The Intestinal Microbiome Primes Host Innate Immunity against Enteric Virus Systemic Infection through Type I Interferon

ABSTRACT Intestinal microbiomes are of vital importance in antagonizing systemic viral infection. However, very little literature has shown whether commensal bacteria play a crucial role in protecting against enteric virus systemic infection from the aspect of modulating host innate immunity. In the present study, we utilized an enteric virus, encephalomyocarditis virus (EMCV), to inoculate mice treated with phosphate-buffered saline (PBS) or given an antibiotic cocktail (Abx) orally or intraperitoneally to examine the impact of microbiota depletion on virulence and viral replication in vivo. Microbiota depletion exacerbated the mortality, neuropathogenesis, viremia, and viral burden in brains following EMCV infection. Furthermore, Abx-treated mice exhibited severely diminished mononuclear phagocyte activation and impaired type I interferon (IFN) production and expression of IFN-stimulated genes (ISG) in peripheral blood mononuclear cells (PBMC), spleens, and brains. With the help of fecal bacterial 16S rRNA sequencing of PBS- and Abx-treated mice, we identified a single commensal bacterium, Blautia coccoides, that can restore mononuclear phagocyte- and IFNAR (IFN-α/β receptor)-dependent type I IFN responses to restrict systemic enteric virus infection. These findings may provide insight into the development of novel therapeutics for preventing enteric virus infection or possibly alleviating clinical diseases by activating host systemic innate immune responses via respective probiotic treatment using B. coccoides.

ABSTRACT Intestinal microbiomes are of vital importance in antagonizing systemic viral infection. However, very little literature has shown whether commensal bacteria play a crucial role in protecting against enteric virus systemic infection from the aspect of modulating host innate immunity. In the present study, we utilized an enteric virus, encephalomyocarditis virus (EMCV), to inoculate mice treated with phosphate-buffered saline (PBS) or given an antibiotic cocktail (Abx) orally or intraperitoneally to examine the impact of microbiota depletion on virulence and viral replication in vivo. Microbiota depletion exacerbated the mortality, neuropathogenesis, viremia, and viral burden in brains following EMCV infection. Furthermore, Abxtreated mice exhibited severely diminished mononuclear phagocyte activation and impaired type I interferon (IFN) production and expression of IFN-stimulated genes (ISG) in peripheral blood mononuclear cells (PBMC), spleens, and brains. With the help of fecal bacterial 16S rRNA sequencing of PBS-and Abx-treated mice, we identified a single commensal bacterium, Blautia coccoides, that can restore mononuclear phagocyte-and IFNAR (IFN-a/b receptor)-dependent type I IFN responses to restrict systemic enteric virus infection. These findings may provide insight into the development of novel therapeutics for preventing enteric virus infection or possibly alleviating clinical diseases by activating host systemic innate immune responses via respective probiotic treatment using B. coccoides. IMPORTANCE While cumulative data indicate that indigenous commensal bacteria can facilitate enteric virus infection, little is known regarding whether intestinal microbes have a protective role in antagonizing enteric systemic infection by modulating host innate immunity. Although accumulating literature has pointed out that the microbiota has a fundamental impact on host systemic antiviral innate immune responses mediated by type I interferon (IFN), only a few specific commensal bacteria species have been revealed to be capable of regulating IFN-I and ISG expression, not to mention the underlying mechanisms. Thus, it is important to understand the cross talk between microbiota and host anti-enteric virus innate immune responses and characterize the specific bacterial species that possess protective functions. Our study demonstrates how fundamental innate immune mediators such as mononuclear phagocytes and type I IFN are regulated by commensal bacteria to antagonize enteric virus systemic infection. In particular, we have identified a novel commensal bacterium, Blautia coccoides, that can restrict enteric virus replication and neuropathogenesis by activating IFN-I and ISG responses in mononuclear phagocytes via an IFNAR-and STAT1-mediated signaling pathway.
KEYWORDS microbiota, B. coccoides, macrophage, type I interferon, enteric virus M ammalian intestines are colonized by trillions of microorganisms, composed of bacteria, viruses, archaea, and fungi, collectively referred to as intestinal microbiota (1, 2). The widely diverse intestinal microbial communities have established an immensely complicated ecosystem that is vital in maintaining host homeostasis. Cumulative evidence strongly supports the view that the dynamic cross talk between the host and its indigenous commensal bacteria is fundamental for the development, induction, education, and tuning of the host immune system (3). As such, perturbation of the microbiota composition (termed "dysbiosis") is linked to a myriad of metabolic and inflammatory diseases, both at intestinal mucosal sites and outside the gastrointestinal tract (2,4,5).
Over the past decade, rapid and extensive advances have shed light on the nature and extreme importance of intestinal microbes in regulating virus replication, transmission, and pathogenesis. Multiple studies utilizing enteric viruses, including poliovirus, retrovirus, and noroviruses, have demonstrated that the intestinal bacteria can promote viral infection (6)(7)(8). The stimulating mechanisms include direct facilitation of viral binding to target cells, virion stabilization, and indirect regulatory pathways of suppressing the mucosal immune responses (7,9,10). However, as every coin has two sides, the signals from the indigenous microbiota are shown to be essential to protect Drosophila from peroral enteric virus infection by priming antiviral innate immunity (11).
