Composition of the Intestinal Microbiota Determines the Outcome of Virus-Triggered Colitis in Mice

The intestinal microbiota is a complex ecosystem implicated in host health and disease. Inflammatory bowel disease (IBD) is a multifactorial chronic disorder of the gastrointestinal mucosa. Even though the exact mechanisms are still unknown, the intestinal microbiota is crucial in IBD development. We previously showed that murine norovirus (MNV) induces colitis in the Il10-deficient (Il10−/−) mouse model of IBD in a microbiota-dependent manner. Thus, in this study we analyzed whether distinct minimal bacterial consortia influence the outcome of MNV-triggered colitis in Il10−/− mice. Gnotobiotic Il10−/− mice associated with Oligo-Mouse-Microbiota 12 (OMM12) or Altered Schaedler Flora (ASF) developed little to no inflammatory lesions in the colon and cecum. MNV infection exacerbated colitis severity only in ASF-colonized mice, but not in those associated with OMM12. Four weeks after MNV infection, inflammatory lesions in ASF-colonized Il10−/− mice were characterized by epithelial hyperplasia, infiltration of inflammatory cells, and increased barrier permeability. Co-colonization of ASF-colonized Il10−/− mice with segmented filamentous bacteria (SFB) abolished MNV-induced colitis, whereas histopathological scores in SFB-OMM12-co-colonized mice stayed unchanged. Moreover, SFB only colonized mice associated with ASF. The SFB-mediated protective effects in ASF-colonized mice involved enhanced activation of intestinal barrier defense mechanisms and mucosal immune responses in the chronic and acute phase of MNV infection. SFB colonization strengthened intestinal barrier function by increasing expression of tight junction proteins, antimicrobial peptides and mucus. Furthermore, SFB colonization enhanced the expression of pro-inflammatory cytokines such as Tnfα, Il1β, and Il12a, as well as the expression of the regulatory cytokine Tgfβ. Altogether, our results showed that MNV-triggered colitis depends on the microbial context.


INTRODUCTION
The intestinal microbiota is a highly complex ecosystem dominated by four bacterial phyla, namely, Bacteroidetes, Firmicutes, Actinobacteria, and Proteobacteria (1,2). Its pronounced influence on host health and disease was corroborated over the last decades by a multitude of publications. The gut microbiota is involved in many physiological processes such as synthesis of vitamins, production of short chain fatty acids (SCFA), bioconversion of complex molecules, degradation of xenobiotic substances, and also in the development and maturation of the mucosal immune system (3,4). On the other hand, the intestinal microbiota is also implicated in the development of many human disorders such as inflammatory, autoimmune and metabolic diseases, as well as tumorigenesis (5)(6)(7)(8). However, it is still unclear whether identified changes in microbiota composition and function are a cause or a consequence of disease, since most of studies are based on associations and lack fundamental proof of causality. In addition, the composition of the intestinal microbiota is complex and non-defined and thus causal microbial effects cannot be appropriately addressed (9)(10)(11). Thus, this emphasizes the need to perform mechanistic studies to understand causal interrelations between the microbiome and the host in health and disease. Gnotobiotic animal models represent a powerful tool for investigating functional effects of host-microbe and microbemicrobe interactions (12). These models include germ-free (GF) animals that are devoid of all other living organisms and animals that are colonized with known microorganisms. The possibility to colonize GF animals with defined microorganisms allows us to analyze complex host-microbiota interactions mechanistically in a simplified way. Utilizing minimal bacterial consortia in animal models reduces the microbiome complexity on a manageable level and supports studies that can evaluate the impact of particular microorganisms on the host physiology. Furthermore, these approaches can contribute to the development of novel therapeutic or prophylactic strategies that would allow noninvasive modulation of the intestinal microbiota.
Inflammatory bowel disease (IBD) is a multifactorial chronic relapsing inflammatory disorder of the gastrointestinal tract and has two main forms-Crohn's disease and ulcerative colitis. The intestinal microbiota was shown to be crucial for the development of IBD, while genetic susceptibility and environmental factors can also play an important role (13,14). The exact mechanisms and involvement of commensal microbiota in IBD development are still not fully understood. Hence, we generated a gnotobiotic model of experimental IBD to dissect the underlying mechanisms of the cross-talk between particular microbes and the host response during IBD development. Our experimental IBD model consists of three defined factors: a susceptible host (GF Il10-deficient mice), defined bacterial microbiomes [Altered Schaedler Flora (ASF) or Oligo-Mouse-Microbiota 12 (OMM 12 )], and murine norovirus (MNV) infection as a specific trigger. Il10-deficient (Il10 −/− ) mice lack regulatory cytokine interleukin 10 (IL10) and spontaneously develop intestinal inflammation (15). The genetic background of Il10 −/− mice is an important factor that determines colitis susceptibility. For example, Il10 −/− mice on a C3H/HeJBir or 129/SvEv background show higher colitis susceptibility than Il10 −/− mice on C57BL/6J background (16,17). However, the intestinal microbiota was shown to be essential for disease development, as GF Il10 −/− mice do not develop colitis (18,19). The composition of the intestinal microbiota can alter colitis severity of Il10 −/− mice (20,21). Previously, we have shown that MNV infection triggers colitis in Il10 −/− mice in a microbiota-dependent manner (22).
In this study, GF Il10 −/− mice were colonized with two different defined minimal bacterial consortia, ASF and OMM 12 . ASF is a well-established model community consisting of eight bacterial species. These bacterial species were isolated from the mouse gut and belong to three bacterial phyla: Bacteroidetes, Firmicutes, and Deferribacteres ( Table 1). The ASF consortium can be stably maintained for generations under gnotobiotic conditions (23,24). This minimal microbiota is widely used in biomedical research including studies addressing perturbations of the microbiota composition upon infection and evaluating the impact of particular microorganisms on the host immune system (24). Moreover, this consortium is devoid of pathobionts and does not cause overt mucosal inflammation (25)(26)(27). The second minimal microbiota used in this study is a defined bacterial consortium OMM 12 (28). This model community contains 12 mouse enteric microbiota-derived bacterial species that represent five major bacterial phyla including Bacteroidetes, Firmicutes, Verrucomicrobia, Proteobacteria, and Actinobacteria (Table 1) (28,29). This defined bacterial community was designed to analyze mechanisms of colonization resistance to enteric infections (28). Additionally, all OMM 12 bacterial strains are included in the publicly accessible catalog, the mouse intestinal bacterial collection (miBC), which makes them attractive for application in mechanistic microbiome studies (30). In this study defined bacterial microbiomes were additionally modulated by segmented filamentous bacteria (SFB). SFB are gram-positive, spore-forming commensal bacteria of the Clostridiaceae family that are found in the gastrointestinal tract of several different species. These bacteria grow attached to epithelial cells and potently stimulate the host's mucosal immune system, especially IL17-mediated immune responses (31,32). Due to their strong immunomodulatory functions, SFB can cause beneficial or adverse effects on the host physiology (32).
Overall, the aim of this study was to assess whether minimal bacterial consortia influence the outcome of MNV-triggered colitis in the Il10 −/− mouse model of IBD and how this phenotype is modulated by specific commensals.

