Distinct Microbial Communities Trigger Colitis

Article

Distinct Microbial Communities Trigger Colitis Development upon Intestinal Barrier Damage via Innate or Adaptive Immune Cells Graphical Abstract

Authors Urmi Roy, Eric J.C. Ga´lvez, Aida Iljazovic, ..., Samuel Huber, Richard A. Flavell, Till Strowig

Correspondence [email protected]

In Brief Alterations in the microbiota contribute to the development of intestinal inflammation. Roy et al. demonstrate that distinct intestinal microbial communities cause colitis via opposing effector mechanisms independent or dependent on adaptive immunity. Their findings suggest that personalized immunomodulatory treatment according to distinct microbial signatures may be beneficial for IBD patients.

Highlights d

Gut microbiota composition modulates colitis severity in immunocompetent hosts

d

Colitogenic microbiota drive colitis via innate or adaptive immunity

d

Distinct microbiota members induce pathogenic CD4+ T cells to drive colitis

Roy et al., 2017, Cell Reports 21, 994–1008 October 24, 2017 ª 2017 The Authors. https://doi.org/10.1016/j.celrep.2017.09.097

Cell Reports

Article Distinct Microbial Communities Trigger Colitis Development upon Intestinal Barrier Damage via Innate or Adaptive Immune Cells 1 Marina C. Pils,2 Ulrike Heise,2 _ Urmi Roy,1 Eric J.C. Ga´lvez,1 Aida Iljazovic,1 Till Robin Lesker,1 Adrian J. B1azejewski, Samuel Huber,3 Richard A. Flavell,4,5 and Till Strowig1,6,* 1Microbial

Immune Regulation Research Group, Helmholtz Centre for Infection Research, Braunschweig, Germany Pathology Platform, Helmholtz Centre for Infection Research, Braunschweig, Germany 3I. Medizinische Klinik und Poliklinik, Universita ¨ tsklinikum Hamburg-Eppendorf, Hamburg, Germany 4Department of Immunobiology, Yale University School of Medicine, New Haven, CT, USA 5Howard Hughes Medical Institute, Yale University, New Haven, CT, USA 6Lead Contact *Correspondence: [email protected] https://doi.org/10.1016/j.celrep.2017.09.097 2Mouse

SUMMARY

Inflammatory bowel disease comprises a group of heterogeneous diseases characterized by chronic and relapsing mucosal inflammation. Alterations in microbiota composition have been proposed to contribute to disease development, but no uniform signatures have yet been identified. Here, we compare the ability of a diverse set of microbial communities to exacerbate intestinal inflammation after chemical damage to the intestinal barrier. Strikingly, genetically identical wild-type mice differing only in their microbiota composition varied strongly in their colitis susceptibility. Transfer of distinct colitogenic communities in gene-deficient mice revealed that they triggered disease via opposing pathways either independent or dependent on adaptive immunity, specifically requiring antigen-specific CD4+ T cells. Our data provide evidence for the concept that microbial communities may alter disease susceptibility via different immune pathways despite eventually resulting in similar host pathology. This suggests a potential benefit for personalizing IBD therapies according to patient-specific microbiota signatures. INTRODUCTION Inflammatory bowel disease (IBD) consists of a complex group of incurable inflammatory disorders comprising Crohn’s disease (CD) and ulcerative colitis (UC). Although the etiopathogenesis of IBD development is not fully understood, numerous studies support the hypothesis of IBD as a pathological immune response against microbial and environmental antigens in genetically predisposed individuals (Imhann et al., 2016; Jostins et al., 2012). The relative contribution of innate and adaptive immune cells and various cytokines to the development of IBD has been controversially debated (Neurath, 2014). Nonetheless, an imbalanced interaction between the host immune system and

