standard plate reader.

Corticosterone ELISA

Corticosterone concentrations in serum isolated by cardiac puncture were measured using the Corticosterone Competitive ELISA Kit (Invitrogen, EIACORT) as per manufacturer’s instructions. Mice composing the 0h control group (no cage change) were euthanized at 0h, 11h, 15h and 24h relative to cage change to account for circadian variation in hormone levels.

Drug administration and other treatments

Mifepristone (MilliporeSigma, M8046) was dissolved in corn oil and administered at 200mg/kg via oral gavage daily from −2 to 0 days relative to cage change. Methylprednisolone acetate (MPA, MilliporeSigma, PHR1718) was dissolved in corn oil and 2mg administered via i.p. once 3 days before cage change. Control mice were administered equal volume of corn oil. Anti-mouse CD4 antibody (BioXCell, BE0003) or control antibody (BioXCell, BE0090) were administered at 0.2mg via i.p. at day −3, 0 and +3 relative to cage change. Anti-mouse IFNγ antibody (BioXCell, BE0055) or control antibody (BioXCell, BE0088) were administered at 1mg via i.p. at day −3 and 0 relative to cage change. Anti-mouse TNFα antibody (BioXCell, BE0058) or control antibody (BioXCell, BE0088) were administered at 0.2mg via i.p. at day −2 and 0 relative to cage change. Recombinant mouse IFNβ (1 μg/mouse, R&D system, 8234-MB-010/CF), mouse IFNγ (5μg/mouse, PeproTech, 315–05-100UG), mouse IFNλ2 (1 μg/mouse, PeproTech, 250–33-5UG), TNFα (5 μg/mouse, PeproTech, 315–01A-20UG) or PBS control were administered via i.p. at 14dpi after sifting the cage contents to remove viral reservoir.

Organ processing

For single cell suspensions of small intestinal and colonic LPL fractions, sections were washed with HBSS (Gibco), fat and Peyer’s patches removed, incubated first with 20 mL of HBSS (Gibco) with 2% HEPES (Corning), 1% sodium pyruvate (Corning), 5mM EDTA, and 1 mM dithiothreitol (DTT; Sigma-Aldrich) for 15 min at 37°C and then washed with 10 mL of HBSS with 2% HEPES, 1% sodium pyruvate, 5mM EDTA for 10 min at 37°C. Tissue sections were then washed in HBSS with 5% FCS, minced, and enzymatically digested in HBSS with 5% FCS, collagenase D (0.5 mg/mL, Roche) and DNase I (0.01 mg/mL, Sigma-Aldrich) for 30 min at 37°C with constant agitation. Digested tissue was passed through 70μm mesh and cells were subjected to gradient centrifugation using 40% Percoll (Sigma-Aldrich). For IEL fraction, supernatant after DTT step was washed with RMPI, filtered and subjected to gradient centrifugation using 40% Percoll. MLNs were collected and passed through 70μm cell strainers (BD). Spleens were passed through a 70μm cell strainer, subjected to red blood cell lysis using ACK buffer (Thermo Fisher) and resuspended in PBS. Blood was collected by cardiac puncture into EDTA coated tubes and subjected to red blood cell lysis using ACK buffer (Thermo Fisher) twice and resuspended in PBS.

For measuring cytokine production, cells were plated in RPMI with 10% FCS and 2% HEPES and incubated with Cell Stimulation Cocktail plus protein transport inhibitors (eBioscience) for 4h at 37°C. For goblet cell sorting, about 10cm of the ileum were collected, washed, digested in HBSS with 5% FCS with collagenase D (1 mg/mL) and DNase I (0.15 mg/mL, Sigma-Aldrich) for 30 min at 37°C with constant agitation and filtered through a 70μm cell strainer. For gene expression measurements in the intestinal epithelium, 7.5cm of the ileum were removed and incubated in 10ml of HBSS with 2% HEPES, 1% sodium pyruvate, 5mM EDTA, and 1 mM DTT for 15 min at 37°C. Cells were stored at −70°C until RNA isolation.