Other than exhibiting a regulatory effect at mucosal sites during enteric viral infection, the gut microbiota has been shown to be essential in limiting viral systemic infection by calibrating innate immune responses mediated by mononuclear phagocytes (12)(13)(14). A role for the commensal bacteria in regulating systemic type I interferon (IFN-I) has been described in these studies, in which impaired production of IFN-I and interferon-stimulated genes (ISGs) was observed in antibiotic (Abx)-treated or germfree (GF) mice infected with influenza A virus, murine cytomegalovirus, or Sendai virus, showing increased mortality and susceptibility (13,15,16). Recently, two compelling back-to-back research articles reported that plasmacytoid dendritic cells (pDCs) are the main cellular sources of microbiota-induced IFN-I. Schaupp and colleagues demonstrated that microbiota-driven IFN-I expression by pDCs primes conventional dendric cells (cDCs) to initiate immune responses following pathogen encounter (17). Meanwhile, Diamond's group pointed out that IFN-I produced by microbiota-enabled pDCs prevents alphaviruses, Chikungunya virus (CHIKV) in particular, from infecting and disseminating in host blood monocytes (18). The latter study demonstrated that a single commensal bacterium, Clostridium scindens, can modulate prompt IFN-I responses through Toll-like receptor 7 (TLR7) and MyD88 signaling in pDCs via primary to secondary bile acid (BA) transformation (18). Although these studies provided important insights into the systemic effect of microbiota to limit viral infection, identification of unknown specific microbiome species that influence type I IFN antiviral responses is still lacking, and the molecular links between the gut microbiota and type I IFN-mediated innate immunity are only beginning to be elucidated.
Here, we describe the impact of the intestinal microbiome on host IFN-I-associated antiviral innate immunity using encephalomyocarditis virus (EMCV), a well-studied picornavirus that targets the central nervous system and is transmitted via the fecaloral route. Depletion of the gut microbiota with an Abx cocktail via oral gavage exacerbated the mortality and neurological symptoms of EMCV infection in mice. EMCV infection of Abx-treated mice resulted in increased viral loads in blood and brain tissues. Coincidental with these results, the Abx-treated mice exhibited diminished innate immune responses reflected by impacted mononuclear phagocyte activation after EMCV infection, and this was correlated with diminished expression of ISGs in the peripheral and brain tissues. We performed bacterial 16S rRNA sequencing of the fecal samples collected from the Abx cocktail-treated and single Abx-treated mice to identify the possible commensal bacteria that may play important roles in priming IFN-I expression. Monocolonization of Abx-treated mice with B. coccoides enabled them to regain the capability of restricting EMCV systemic infection by activating mononuclear phagocytes through type I interferon expression. Thus, in the present study, we have identified a commensal bacterium that, as far as we know, has not been previously documented to be able to regulate type I IFN-mediated innate immunity in the context of enteric systemic virus infection.

RESULTS
Microbiota depletion alters the neuropathogenesis and mortality of EMCV infection. Groups of wild-type C57BL/6J mice (referred to here as WT B6 mice) were subjected to oral gavage with a broad-spectrum Abx cocktail, composed of vancomycin (Van), neomycin (Neo), ampicillin (Amp), and metronidazole (Metro), consecutively for 5 days and infected orally with different doses of EMCV; the same Abx were added to the animals' drinking water until the end of the experiments postinoculation (6,19,20). The Abx-treated mice and phosphate-buffered saline (PBS)-treated control animals were subsequently inoculated with different doses of EMCV in a 25-ml inoculum orally to mimic the route of natural infection. The clinical symptoms and mortality rates of different groups were observed and documented for 14 days postinfection (dpi). As shown in Fig. 1A, 60% of Abx-treated mice inoculated with high dose of 2 Â 10 7 50% tissue culture infective doses (TCID 50 s) succumbed to infection by 9 dpi, only about 10% of Abx-treated mice succumbed at the dose of 2 Â 10 5 TCID 50 s, and no mortality occurred at the low dose of 2 Â 10 3 TCID 50 s. However, all PBS-treated control mice survived without clinical symptoms during the entire observation period. In addition to lethality, the Abx-treated mice infected with 2 Â 10 7 TCID 50 s of EMCV developed various clinical symptoms, including hunched posture, trembling, hind limb paralysis, dyspnea, and death at 5 dpi, while the PBS-treated mice exhibited no obvious clinical signs (Fig. 1A).
We then harvested the blood and brain tissues of the infected mice at 3 and 5 dpi for virus load detection using real-time reverse transcription-PCR (RT-PCR). It was shown that peroral EMCV inoculation resulted in significantly higher viral burdens in the blood and brain tissues of Abx-treated mice than the PBS-treated mice, indicating that viral replication in the circulatory system and the target tissues was controlled by the presence of microbiota during acute EMCV infection via oral delivery (Fig. 1B). Consistent with the clinical symptoms and viral burden, immunofluorescent assays revealed that there were more viral signals (minor capsid protein VP2; Fig. 1C, left, magenta) in the brain sections of Abx-treated mice than the PBS-treated mice at 5 dpi (Fig. 1C, right). It seems that alteration of microbiota composition does not alter the cellular tropism of EMCV in the brain, because the majority of viral signals were still costained with the marker of astrocytes (glial fibrillary acidic protein [GFAP]; Fig. 1C, left, red) but not with the marker of microglia (IBA-1; green) in either PBS-or Abxtreated mice. Collectively, these data indicate that the intestinal microbes are essential for limiting EMCV replication in the target cells in the brain and are required to protect against neurological disease and cerebral tissue lesions from EMCV peroral infection.