gut microbiota is thought to play a pivotal role in disease manifestation and maintenance (Cho, 2008; Gevers et al., 2014; Honda and Littman, 2012). Notably, various human disease conditions have been associated with imbalances in the composition of the gut microbiota, so-called dysbiosis; however, whether these changes contribute directly to the development of the disease or reflect an altered physiology of the host remains debated in many instances (Kamada et al., 2013; Ley et al., 2005; Turnbaugh et al., 2008). In various mouse models of IBD, the microbiota and, in some cases, specific members have been shown to influence disease outcome (Saleh and Elson, 2011). Examples of IBD mouse models that lack disease development in the absence of any microbiota are the Il10 / model of colitis and the TNFdeltaARE model of ileitis (Keubler et al., 2015; Schaubeck et al., 2016). Furthermore, disease development in these models is impaired or delayed under specific pathogen-free (SPF) conditions compared with conventional housing conditions, which potentially contain pathogenic bacteria, demonstrating that particular microbiota members or distinct communities only present in conventionally housed mice modulate disease onset (Laukens et al., 2016). Specifically, Enterobacteriaceae in Tbet / Rag2 / mice (Garrett et al., 2010) as well as Bacteroides spp. (Bloom et al., 2011), Helicobacter spp. (Fox et al., 2011), and Bilophila wadsworthia (Devkota et al., 2012) in Il10 / have been shown to enhance intestinal inflammation. The acute dextran sulfate sodium (DSS) colitis model of human UC is considered to be largely dependent on innate immunity (Chassaing et al., 2014). We previously demonstrated that the dysbiotic microbiota of Nlrp6 inflammasome-deficient mice has the ability to directly enhance DSS colitis severity, but the effector mechanism remained unknown (Elinav et al., 2011). Notably, a recent study identified that specific metabolites of this dysbiotic community actively modulate innate immune signaling and, subsequently, the host-microbiota interface (Levy et al., 2015). Subsequently, similar dysbiotic communities with the ability to modulate the severity of DSS colitis have been described in other gene-deficient mice (Couturier-Maillard et al., 2013; Hu et al., 2015; Roberts et al., 2014). However, it remains to be examined whether different colitogenic communities

994 Cell Reports 21, 994–1008, October 24, 2017 ª 2017 The Authors. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

trigger intestinal pathologies via shared or distinct immune pathways. This knowledge could potentially explain the variable roles that have been suggested for various immune effectors and pathways for IBD pathogenesis. In the present study, we have characterized the susceptibility of mouse lines differing only in their microbiota composition toward DSS colitis. Besides the dysbiotic community (DysN6) from Nlrp6 / mice, interestingly, also certain but not all SPF communities demonstrated the ability to cause severe intestinal inflammation in immunocompetent mice. Strikingly, mice displayed different inflammatory responses depending on their intestinal microbiota composition, either characterized by infiltration of neutrophils or the presence of proinflammatory CD4+ T cells. By utilizing gene-deficient mice and antibody-mediated depletion of T cell subsets, we demonstrated that the DysN6 community, but not another colitogenic community, depends on CD4+ T cells to exacerbate DSS colitis severity. Our data identify that specific interactions between colitogenic communities and host immune pathways drive colitis development via distinct mechanisms. RESULTS DSS Colitis Severity Is Strongly Influenced by Microbiota Composition in SPF Mice Distinct differences in microbiota composition between isogenic mice from commercial vendors—e.g., the presence of segmented filamentous bacteria (SFB)—have been found to influence the outcome of disease models in mice (Ivanov et al., 2009). To investigate whether C57BL/6N mice differ in their susceptibility to intestinal inflammation after chemically induced damage to the intestinal barrier, we induced DSS colitis in SPF mouse lines obtained from vendors or bred in-house (Figure 1A; Table S1). The severity of disease was compared within lines of SPF mice and with previously described dysbiotic Nlrp6 / mice that were obtained from the original vivarium and subsequently bred in our animal facility without rederivation (Figure 1B; Figure S1A; Elinav et al., 2011). SPF-1, SPF-5, and SPF-6 mice were characterized by mild colitis with moderate weight loss and no mortality, but SPF-2, SPF-3, and SPF-4 mice as well as dysbiotic Nlrp6 / mice developed a similar severe colitis with profound loss of body mass and mortality (Figure 1B; Figure S1A). Colitis severity in each representative isogenic mouse line from different commercial or in-house sources (SPF-1, SPF-2, SPF-4, SPF-6, and DysN6) was also illustrated by measuring colon shortening and supported by histological characterization of tissue damage (Figures S1C and S1D). Next we investigated fecal microbiota composition before induction of DSS colitis using 16S rRNA gene sequencing. Analysis of b diversity using principle coordinates analysis (PCoA) showed that mice with mild colitis severity (SPF-1, SPF-5, and SPF-6) clustered separately from mice featuring a high severity of colitis (SPF-2, SPF-3, SPF-4, and DysN6). We noted a high similarity between SPF-2, SPF-3 (both from different barriers of the same vendor), and SPF-4 mice as well as between SPF-5 and SPF-6 mice (both from different barriers of the same vendor), respectively, whereas SPF-1 and DysN6 mice clustered distinctly (Figure 1C). A more detailed analysis revealed that species richness (Chao index) was lower in SPF-1 mice but that the complexity of the community structure (Shannon index) was not significantly different