Mice
Germ-free (GF) male and female C57BL/6J.129P2-Il10 tm1Cgn / JZtm (B6-Il10 −/− ), C3H/HeJBir.129P2-Il10 tm1Cgn /JZtm (C3H-Il10 −/− ), C57BL/6;129Sv-Rag2 tm1Fwa /Ztm (B6-Rag2 −/− ), and gnotobiotic C57BL/6JZtm OMM12 , C.B-Igh1 b /IcrTac ASF (C.B-17 ASF ) and NOD/LtSz.CB17-Prkdc scid /JZtm SFB (NOD-scid SFB ) mice were obtained from the Central Animal Facility (Hannover Medical School, Hanover, Germany). Breeding of gnotobiotic animals was performed in plastic film isolators (Metall+Plastik GmbH, Radolfzell-Stahringen, Germany) located in a room with a controlled environment and 12 h light/dark cycles. For experiments, mice were maintained in airtight cages with positive pressure (IsoCage P, Tecniplast Deutschland GmbH, Bavaria, Germany) to keep their gnotobiotic status. Mice received pelleted 50 kGy gamma-irradiated feed and autoclaved water ad libitum. Mice were sacrificed by CO 2 inhalation followed by exsanguination at 12 weeks of age (chronic MNV infection) . For each analysis, sectionmatched tissue was used. Each cohort consisted of at least 5 mice per group and each experiment was performed one, two or three times. Gnotobiotic animals bred at the Central Animal Facility were monitored according to recommendations for maintaining gnotobiotic colonies (33) and FELASA recommendations (34) and were proven to be free of contaminants or infection with common murine pathogens. Furthermore, all experimental groups were screened for the presence of microbial contaminants at the end of the experiment. Animals confirmed to have unexpected microbes were omitted from analyses. The presence of contaminating bacteria was controlled in DNA isolated from feces by 16S rRNA gene sequencing analysis. Furthermore, to rule out contamination with MNV, non-infected groups were randomly tested for the presence of MNV.

Colonization of Germ-Free Mice
Four week old GF B6-Il10 −/− and C3H-Il10 −/− mice were colonized with one of the two minimal consortia (ASF or OMM 12 ) via co-housing for 4 weeks with gnotobiotic donor animals (C57BL/6JZtm OMM12 and C.B-17 ASF ). Juvenile GF B6-Rag2 −/− mice were also associated with OMM 12 via co-housing over a period of 4 weeks. To generate SFB gut content aliquots for inoculation, three SFB monoassociated mice (NOD-scid SFB ) were sacrificed. Subsequently, ileum and cecum were harvested and placed in a cold sterile petri dish. The lumen was flushed with sterile Luria-Bertani media (1:2/ intestinal content:media).
Shortly after larger fecal particles settled, intestinal content was transferred into cryotubes. Per cryotube, 850 µL of intestinal content was mixed with 150 µL glycerol and stored at −80 • C until needed. From these aliquots, mice were colonized with SFB via oral gavage of 50 µL of intestinal content on 2 consecutive days.

Polymerase Chain Reaction (PCR)
DNA extraction from feces of ASF-colonized mice was performed using the PSP R Spin Stool DNA Kit (Stratec Molecular GmbH, Berlin, Germany) following the manufacturer's instructions. To detect the ASF consortium, species-specific forward (FW) and reverse (RV) primers synthesized by Eurofins (Eurofins Genomics, Ebersberg, Germany) were used to detect 16S rRNA genes (

Quantitative Real-Time PCR (qPCR)
For gene expression analyses, the proximal colon was collected. The colon tissue was flushed with sterile PBS, snap frozen in liquid nitrogen, and kept at −80 • C until further processing. Total RNA was extracted from the proximal colon tissue using the The absolute quantification using a standard curve performed with the QuantStudio 6 Flex Real-Time PCR System was used to determine MNV and SFB gene copy numbers/µL in the colon tissue or feces. For MNV quantification, the proximal colon tissue was harvested and prepared as described above including total RNA and cDNA isolation. MNV load was determined using a TaqMan assay with an MNV-specific primer/probe set ( Table 4) and the TaqMan R Fast Advanced Master Mix as recommended  step using zirconia beads. Isolated DNA was diluted to a final concentration of 25 ng/µL using HPLC grade water (J. T. Baker) and stored at −20 • C until further processing. SFB 16S rRNA gene copy numbers were determined using a TaqMan assay with an SFB-specific primer/probe set ( Table 4) and Taqman R Universal Master Mix according to the manufacturer's instruction. The thermocycling conditions for SFB detection were: (i) an incubation step of 2 min at 50 • C; (ii) a polymerase activation step of 10 min at 95 • C; and (iii) 40 cycles of 15 s at 95 • C and 1 min at 60 • C (annealing and elongation step). Both qPCR assays were performed using 900 nM of each primer and 200 nM of a specific probe. The qPCR standards for quantifying MNV and SFB abundance were generated using plasmids containing MNV or SFB amplicon products generated with specific primers listed in Table 4. The MNV and SFB amplicon products were cloned into a pSC-A-amp/kan vector using the StrataClone PCR Cloning Kit (Agilent Technologies, La Jolla, CA, USA) following manufacturer's instructions. The plasmid DNA was purified by NucleoSpin R Plasmid Kit followed by NucleoBond R Xtra Maxi Kit (Macherey-Nagel GmbH & Co. KG, Duren, Germany) according to the manufacturer's protocol. Standard curves for MNV and SFB were generated using a 10-fold dilution series of the plasmids ranging from 1 to 10 4 copies/µL for MNV and 1 to 10 7 copies/µL for SFB. The detection of 16S rRNA gene copy numbers of particular OMM 12 members was performed as described by Brugiroux et al. (28).