To further confirm that the intestinal microbiome had an extraintestinal effect on limiting EMCV systemic infection, we next infected both PBS-and Abx-treated mice systemically with 200 TCID 50 s of EMCV in 100 ml inoculum via intraperitoneal injection to analyze the viral neurovirulence through this infectious route. Indeed, Abx-treated mice were more susceptible to intraperitoneal infection than PBS-treated animals, and reintroduction of fecal bacteria into Abx-treated animals alleviated EMCV mortality and disease (Fig. 1D). As the onset of viral dissemination and in vivo replication is faster in the context of intraperitoneal inoculation than peroral infection, we measured the viral loads in the blood and brain tissues earlier at 1 and 3 dpi and discovered that the Abxtreated mice supported more significant EMCV replication than PBS-treated mice, whereas viral titers of Abx-treated mice that had undergone fecal microbiota transplantation (FMT) were similar to those of PBS-treated mice (Fig. 1E). It was demonstrated that EMCV viremia and replication in target tissues were greatly increased in the absence of microbiota and that FMT could restore the capability for systemic viral clearance. Taken together, these data indicate that commensal bacteria are critical in Microbiota deficiency severely diminishes innate mononuclear phagocytes' antiviral immune responses to systemic EMCV infection. The impaired viral clearance in the blood and brains of Abx-treated animals after EMCV infection led to the hypothesis that the early systemic innate immune response was crippled by microbiota depletion. Thus, we first evaluated the recruitment and activation of early responding innate immune cells after EMCV infection by infecting PBS-or Abx-treated mice intraperitoneally with 200 TCID 50 s of EMCV and extracting the peritoneal blood mononuclear cells (PBMCs) and splenocytes at 3 dpi. The cell extracts were stained with different antibodies specific for various cell types and quantified by flow cytometry. Compared with the respective mock-infected control mice, infected PBS-treated mice exhibited a significant increase in the percentages of inflammatory monocytes, NK cells, and mononuclear phagocytes, whereas there was a comparable frequency of pDCs and cDCs. In contrast with these data, it seems that EMCV infection did not activate any of these cell subsets in infected versus mock-infected Abx-treated mice ( Fig. 2A). Also, analysis of the T cells and B cells suggests that depletion of commensal bacteria did not alter the frequencies of T/B lymphocytes in the context of EMCV infection (Fig. 2B).
Because of its crucial role as a cytotoxic effector molecule produced by NK cells, we next tested whether the production of granzyme B (GrB) was hindered in the microbiota-deficient mice. Respective gating strategies are presented in Fig. S1A in the supplemental material. As shown in Fig. S1B, NK cells from Abx-treated infected mice did not excrete upregulated GrB at 3 dpi compared to Abx-treated mock-infected mice, while EMCV infection induced remarkable excretion of GrB in PBS-treated mice. These data indicated that the cytotoxicity of NK cells was greatly impaired without commensal bacteria during viral systemic infection.
Previous studies have shown that NK cell priming is dependent on cytokines expressed by IFN-I-stimulated mononuclear phagocytes, including DCs and macrophages (13,21). Macrophages, but not DCs, had decreased expression of surface molecules that are critical during the early responses to lymphocytic choriomeningitis virus (LCMV) or influenza virus infection under the condition of Abx oral treatment (15). Consistent with these findings, we discovered that surface molecules reflecting macrophage activation, like major histocompatibility complex class I (MHC-I) and CD80, were greatly augmented in the PBMCs from infected PBS-treated mice but not in those from infected Abx-treated mice at 3 dpi compared to mock-infected animals ( Fig. 2C and Fig. S2). To further determine the requirement for mononuclear phagocytes in the systemic antiviral innate immunity and verify the necessity of the intestinal microbiome in the mononuclear phagocytes' response to viral infection, we treated mice given PBS or Abx systemically with clodronate liposomes to deplete mononuclear phagocytes and analyzed survival kinetics after intraperitoneal EMCV inoculation. Depletion of mononuclear phagocytes was at least 90% effective, as assessed by flow cytometry (Fig. S3). For the mice that received control reagent, Abx treatment still resulted in significantly higher death rates than PBS controls; however, for the mice injected with clodronate liposomes, Abx treatment did not make the animals more susceptible to EMCV infection (Fig. 2D). In support of this observation, at 1 dpi, higher titers of EMCV were detected in the blood of PBS-treated mice than Abx-treated mice both given the control reagent, and no difference was observed between PBS-treated and Abx-treated and pathological scores (n = 10 to 15) were documented. (B) Blood and brain samples of PBS-or Abx-treated mice inoculated with 2 Â 10 7 TCID 50 s of EMCV orally were harvested and tested for viral loads at 3 and 5 dpi by qPCR (n = 6). (C) IFA of brain sections from PBS-or Abx-treated mice at 5 dpi. EMCV capsid protein VP2 and cellular surface markers were double stained with the respective antibodies (representative images from 6 per group). (D) Groups of PBS-treated, Abx-treated, or Abx-treated FMT recipient mice (mice that received FMT from conventionally housed PBS-treated naive mice) were intraperitoneally inoculated with 200 TCID 50 s of EMCV for survival kinetics analysis (n = 13 to 17). (E) Viral titers of brain tissues collected from PBS-treated, Abx-treated, or FMT recipient mice at 1 and 3 dpi (same dose and route as for panel D; n = 5 or 6). Data are from two independent experiments. Broken lines indicate the limit of detection (LD). *, P , 0.05; **, P , 0.01; ***, P , 0.001; ****, P , 0.0001; ns, not significant.  (Fig. 2E). Colonized mice treated with clodronate liposomes had almost 100-fold-higher blood viral loads than control reagent-treated colonized mice, indicating that mononuclear phagocytes are absolutely important for the early control of EMCV systemic infection. Additionally, blood viral titers in mice depleted of mononuclear phagocytes and commensal bacteria were also significantly higher than those in Abx-treated mice without mononuclear phagocyte depletion, suggesting that commensal bacteria are less important than mononuclear phagocytes in conferring protection against EMCV infection (Fig. 2E).
Together, these data suggest that gut microbiota might be correlated with the activation of mononuclear phagocytes, while disrupting the bacterial composition by antibiotic administration might result in a negative effect on the mononuclear phagocytemediated protective innate immunity that limits viral systemic infection.
