Gut Microbial Dysbiosis Due to Helicobacter Drives an Increase in Marginal Zone B Cells in the Absence of IL-10 Signaling in Macrophages

Interleukin-10 is an important anti-inflammatory cytokine produced in the gut by a variety of immune cells, including B cells, T cells, and macrophages, as well as nonhematopoietic cells (1). Genome-wide association studies revealed a linkage between IL-10 polymorphisms and susceptibility to inflammatory bowel disease in humans (2, 3). Humans harboring loss-of-function mutations in IL10RA and IL10RB, whose gene products encode for the heterodimer IL-10R1 and IL-10R2 subunits, respectively, succumb to early-onset colitis (4–6). In mice, spontaneous onset of severe intestinal inflammation also occurs in mice deficient in IL-10 or its receptor (7, 8), and is prevented by IL-10 signaling in macrophages (9, 10). Collectively, these studies illuminate the essential role for IL-10 in dampening intestinal inflammation and the maintenance of gut homeostasis.

It has become clear that the gut microbiome plays an essential role in the health of the host and that microbial dysbiosis can contribute to disease (11). For example, increases in Enterobacteriaceae and Gram-negative anaerobes, including Bacteroides, have been associated with inflammatory bowel disease in humans (12–14). A decrease in presumed beneficial bacteria, such as Bifidobacteria species, has also been reported in inflammatory bowel disease (14, 15). In mice, spontaneous enterocolitis in Il10^−/− mice only occurs in some animal colonies in a microbiome-dependent manner (7, 16, 17). Mice lacking Helicobacter do not succumb to spontaneous colitis in the absence of IL-10 (18). In addition, Helicobacter hepaticus-induced colitis susceptibility has been shown to differ between mouse colonies with markedly different microbiomes (16). Although the gut microbiome can influence immune responses at distal sites (11, 19), the role of specific bacterial species in systematic inflammation remains only partially understood.

The Il10^−/− mouse has not been reported to have defects in hematopoietic lineage development, but at 8 wk an elevation in IgM, IgG1, IgG2b, IgG2a, and IgA was reported (7). Whether this is due to altered B cell function is not known, and a comprehensive analysis of B cell subsets has not been performed in the Il10^−/− mouse. Whereas T-dependent immune responses in Il10^−/− mice were reported to be normal (7), the slight elevation in IgM is suggestive that T-independent immune responses may be altered. T-independent Ab responses are in part mediated by marginal zone (MZ) B cells that respond very rapidly to blood-borne pathogens due to their localization within the splenic MZ (20). The MZ is situated between the red and white pulp through which most of the blood that enters the spleen flows through before re-entering the circulation (20). Thus, a breach in the epithelial layer during gut inflammation allowing bacterial translocation would potentially prompt a MZ B cell response. In Il10^−/− mice, bacteria were found to penetrate the colon mucus layer and directly contact the epithelial barrier, and Enterococcus faecalis, through the production of a metalloprotease, was shown to compromise the barrier (21, 22). These effects on the barrier are thought to contribute to inflammation within the gut, but whether they alter immune cell function and homeostasis at distal sites is unknown.

In this study, we asked whether Il10^−/− mice with susceptibility to spontaneous enterocolitis had an imbalance in immune cell homeostasis outside of the gut compared with resistant mice raised in a separate facility. We found that susceptible mice exhibited a significant increase in both MZ B cells and CD11b⁺ myeloid cells in the spleen that was inhibited by IL-10 signaling in macrophages. Cohousing, antibiotic treatment, and 16S rRNA sequencing studies indicated that the immune homeostasis alteration was microbiome dependent. In addition, introduction of H. hepaticus to nonsusceptible mice was sufficient to drive MZ B cell differentiation and macrophage expansion. These results indicate that introduction of a single bacterial species can produce dysbiosis in the gut and drive a functional imbalance in immune homeostasis in the spleen when the gatekeeper function of IL-10 is compromised.

C57BL/6J and B10.PL (H-2^u) wild-type (WT) mice and B6.129P2-Il10^tm1Cgn/J mice (Il10^−/−) were purchased from The Jackson Laboratory (Jax; Bar Harbor, ME). IL-10Rα^fl/flCD19Cre⁺ and IL-10Rα^fl/flCD19Cre⁻ mice were provided by T. Geiger (St. Jude’s Children’s Research Hospital, Memphis, TN) (23). IL-10Rα^fl/flLysMCre⁺ and IL-10Rα^fl/flLysMCre⁻ mice were provided by W. Muller (University of Manchester, Manchester, U.K.) (24). B6.129P2-Il10^tm1Cgn/J mice were backcrossed to B10.PL for three generations and then intercrossed to generate B10.PL H-2^uxu Il10^+/+, Il10^+/−, and Il10^−/− littermates. All animals were housed and/or bred at the Translational Biomedical Research Center of the Medical College of Wisconsin. All animal protocols were approved by the Medical College of Wisconsin Institutional Animal Care and Use Committee. At the initiation of all experiments, including cohousing, mice were between 6 and 8 wk of age.

The 2.4G2 Ab was produced locally. Mouse-specific CD45R-PE-Texas Red, CD45R-PE, CD5-allophycocyanin, CD86-V450, Ki-67-FITC, caspase 3-FITC, and CD40 Abs were purchased from BD Biosciences (San Diego, CA). Mouse-specific CD21-eFluor 450, CD23-PE-Cy7, CD23-FITC, CD1d-PE, CD93-biotin, CD93-allophycocyanin, CD93-PE, TCR-β-FITC, TCR-β-PE, CD4-biotin, CD4-FITC, CD4-allophycocyanin-eFluor 780, CD8-PE-Cy7, CD11b-biotin, CD11b-eFluor 450, and Foxp3-PE Abs were purchased from eBioscience (San Diego, CA). Mouse-specific CD11b-Alexa Fluor 488, CD45R-Alexa Fluor 594, CD80-PE-Cy5, CD40-Alexa Fluor 647, MHC class II-PE-Cy7, Ly6C-allophycocyanin, Ly6G-allophycocyanin-Cy7, Ly6G-Alexa Fluor 647, F4/80-PE-Cy7, CD138-allophycocyanin, IgM-allophycocyanin-Cy7, IgD-Pacific Blue, Notch 2-PE, Delta-like 1-Alexa Fluor 647 Abs, and the LEGENDplex multianalyte flow assay kit were purchased from BioLegend (San Diego, CA). Mouse-specific Marco-FITC and MOMA-FITC Abs were purchased from AbD Serotec (Raleigh, NC). Anti–Bc1-2 and anti–Bcl-x_L were purchased from Cell Signaling Technology (Danvers, MA). Anti-mouse IgM-FITC was purchased from Southern Biotech (Birmingham, AL). Anti-mouse IgM F(ab′)₂ was purchased from Jackson ImmunoResearch Laboratories (West Grove, PA). Streptavidin-PE-Cy5.5 was purchased from eBioscience. Anti–BrdU-allophycocyanin was purchased from BD Biosciences. CFSE and DAPI were purchased from Molecular Probes (Eugene, OR). LPS was obtained from Sigma-Aldrich (St. Louis, MO), and CpG from Invivogen (San Diego, CA). Ampicillin and neomycin were purchased from LKT Laboratories (St. Paul, MN), and metronidazole and vancomycin were obtained from Sigma-Aldrich (St. Louis, MO).