Histology
The small intestine, cecum, and colon were collected and fixed in neutral buffered 4% formalin. Subsequently, samples were dehydrated, embedded in paraffin, sectioned at 3 µm, and stained with hematoxylin and eosin (H&E). H&E stained cecum and colon sections were scored as described previously (37). Briefly, histopathological lesions were scored blindly for ulceration, hyperplasia, severity, and the involved area. Each parameter was graded from 0 (physiological) to 3 (severe changes) and added in a total score from 0 to 12. Colon sections were scored separately for the proximal, middle, and distal part. A total colon score was calculated by adding all three colon sections (maximal score 36).

Immunofluorescence
Immunofluorescence staining for CD3 and CD45R was performed on formalin-fixed paraffin-embedded colon tissue sections. Sections were deparaffinized using xylol and rehydrated using decreasing concentrations of ethanol (100, 95, and 70%) followed by a short wash in distilled water. Heat-induced antigen retrieval was performed in citrate-based buffer (Target Retrieval Solution, Agilent Dako, Santa Clara, CA, USA) in 700 W microwave. Sections were blocked and permeabilized in PBS containing 10% horse serum and 0.1% Triton X-100 for 1 h at room temperature followed by overnight incubation at 4 • C with either rabbit anti-CD3 monoclonal antibody (1:50; clone SP7, Abcam, Cambridge, UK) or rat anti-CD45R monoclonal antibody (1:200; clone RA3-6B2, Abcam). After three 5 min washes in PBS, sections were incubated for 2 h at room temperature with either DyLight R 594 conjugated donkey anti-rabbit polyclonal secondary antibody (1:500, Abcam) or Alexa Fluor 555 conjugated goat anti-rat polyclonal antibody (1:500, Invitrogen, CA, USA). Immunofluorescence staining for mucin 2 was performed on Carnoy's solution (60% absolute ethanol, 30% chloroform, and 10% acetic acid) fixed and paraffin embedded colon tissue sections using rabbit anti-mucin 2 polyclonal primary antibody (1:100; antibodies-online GmbH, Aachen, Germany) and DyLight R 594 conjugated donkey antirabbit polyclonal secondary antibody (1:250; Abcam). Nuclear counterstaining was performed with a mounting medium containing DAPI (Vectashield, Vector Laboratories, USA). Stained tissue sections were examined using the Zeiss Axioskop 40 microscope (Carl Zeiss Microscopy GmbH, Göttingen, Germany) connected to an AxioCam MRm (Carl Zeiss). For analyses, the number of CD3+ or CD45R+ cells was determined by counting stained cells per visual field (ten fields per slide). Mucin 2 staining was analyzed by measuring the mucus layer thickness using Zeiss ZEN blue software (seven animals per group, six images per animal, and five measurements per image). Mean value of five measurements per image was generated and data were presented as six technical replicates per animal. All images were taken and scored blindly.

Western Blot
Western blot analyses were performed as described previously (39). Briefly, total proteins were extracted from colon tissue and measured by Bradford Assay using Biorad-Dye-Reagent-Concentrate (Bio-Rad, Hercules, CA, USA) on the Victor TM X3 reader (Perkin Elmer, Waltham, MA, USA). One hundred microgram proteins per sample were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis on 15% gels and transferred to a nitrocellulose membrane (GE Healthcare) using a semidry system ("Pegasus, " Gesellschaft für Phorese, Analytik und Separation GmbH). After blocking with 5% nonfat milk in TBST (10 mM

Statistical Analysis
All data were analyzed using GraphPad Prism 6 R software (GraphPad Software, La Jolla, USA). Parametric data were shown as mean ± SEM and non-parametric data as median ± interquartile range. All data were tested with the Shapiro-Wilk or Kolmogorov-Smirnov normality test for normal distribution.
When the assumption about normally distributed data was not met, a non-parametric test was used. Statistical analysis for nonparametric data was performed using one-way ANOVA Kruskal-Wallis test with Dunn's multiple comparison test. For parametric data, one-way ANOVA with Sidak's test as post-hoc test was carried out. Statistical analysis of mucus layer thickness was performed using repeated measures one-way ANOVA Friedman test with Dunn's multiple comparison test. Comparison of data with two factors was analyzed with two-way ANOVA with Sidak's multiple comparisons test. P < 0.05 was considered significant ( * P < 0.05, * * P < 0.01, * * * P < 0.001, * * * * P < 0.0001).

RESULTS
Minimal Bacterial Consortia, ASF, and OMM 12 , Stably Colonize GF Il10-deficient Mice GF Il10 −/− mice were colonized with minimal bacterial consortia, ASF or OMM 12 , by co-housing them with minimal microbiota donor mice for 4 weeks (Figure 1A). The microbiota transfer (co-housing) was initiated after weaning at the age of 4 weeks. Subsequently, these two minimal bacterial consortia were modulated by SFB and MNV infection. One week after cohousing began, mice were inoculated with SFB on 2 consecutive days. At the end of co-housing, 8 week old Il10 −/− mice were infected with MNV for 4 weeks (chronic MNV infection) or 48 h (acute MNV infection) and subsequently sacrificed (Figure 1A).
At the age of 8 weeks, fecal samples were collected from gnotobiotic Il10 −/− mice and screened for the presence of minimal microbiota members. All ASF strains except ASF360 were detected in fecal samples by PCR analysis (Figure 1B).