Commensal bacteria depletion greatly impairs the systemic and cerebral IFN-I response following EMCV infection. Because IFN-I is the key factor that contributes to the stimulation of macrophages, which subsequently prime NK cells for antigen encounter via cytokine expression, we hypothesized that the microbiota could regulate the IFN-I response in our model. To test whether the commensal bacteria altered the profile of IFN-I production at different time points postinfection, we inoculated groups of Abx-treated WT mice and PBS-treated controls with 200 TCID 50 s of EMCV intraperitoneally, collected blood and spleen at various time points postinfection, and probed for IFN-b and ISGs by real-time PCR. As early as 12 h postinfection (hpi), there was significantly upregulated expression of IFN-b and ISGs, including Isg15, Isg56, Oas1a, and Mx1, in the PBMC extract (Fig. 3A) and spleens (Fig. 3B) of PBS-treated mice compared to Abx-treated mice, suggestive of an impaired innate immune response mediated by IFN-I. Consistent with the affected systemic IFN-I response postinfection, there was reduced expression of IFN-b and associated antiviral defense genes like Isg15, Irf7, Irf9, and Stat1/2 in the brains of Abx-versus PBS-treated mice at 1 and 3 dpi, indicating that the antiviral innate immunity is impeded not only systemically but also locally at the target site of dissemination (Fig. 3C). In contrast, no differences in expression of proinflammatory cytokine tumor necrosis factor alpha (TNF-a), type II interferon, and the immune regulatory cytokines interleukin 17 (IL-17) and IL-10 were observed in the brain tissues of PBS-and Abx-treated mice at 3 dpi, implying that the impaired antiviral immunity in microbiome-deficient Abx-treated mice was not associated with the production of these cytokines (Fig. 3D).
To further confirm that the commensal-bacterium-driven type I interferon response is critical in antiviral innate immunity, we examined the requirement for the IFN-I induction and amplification pathway in early EMCV control with or without antibiotic suppression. Irf3 2/2 or Ifnar 2/2 mice, which lack key mediators in either the type I interferon upstream induction pathway or the amplification loop, respectively, were utilized to clarify the effects of microbiota on different phases of IFN-I production during EMCV systemic infection. Despite a narrower difference between PBS-and Abx-treated wildtype mice (P = 0.0359), the lethality rate discrepancy in Irf3 2/2 mice was still above 30% (P = 0.106). In contrast, all Ifnar 2/2 mice succumbed to EMCV inoculation by 5 dpi, showing identical susceptibility to EMCV intraperitoneal infection regardless of PBS or antibiotic treatment (Fig. 3E). Consistent with the survival kinetics, the dissimilarities between PBS and Abx wild-type mice in viremia (20-fold) and brain viral replication (10-fold) at 1 dpi narrowed to 6.3-fold and 3-fold in Irf3 2/2 mice, whereas the differen-

cells, or B cells from
PBMCs or spleens of mock-infected or infected PBS-or Abx-treated mice at 3 dpi (n = 6). (C) Expression of MHC-I and CD80 on PBMC mononuclear phagocytes from mock-infected or infected PBS-or Abx-treated mice at 3 dpi (top two panels, histograms of MHC-I-or CD80expressing mononuclear phagocytes; bottom two panels, frequency of mononuclear phagocytes expressing MHC-I or CD80 in total PBMC mononuclear phagocytes; n = 6). (D) Survival analysis of PBS-or Abx-treated mice treated with clodronate liposomes or control reagent following EMCV intraperitoneal inoculation (n = 10 or 11). (E) Blood viral titers of PBS-or Abx-treated mice treated with clodronate liposomes or control reagent at 1 dpi (n = 6). Data are from two independent experiments, and P values were determined by unpaired twotailed Student's t test. *, P , 0.05; **, P , 0.01; ***, P , 0.001; ****, P , 0.0001. and spleen (B) of PBS-or Abx-treated mice at 12 hpi following EMCV infection relative to respective mock controls (n = 5). (C) Expression of Ifnb and ISGs in the brains of infected PBS-or Abxtreated mice (fold change compared to respective mock-infected mice) at 1 or 3 dpi (n = 5 or 6). IFN-b levels in the brain at 3 dpi as detected by ELISA (middle). (D) Fold induction of proinflammatory or immune regulatory cytokine expression in the brain at 1 or 3 dpi relative to respective mock-infected mice (n = 6). (E) Survival kinetics of PBS-or Abx-treated Irf3 2/2 or Ifnar 2/2 mice infected with EMCV intraperitoneally; wild-type C57BL/6J mice served as controls (left, n = 10 to 12; middle, n = 9 to 14; right, n = 13 to 15). (F) Viral titers in blood and brain collected from infected PBS-or Abx-treated WT, Irf3 2/2 , or Ifnar 2/2 mice at 1 dpi (n = 5 or 6). All data are from two independent experiments. *, P , 0.05; **, P , 0.01; ***, P , 0.001; ****, P , 0.0001; ns, not significant. ces were completely lost in Ifnar 2/2 mice (Fig. 3G). These findings indicate that the Irf3mediated IFN-I induction pathway was partially affected by Abx treatment. Given the fact that Ifnar 2/2 mice are extremely susceptible to EMCV infection even at a low dose of 50 TCID 50 s with or without having an intestinal microbiome, we would assume that the IFN-I amplification pathway is absolutely essential for preventing EMCV systemic infection and the IFNAR-mediated pathway might be correlated with the integrity of the host intestinal microbiome.
B. coccoides monocolonization alleviates EMCV pathogenesis and restricts viral infection. We next investigated whether specific commensal bacteria taxa are responsible for the protein against systemic EMCV infection. First, we tested this idea by determining which antibiotic in the cocktail was required for the effect on IFN-I modulation and whether the effect could be narrowed down to a single antibiotic treatment or to a certain kind of alteration in the intestinal microbiome per se. Groups of WT B6 mice were treated with a combination of four antibiotics or with Van, Neo, Amp, or Metro singly and inoculated intraperitoneally with EMCV, and the brain tissues were dissected at 3 dpi for EMCV viral burden measurement and IFN-b expression detection. Generally, single-Abx treatment significantly increased viral replication in the brain compared to PBS-treated controls except Metro or Neo. Van treatment exhibited a partial suppressive effect in early EMCV control, as brain viral titers of these mice were not as high as those in Abx cocktail-treated animals. Amp was the most effective single antibiotic, as the viral replication was comparable to that seen with the Abx combination (Fig. 4A).