Single-cell suspensions were prepared from bone marrow (BM), thymus, Peyer’s patches, inguinal lymph nodes, and spleens. Peritoneal cavity cells were isolated, as previously described (25). A total of 1 × 10⁶ cells was incubated with anti-CD16/CD32 (Fc block) (clone 2.4G2) for 15 min, followed by cell surface staining with specific mAb. Intracellular Ki-67 was performed using the anti-mouse/rat Foxp3 staining buffer set from eBioscience. Cells were acquired on a LSRII flow cytometer (BD Biosciences), and data were analyzed using FlowJo software (Tree Star, Ashland, OR). Splenic B cell subsets were characterized, as described (26). For in vitro culture and real-time PCR, B cell subsets were sorted using a FACSAria cell sorter (BD Biosciences), as described (27).

Spleens from 8-wk-old mice were embedded in Tissue-Tek OCT compound (Sakura Finetek, Torrance, CA) and snap frozen. The 7-μm sections were stained with B220-PE and MOMA-FITC, and images were captured using a Nikon Eclipse TE200 inverted fluorescent microscope, as described (28). Sections stained with B220-Alexa Fluor 594, CD11b-Alexa Fluor 488, and LysG-Alexa Flour 647 were imaged by confocal microscopy with an Olympus Fluoview FV1000 MPE Multiphoton Scanning Microscope.

Serum and colon tissue were collected from 7- to 8-wk-old naive mice. Colons were homogenized in PBS containing 0.1% IGEPAL CA-630 (Sigma-Aldrich) and mini protease inhibitor (Roche, Indianapolis, IN) using the gentleMACS Dissociator (Miltenyi Biotec, San Diego, CA). Cytokines and chemokines in serum and colon lysates were determined using LEGENDplex multianalyte flow cytometry assay kit (BioLegend).

Mice were immunized with 30 μg 4-hydroxy-3-nitrophenylacetyl (NP)-Ficoll i.p., and sera were collected 7 and 14 d later. Sera from 8-wk unimmunized mice were used for detecting resting IgM and IgG3 levels. ELISA was performed to determine NP-specific Ig titers and total resting IgM and IgG3 levels in the serum, as described previously (28, 29).

Culture of the MU-94 H. hepaticus strain used in these studies was performed at the University of Missouri in the laboratory of C. L. Franklin (30, 31). The culture suspension was pipetted into 15-ml conical tubes, wrapped in bubble wrap, and placed in a foam shipping container lined with ice packs. A previous experiment demonstrated that H. hepaticus was viable in these conditions for ∼16 h (data not shown). Cultures were shipped by overnight courier to ensure arrival in Milwaukee, WI, within 16 h and gavaged immediately upon receipt.

Il10^−/− Jax, IL-10Rα^fl/flCD19Cre⁺, or IL-10Rα^fl/flLysMCre⁺ mice were cohoused with Il10^−/− mice from our colony (B10.PL background) in a 1:1 ratio for 4 wk for feces-mediated transfer of the microbiota. Il10^−/− Jax mice were orally gavaged with ∼5 × 10¹⁰ CFU H. hepaticus suspended in Brucella broth. Control mice received sterile broth only. Cellular analysis of splenic B cell subsets and CD11b⁺ myeloid cells was performed 4 wk after cohousing or bacterial gavage.

The following antibiotics were provided ad libitum in drinking water for 4 wk: ampicillin trihydrate (1.0 g/L), vancomycin hydrochloride (0.5 g/L), neomycin sulfate (1.0 g/L), and metronidazole (1.0 g/L). This combined antibiotic regimen provides effective broad-spectrum coverage against Gram-positive and -negative bacteria (32).

Mice were i.p. injected with 1 mg BrdU, twice daily, for 3 d. BrdU⁺ B cells were tracked 1, 30, and 45 d after the last BrdU injection using the allophycocyanin BrdU flow kit (BD Biosciences). For the B cell turnover assay, mice were provided with BrdU in the drinking water (0.8 mg/ml) for 9 d, and the next day BrdU⁺ B cells were detected using flow cytometry (33, 34).

FACS-purified B cells (1 × 10⁵) were cultured in vitro in the presence of LPS (1 and 10 μg/ml), CpG (0.1 and 1 μg/ml), goat anti-mouse IgM F(ab′)₂ (0.5 and 5 μg/ml), or anti-CD40 (0.1 and 1 μg/ml) for 72 h, and cell survival was determined by flow cytometry gating on live DAPI⁻ cells.

Total RNA was isolated from FACS-purified cells using the RNAqueous -4PCR kit (Ambion, Austin, TX), and first-strand cDNA was synthesized using Superscript II Reverse Transcriptase (Invitrogen, Carlsbad, CA). The quality of the cDNA was assessed by Gapdh PCR. The expression of Gapdh, Hes1, Dtx1, and Rbpj was quantified by real-time PCR using TaqMan primers and probes from Invitrogen and an ABI Prism 7500 PCR machine (Applied Biosystems, Foster City, CA). The amplification cycle was as follows: 10-min hold at 95°C, and then 40 cycles of 15 s at 95°C and 60 s at 60°C. The expression of the Hes1, Dtx1, and Rbpj genes was normalized to the housekeeping gene Gapdh by calculating the ΔC_T. The relative expression of each gene is represented as 2^(−ΔCT).