Flow cytometry and cell sorting

Cells were pre-incubated with TruStain FcX™ PLUS (anti-mouse CD16/32) Antibody (BioLegend) and Zombie UV Fixable Viability dye (BioLegend). For Figure 3 and Extended Data Figure 2–3 three separate staining panels were adapted from Dallari et. al⁵⁰. The first panel included antibodies against CD90.2 (1:200), CD64 (1:200), CD11c (1:200), CD45 (1:200), MHCII (1:200), B220 (1:200), CD11b (1:200), LY6C (1:200) and CD103 (1:200). The second panel included antibodies against CD19 (1:100), GATA3 (1:100), TCRγδ (1:200), T-bet (1:50), CD3 (1:100), CD44 (1:200), NK1.1 (1:200), CD127 (1:100), CD62L (1:100), CD8β (1:200), RORγt (1:100), CD45 (1:200), CD4 (1:200), CD8α (1:200) and FoxP3 (1:100). The third panel included antibodies against CD19 (1:100), IFNγ (1:100), Granzyme B (1:67), TCRγδ (1:200), IL-22 (1:100), CD3 (1:100), IL-17 (1:100), CD127 (1:100), TCRβ (1:100), IL-10 (1:100), CD45 (1:200), CD4 (1:100) and CD8α (1:100). Samples were fixed with either Fixation Buffer (BioLegend) or FoxP3/Transcription Factor Staining Buffer Set (eBioscience). For transcription factors staining cells were permeabilized and stained in the Foxp3/Transcription Factor Staining Buffer Set at 4°C. For cytokine staining cells were permeabilized and stained with the Intracellular Staining Permeabilization Wash Buffer (BioLegend) at 4°C. Samples were recorded on a BD LSR II (BD Biosciences) and analyzed using FlowJo software (FlowJo LLC).

For Figure 5 and Extended Data Figure 4–5 two separate staining panels were used. These panels included antibodies against CD8α (1:100), CD45 (1:200), CD44 (1:100), PD-1 (1:100), CD62L (1:100), CD8β (1:200), MuAstV Tetramer (1:150), TCRβ (1:200), CD45RB (1:100), CD103 (1:100), CD38 (1:100), Tbet (1:50), Ki67 (1:200), CD39 (1:100), GzmB (1:30), TNFα (1:100), and IFNγ (1:100). Samples were processed as before, recorded on a Cytek Aurora (Cytek Biosciences) and analyzed using FlowJo.

The antibody panel used for sorting of goblet cells contained antibodies against CD45 (1:200), EpCAM (1:1000), and CLCA1 (1:100). For cell sorting, cells were stained with extracellular markers and sorted using a Cytek Aurora CS by the Flow Cytometry Core Laboratory at the Children’s Hospital of Philadelphia.

Peptide library screening

A peptide library spanning the entire MuAstV NYU-1 capsid protein was created with 15aa peptides and 11aa overlap totaling 228 peptides. MHC I and MHC II epitopes were predicted using the Immune Epitope Database Analysis Resource (http://tools.iedb.org/main/). 50 peptides were chosen that contained the major MHC I and MHC II epitopes and synthesized by GenScript (see supplemental table 2). Peptides were resuspended in DMSO to 50 mg/ml and combined for 1 mg/ml/peptide. Small intestine LPL fraction was isolated and stimulated with peptides at final concentration of 2.5 μg/ml/peptide in RPMI for 5h at 37°C. After 1h of stimulation, Protein Transport Inhibitor cocktail containing Brefeldin A and Monensin (BD) was added. Cell Stimulation Cocktail plus protein transport inhibitors (eBioscience) was used as positive control while DMSO alone served as negative control. Cells were analyzed by flow cytometry for IFNγ production. The MuAstV tetramer based on the final 9aa peptide was produced by the NIH Tetramer Facility.

CD8 T cell transfer

LPL and IEL fractions from the small intestine were isolated from infected WT donor mice at 14dpi as described in the ‘Organ processing’ section above. CD8 T cells were enriched using a negative selection kit (Miltenyi Biotec, 130–104-075). At least 200,000 CD8 T cells or PBS control were transferred into infected Rag1−/− recipient mice (also 14dpi) using retro-orbital injection. MuAstV levels were monitored in stool and CD8 engraftment was confirmed 14d post transfer by isolation of LPL and IEL small intestinal fractions followed by flow cytometry.