Severity of MNV-triggered Colitis Depends on the Presence of Specific Bacteria
Il10 −/− mice develop spontaneous colitis, which is strongly dependent on the microbiota (19). Additionally, MNV has been shown to trigger colitis in these mice in a microbiotadependent manner (22). In our study, we first assessed the impact of minimal bacterial consortia alone on colitis development in Il10 −/− mice. ASF and OMM 12 induced none to mild histological changes in the colon and cecum of 12 week old B6-Il10 −/− mice (Figures 2A-D). These changes were mainly characterized by infiltration of myeloid cells such as lymphocytes in the lamina propria. MNV infection exacerbated the severity of colitis only in B6-Il10 −/− mice that were colonized with ASF, causing moderate intestinal inflammation (Figures 2A-C).
In contrast, B6-Il10 −/− mice colonized with OMM 12 were unaffected by MNV infection (Figures 2A,B,D). Pathological lesions in ASF-associated B6-Il10 −/− mice were predominantly located in the proximal colon and characterized by hyperplasia of the crypt epithelium and infiltration of inflammatory cells such as lymphocytes and granulocytes to the tela submucosa and tunica muscularis ( Figure 2C). The middle and distal part of the colon were less affected.
with MNV. Uninfected ASF-colonized C3H-Il10 −/− mice showed no histopathological changes in either the colon or cecum (Figures 3A-C). In contrast, uninfected OMM 12associated C3H-Il10 −/− mice showed mild histopathological changes mainly in the colon characterized by infiltration of lymphocytes (Figures 3A,B,D). MNV infection triggered colitis only in C3H-Il10 −/− mice colonized with ASF, but not in those carrying OMM 12 (Figures 3A-D). The histopathological scores between ASF-colonized B6-Il10 −/− and C3H-Il10 −/− mice were comparable with or without MNV infection, indicating that under these conditions both strains show similar colitis susceptibility. The same results were also observed between OMM 12 -colonized B6-Il10 −/− and C3H-Il10 −/− mice. In addition, and in line with our previous report, MNV alone did not cause pathological changes in the gut of GF B6-Il10 −/− or C3H-Il10 −/− mice (Figures 2E-G, 3E-G). These results indicated that the ability of MNV to trigger colitis depends on the presence of specific bacteria.

SFB Co-colonization Abolishes MNV-induced Colitis Only in Mice Colonized With ASF
Subsequently, we modulated ASF and OMM 12 consortia by cocolonizing Il10 −/− mice with SFB at the age of 5 weeks. These commensal bacteria show potent immunostimulatory effects, but are also discussed to be pathobionts (32). SFB co-colonization did not exacerbate the histopathology score in the colon or cecum of 12 week old B6-Il10 −/− mice colonized with ASF or OMM 12 consortium. Moreover, when ASF-associated B6-Il10 −/− mice were infected with MNV, the presence of SFB abolished MNVtriggered colitis in these mice. The histopathology score was  significantly reduced in both the colon and cecum of B6-Il10 −/− mice (Figures 2A,B). Furthermore, microscopic analysis of H&E stained colon tissue sections showed restoration of normal intestinal morphology (Figure 2C). In contrast, histological changes in the gut of OMM 12 -associated B6-Il10 −/− mice were not markedly modulated by either MNV or SFB (Figures 2A,B,D). Similar results were also observed in C3H-Il10 −/− mice, however, the SFB-mediated protective effect was less pronounced in these mice (Figures 3A-D). Furthermore, SFB monocolonization as well as SFB and MNV co-infection did not cause pathological changes in the gut epithelium of both analyzed Il10 −/− mouse strains (Figures 2E-G, 3E-G). Together, these results suggested that SFB co-colonization protects against pathological changes.

Host Response Is Distinctively Modulated by Different Microbiota Compositions
Next, we determined the expression of pro-inflammatory cytokines in the proximal colon to investigate the signaling molecules that are produced by the host upon encountering specific microorganisms. B6-Il10 −/− mice harboring different minimal bacterial consortia expressed partially distinct cytokines (Figures 4A-E). Twelve week old ASF-associated B6-Il10 −/− mice showed higher expression of Tnfα and lower expression of Ifnγ than OMM 12 -colonized B6-Il10 −/− mice (Figures 4A,B). In mice colonized either with ASF or OMM 12 , gene expression of Il1β and Il12a was similar (Figures 4C,D). Four weeks after MNV infection, OMM 12 -associated mice produced more Tnfα than ASF-associated mice in which no upregulation of this cytokine was observed upon infection ( Figure 4A). However, ASF-colonized mice showed slight increase of Ifnγ expression upon infection ( Figure 4B). Co-colonization of ASFassociated mice with SFB induced higher expression of proinflammatory cytokines such as Tnfα, Il1β, and Il12a upon MNV infection compared to infected ASF-associated mice without SFB (Figures 4A,C,D). The expression of these cytokines was not elevated in infected OMM 12 -associated mice co-colonized with SFB. Expression of Ifnγ was slightly reduced in uninfected and infected ASF-colonized mice upon SFB co-colonization, but remained unchanged in mice associated with OMM 12 (Figure 4B). Il17a was expressed in ASF-colonized B6-Il10 −/− mice co-colonized with SFB with and without MNV infection, but also in mice only carrying ASF. In contrast, no Il17a expression was detected in all groups of OMM 12 -associated B6-Il10 −/− mice. Upon MNV infection, Il17a expression was abolished in ASF-colonized B6-Il10 −/− mice without SFB ( Figure 4E).
To determine the cells present in the gut epithelium of these mice, colon tissue sections were stained immunohistochemically for the presence of myeloid CD3+ (predominantly T cells) and CD45R+ cells (naïve B cells). MNV infection elevated the number of CD3+ cells in the lamina propria of mice colonized with either minimal consortium ( Figure 4F). SFB co-colonization increased the number of CD3+ cells in ASFcolonized B6-Il10 −/− mice, but not in OMM 12 -associated B6-Il10 −/− mice. Furthermore, when comparing MNV-infected groups only, SFB colonization reduced the number of CD3+ cells in ASF-and OMM 12 -colonized mice 4 weeks after MNV infection. However, this effect was more pronounced in ASFcolonized mice ( Figure 4F). Moreover, staining for CD45R+ cells showed no difference among all groups ( Figure 4G).
As MNV-triggered inflammation is associated with barrier disruption, we analyzed whether these two minimal consortia differently affect MNV-induced barrier disruption by downregulating gene expression of tight junction (TJ) components. Therefore, we measured the gene expression of claudin 4 and 8, whose downregulation is associated with increased intestinal permeability (22,40,41). MNV infection significantly reduced gene expression of claudin 4 and 8 in the proximal colon of 12 week old ASF-associated B6-Il10 −/− mice. MNV was not able to downregulate the gene expression of these claudins in mice colonized with OMM 12 . Furthermore, when ASF-associated mice were co-colonized with SFB, MNV infection did not downregulate the gene expression of these TJ proteins ( Figure 4H). Co-colonization with SFB did not significantly change the expression of claudin 4 and 8 in mice carrying OMM 12 ( Figure 4H). This indicates that the protective effect provided by SFB is partially mediated by strengthening the epithelial barrier.