Consistent with these data, Metro-treated mice exhibited levels of IFN-b expression very similar to those of the Abx cocktail-treated controls, suggesting that Metro was not required for the IFN-b suppressive effect. Amp, Van, and Neo treatments all dampened IFN-b expression significantly compared to PBS. Among them, Amp treatment conferred an optimal suppression on IFN-b expression level, Van treatment showed an intermediate effect, while Neo treatment had the least impact on IFN-b expression (Fig. 4B). As expected, successful inhibition of IFN-b expression in brain tissues in general correlated with a substantial reduction in detectable 16S rRNA readouts upon antibiotic treatment (Fig. 4C). In comparison, Van, Neo, or Amp treatment had an effect similar to that of the antibiotic cocktail, whereas Metro exhibited the least effective bacterial depletion (Fig. 4C). Based on these data, one would hypothesize that certain bacterial communities which were present in the PBS control and resistant to Metro treatment but significantly lower in the Neo-, Van-, Amp-, and Abx-treated mice mediated the protective effect against EMCV systemic infection as an IFN-I inducer. Thus, we then carried out 16S rRNA gene sequencing and analysis on fecal samples collected from PBS-, Abx cocktail-, and single-antibiotic-treated mice at 0 and 3 dpi (Fig. S4 and Fig. 4D).
On the basis of the hypothesis mentioned above, we chose three bacterial taxa, namely, Blautia, Akkermansia, and Lactobacillus, because they had similar abundances in PBS-and Metro-treated mice at 0 and 3 dpi in general (Fig. 4E). To explore whether they are the specific bacteria that mediated the protective effect of the microbiota against viral systemic infection, we subjected Abx-treated mice to gavage with B. coccoides, Akkermansia muciniphila, Lactobacillus reuteri (an unrelated Gram-positive human symbiont), Clostridium butyricum, or mouse fecal contents (fecal microbiota transplantation [FMT]) prior to EMCV intraperitoneal inoculation. Strikingly, colonization of Abx-treated mice with B. coccoides alone fully alleviated EMCV mortality to the extent seen in mice that underwent FMT and that of PBS-treated controls (Fig. 4F), whereas gavage with C. butyricum, L. reuteri, or A. muciniphila did not alter mortality, even though these bacteria exhibited efficient colonization (Fig. S4). Similarly, we observed significantly lower viral burdens in the brain tissues of B. coccoides-colonized Abx-treated mice than in Abx-treated, FMT recipient, or C. butyricum-colonized Abxtreated mice at 3 dpi (Fig. 4G). Taken together, these data suggest that the specific commensal B. coccoides protects hosts from systemic EMCV infection by restricting viral replication.

B. coccoides monocolonization promotes macrophage activation and IFN-I responses to systemic EMCV infection.
To further elucidate whether B. coccoides restricted EMCV infection by promoting macrophage activation, we then subjected Abxtreated mice to gavage with B. coccoides, C. butyricum, or fecal contents to evaluate GrB-secreting NK cells and mononuclear phagocytes with activation of the surface markers MHC-I and CD80 in PBMC and spleen. Indeed, B. coccoides colonization in Abxtreated mice restored frequency of GrB-secreting NK cells in PBMCs following EMCV infection to the level of PBS-treated controls at 3 dpi, whereas C. butyricum oral gavage did not improve the impairment of GrB secretion in NK cells exhibited in Abx-treated controls during infection (Fig. S5).
B. coccoides colonization in Abx-treated mice remarkably increased the frequency of MHC-I mononuclear phagocytes by 45% in PBMC (7.2% to 52%) and by 13% in spleen (from 13.5% to 26%) monocytes/macrophages following EMCV infection at 3 dpi, comparable to FMT recipient mice and PBS-treated controls (Fig. 5A). In a similar way, levels of CD80-upregulated mononuclear phagocytes were elevated from 10% to 26.4% in PBMCs and from 7.36% to 53.7% in spleen mononuclear phagocytes after B. coccoides colonization in the context of EMCV infection (Fig. 5B). B. coccoides colonization recovered MHC-I expression in both PBMC and spleen mononuclear phagocytes almost to the same extent as in FMT recipient mice and PBS-treated controls, whereas it only partially rescued CD80 expression, to a lesser extent than FMT, especially in PBMC mononuclear phagocytes (Fig. 4B) (16.4% versus 37.5%), indicating that the expression of this surface molecule might not be regulated only by B. coccoides but might also be dependent on other intestinal bacteria. Likewise, C. butyricum-colonized mice retained a low level of either MHC-I-or CD80-upregulated PBMC and spleen mononuclear phagocytes compared to Abx-treated mice following viral infection at 3 dpi. To determine whether B. coccoides mediates protection from systemic EMCV infection through a mononuclear phagocyte-dependent mechanism, we then colonized mononuclear phagocyte-depleted Abx-treated mice with B. coccoides before intraperitoneal EMCV inoculation and observed the survival curves for 14 days. Clodronate treatment abolished B. coccoides-mediated protection, demonstrating that B. coccoides is required for conferring mononuclear phagocyte-mediated protection against EMCV infection (Fig. 5C).