Single-cell RNA sequencing

About 10cm of ileum were removed, washed, digested in HBSS with 5% FCS with collagenase D (1 mg/mL) and DNase I (0.15 mg/mL, Sigma-Aldrich) for 30 min at 37°C with constant agitation and filtered through a 70μm cell strainer. Cells were counted and equal numbers from two mice were pooled. Cells were purified using Debris Removal Solution (Miltenyi Biotec) and Dead Cell Removal Kit (Miltenyi Biotec). Libraries were generated by the Single Cell Technology Core at the Children’s Hospital of Philadelphia using an 10X Genomics platform. Libraries were sequenced by the High Throughput Sequencing Core at the Children’s Hospital of Philadelphia using a NovaSeq6000 sequencing system (Illumina).

Single-cell RNA sequencing data analysis

The Cell Ranger pipeline version 7.1. was used to demultiplex cellular barcodes and align reads against the mouse reference genome mm10 (GRCm38) using the GENCODE vM23/Ensembl 98 annotation. Downstream RNA-seq analysis was conducted using Seurat version 5.1. on R version 4.2.1 with the filtered RNA and HTO feature counts. Cells containing more than 10% mitochondrial DNA and fewer than 300 feature genes were filtered out during initial quality control. HTOs were normalized using centered log-ratio transformation and demultiplexed. Doublets were removed using the scDblFinder package with a calculated doublet rate of 7.5%. Regularized negative binomial regression was performed for RNA normalization using the v2 SCTransform, and integrated samples using Harmony v 1.2.3 using sequencing lanes as co-variates to account for potential batch effects. Principal component analysis was conducted, and the Louvain algorithm was used for unsupervised clustering, with dimensions determined via the ElbowPlot method. UMAP representation was used to visualize the data.

Cell types were determined by a combination of unbiased clustering, canonical cell type marker signatures, and cell type annotation via the SingleR v 2.6.0 package with the ImmGenData open-source reference databases in the CellDex. v1.14.0 package. The Wilcoxon test was used to assess differentially expressed genes between clusters and treatment groups, with a Benjamini-Hochberg p-value adjustment. Differentially expressed genes were required to be expressed in a minimum of 20% of cells per group, have a log fold change of at least 1, and a p_val_adj < .01. Enrichment of Gene Ontology (GO) Biological Processes pathways was performed using the enrichGO function in the clusterProfiler v4.12.6 package. Significant pathways required a minimum of 15 genes per pathway and p and q-values of < .01. GO categories with p < .01, q < .01, and a minimum occurrence of ≥15 genes per pathway were considered significant.

Statistics and Reproducability

Statistical tests were selected according to appropriate assumptions to data distribution and variance. Statistics performed as noted in figure legends and performed using R, Microsoft Excel, or GraphPad Prism 10 software (La Jolla, CA, USA). No statistical methods were used to predetermine sample size. No data was excluded from analysis. Data distribution was assumed to be normal but this was not formally tested. The experiments were not randomized. Data collection and analysis were not performed blind to the conditions of the experiments.

Extended Data

Extended Data Fig. 1:

a, MuAstV genome levels in stool of GF versus SPF B6 mice over time without CC detected by RT-qPCR. n = 3 mice. b, Ct values for MuAstV negative strand RT-qPCR mock mice and those at peak of infection at 7dpi in different gut sections normalized per 100ng total RNA. D - Duodenum, J - Jejunum, I - Ileum, C - Colon. Each dot represents data from one animal. n = 2 mice. c, representative flow plots illustrating the gating strategy used for sorting of the following population in Fig. 1e: Immune - CD45+, Epithelial (-GC) - EpCAM+CLCA1-, Goblet Cells - EpCAM+CLCA1+. d, MuAstV levels for individual mice infected with 1E6 viral genomes. n = 4 mice. e, representative images of cages before and after removal of the environmental viral reservoir through sifting. GF mice used in panel (a), both SPF and GF mice in (d), and SPF mice in all other panels. Data in (a) are represented as mean ±SEM. ns = not significant, *p≤0.05, by 2way ANOVA and Holm-Šídák's comparison.

Extended Data Fig. 2:

a, 3D structure of MuAstV Orf2 as predicted with AlphaFold and corresponding confidence. Amino acids 425–686 are the spike portion of the MuAstV capsid and represent the immunodominant domain for humoral responses. b, raw values for anti-spike IgA data shown in Figure 3B. n = 4 mice. c, Representative flow plots illustrating the gating strategy used for immune profiling shown in Fig. 3 and Extended Data Fig. 3. Each dot represents data from one animal in panels (b). Data in (b) are represented as mean ±SEM.