SFB Co-colonization Depends on the Microbiota Composition
To exclude that differences in the MNV infection rate are responsible for observed differences in intestinal pathology, we measured the viral load in total RNA isolated from the proximal colon using virus-specific qPCR. MNV copy numbers decreased when mice were colonized with bacteria ( Figure 5A). However, no correlation between the MNV load and intestinal pathology was found, as the highest viral load was detected in GF B6-Il10 −/− mice that showed no intestinal pathology (Figures 2E,G,  5A). Additionally, SFB inoculation did not significantly decrease the viral load in mice carrying minimal consortia suggesting that the SFB-mediated protective effect is not facilitated by preventing MNV infection ( Figure 5A).
As the SFB-mediated protective effect was absent in B6-Il10 −/− mice colonized with OMM 12 , we investigated the SFB colonization rate. Using an SFB-specific qPCR assay we quantified SFB in feces of GF, ASF-, and OMM 12 -colonized B6-Il10 −/− mice 7 weeks after SFB co-colonization. The presence of SFB was confirmed only in GF and ASF-colonized mice ranging from ∼10 2 to 10 3 copies/µL in ASF-colonized mice and 10 5 copies/µL in SFB monocolonized mice. In contrast, SFB remained largely undetected in mice carrying OMM 12 ( Figure 5B). To confirm the absence of SFB in OMM 12 -colonized mice, we performed fluorescence in situ hybridization (FISH) by staining ileum and colon sections with SFB-specific and bacteria domain-specific (EUB338) probes. SFB were detected and localized near the intestinal epithelium mainly in the ileum, but also in the colon of SFB monocolonized mice. SFB were not detected in OMM 12 -associated mice ( Figure 5C). As we could not detect SFB 7 weeks post inoculation, we questioned whether SFB could colonize OMM 12 -associated mice at all or whether Il17a measured by qPCR in total RNA isolated from proximal colon of ASF-colonized or OMM 12 -colonized B6-Il10 −/− mice 4 weeks after MNV infection with or without SFB co-colonization. Relative differences in gene expression were calculated by the comparative 2 − Ct method. Parametric data were shown as mean ± SEM (n = 5-7). Non-parametric data were shown as median ± interquartile range (n = 5-7). Data were obtained from one to two independent experiments. (F,G) Quantification of (F) CD3+ and (G) CD45R+ cells in the colon tissue of ASF-or OMM 12 -colonized B6-Il10 −/− mice with or without SFB co-colonization 4 weeks after MNV infection. Data presented in box and whiskers plots are medians with minimum, maximum, and individual values obtained from one to two independent experiments (n = 5-7). (H) Gene expression of tight junction genes claudin 8 and claudin 4 measured by qPCR in total RNA isolated from proximal colon of ASF-or OMM 12 -colonized B6-Il10 −/− mice with or without SFB co-colonization 4 weeks after MNV infection. Parametric data were shown as mean ± SEM (n = 5-7). Non-parametric data were shown as median ± interquartile range (n = 5-7). Data were obtained from one to two independent experiments. ASF, Altered Schaedler Flora; OMM 12 , Oligo-Mouse-Microbiota 12; SFB, segmented filamentous bacteria; MNV, murine norovirus; Tnfα, Tumor necrosis factor alpha; Ifnγ , Interferon gamma; Il1β, Interleukin 1 beta; Il12a, Interleukin 12a; Il17a, Interleukin 17a. Statistically significant differences are indicated as follows: *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001.