Small and large intestines and cecum were isolated from Il10^−/− Jax mice that were or were not cohoused with Il10^−/− mice from our colony, as described (35). The isolated tissues were homogenized, bacterial DNA was isolated using MO BIO Power Analyzer kit (MO BIO Laboratories, Carlsbad, CA), and PCR was used with unique bar-coded primers to amplify the V4 region of the 16S rRNA gene to create an amplicon library from individual samples (36, 37). The 5′ primer was 5′-AATGATACGGCGACCACCGAGATCTACACTATGGTAATTGTGTGCCAGCMGCCGGGTAA-3′, and the 3′ primer was 5′-CAAGAGAAGACGGCATACGAGATNNNNNNAGTCAGTCAGCCGGACTACHVGGGWTCTAAT-3′. The individual PCRs were set up as follows: 10 μL 5× reaction buffer, 1.5 μL dNTPs, 2 μL 5′ primer, 2 μl 3′ primer, 1.5 μl LongAmp enzyme kit from New England Biolabs (Ipswich, MA), 30 μL DNA with the concentration of DNA at 2–5 ng/μl, and 3 μl H₂O, as described (37). The amplification cycle was as follows: denature at 94°C for 1 min, followed by 32 cycles of 94°C for 30 s, 50°C for 1 min, 65°C for 1 min, with a final extension of 65°C for min. The PCR products were sequenced using NextGen sequencing Illumina GAIIx platform (36). The samples were first quantitated using Pico Green, adjusted to a concentration of 4 nM, and then used for sequencing (37). Fastq conversion of the raw data files was performed following demultiplexing. Overall reads quality was assessed using the program FASTQC, and quality filtering of fastq files was performed using the FASTX toolkit.

A total of 11,321,588 sequencing reads of 101-bp length was generated from 12 animals (36 samples total) using Illumina GAIIx sequencing technology. Quality filtering of raw data was performed in two steps, as follows: 1) due to low quality of bases toward the 3′ ends of the read, 11 bases were trimmed from the 3′ end for all reads and 2) only high-quality reads having >90% bases with a Q Score >30 are retained. The quality filtering step retained 78% of total reads (8,928,469) with average of 248,013 reads per sample (range: 127,940–484,143 reads per sample). We clustered those sequences using AbundantOTU⁺ v.0.93b with the “-abundantonly” option at 97% similarity and generated 824 operational taxonomic units (OTUs) incorporating 99.38% of the total input sequences. We used USEARCH (38) (v. 6.1.357) and the Gold reference database to screen for the presence of chimeras in our OTU sequences, and a total of 135 OTUs was removed. The remaining 689 OTUs (representing 98.73% of the input sequences) were used for downstream analysis. We assigned taxonomic ranks through the Quantitative Insight into Microbial Ecology (QIIME) (39) (v. 1.8.0) using both uclust consensus taxon assigner and the Ribosomal Database Project (40) Classifier (v. 2.2) (confidence set to 80%). OTU consensus sequences were then aligned to the GreenGenes core reference alignment (41) using PyNAST (42). We built a phylogenetic tree from those alignments using FastTree (43) (v. 2.1.3) as implemented in the QIIME pipeline (v. 1.8.0). We generated BIOM files from our OTU counts and fed those along with the phylogenetic tree to PhylosEquation (44) (v. 1.7.24). Raw OTU counts were normalized and log₁₀ transformed according to the following formula:

The normalized and log₁₀-transformed reads were used in Phyloseq to produce principal coordinate analysis (PCoA) plots from Bray-Curtis and UniFrac (weighted and unweighted) (45). The α diversity measures (Chao1 richness estimate and Shannon diversity index) were also calculated using Phyloseq after rarefying the raw counts to an even depth of 97,988 reads (the minimum count in all samples). We used a mixed linear model utilizing SAS (v. 9.3) software to analyze the data and accounting for possible effects that may arise from cohousing the mice in the same cage (46).

Data were analyzed using GraphPad prism software and were presented as mean ± SE. Statistical significance was determined using the nonparametric Mann–Whitney U test or unpaired t test. The p values <0.05 were considered significant. Our mixed linear model, in which origin and tissue are a fixed effect and cage is a random effect (46), takes the form of the following:

where Y_ijkl represents either PCoA axis value OTU log normalized count Shannon or Chao1 value for Origin i, Tissue j, Cage k, and replicate l. O_i is the effect of the i^th origin, where origin is set to either The Jackson Laboratory animals, The Jackson Laboratory animals cohoused with our animals, or our Il10^−/− animals. T_j is effect from the j^th tissue (cecum, small intestine, or large intestine). (OT)_ij is the interaction effect between treatment i and tissue j. C_k(i) is the effect from the k^th cage that is nested within the i^th origin, and $ε$_ijkl$εijkl$ denotes the error associated with measuring Y_ijkl$Yijkl$. We controlled for the false discovery rate (FDR) by correcting the p values using Benjamini and Hochberg approach (47).

To determine whether the gut microbiota alters distal immune cell homeostasis in mice susceptible to colitis, we quantitated immune cell populations in Il10^−/− mice. For this analysis, we used Il10^−/− mice acquired from The Jackson Laboratory (Jax) in the year 2000 that were crossed to the B10.PL background (H-2^u) and maintained in our colony by heterozygous crossings under specific pathogen-free conditions. Because our Il10^−/− mice are susceptible to spontaneous enterocolitis, we used mice at 6–8 wk of age prior to enterocolitis onset and evidence of T cell activation in the gut (data not shown). First, we compared the total cellularity of Il10^+/+ (WT) and Il10^−/− littermates and found no difference in the spleen (Fig. 1A), lymph node, thymus, Peyer’s patch, or BM (Supplemental Fig. 1A). We next performed a comprehensive immune phenotyping to determine whether the homeostasis of specific immune cell populations was altered in Il10^−/− mice. In the spleen (Fig. 1B) and lymph node (Supplemental Fig. 1B), we found that percentage of αβ T cells was significantly reduced, whereas no difference was observed in the Peyer’s patch or peripheral blood (Supplemental Fig. 1B). We also observed a significant reduction in the absolute number of αβ T cells in the spleen (Fig. 1B). Overall, we did not observe a difference in the ratio of CD4:CD8 T cells in the spleen (Fig. 1C) or in the lymph node, thymus, or peripheral blood (Supplemental Fig. 1C, 1D). For B cells, whereas the percentage of total B cells was significantly decreased in the spleen of Il10^−/− mice, the reduction in absolute number did not reach significance (Fig. 1D). Interestingly, the percentage of B cells was significantly increased in the lymph node and decreased in the peripheral blood, although no change was observed in the Peyer’s patch or BM (Supplemental Fig. 1E).