Extended Data Fig. 3:

a-b, Heatmaps of average fold changes compared to mock-CC for immune populations (c) or cytokine production (d) in the small intestine and colon as frequencies of LPL in GF mice over time as measured by flow cytometry. n = 2–6 mice. c, Total number of CD8 T cells in the SI LPL fraction at 28dpi. n = 4–6 mice. d, Percentage of CD8 T cells producing IFNγ after stimulation at 28dpi. n = 4–6 mice. e, MuAstV levels in mice treated with control and anti-CD4 antibody +/- cage change at 14dpi measured by RT-qPCR. GF mice used in all panels. Each dot represents data from one animal in panels (c-e). Data in (c-e) are represented as mean ±SEM.ns = not significant, **p≤0.01, ***p≤0.001, ****p≤0.0001 by 2way ANOVA and Holm-Šídák's comparison for (c) and (e), and two-sided Student’s T-test for (a-b). For panel (a-b) ‘+’ only denotes a fold change with any p-value less than 0.05.

Extended Data Fig. 4:

a, Percentage of IFNγ+ CD8 T cells in SI LPL of mock or MuAstV mice after 5h of stimulation with either DMSO (-), a small peptide pool of MuAstV capsid peptides containing the 50 peptides with the highest binding capacity to mouse MHC I/II (pool), or cell stimulation cocktail (+) as measured by flow cytometry. n = 2–4 mice. b, Percentage of IFNγ+ CD8 T cells in SI LPL of MuAstV mice stimulated with smaller peptide pools or controls. n = 3 mice. c, percentage of IFNγ+ CD8 T cells in SI LPL of MuAstV mice stimulated with individual peptides or controls. n = 3 mice. d, Rank of MHC-I binding prediction of MuAstV Orf2 capsid protein. Identified immunodominant peptides are highlighted with the red circle. e, Representative flow plots illustrating the gating strategy using the MuAstV-specific Tetramer to characterize CD8 T cell responses after viral infection. Each dot represents data from one animal in panels (a-c) and data are represented as mean ±SEM from at least two independent experiments.

Extended Data Fig. 5:

a, Percentage of Tet+ CD8ab T cells in SI IEL over time. b, total number of Tet+ cells in SI LPL and IEL over time. c, Percentage of Tet- and Tet+ CD8ab T cells out of CD45+ immune cells in SI IEL over time. d, Total number of Tet+ and Tet- CD8ab T cells in SI LPL and IEL over time. e, Total number of CD8ab in SI LPL and IEL over time. f-o, Percentage of Tet- and Tet+ CD8ab T cells or mean fluorescence intensity (MFI) in SI LPL or IEL over time that are (f) CD62L+CD44-, (g) CD38+, (h) CD103+, (i) Ki67+, (j) T-bet+, (k) GzmB MFI (l) IFNγ MFI, (m) IFNγ+, (n) TNFα MFI and (o) PD-1+. GzmB, IFNγ, and TNFα after stimulation. n = 6–16 mice in (a-o). p, Levels of CD8ab, Tet+ and CD39+PD-1+ cells in the SI IEL at input and after transfer into Rag1-/-. SPF mice were used in all experiments. o, Representative flow plots for CD39 and PD-1 levels on CD8ab T cells before and after transfer into Rag1-/-. n = 3 mice. r-s, Comparison of Tet- and Tet+ populations in the SI LPL (r) and IEL (s) in MuAstV infected mice 1d post CC as compared to the average of the -CC controls. n = 4–10 mice. SPF mice were used in all experiments. Each dot represents data from one animal in panels (a-b), (e-p) and (r-s). All data are represented as mean ±SEM from at least two independent experiments. ns = not significant, *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001 by 2way ANOVA and Holm-Šídák's comparison for (a-o), two-sided Student’s T-test (r-s).