FIGURE 5 | SFB co-colonization depends on the microbiota composition. (A)
Using virus-specific qPCR, viral load was measured in total RNA isolated from the proximal colon of GF, ASF-and OMM 12 -colonized B6-Il10 −/− mice infected with MNV for 4 weeks with or without co-colonization with SFB. Data were shown as median ± interquartile range (n = 4-6). Data were collected from one to two independent experiments. (B) SFB copy numbers were measured in total DNA isolated from feces of 12 week old GF, ASF-, or OMM 12 -associated B6-Il10 −/− mice co-colonized with SFB with or without MNV infection for 4 weeks. Data were shown as median ± interquartile range and collected from three to nine independent experiments (n = 8-27). (C) Representative images of FISH staining of ileal and proximal colon tissue sections of GF and OMM 12 -associated B6-Il10 −/− mice co-colonized with SFB (n = 5). DNA was stained using DAPI (blue), SFB were detected using an SFB-specific probe (yellow) and all bacteria were detected by the EUB338 probe (red). Scale bars: 50 µm. (D) SFB colonization kinetic was determined by using an SFB-specific qPCR measuring SFB copy numbers in the total DNA isolated from feces of OMM 12 -colonized B6-Il10 −/− mice over a period of 20 days. OMM 12 -colonized B6-Il10 −/− mice were inoculated with the gut content of SFB monocolonized mice at day 0 and 1. Data were shown as median ± interquartile range (n = 7). Data were obtained from two independent experiments. (E) SFB colonization kinetics were determined by an SFB-specific qPCR measuring SFB copy numbers in the total DNA isolated from feces of OMM 12 -colonized B6-Rag2 −/− mice over a period of 39 days. OMM 12 -colonized B6-Rag2 −/− mice were inoculated with the gut content of SFB monocolonized mice at day 0 and 1. Data were shown as median ± interquartile range (n = 7). Data were obtained from two independent experiments. (F) SFB colonization kinetics were determined by an SFB-specific qPCR measuring SFB copy numbers in the total DNA isolated from feces of B6-Il10 −/− mice first monocolonized with SFB and subsequently colonized with OMM 12 . GF B6-Il10 −/− mice were inoculated with the gut content of SFB monocolonized mice at day 0 and 1. OMM 12 consortium transfer started when stable SFB colonization was established. Data were shown as median ± interquartile range (n = 7). Data were obtained from two independent experiments. GF, germ-free; ASF, Altered Schaedler Flora; OMM 12 , Oligo-Mouse-Microbiota 12; SFB, segmented filamentous bacteria; MNV, murine norovirus; FISH, fluorescent in situ hybridization. Dash-dotted line: detection limit. Statistically significant differences are indicated as follows: *P < 0.05, **P < 0.01 and ****P < 0.0001. they were eradicated over the course of the experiment. To test this, we followed SFB colonization kinetics in OMM 12 -associated mice over 20 days post SFB inoculation. OMM 12 -associated B6-Il10 −/− mice were inoculated with the gut content of SFB monocolonized mice at day 0 and 1. Feces samples were collected every few days and SFB-specific qPCR was used to determine the presence of SFB. Over the course of 20 days, SFB were below the detection limit, demonstrating that under this experimental condition SFB are unable to colonize OMM 12 -associated mice ( Figure 5D). These results indicated that the absence of the SFBmediated protective effect in mice colonized with OMM 12 is likely due to lack of SFB colonization. Furthermore, to analyze whether the adaptive immunity in OMM 12 -associated mice plays a role in SFB colonization, we colonized GF B6-Rag2 −/− mice with the OMM 12 consortium and SFB, and again followed SFB colonization kinetics. SFB successfully colonized GF B6-Rag2 −/− mice. However, SFB could not colonize OMM 12 -associated B6-Rag2 −/− mice indicating that the adaptive immune response is not responsible for preventing SFB colonization ( Figure 5E). Additionally, to analyze whether the absence of SFB is due to colonization resistance provided by OMM 12 consortium, we reversed the order of colonization. Juvenile GF B6-Il10 −/− mice were first inoculated with SFB and co-housing with an OMM 12 donor was initiated when GF B6-Il10 −/− mice were stably monocolonized with SFB. The presence of SFB was detected in feces of monocolonized B6-Il10 −/− mice starting at day 3 post inoculation (p.i.) and reached its maximum at day 5 p.i. At day 9 p.i., SFB monocolonized mice were co-housed with an OMM 12 donor. The SFB load exponentially decreased but stabilized at 10 2 -10 3 copies/µL from day 12 p.i., indicating that SFB can compete with OMM 12 members only when they colonize first ( Figure 5F). In addition, SFB copy numbers detected upon colonization with OMM 12 were comparable with those that were observed in ASF-SFB co-colonized mice ( Figure 5B). SFB Co-colonization of ASF-associated B6-Il10 −/− Mice