To determine whether colonizing Abx-treated mice with B. coccoides restores systemic IFN-I responses to viral infection, we harvested PBMCs from Abx-treated mice with B. coccoides, C. butyricum, or fecal content colonization at 3 dpi and measured IFN-b expression by qPCR. Apparently, B. coccoides colonization in Abx-treated mice restored IFN-b expression in PBMCs to the level detected in FMT recipient and PBStreated controls upon EMCV infection at 3 dpi, whereas C. butyricum colonization did not (Fig. 5D). Since we have demonstrated that IFN-I amplification pathway is absolutely essential for antagonizing viral systemic infection, we next investigated whether B. coccoides mediates protection from EMCV infection in an IFN-I-dependent manner. We colonized Ifnar 2/2 mice with B. coccoides prior to virus inoculation and discovered that these mice did not display less susceptibility to EMCV infection than Abx-treated Ifnar 2/2 mice (Fig. 5E). Furthermore, B. coccoides colonization of Abx-treated Ifnar 2/2 mice did not reduce viremia or brain replication in comparison to Abx-treated Ifnar 2/2 controls at 1 dpi (Fig. 5F). As a whole, these data demonstrate that the protective effect of B. coccoides against viral systemic infection requires mononuclear phagocytes and IFN-I amplification signaling.
B. coccoides colonization promotes type I IFN responses in mononuclear phagocytes to limit EMCV infection. To directly clarify whether B. coccoides can pro- Expression level of Ifnb in PBMCs isolated from infected PBS-treated mice, Abx-treated mice, or Abx-treated mice subjected to gavage with FMT, B. coccoides, or C. butyricum, compared to their respective mock-infected controls (n = 5). (E) Survival kinetics of infected Abx-treated and B. coccoides-colonized Abx-treated Ifnar 2/2 mice (n =6 to 9). (F) Viral titers in blood and brain collected from infected PBS-treated, Abx-treated, or B. coccoides-colonized WT or Ifnar 2/2 mice at 1 dpi (n =5 or 6). All data are from two independent experiments, and P values were determined by unpaired two-tailed Student's t test. *, P , 0.05; **, P , 0.01; ***, P , 0.001; ****, P , 0.0001; ns, not significant. mote systemic type I IFN responses through monocyte/macrophage activation, we infected bone marrow-derived macrophages (BMDMs) isolated from PBS-treated, Abxtreated, B. coccoides-colonized, or C. butyricum-colonized mice with EMCV at a multiplicity of infection (MOI) of 5 in vitro and detected induction of Ifnb and associated ISGs at 8 hpi using quantitative RT-PCR (qRT-PCR). Expression of Ifnb and ISGs, including Oas1a, Isg15, and Mx1, was drastically reduced in BMDMs isolated from Abx-treated mice compared to PBS mice, suggestive of an intrinsic inability to respond to viral infection, but this responsiveness was partially restored by B. coccoides, but not C. butyricum, colonization (Fig. 6A). Furthermore, B. coccoides colonization in Abx-treated mice endowed their BMDMs with the ability to induce ISG expression in an Ifnar-dependent manner, because ISG expression was undetectable (or extremely low) in BMDMs isolated from B. coccoides-colonized Ifnar 2/2 mice following EMCV infection at 8 hpi (Fig. 6B).
Because STAT1 conveys signals downstream of IFN receptor engagement and its phosphorylation and nuclear translocation mediate transcription of ISGs in the IFN-I amplification loop, we stimulated mononuclear phagocytes isolated from spleens of PBS-treated, Abx-treated, or B. coccoides-colonized Abx-treated mice with IFN-g in vitro and tested STAT1 phosphorylation using flow cytometry. IFN-g-stimulated mononuclear phagocytes isolated from Abx-treated mice exhibited significantly low levels of pSTAT1 compared to those from PBS-treated mice, while B. coccoides colonization in Abx-treated mice restored STAT1 phosphorylation in splenic mononuclear phagocytes to levels comparable to those in PBS controls ( Fig. 6C and Fig. S7). We also examined expression of multiple antiviral defense genes in mononuclear phagocytes isolated from the spleens of infected PBS-treated, Abx-treated, and B. coccoides-colonized animals to confirm whether the same phenotype exists in vivo after viral infection. At 1 dpi, splenic mononuclear phagocytes from Abx-treated mice exhibited significantly decreased antiviral gene expression compared to those from PBS-treated mice, but B. coccoides colonization in Abx-treated mice induced significantly higher levels of expression of Ifnb and ISGs, although they were not fully restored to the level of PBS controls (Fig. 6D). This is consistent with the data showing that B. coccoides colonization partially restored the frequency of CD80-expressing mononuclear phagocytes (Fig. 5C) in PBMCs, implying that other commensal bacteria species also possess the ability to induce type I IFN and ISG responses in circulating mononuclear phagocytes.
Again, B. coccoides exerts its effects through IFNAR-mediated IFN-I signaling, because no detectable or obviously upregulated antiviral gene expression was observed in mononuclear phagocytes isolated from spleens of B. coccoides-colonized Abx-treated Ifnar 2/2 mice (Fig. 6E). Overall, these findings suggest that B. coccoides is sufficient for maintaining optimal mononuclear phagocyte responsiveness to stimuli (cytokine and viral infection), and it seems very likely that B. coccoides augments type I IFN signaling in mononuclear phagocytes via IFNAR and STAT1 phosphorylation to protect the host from virus systemic infection.

DISCUSSION
In the present study, we describe a crucial role for the commensal microbiota in modulating antiviral innate immunity to systemic enteric virus infection. In particular, the intestinal microbiome protects against systemic viral infection by promoting type I IFN responses in peripheral innate immune cells, mainly mononuclear phagocytes, via an Ifnar-correlated signaling pathway. Furthermore, we have identified the mechanism by which an understudied commensal bacterium, B. coccoides, restricts enteric virus systemic infection by enhancing type I IFN responses in mononuclear phagocytes.