We then asked whether the homeostasis of specific B cell subsets was altered in Il10^−/− mice. Using a gating strategy to identify peritoneal cavity B cell subsets, there was no difference in the absolute number of B2, B1, B1a, or B1b cells (Supplemental Fig. 2A). When we gated on splenic B cell subsets (26) (Supplemental Fig. 2B), we found that both the percentage and absolute number of the transitional 1 (T1) and transitional 2 (T2) precursors and the mature follicular (Fo) populations were significantly decreased in Il10^−/− mice (Fig. 1E). There was no difference in the percentage or absolute number of T2–MZ precursors, the immediate precursor to MZ B cells (Fig. 1E). However, there was a significant increase in MZ B cells (Fig. 1E), which was confirmed using an alternative gating strategy (Supplemental Fig. 2C) (48, 49). Because MZ B cell retention is regulated in part by macrophages and perhaps neutrophils (50, 51), we quantitated splenic CD11b⁺ myeloid cells and found that both their percentage and absolute number were significantly increased in Il10^−/− mice (Fig. 1F). We next used immunofluorescence to determine whether IL-10 deficiency altered the splenic architecture. By detecting B220⁺ B cells and MOMA-1⁺ macrophages to demarcate the MZ, we found an evident increase in the size of the MZ in Il10^−/− mice, but no obvious difference in the splenic architecture (Fig. 1G). In measuring CD11b⁺ myeloid cells at other anatomical sites, we found a similar increase in their percentage in the peripheral blood and BM (Fig. 1H), but not the lymph node (data not shown). These data indicate that the loss of IL-10 has organ-specific effects on immune cell homeostasis.

We next determined whether the increase in myeloid cells and MZ B cells was associated with signs of inflammation. To measure local inflammation in the gut, we quantitated the protein expression level of two chemokines and six proinflammatory cytokines known to be associated with intestinal inflammation in the colon. As compared with WT mice, Il10^−/− mice had significantly increased levels of the chemokines CCL2 and CXCL10 and the cytokines TNF-α, IL-1β, IL-6, and IL-17, but not IL-23 or IFN-γ (Fig. 1I). These data indicate that local inflammation is occurring in Il10^−/− mice in our animal colony, which is consistent with their eventual development of spontaneous enterocolitis. To determine whether inflammation was detectable systemically, we measured the same panel of chemokines and cytokines in the serum and found no difference between the two cohorts of mice (Table I). These data suggest that the increase in MZ B cells and myeloid cells in Il10^−/− mice is not due to a systemic immune response, but stems from inflammation in the colon.

We next examined the immunological consequence of an increase in MZ B cells and found that resting levels of IgM, but not IgG3, were significantly elevated in Il10^−/− mice (Fig. 1J). To determine whether the increase in MZ B cells impacted the amplitude of the humoral immune response, we immunized with the T-independent Ag NP-Ficoll. On day 0, there was a significant increase in NP-specific IgM in Il10^−/− mice that was retained on day 7 (Fig. 1J). NP-specific IgG3 was not detectable on day 0 (data not shown), and there was no difference in its generation on day 14 in WT versus Il10^−/− mice (Fig. 1J). The increase in NP-specific IgM was not accompanied by an increase in the percentage of plasma cells in the spleen (Supplemental Fig. 2D). However, there was an increase in the absolute number of plasma cells in the spleen of Il10^−/− mice (Supplemental Fig. 2D). The increase in IgM was not likely due to B1 cells (52) because we found no difference in the absolute number of peritoneal cavity B cell subsets (Supplemental Fig. 2A). Thus, the increase in resting levels of total IgM in Il10^−/− mice is most likely due to increased numbers of MZ B cells contributing to the IgM pool. Following immunization, the increase in NP-specific IgM is consistent with an increase in the absolute number of plasma cells.

To determine whether the alterations in B cell and myeloid homeostasis in Il10^−/− mice were unique to our colony, we performed the same splenic phenotypic analysis on mice purchased directly from Jax and found no significant difference in the absolute number of T cells, B cell subsets, and myeloid cells between WT Jax and Il10^−/− Jax mice (data not shown). The contrast between mice housed long-term in our colony and mice acquired directly from Jax supports the hypothesis that the microbiome acquired after housing in isolated animal colonies can specifically alter immune cell homeostasis. To evaluate this hypothesis, we determined whether Il10^−/− Jax mice cohoused with Il10^−/− mice from our colony would acquire the immune homeostasis defect. We found that 4 wk of cohousing was sufficient to induce an increase in both MZ B cells and CD11b⁺ myeloid cells, but not Fo B cells (Fig. 2A). Kinetically, the increase in the percentage of MZ B cells and CD11b⁺ myeloid cells was evident as early as 1 wk following cohousing, which peaked at 4 wk (Fig. 2B). In contrast, the percentage of Fo B cells within the total B cell population did not change (data not shown). To further demonstrate that the acquisition of an imbalance in immune homeostasis was driven by the microbiome, we included a treatment group whereby Il10^−/− Jax cohoused mice were provided a combination of nonabsorbable broad-spectrum antibiotics in the drinking water (32). The inclusion of antibiotics from the outset of the experiment prevented the increase in the percentage and absolute number of MZ B cells and CD11b⁺ myeloid cells (Fig. 2C).