Boosted Epithelial Barrier Defense and Immune Response in the Chronic and Acute Phase of MNV Infection Next, we were interested in elucidating SFB-mediated protective effects that prevented MNV to induce intestinal inflammation in B6-Il10 −/− mice associated with ASF. As we already mentioned previously, SFB promoted barrier integrity by preventing downregulation of tight junction components claudin 4 and 8 in the chronic phase of MNV infection (4 weeks post infection) ( Figure 4H). However, the presence of SFB was also associated with increased expression of pro-inflammatory cytokines such as Tnfα, Il1β, Il12a, and Il17a even though no inflammatory lesions were observed in the gut of these animals (Figures 2A,C, 4A). Therefore, we analyzed expression of other barrier-determining factors such as mucus 2 and an antimicrobial peptide produced by epithelial cells, Reg3γ . Furthermore, we also measured expression of the regulatory cytokine Tgfβ. These analyses showed that in the chronic phase of MNV infection in ASFcolonized mice, the presence of SFB increased production of Muc2, a major constituent of the mucus layer ( Figure 6A).
The expression of Reg3γ was slightly increased in ASF mice cocolonized with SFB 4 weeks after MNV infection ( Figure 6B). Furthermore, the presence of SFB also significantly increased the expression of Tgfβ in uninfected and chronically MNV-infected ASF-associated mice (Figure 6C). These results strengthened the hypothesis that the protective effects of SFB are mediated by enhancing intestinal barrier defense and host immune response.
As the ileum is the predominant SFB habitat, we investigated whether SFB or MNV cause pathological changes in the small intestine by staining ileal tissue with H&E. ASF-colonized and ASF-SFB co-colonized mice did not show any signs of ileitis or changes in ileal histomorphology 4 weeks after MNV infection ( Figure 6D).
In our previous study, we showed that MNV infection at an early stage (48 h) post infection induces structural and functional intestinal barrier changes that are hypothesized to be the initiating factor in the development of observed intestinal inflammation. Therefore, we analyzed whether the presence of SFB prevents MNV-induced barrier damage in the acute phase of MNV infection. Thus, ASF-SFB co-colonized B6-Il10 −/− mice were sacrificed 48 h post MNV infection and expression of barrier-determining factors as well as the host immune response were analyzed. Acute MNV infection resulted in reduced thickness of the mucus layer in the colon of ASFcolonized mice, which was prevented when these mice were cocolonized with SFB (Figures 7A,B). Furthermore, the presence of SFB enhanced mucin 2 gene expression and MUC2 abundance in the colon of ASF-colonized mice (Figures 7A-C). Additionally, MNV infection reduced the gene expression of ß-defensin 2, an antimicrobial peptide, in ASF-colonized mice. However, when SFB were present, MNV did not downregulate ß-defensin 2 expression (Figure 7D). SFB co-colonization increased Reg3γ gene expression and protein production (Figures 7E,F). Even though no changes in the gene expression of claudin 4 and 8 were observed, at the protein level MNV infection reduced the expression of claudin 8 (Figures 7G,I,J). Furthermore, the expression of tight junction proteins, claudin 4 and 8, was upregulated 48 h after MNV infection only when SFB were present (Figures 7H,J).
The impact of SFB co-colonization on the host immune system in the acute phase of MNV infection was also investigated. In ASF-SFB co-colonized mice strongly activated immune response was detected. SFB enhanced expression of proinflammatory cytokines such as Tnfα, Il1β, and Ifnγ in the colon ( Figure 8A). However, the presence of SFB markedly upregulated expression of the regulatory cytokine Tnfβ, but also enhanced the expression of Foxp3 that encodes for a transcriptional regulator protein involved in the development and function of regulatory T cells (Figures 8B,C). Surprisingly, 48 h post MNV infection elevated levels of measured pro-inflammatory cytokines were not observed. Furthermore, the presence of SFB also increased the number of CD3+ and CD45R+ cells in the colon lamina propria (Figures 8D,E). As interferon type III is important for epithelial defense against viruses, the expression of Ifnλ 2 in the colon was measured 48 h post MNV infection. The presence of SFB strongly enhanced expression of Ifnλ 2 , whereas MNV infection alone did not ( Figure 8F). In addition, expression of Mmp7, a gene that is predominantly induced by Ifnλ 2, was strongly upregulated in ASF mice that were co-colonized with SFB ( Figure 8G). Altogether, our results showed that SFB-mediated protective effects in both chronic and acute phase of MNV infection were linked to improved intestinal epithelial barrier defense and also enhanced host immune response.