Extended Data Fig. 6:

a, Cell numbers for singe-cell RNA sequencing (scRNA-Seq) for the different conditions that passed quality control. b, UMAP visualization of all cells in the scRNA-Seq data set. c, Heatmap of lineage marker expression used for defining the different immune cell clusters. d, Proportion of cytotoxic CD8 under the different conditions in the immune cell data set. e, UMAP visualization of the T cell subclusters. f, UMAP visualization of T cell subsets in the different samples. g, Heatmap of lineage marker expression used to define the different T cell subclusters. h, Proportion of CD8 TCM and TEM under the different conditions in the T cell subset.

Extended Data Fig. 7:

a, Pathway enrichment analysis of genes upregulated in both cage change conditions showing the most upregulated pathways in mock (top) and MuAstV (bottom) in the T cell subset with GeneRatio for each. b, Pathway enrichment analysis of genes upregulated in both cage change conditions showing the seven most upregulated pathways in mock (left) and MuAstV (right) in the CD8 TEM cluster with GeneRatio for each. c, UMAP visualization of EpCAM+ cells in the data set color coded by sample. d, Heatmap of lineage marker expression used for defining the different epithelial cell clusters.

Extended Data Fig. 8:

a, Pathway enrichment analysis of genes upregulated in both cage change conditions showing the six most upregulated pathways in mock (left) and MuAstV (right) in the EpCAM+ subset with GeneRatio for each. b, Module Score of the ISG signature in epithelial cell clusters in the individual samples.

Extended Data Fig. 9:

a-e, Normalized RNA levels in mock infected WT, MuAstV infected WT and MuAstV infected CD8 KO mice 24h +/-CC measured by RT-qPCR for the following genes: (a) Ifit1, (b) Isg15, (c) Mx2, (d) Oasl2, and (e) Tnfa. Square (□) indicates the same mouse that behaved as an outlier and was excluded from statistics. n = 4–14 mice. f, MuAstV levels over time in Ifngr-/- mice +/-CC at 14dpi as measured by RT-qPCR. n = 6 mice. g, MuAstV levels before and after CC at 14dpi in GF or SPF mice that were either pre-treated with a control or anti-IFNγ antibody. n = 2 mice. h, MuAstV levels over time in IFNAR ΔIEC mice +/-CC at 14dpi as measured by RT-qPCR. n = 3–5 mice. i, MuAstV levels over time in IFNLR ΔIEC mice +/-CC at 14dpi as measured by RT-qPCR. n = 5–7 mice. j, MuAstV levels in mice treated with control or anti-TNF antibody before and after CC at 14dpi. n =3–4 mice. k, MuAstV levels over time in Prf1-/- mice +/-CC at 14dpi as measured by RT-qPCR. n = 5–6 mice. l, MuAstV levels in mice before and after treatment with PBS or the indicated cytokine at 14dpi. n = 3–8 mice. SPF mice were used in all experiments. Each dot represents data from one animal in all panels. All data are represented as mean ± SEM from at least two independent experiments. ns = not significant, *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001 via two-sided Student’s T-test in panels (a-e) and by 2way ANOVA and Holm-Šídák's comparison for panels (f-l).

Extended Data Fig. 10:

a, Heatmap showing changes in gene expression of ISGs and Tnfa at 12h and 24h post cage change normalized to the respective -CC controls from the same time point measured by RT-qPCR. n = 3–10 mice. b, MuAstV levels before and after cage change in the morning or at night as measured by RT-qPCR. n = 3 mice. c, Heatmap showing changes in gene expression of ISGs and TNFα 24h after cage change (CC), cage trade (CT), or sifting (Sift). n = 3–10 mice. d, Heatmap showing changes in gene expression of ISGs and TNFα 24h post cage change in different organs. n = 3–10 mice. e-j, Normalized RNA levels in control, Mifepristone and Methylprednisolone acetate treated mice 24h post cage change measured by RT-qPCR for the following genes: (e) Ifit1, (f) Ifit2, (g) Isg15, (h) Mx2, (i) Oasl2, and (j) Tnfa. n = 4–6 mice. In all the panels each dot represents one animal. SPF mice were used in all experiments. Each dot represents data from one animal in panels (b) and (e-j) and data are represented as mean ± SEM. ns = not significant, *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001 by 2way ANOVA and Holm-Šídák's comparison for panels (b) and via two-sided Student’s T-test in panels (e-j).

Data availability

Sequencing data was deposited to Gene Expression Omnibus (GEO) under the accession number GSE279634.