To determine the phenotype of the expanded CD11b⁺ myeloid population in Il10^−/− mice, we subdivided CD11b⁺ cells by Ly6C and Ly6G expression. Specifically, we quantitated Ly6C⁺Ly6G^high neutrophils, Ly6C^highLy6G⁻ inflammatory and Ly6C^lowLy6G⁻ resident monocytes/macrophages, and Ly6C⁻Ly6G⁻ cells that are most likely a combination of resident monocytes/macrophages, dendritic cells, and NK cells (Fig. 3A, Supplemental Fig. 2E) (53, 54). When we measured the absolute number of each CD11b⁺ myeloid cell population in Il10^−/− Jax mice following cohousing with Il10^−/− mice from our colony, we found a significant increase in the Ly6C⁺Ly6G^high neutrophil and in the Ly6C^highLy6G⁻ and Ly6C^lowLy6G⁻ monocyte/macrophage populations as compared with both WT mice and noncohoused IL10^−/− Jax mice (Fig. 3B). No change in the Ly6C⁻Ly6G⁻ population was observed (Fig. 3B). Antibiotic treatment during the cohousing period prevented the increase in the absolute number of neutrophils and monocytes/macrophages (Fig. 3B). When we measured the relative distribution of each cell type within the CD11b⁺ population, we observed a significant increase in the percentage of neutrophils and monocytes/macrophages with cohousing (Fig. 3C) that correlated with their increase in absolute number (Fig. 3B). Interestingly, antibiotic treatment did not affect the percentage of neutrophils, but did prevent the increase in the percentage of the monocytes/macrophages (Fig. 3B), which was offset by the reciprocal effect on the Ly6C⁻Ly6G⁻ population (Fig. 3C). We confirmed that the Ly6C^highLy6G⁻ inflammatory macrophages expressed CCR2, which was not expressed by the Ly6C^lowLy6G⁻ monocytes/macrophages (data not shown) (55). We next determined the localization of the increased myeloid cells in the spleen by immunofluorescence. As shown in Fig. 3D, the increased numbers of both CD11b⁺Ly6G⁻ macrophages (green) and CD11b⁺Ly6G⁺ neutrophils (yellow) in Il10^−/− mice were clearly evident. Interestingly, the myeloid cells were restricted to the red pulp and were completely excluded from the B and T cell zones. We then examined the activation status of the macrophages and neutrophils by flow cytometry staining for MHC class II and CD86, and found that the two cell populations in Il10^−/− Jax cohoused mice were not more activated as compared with WT mice and Il10^−/− Jax cohoused antibiotic-treated mice (Fig. 3E). Surprisingly, Il10^−/− Jax cohoused mice expressed lower levels of MHC class II with no difference in CD86 (Fig. 3E). These data indicate that loss of IL-10 signaling outside of the gut does not lead to the overactivation of myeloid cells. This finding is consistent with the histology, because unactivated macrophages would not be expected to migrate into the white pulp.

Because MZ B cell retention within the MZ has been reported to be regulated by MARCO⁺ macrophages resident in the MZ (56), we also determined the absolute number of MARCO⁺ macrophages and found that their absolute number increased upon cohousing (Fig. 3F). Interestingly, antibiotic treatment did not prevent the increase in MARCO⁺ macrophages (Fig. 3F), suggesting that their increase alone is not responsible for the increase in MZ B cells.

Next, we investigated the mechanism driving the immune homeostasis defect. First, we examined proliferation and found that total B cells from Il10^−/− mice exhibited a slight proliferative advantage over WT when stimulated with LPS, CpG, or anti-CD40, but not anti-IgM (data not shown). However, this did not translate into increased steady state proliferation of Il10^−/− MZ B cells in vivo as measured by BrdU incorporation (Fig. 4A), nor did cohousing increase the number of Ki-67⁺ MZ B cells (Fig. 4B). Interestingly, MZ B cells were proliferating less in the Il10^−/− mice (Fig. 4A, 4B). These findings are consistent with similar expression of the activation markers CD80, CD86, MHC class II, and CD40 by WT and Il10^−/− MZ B cells (data not shown) (57). To determine whether Il10^−/− MZ B cells have a survival advantage, we repeated the in vitro culture using FACS-purified MZ B cells and found that in the absence of stimulation Il10^−/− MZ B cells had a higher level of survival after 3 d in culture (Fig. 4C). Following stimulation with LPS or CpG, both the WT and Il10^−/− MZ B cells expanded with a similar dose-dependent kinetics that was proportional to the percentage of surviving cells without stimulation, which resulted in a significant increase in the percentage of surviving Il10^−/− MZ B cells (Fig. 4C). Although anti-CD40 induced expansion of both WT and Il10^−/− MZ B cells, there was not a significant difference in their survival (Fig. 4C). Anti-IgM only led to a marginal expansion of both populations (Fig. 4C). The survival advantage was marginally evident following a BrdU pulse-chase assay after 30 and 45 d (Fig. 4D). This finding is consistent with the similar expression levels of the cell survival proteins Bcl-2 and Bcl-x_L in MZ B cells from WT and Il10^−/− mice (data not shown). In addition, Il10^−/− MZ B cells were not undergoing reduced levels of apoptosis (data now shown). These cumulative data indicate that the increased numbers of MZ B cells in Il10^−/− mice are not due to increased cell survival or proliferation. Thus, we examined whether B cell differentiation is altered in Il10^−/− mice. In the BM, we found no alteration in the proportion of pro/pre and immature B cells in Il10^−/− mice, indicating that IL-10 is not required for early B cell differentiation (Fig. 4E). To specifically examine whether enhanced MZ lineage differentiation is occurring in the spleen, we measured Notch-2 expression in the T1 subset because it is required to drive precursor maturation into the MZ B cell lineage (58, 59). Using flow cytometry, we found that Notch-2 expression was significantly increased on Il10^−/− T1 cells (Fig. 4F). However, Notch-2 expression was not altered in the mature Fo and MZ subsets (data not shown). We then determined whether the increase in Notch-2 expression led to increased signaling by measuring the transcript levels of its downstream effector molecules (60, 61). Although we saw no difference in mRNA levels of Rbpj and Hes1, we found that Deltex 1 (Dtx1) transcripts were significantly increased in Il10^−/− T1 B cells (Fig. 4G). In addition, the cell surface expression of CD21, another Notch-2–regulated protein (59), was significantly increased on T1 B cells in Il10^−/− mice (Fig. 4H). Finally, we asked whether the expression of the ligand for Notch-2, Delta-like 1, was increased on CD11b⁺ myeloid cells in Il10^−/− mice. Although we did not detect a difference in the expression level of Delta-like 1 on myeloid cells (Fig. 4I) or in the percentage of CD11b⁺ cells that express Delta-like 1 (data not shown), there was a ∼15-fold increase in the absolute number of CD11b⁺Delta-like 1⁺ cells in the Il10^−/− mice (Fig. 4J). When we examined the proliferation of Il10^−/− CD11b⁺ myeloid cells in the spleen in the steady state, we found that in Il10^−/− Jax cohoused mice they underwent a significantly higher rate of proliferation as compared with WT Jax mice (Fig. 4K). Identical results were obtained in WT and Il10^−/− mice from the home colony (Fig. 4L). These cumulative data suggest that the increase in MZ B cells in Il10^−/− mice is due to increased differentiation of the T1 subset into the MZ lineage via upregulation of Notch-2 and increased availability of its ligand Delta-like 1, whereas the increase in myeloid cells is due to increased proliferation.