DISCUSSION
Mucosal surfaces are colonized with a complex community of microorganisms that play an important role in host health and disease (5). However, underlying mechanisms of how the intestinal microbiota and its particular members make or break intestinal homeostasis are still mainly unknown. Thus, gnotobiotic models evolved as powerful tools to mechanistically analyze host-microbiome interactions under standardized conditions. In this study, we employed a gnotobiotic model of experimental IBD and demonstrated that the outcome of virusinduced colitis in Il10 −/− mice depends on the composition of the intestinal microbiota.
Murine noroviruses (MNVs) are single-stranded RNA viruses that are highly prevalent in mouse colonies (42). This group of viruses is related to human noroviruses that are the main cause of non-bacterial gastroenteritis and are implicated in disease severity of IBD patients (43)(44)(45). Furthermore, MNVs were shown to induce and exacerbate intestinal inflammation in experimental models of IBD such as Il10 −/− , Atg16L1 −/−, and Mdr1a −/− mice (22,46,47). MNV-induced intestinal pathology is microbiota-dependent, as GF or antibiotic treated mice do not develop inflammation after MNV infection (22,48,49). However, it is still unknown how different microbial communities help to facilitate norovirus infection. Here, we showed that the severity of MNV-triggered colitis in B6-Il10 −/− and C3H-Il10 −/− mice depends on the presence of specific bacteria. MNV exacerbated colitis and induced moderate intestinal inflammation in Il10 −/− mice colonized with ASF, a minimal microbiota composed of eight bacterial species. However, the histopathology score in mice associated with OMM 12 , a minimal bacterial consortium of 12 bacterial species, was not affected by MNV infection. Inflammatory lesions in Il10 −/− mice normally first appear in the cecum and proximal colon and involve the lower colon and rectum as disease progresses (50). Interestingly, Kuhn et al. showed that the localization of inflammatory lesions in Il10 −/− mice can be affected by the intestinal microbial composition. They described that Il10 −/− mice housed in a specific pathogen free environment develop an attenuated disease with lesions restricted to the proximal colon, while Il10 −/− mice housed under conventional conditions develop enterocolitis affecting the entire large intestine (15). In our model, MNV-induced inflammatory lesions in ASF-colonized mice predominately localized to the cecum and proximal colon, whereas the middle and distal colon were less affected. Therefore, a defined microbial environment represents a low colitogenic stimulus as defined by the degree and extent of inflammation.
Colonization processes and infections modulate host immunity. MNV infection activates interferon (IFN) and TNFα immune response pathways (58). Moreover, interferon type I and II responses are important in limiting MNV pathogenesis (59)(60)(61). In contrast, colitis development in Il10 −/− mice is driven by an aberrant response of Th1 cells and exaggerated production of pro-inflammatory cytokines such as IFNγ, IL17, and IL12 to microbiota-derived antigens (18). In our study, immunohistological staining of the colon showed increased infiltration of lymphocytes that were mainly characterized as CD3+ cells supporting the T cell mediated response to chronic MNV infection. Furthermore, MNV slightly increased expression of Ifnγ in ASF-colonized Il10 −/− mice, whereas in OMM 12 -assciated Il10 −/− mice, Tnfα was upregulated. In contrast, SFB co-colonization strongly upregulated expression of pro-inflammatory cytokines Tnfα, Il1β, and Il12a, but not Ifnγ in ASF-colonized mice 4 weeks post MNV infection. Expression of these cytokines was mainly unchanged in OMM 12associated mice. Discrepancies in Il12a and Ifnγ expression in ASF-colonized mice after SFB colonization suggest an IFNγ-independent production of IL12. Production of IL12 in an IFNγ-independent manner is hypothesized to be relevant for physiological and pathological immune responses (62). However, whether particular bacterial species play a role in downregulation of Ifnγ expression in chronic MNV infection needs to be further investigated. SFB colonization also stimulated expression of the regulatory cytokine Tgfβ, while no increase in the expression of Il17a was observed in these mice. IL17A is a signature cytokine of Th17 cells, which can be stimulated by SFB. However, Th17 cells can acquire a regulatory phenotype and lose the Il17a expression signature, which is promoted by TGFβ (63). Furthermore, the presence of TGFβ, IL6, and IL1β supports differentiation of Th17 cells devoid of pathogenicity (64). In addition, a recent publication demonstrated that the epithelial cell endocytosis of SFB antigens is important for regulation of T cell homeostasis (65).
The intestinal epithelium is crucial for the maintenance of gut homeostasis (66). We previously showed that MNV induces intestinal barrier disruption by downregulating expression of TJ components, which regulate intestinal permeability (22). Moreover, altered expression and distribution of TJ proteins are accompanied by increased paracellular permeability which is associated with the development of the intestinal inflammation (67)(68)(69). Claudin 4 and 8 in particular are important for sealing TJs. In our study, MNV decreased expression of claudin 4 and 8 in mice colonized with ASF, but not OMM 12 , indicating that specific bacteria synergize with MNV to cause barrier dysfunction by influencing expression of TJ components. Furthermore, SFB colonization in mice carrying ASF prevented MNV-induced downregulation of these TJ proteins. However, SFB colonization did not influence their expression in OMM 12 -associated mice. Additionally, SFB co-colonization increased gene expression of Muc2 and antimicrobial peptide Reg3γ in ASF-colonized mice. Mucin 2 is the main component of the mucus layer that keeps intestinal bacteria at a distance from the intestinal epithelium. The inner mucus layer contains various antimicrobial peptides produced by epithelial cells that exhibit bactericidal effects (70,71). Colonization with SFB was shown to ameliorate colitis induced by the intestinal pathogen Citrobacter rodentium by increased expression of pro-inflammatory cytokines and antimicrobial defenses (31). Furthermore, several studies showed that animal models with an impaired MUC2 or REG3γ production pose an increased risk of developing intestinal inflammation (72)(73)(74).
As the phenotype of OMM 12 -colonized mice was not modulated by SFB colonization, we analyzed SFB colonization dynamics. We showed that SFB cannot colonize OMM 12 -associated mice and hypothesized that the absence of SFB-mediated protective effects in these mice is due to lack of SFB. Both the host immune response and intestinal microbiota itself regulate colonization processes in the gut. The host immune response can limit colonization by inducing secretion of cytokines and immunoglobulins (75). The intestinal microbiota confers colonization resistance mechanisms for protection against new and harmful organisms by competing for essential nutrients or secreting inhibitory bacteriocins (76,77). Our results showed that SFB colonization depends on the presence of specific bacterial species. SFB could not colonize OMM 12 -associated B6-Rag2 −/− mice, excluding the adaptive immune response as a reason for unsuccessful SFB colonization. However, SFB could compete with OMM 12 members when we reversed the order of colonization demonstrating that established OMM 12 consortium prevents SFB colonization. The OMM 12 consortium, despite its reduced complexity, provides partial colonization resistance to Salmonella enterica serovar Typhimurium infection (28). However, the exact mechanism of how OMM 12 bacteria interfere with SFB colonization needs to be determined in future studies.
As MNV was shown to initiate barrier disruption in an early phase of infection, we analyzed whether SFB could prevent adverse effects of MNV in ASF-colonized B6-Il10 −/− mice 48 h post infection. In the acute phase of MNV infection, the presence of SFB strengthened the intestinal epithelium by increasing the production of antimicrobial peptides (ß-defensin 2 and REG3γ), MUC2 and tight junction proteins claudin 4 and claudin 8. Moreover, SFB modulated host immune response by upregulating expression of pro-inflammatory cytokines such as Tnfα, Ifnγ , and Il1β. However, the expression of the antiinflammatory cytokine Tgfβ, as well as Foxp3, a gene coding for a transcription factor involved in the development of Tregs, were also markedly upregulated in these mice. These results supported our conclusion that SFB can induce mucosal regulatory responses in the absence of IL10. Furthermore, type III interferons protect the epithelial barrier from viral infections and damage induced by bacteria, but also suppress intestinal inflammation in mice (78,79). MNV did not strongly upregulate the expression of interferon lambda 2 (Ifnλ 2 ) 48 h post infection, whereas ASF-SFB co-colonized mice showed higher expression of this cytokine. This suggests that upregulated Ifnλ 2 expression is part of the SFB-mediated protective effect in the acute phase of MNV infection. Moreover, the presence of SFB resulted in the increased expression of Mmp7, which is predominantly induced by IFNλ 2 (80). Mmp7 is a member of the matrix metalloproteinase family of enzymes that are involved in tissue remodeling and wound repair. Altogether, our results suggested that SFB prevented development of inflammatory lesions in Il10 −/− mice by enhancing intestinal barrier defense mechanisms and inducing regulatory immune response. We showed that the severity of MNV-induced colitis in Il10 −/− mice depends on the intestinal microbial context and that SFB-mediated protective effects are multifaceted. Ultimately, it is essential to understand the finetuned interplay between the host and specific commensals, as this will deliver new strategies that can be used for disease interventions.

DATA AVAILABILITY
The raw data supporting the conclusions of this manuscript are available from the corresponding author on reasonable request to any qualified researcher.

ETHICS STATEMENT
This study was conducted in accordance with German animal protection law and with the European Directive 2010/63/EU.