between mouse lines (Figure S1B). Global changes in the composition of microbiota have been associated with IBD (Gevers et al., 2014), such as a decrease in the level of resident Firmicutes and/or Bacteroides and an overabundance of Proteobacteria (Frank et al., 2007). We observed a significant expansion of Bacteroides over Firmicutes in colitogenic SPF-2, SPF-3, SPF-4, and DysN6 mice compared with SPF-1, SPF-5, and SPF-6 mice (Figure 1D). Overgrowth in Proteobacteria was highest in DysN6 mice, followed by SPF-2, SPF-3, SPF-4, and SPF-5 mice, and was mostly absent in SPF-1 and SPF-6 mice (Figure 1D; Table S2). To exclude the effect of genetic drift in inbred mice from different sources, we performed cohousing experiments with microbiota donor and germ-free recipient mice. We focused on SPF-1 (low susceptibility, higher Firmicutes), SPF-2 (high susceptibility, higher Bacteroides), and DysN6 mice (high susceptibility, higher Bacteroides, and higher Proteobacteria) representing the different colitis outcomes and microbiota compositions. Transfer of the donor microbiota into germ-free (GF) recipient (exGF) mice was confirmed by 16S rRNA gene sequencing (Figure 1E). Upon induction of DSS colitis, exGF mice phenocopied the respective donor mice, supporting that the differences in colitis severity were dependent on the microbiota (Figure 1E). Similar microbiota-driven phenotypes were confirmed for the SPF-5 and SPF-6 communities (data not shown). These data demonstrate that distinct types of microbial communities that are stably maintained in wild-type (WT) mice are able to alter the host’s susceptibility to DSS colitis. Transfer of Colitogenic Microbial Communities into an Immunocompetent Host Induces Distinct Patterns of Host Gene Expression and Alters Colitis Susceptibility Next we investigated whether the degree of colitis severity was also transferable between SPF mice with variable DSS colitis susceptibility, similar to what has been observed for Nlrp6 inflammasome-deficient mice (DysN6) (Elinav et al., 2011). Therefore, we performed cohousing experiments of mice featuring mild colitis (SPF-1) with mice having high colitis severity (SPF-2 and DysN6). Cohousing for 4 weeks resulted in reshaping of the microbiota in SPF-1 mice cohoused with SPF-2 mice (SPF-1 + SPF-2) and DysN6 mice (SPF-1 + DysN6) compared with SPF-1 control mice, respectively (Figure 2A). Moreover, cohousing also transferred colitis susceptibility (Figure 2B; Figure S2A). Because SPF-1 + SPF-2 and SPF-1 + DysN6 mice behaved like SPF-2 and DysN6 mice, we refer to them hereafter as cSPF-2 and cDysN6 (cohoused SPF-2 or DysN6), respectively. A similar transfer of colitis severity was also achieved by cohousing SPF-2 with SPF-6 mice (Figure S2C) and after fecal transplantation (FT) from SPF-2 and DysN6 mice into SPF-1 mice (data not shown). Increased colitis severity in cSPF-2 and cDysN6 mice was also illustrated by enhanced colon shortening and corroborated by histological characterization of tissue damage as well as endoscopy (Figures 2C and 2D; Figure S2B). These data demonstrate that distinct types of microbial communities are able to alter the host’s susceptibility to DSS colitis even in already colonized immunocompetent recipients. Induction of DSS colitis has been shown to alter the composition of the intestinal microbiota (Schwab et al., 2014). To identify whether a shared group of commensals alters their abundance