Over the past 10 years, it has been gradually acknowledged that the intestinal microbiome can shape host antiviral immunity. Accumulating literature has shown that animals given cocktails of broad-spectrum antibiotics and GF mice have impaired innate and adaptive immune responses following infection with various viral pathogens, including LCMV, influenza virus, murine cytomegalovirus (MCMV), flaviviruses,  (13,15,16,18,22). However, until now, all studies on mammalian enteric viruses have reported that the intestinal microbiome promotes virus infection (6,7,10,20,23), with two exceptions. One study, conducted by Grau et al., reported that C. scindens primes type III interferon induction to suppress murine norovirus (MNV) infection in proximal small intestines by bile acid biotransformation (19). Another example of commensal bacterium-mediated anti-enteric virus protection was provided by Shi et al., who demonstrated that segmented filamentous bacteria (SFB) prevent and cure rotavirus infection by accelerating epithelial cell turnover (24). Unlike norovirus and rotavirus infection, which cause only self-limited diseases, some enteric viruses, enterovirus 71 and EMCV for instance, can result in systemic infection and lead to a diverse array of neurological diseases (25,26).
To determine whether the intestinal microbiota could be protective indirectly by regulating host antiviral innate immunity during enteric virus infection, we infected wild-type mice with EMCV orally, to mimic natural infection, or intraperitoneally, to immediately establish systemic infection. To our surprise, microbiota depletion with antibiotics exacerbated the neurological disease and enhanced viral replication upon EMCV infection for both routes of inoculation (Fig. 1). This outcome resembles the results of previous studies on the nonenteric viral pathogens mentioned above but not the milestone study conducted by Kuss et al. using poliovirus and reovirus, in which intestinal microbiomes and LPS were reported to promote viral replication and systemic pathogenesis (6). These discrepancies suggest that microbiota acts very differently in interacting with even two closely related enteric viruses (EMCV and poliovirus are both picornaviruses and have very similar viral structures). It is highly likely that the innate immune responses were also severely diminished by antibiotic treatment in their model (we used the same antibiotic regimens), whereas poliovirus replication in vivo was barely established to reflect the defects on host innate immunity in the absence of microbiota. Conversely, EMCV in vivo replication and dissemination do not require the microbiota in our model ( Fig. 1B and C), so that the neuropathogenesis was diminished in the context of antiviral innate immunity being dampened by microbiota depletion.
Collectively, multiple studies have demonstrated that commensal bacteria are responsible for shaping host antiviral immunity beyond mucosal sites. GF mice and Abxtreated mice exhibited blunted type I interferon responses, which are required for optimizing functions of nonmucosal immune cells, including NK cells, macrophages, and CD8 1 T cells (13,14,16,22). In our study, we showed that mononuclear phagocytes isolated from Abx-treated mice did not display a higher frequency or upregulated expression of activating surface markers like MHC-I and CD80 at 3 dpi ( Fig. 2A and D). Additionally, mononuclear phagocyte depletion equalized the discrepancies in mortality and viral replication between Abx-treated mice and PBS controls, indicating that mononuclear phagocytes are essential for microbiota-mediated protection from EMCV systemic infection (Fig. 2E and F). Our data are consistent with previous studies demonstrating that the microbiome can regulate antiviral macrophage responses by inducing the expression of type I IFN and ISG genes after influenza virus infection (15,16). Similar to these studies, we also showed that microbiota depletion results in severely diminished type I IFN and ISG responses in both peripheral tissues and brain (Fig. 3A to D), which causes unrestricted virus replication and exacerbates neuropathogenesis and mortality. The impaired IFN-I production and ISG expression in the brain tissues of Abx-treated mice suggests major defects in responsiveness of microglia to viral infection, because they are the main sources of IFN production in the brain. The results are consistent with one paper reporting that microglia of GF mice were unable to produce various cytokines and chemokines upon stimulation (27). Our model revealed that the microbiota-driven IFN and ISG responses are partially dependent on Irf3-mediated IFN-I induction and also probably rely on the Ifnar-mediated signaling pathway (Fig. 3E  and F).
Although accumulated evidence shows that the microbiota can promote ISG responses in multiple cell types, the cellular sources of type I IFN were less well defined until very recently, when two papers demonstrated that the microbiota constitutively induces type I IFN production and a basal level of ISG expression in pDCs at systemic sites (17,18). In our study, we examined only the frequency of pDCs in PBMC or spleen following EMCV infection at 3 dpi, and we observed no differences between PBS-and Abx-treated mice ( Fig. 2A). Similarly, Abt et al. reported that at 3 dpi following influenza virus infection, there was a comparable influx of pDCs into bronchiole alveolar lavage and the pDCs exhibited similar activation profiles in PBS-and Abx-treated mice (15). These seemingly discordant observations could be explained by the timing of detection, because pDCs respond to virus nucleic acids with massive and rapid secretion of IFN-I (1 to 3 h poststimulation) independently of the IFNAR-based feedback signaling, which is always required for most cell types in IFN-I production (28). Also, profiling the IFN-I and ISG expression in pDCs could be very subtle, especially at steady state (0 dpi), and using different microbiota-deficient mouse models (GF versus Abx-treated mice) may result in distinct phenotypic differences (17,18). Generally, it is plausible that signals of commensal bacteria drive a very swift first wave of IFN-I (especially IFN-b) production in poised pDCs to activate NK cells, macrophages, and overall innate immune responses to viral infection.
Until now, only a few publications had begun to identify specific commensal bacterial species that play an important role in inducing type I IFN-mediated innate antiviral immunity (16,18,29,30). Although the study by Abt and colleagues revealed an interplay between commensal bacteria and poised low-level tonic antiviral interferon signaling, they did not identify the specific bacterial species that regulates the "steadystate" readiness of antiviral pathways in macrophages (15). In order to identify the specific commensal bacteria that are responsible for the protein against EMCV systemic infection, we performed 16S rRNA gene sequencing and analysis on fecal samples collected from PBS-, Abx cocktail-, and single-antibiotic-treated mice. Interestingly, we noted that EMCV infection has a dramatic effect on the prevalence of certain microbes, including Blautia, Clostridioides, Lachnoclostridium, and others, even in the PBS control ( Fig. 4D and Fig. S4). This result is distinct from the data reported by Nelson et al., who showed that the infection with the enteric virus MNV does not cause major disruptions in the murine intestinal microbiota (31). We speculated that this might be attributed to the peripheral IFN-I production induced by EMCV systemic infection, because Deriu et al. reported that influenza virus-induced type I IFNs produced in the lungs promoted the depletion of obligate anaerobic bacteria and the enrichment of Proteobacteria in the gut, leading to a secondary Salmonella infection (32).