We next determined whether the increase in MZ B cells and myeloid cells was due to a loss of IL-10 signaling in one or both cell types. To examine the necessity for IL-10 signaling in B cells, we used IL-10R^fl/flCD19Cre⁺ mice and found that upon cohousing these mice exhibited no alteration in the percentage of total (Fig. 5A) or Fo B cells (Fig. 5B) as compared with IL-10R^fl/flLysMCre–negative control mice. Interestingly, the percentage of MZ B cells was significantly decreased (Fig. 5C). In addition, the percentage of CD11b⁺ cells was not altered (Fig. 5D). To render monocytes/macrophages and other myeloid subsets deficient in IL-10R signaling, we used IL-10R^fl/flLysMCre⁺ mice and found that upon cohousing these mice had an increase in both the percentage and absolute number (data not shown) of both MZ B cells and CD11b⁺ cells at a level similar to Il10^−/− mice from our colony (Fig. 5E). WT levels of both populations were observed in the IL10R^fl/flLysMCre⁻ and IL10R^fl/flLysMCre⁺ noncohoused controls (Fig. 5E). These cumulative data indicate that IL-10R signaling in monocytes/macrophages, but not B cells, is required to regulate the steady state homeostasis of MZ B cells and CD11b⁺ myeloid cells in the spleen.

To identify the microbial community associated with the transition from The Jackson Laboratory, we performed Illumina sequencing targeting the V4 region of the 16S rRNA gene, after quality filtering 8,928,469 sequences of length = 90 from 12 animals (four in each group sampled from three different tissues). Those sequences were then clustered into OTUs, groups of sequences with an average percent identity of 97%. A PCoA analysis of these OTUs revealed that Il10^−/− Jax mice upon immediate receipt from Jax showed distinct clustering from Il10^−/− mice housed in our facility and Il10^−/− Jax mice cohoused with our animals (Fig. 6A). The FDR was p < 0.05 for Il10^−/− Jax mice compared with IL10^−/− mice in our facility and cohoused Il10^−/− Jax mice.

To separate the effect of the Jax microbiota from the tissue type from which we sampled (Fig. 6B) and the cage (Fig. 6C) in which the animals were housed in, we used a mixed linear model, as described previously (46), with the following terms: origin, indicating Il10^−/− Jax mice, Il10^−/− Jax mice cohoused with our mice or our Il10^−/− mice; tissue, indicating the source tissue of the sample (cecum, small intestine, or large intestine); and cage, indicating the cage at which the animals were housed. The results from this analysis have been deposited at http://metagenomics.anl.gov/linkin.cgi?project=14383. Among the first five PCoA axes, we observed many highly significant effects for tissue, tissue × origin, and origin. These results are consistent with the Il10^−/− Jax mice having a distinct microbiome that shifts over time and with cohousing in our animal facility. It is interesting to note that the differences between the Jax mice gut microbiome and our facility’s mice gut microbiome (Fig. 6A) appear to be larger than the differences that are attributable to the tissue that is sampled (Fig. 6B). This emphasizes the profound influence of the animal facility environment over the gut microbiome. Analysis of these data utilizing alternative algorithms (for example, utilizing UNIFRAC distance as opposed to Bray-Curtis or assigning OTUs through the open-reference algorithm in QIIME) yielded nearly identical patterns of clustering (Supplemental Fig. 3A–I). Our conclusion of a distinct microbial community in the Il10^−/− Jax mice is therefore unlikely to be the result of particular analysis choice that we made in our pipeline. However, α diversity analysis (Fig. 6D, 6E) showed no significant differences in terms of richness or diversity between the three groups.

To determine the biological impact of differences between Jax mice and mice in our facility, we focused on Helicobacter because of its known association with colitis (11, 18). In addition, our Il10^−/− mice harbor Helicobacter and succumb to spontaneous colitis, whereas the Il10^−/− Jax mice do not. We found that Consensus3 classified to genus level as Helicobacter by both Ribosomal Database Project Classifier (confidence score > 0.8) and uclust was significantly lower in Il10^−/− Jax mice FDR, p < 0.05) (Fig. 7A). To determine whether the presence of Helicobacter was responsible for the immune homeostasis defect in Il10^−/− mice, we inoculated Il10^−/− Jax mice once with H. hepaticus by gavage. We found that Il10^−/− Jax mice gavaged with H. hepaticus had a significant increase in both MZ B cells and CD11b⁺ myeloid cells as compared with Il10^−/− mice that were gavaged broth (Fig. 7B). WT Jax mice gavaged with H. hepaticus did not exhibit an increase in either MZ B cells or CD11b⁺ myeloid cells (Fig. 7B). We next asked whether the immune homeostasis defect was permanent and found that antibiotic treatment of our Il10^−/− mice reversed the increase in MZ B and CD11b⁺ myeloid cells (Fig. 7C). Together, these results indicate that a single bacterial strain upon introduction to the intestinal track can modulate immune cell homeostasis at distal sites in addition to contributing to a disease state in immunocompromised individuals.

We have demonstrated that differences in the microbiota between genetically similar mice can vary profoundly depending on the animal husbandry history, leading to an imbalance in immune cell development and homeostasis at distal sites outside of the intestinal track. Specifically, we showed that MZ B and CD11b⁺ myeloid cell numbers were significantly increased in the spleen of Il10^−/− mice from our colony as compared with WT mice from our colony and Il10^−/− Jax mice. Sequencing of the microbiome revealed that our mice were colonized with a distinct bacterial microbiota, which was transferred to Il10^−/− Jax mice upon cohousing with our mice in conjunction with the acquisition of the homeostasis defect in MZ B and myeloid cells. We also found that the transfer of a single bacterial species was sufficient to drive the imbalance in immune homeostasis seen in our colony. These results have profound implications for the reproducibility of experimental data when animals with deficiencies in immune regulatory mediators are used and emphasize the importance of controlling the composition of the microbial community.

When genetically altered mice are phenotyped for deficiencies in immune cell populations, it is very uncommon for specific B cell subsets to be examined. Thus, even profound differences in the relatively rare population of MZ B cells (4–8% of splenic B cells) can be missed. This is clearly demonstrated in our study, in which we found that total B cells were decreased in Il10^−/− mice from our colony (Fig. 1D). This reduction was largely due to a decrease in Fo B cells (Fig. 1E), and, because Fo B cells are the major B cell subset in the spleen, their decrease was able to offset the increase in MZ B cells in the total B cell pool (Fig. 1E). We found that in Il10^−/− mice from our colony, there were increased levels of resting IgM and Ag-specific IgM following immunization with a T-independent Ag (Fig. 1J). Although we attribute this finding to an increase in MZ B cells, if the increase in MZ B cells had not been observed, an alternate explanation would be that IL-10 directly regulates IgM production. If the same experiment had been performed with Il10^−/− Jax mice, no alteration in IgM levels would have been observed, which would be interpreted as IL-10 playing no role in IgM production. Thus, two laboratories doing the same experiment using the same mouse strain would come to very different conclusions. In support of the importance of phenotyping B cell subsets, in previous studies in mice deficient in the cannabinoid receptor 2, we found that whereas the percentage of Fo B cells was not altered in the spleen, the percentage of MZ B cells was significantly reduced (28). These data indicate that assuming B cell development and homeostasis is normal based on using global B cell markers such as B220 and CD19 is not sufficient when phenotyping genetically modified mice.

Using two different gating strategies (26, 48), we were able to definitively demonstrate that MZ B cell numbers are increased in the spleen of Il10^−/− Jax mice as early as 7 d after cohousing with mice from our colony (Fig. 2B). Using a variety of approaches, we determined that the increase in MZ B cells in Il10^−/− mice was not due to increased levels of proliferation or enhanced cell survival (Fig. 4). However, we cannot rule out a role for either increased expression of B cell survival factors or increased MZ B cell responsiveness to survival factors such as BAFF in Il10^−/− mice. We do not favor the increased survival factor hypothesis because it was previously shown that serum levels of the B cell survival factor BAFF were not increased in Il10^−/− mice (62). Given the requirement for Notch-2 signaling for the differentiation of the MZ lineage (59), our data demonstrating an increase in Notch-2 expression by the T1 precursors (Fig. 4F) suggest that the increase in MZ B cells is due to increased lineage commitment to the MZ lineage. This would mean that fewer progenitors would be available to commit to the Fo lineage, which is what we observed (Fig. 1E). The increase in the mRNA expression of the Notch-2 target gene Dtx1 (Fig. 4G) and surface expression of CD21 (Fig. 4H) indicates that the increase in Notch-2 protein has a functional outcome most likely due to enhanced signaling. Little is known regarding how Notch-2 expression is regulated in MZ B cells, but its function is dependent upon binding to its ligand Delta-like 1 (20, 58). Delta-like 1 expression by splenic nonhematopoietic cells, most likely a stromal cell, is sufficient to drive MZ B cell differentiation (63). The site of MZ commitment is unknown, but it most likely occurs outside of the MZ. Our finding that Delta-like 1 is expressed by monocytes/macrophages in the spleen (Fig. 4I) is consistent with the literature (64). We found that neither the mean fluorescence intensity of Delta-like 1 on CD11b⁺ cells (Fig. 4I) nor their percentages expressing Delta-like 1 (data not shown) were altered in Il10^−/− mice. However, the increase in total myeloid cells in the Il10^−/− spleen resulted in an increase in the total number of Delta-like 1⁺ cells by ∼15-fold (Fig. 4J). Thus, although in the steady state hematopoietic cells do not promote MZ B cell differentiation, we propose that during inflammation the increase in CD11b⁺Delta-like 1⁺ cells in the spleen increases Notch-2 ligand availability, transiently increasing the number of MZ B cells most likely by promoting both their differentiation and retention. This speculative hypothesis is supported by our finding that antibiotic treatment of Il10^−/− mice, presumably dampening the inflammatory response, was sufficient to reduce the number of MZ B cells and CD11b⁺ myeloid cells in the spleen (Fig. 7C). Further support of this hypothesis is the finding that intradermal injection of Borrelia burgdorferi, the causative agent of Lyme disease, led to a transient increase in MZ B cells in mice that correlated with the first appearance of the spirochete in other tissues (65).

Although we favor the hypothesis that the increase in MZ B cells in Il10^−/− mice is due to enhanced differentiation, we cannot rule out the possibility that MZ B cells are being retained within the spleen. Using a Delta-like 1 blocking Ab, it was shown that Notch-2 engagement with its ligand was required for retention of the pre-established long-lived MZ B cell pool within the spleen in adult mice (64). In addition, MARCO⁺ macrophages within the MZ are also thought to contribute to MZ B cell retention (56). However, the increase in MARCO⁺ macrophages in cohoused Il10^−/− mice with and without antibiotics suggests that retention is not likely the major mechanism that drives the increase in MZ B cells (Fig. 3F). An unanswered question is the myeloid cell type that potentially drives MZ B cell differentiation. By flow cytometry phenotyping, we found that both neutrophils and monocytes/macrophages are significantly increased in Il10^−/− Jax mice upon cohousing with mice from our colony (Fig. 3B, 3C). Based on the composition of the myeloid cells, we favor the hypothesis that the Ly6C^lowLy6G⁻ monocytes/macrophages are the required cell. Cohousing of Il10^−/− Jax mice significantly increased the absolute number of neutrophils (Ly6C⁺Ly6G^high), inflammatory macrophages (Ly6C^highLy6G⁻), and the Ly6C^lowLy6G⁻ monocytes/macrophages, the expansion of which was prevented by the inclusion of antibiotics (Fig. 3B). However, when the composition of the myeloid cell population was determined, the percentage of neutrophils within the CD11b⁺ population was identical in the cohoused with and without antibiotic groups (Fig. 3A, 3C). The most dramatic change in composition was in the Ly6C^lowLy6G⁻ population, which was increased by >50% after cohousing. We cannot rule out a role for neutrophils given that Puga et al. (51) recently demonstrated that they are located peri-MZ in the spleen. That elegant study largely conducted in humans further demonstrated that neutrophils provide a helper function to MZ B cells by producing factors promoting isotype class switching, somatic hypermutation, and Ig production (51). Interestingly, a broad group of neutropenic patients was shown to have reduced numbers of circulating MZ B cells with a poorly developed MZ (51). However, the study was not designed to determine whether neutrophils are required for development of the MZ and/or lineage commitment of MZ B cells.