Our work builds on the limited existing literature by identifying a less well documented commensal bacterium, B. coccoides, that plays a crucial role in protecting against enteric virus systemic infection through IFN-I induction. Abx-treated mice subjected to gavage with 10 8 CFU of B. coccoides reached a relative bacterial abundance similar to that in the PBS-treated control, suggesting that this dose results in an effective colonization close to a physiological level (Fig. S6). Above that, colonization with 10 10 CFU of B. coccoides appeared to reach a level almost 40-fold higher than that in the PBS control (Fig. S6), and colonialization almost fully restored the upregulation of MHC-I and CD80 on the surfaces of PBMC and spleen mononuclear phagocytes in Abxtreated animals following EMCV infection ( Fig. 5B and C). Single colonization of B. coccoides in Abx-treated mice can also rescue the intrinsic defects in expression of IFN-I and key ISGs in BMDMs in an Ifnar-dependent manner following EMCV infection in vitro (Fig. 6A and B) or restore the STAT1 phosphorylation in splenic mononuclear phagocytes following IFN-g stimulation in vitro (Fig. 6C). The present study adds to the previous work by demonstrating that colonization of Abx-treated mice with a single bacterium, B. coccoides, can restore the ability of mononuclear phagocytes to respond to type II IFN or virus stimulation.
Since B. coccoides monocolonization in Abx-treated mice restored protection of EMCV systemic infection by restricting in vivo viral replication in a type I interferon-dependent manner (Fig. 4F and G and 5D and F), a next question will be, what does B. coccoides utilize to induce IFN-I responses in mononuclear phagocytes to confer antiviral systemic protection? Two recent reports suggest that C. scindens-derived metabolite deoxycholic acid or glycolipids of Bacteroides fragilis can impact host antiviral innate immunity by inducing type I IFN production and ISG expression in pDCs or a subset of colonic cDCs (18,30). Also, Gutierrez-Merino et al. have reported that beneficial commensal lactic acid bacteria can trigger type I interferon production via direct recognition by the intracellular sensors of STING and MAVS (33).
Short-chain fatty acids (SCFA), primarily produced by the Gram-positive bacteria Firmicutes and Bacteroidetes through fermentation of undigested polysaccharides, are not only important local energy sources for gut microbiota and intestinal epithelial cells but also crucial regulators for shaping immune systems at extraintestinal sites (5). As the most abundant SCFA in the colon, acetogenic bacterial communities such as Blautia produce nearly one-third of the acetate (34). It was shown that acetate derived from high-fiber diets protects against respiratory syncytial virus infection by promoting type I IFN production and ISG expression in pulmonary epithelial cells through GPR43 (35). Future studies will be carried out to examine the effects of specific SCFAs on host innate immunity to enteric viruses using a microbiota-deficient mouse model.
B. coccoides was purchased from ATCC (ATCC 29236) and cultured in modified chopped-meat medium (ATCC medium 1490; ELITE-MEDIA) at 37°C under anaerobic conditions. C. butyricum was purchased from ATCC (ATCC 19398) and cultured in thioglycolate medium (Hopebio) at 37°C under anaerobic conditions. A. muciniphila was kindly provided by Lan-juan Li (The First Affiliated Hospital, College of Medicine, Zhejiang University, Hangzhou, People's Republic of China) and cultured in brain heart infusion medium (BHI; Oxoid) at 37°C under anaerobic conditions. L. reuteri was purchased from China Center for Type Culture Collection (CCTCC; AB 2014289) and cultured in DeMan-Rogosa-Sharpe medium (Oxoid) at 37°C under anaerobic conditions. The concentration of each bacterial species was quantified based on the optical density at 600 nm (OD 600 ).
Recombinant virus derived from full-length clone of EMCV strain BJC3 (GenBank accession no. DQ464062) was used in this study (36), and viral titers of EMCV stocks were determined by a standard TCID 50 assay described previously (37).
Mice and infections. Mice were maintained in a specific-pathogen-free (SPF) facility with a temperature-and humidity-controlled environment (22 6 2°C, 50% 6 10% humidity), and all animal experiments were strictly carried out in accordance with protocols approved (no. 117113) by the Institutional Animal Care and Use Committee of Zhejiang University. Six-to 8-week-old, sex-matched mice were used for all experiments. C57BL/6J wild-type mice were purchased from the Model Animal Research Center of Nanjing University (Nanjing, China). Type 1 interferon receptor knockout mice (referred to as Ifnar 2/2 ) were a kind gift from Yu Chen (Wuhan University, Hubei, People's Republic of China), and interferon regulatory factor 3-deficient mice (referred to as Irf3 2/2 ) mice were provided by Jin Jin (Zhejiang University, People's Republic of China).
For viral titer determination, tissue samples taken at different time points were harvested, weighed, and homogenized with stainless steel beads in 1 ml of DMEM supplemented with 2% fetal bovine serum information. The richness of certain commensal bacteria taxa was calculated using the usearch -alpha_div (version 10; http://www.drive5.com/usearch/) according to the OTU abundance. Based on the relative abundance of species at each classification level in otu_table, R software was used to draw the histogram, heat map, and ternary-phase diagram.
Statistical analysis. Statistical analyses were performed with Prism GraphPad software v 8.0. Error bars represent standard errors of the means in all figures and P values were determined by unpaired, two-tailed Student's t test. A log-rank test was used for survival curves.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.