In our kinetic study, upon cohousing MZ B cells and CD11b⁺ cells increased in parallel in the Il10^−/− Jax mice with a significant increase evident as early as day 7 postcohousing (Fig. 2B). From these data, it was not clear whether the expansion of the two populations is linked or are independent events. We investigated this question by conditionally knocking out the IL-10R in B cells or macrophages. For B cells, we used CD19-Cre that achieves close to 100% depletion in splenic B cells (66). For macrophages, we used LysM-Cre that largely targets macrophages with a reported depletion rate of the IL-10R in 90% of monocytes/macrophages (24). Depletion in neutrophils has also been reported (24). Loss of IL-10R signaling in B cells had no impact on B cell or myeloid cell homeostasis (Fig. 5A–D) and supports our findings in global Il10-deficient mice that IL-10 has no major intrinsic role in B cell development, function, or homeostasis. In contrast, loss of IL-10 signaling in myeloid cells was sufficient to drive an increase in both MZ B cells and myeloid cells, further supporting our hypothesis that macrophages and/or neutrophils can promote MZ B cell differentiation (Fig. 5E). The expansion in splenic myeloid cells was due to increased proliferation with up to 90% of the cells proliferating in the absence of IL-10 (Fig. 4K, 4L). Because peripheral neutrophils do not divide and they make up 10% of the myeloid population in the Il10^−/− mice, this indicates that 100% of the monocytes/macrophages were undergoing cell division. This remarkable finding indicates that IL-10 is a master regulator of monocyte/macrophage proliferation. The increase in CD11b⁺ cells in the BM (Fig. 1H) indicates that IL-10 may exert this control even during myeloid cell development. In support of this hypothesis are studies in the Rag1^−/− CD4⁺CD25⁻CD45RB^high T cell transfer model of colitis in which increased numbers of both macrophages and neutrophils were observed in the spleen and BM. This study concluded that the increase in myeloid cells was largely due to an increase in proliferative hematopoietic stem cells and skewed differentiation toward granulocyte–monocyte progenitors, resulting in a reduction in the percentages of both lymphoid and erythroid progenitors (67). Of importance to our study is the finding that H. hepaticus-induced colitis in Rag1^−/− mice also resulted in the accumulation of myeloid cells in the spleen and BM (67). The increase in myeloid cells was shown to be IFN-γ dependent (67). Although we did not specifically examine myeloid cell commitment in the BM, it is possible that the combination of IL-10 deficiency and H. hepaticus also results in skewed myeloid commitment in the BM, leading to increased numbers of neutrophils in the spleen and peripheral blood. It is not clear whether this would be the result of altered production of proinflammatory cytokines, such as IFN-γ, due to the loss of IL-10 immunosuppression. However, we do not favor this hypothesis because it has been shown that serum levels of a large number of cytokines were not increased in Il10^−/− mice as compared with WT (62), which we further confirmed in mice from our colony (Table I). Although there were no signs of systemic inflammation (Table I), levels of a number of proinflammatory cytokines and chemokines were significantly elevated in the colon of Il10^−/− mice (Fig. 1I). This suggests that local colonic inflammation drives the expansion of myeloid cells in Il10^−/− mice.

Our finding that the introduction of H. hepaticus to the microbiome of normal Il10^−/− mice can drive myeloid cell expansion (Fig. 7B) indicates that the introduction of a single bacterial species to immune-compromised individuals has the potential to alter their immune status in essentially every tissue. For infectious disease, overcoming the IL-10–negative feedback loop would be beneficial with increased numbers of myeloid cells better able to clear infections. However, in autoimmunity, such as in multiple sclerosis, or in colitis, in which macrophages are thought to be drivers of disease, increased numbers of myeloid cells that cannot be negatively regulated would be detrimental. Indeed, it was recently shown that loss of IL-10 signaling in macrophages was sufficient to lead to spontaneous colitis in mice (9, 10). The ability of one bacterial species that is seemingly harmless upon introduction to alter the host immune status is likely why it has been difficult to link particular micro-organisms with the onset of autoimmunity and other diseases.

Although the phenomenon of irreproducibility is not new to science or immunology, in this study we clearly show that it can be affected by the introduction of a single bacterial species (Fig. 7B). Another clear example of a single bacterial species altering immune homeostasis is segmented filamentous bacteria (SFB) that when introduced to the gut induced lamina propria CD4 T cells to produce IL-17 (68, 69). Introduction of SFB into germfree K/B×N mice that exhibit attenuated arthritis was sufficient to induce lamina propria Th17 cells and the onset of arthritis (70). Whereas SFB is linked to pathogenesis, both Bacteroides and Clostridium species have been implicated in immunoprotection (71, 72). Although the mechanisms whereby particular bacteria alter immune homeostasis and function are largely unknown, Bacteroides fragiles immune regulatory activity is dependent upon expression of the capsular polysaccharide A (PSA) (72). In experimental autoimmune encephalomyelitis, B. fragiles expressing PSA could not restore susceptibility in antibiotic-treated mice, whereas mice that received PSA⁻ bacteria exhibited experimental autoimmune encephalomyelitis similar to controls (72). The protection was dependent upon Foxp3⁺ T regulatory cells (Treg), suggesting that PSA plays a role in Treg conversion (72). Likewise, Clostridium most likely via the induction of Treg in the lamina propria can provide protection against colitis (71).

In this study, we demonstrated that the introduction of a single bacterial species to the intestinal track altered immune cell homeostasis at a distal site in the spleen. To our knowledge, this is the first time that introduction of a single taxa to the microbiome has been shown to affect MZ B cell homeostasis in the spleen. Moreover, our results have profound implications for scientific reproducibility, as we have demonstrated that differences in the history of animal husbandry can directly impact measurements of the immune system via differences in the microbial community. Failure to explicitly consider these differences will most likely lead to results that are difficult to reproduce. Although we have demonstrated a specific role for an IL-10 regulatory loop in controlling activation of macrophages and subsequent MZ B cell differentiation by the microbiota, similar regulatory pathways most likely exist for other cytokines and inflammatory mediators. MZ B cells are poised to rapidly respond to blood-borne pathogens given their localization within the MZ in the spleen and their partially activated phenotype (57). Thus, a specific increase in MZ B cells would facilitate faster clearance of potential pathogens that transverse the gut endothelium and enter the blood during inflammation. Future work will undoubtedly continue to define the role of gut bacteria in signaling to the immune system via breaches in the gut barrier.

We thank Shelley Morris for help with the animal studies, Nichole Miller and Michael Hayward for technical assistance, Dr. Sridhar Rao for help with the design and implementation of the microbiome studies, and Dr. Monica Mann for the initial suggestion to more carefully examine the Il10^−/− mouse.

The authors have no financial conflicts of interest.