Cell Reports 21, 994–1008, October 24, 2017 995

Experimental procedure

in-house Nlrp6-/-

Vendor 3

0

2

4 6 Time (d)

8

10

0

2

4 6 Time (d)

8

10

D 214.L11

Microbiota

101.L19 149.L19 106.L19 105.L19 143.L19 104.L19 103.L19 173.L04 142.L19 141.L19 108.L19 102.L19 162.L04 168.L04

148.L19

161.L04

151.L04 138.L19144.L04 233.L08 156.L04 176.L19 174.L19 177.L19 134.L19 173.L19 140.L19 157.L04 131.L19 159.L04 226.L08 145.L04 158.L04 150.L04 230.L08 155.L04 130.L19 136.L19 129.L19 234.L08 139.L19 135.L19 232.L08 149.L04 154.L04 227.L08 175.L19 143.L04 148.L04 225.L08 152.L04 147.L04 160.L04 137.L19 133.L19 153.L04 224.L08 220.L08 166.L04221.L08 231.L08 222.L08 216.L08 228.L08 217.L08 223.L08 229.L08 167.L04 163.L04 165.L04 164.L04 219.L08

a

a

218.L08

a

a

−0.2

a

a

a

−0.4

SPF-1 SPF-2 SPF-3 SPF-4 SPF-5 SPF-6 Nlrp6-/(DysN6)

4 2 0

139.L26 148.L26 194.L04 198.L04 195.L04

8000 4000 200 100 0

SP F1 SP F2 SP FSP 3 F4 SP F5 SP F6 D ys N 6

6

212.L11

193.L04

Firmicutes/Proteobacteria ratio

211.L11 215.L11

213.L11 110.L19 217.L11 115.L19 114.L19 116.L19 113.L19 112.L19

117.L19

Axis.2 [13.8%]

0 0

o + + + o o +

172.L19 171.L19 159.L19

158.L19 157.L19

147.L19 175.L04 174.L04 176.L04 146.L19 100.L19 145.L19 177.L04

70

50

SPF-1 SPF-2 SPF-3 SPF-4 SPF-5 SPF-6 Nlrp6-/(DysN6)

170.L19

153.L19

0.0

80

Firmicutes/Bacteroides ratio SP F1 SP F2 SP F3 SP F4 SP F5 SP F6 D ys N 6

156.L19

154.L19

90

155.L19

152.L19

0.2

100

100

DSS severity

DysN6

SPF-5 SPF-6

150.L19 151.L19

10

7 Time (day)

SPF-1

C

Drinking water

2% DSS

SPF-4

110

Survival (%)

in-house SPF-2 SPF-3

Microbiota

B

Vendor 2

Vendor 1

Body weight (%)

A

199.L04

197.L04 155.L26 138.L26 151.L26152.L26 196.L04 140.L26 137.L26

−0.50

−0.25

0.00

0.25

Axis.1 [23.8%]

a

0.2

a

0.0 −0.2 −0.25 0.00 0.25 Axis.1 [36.9%]

Cohousing Donor ex-GF

S1 vs S2

110 S1 vs Dys

** *** *** *** ** ** *** *** ***

100

Survival (%)

Microbiota SPF-1 SPF-2 DysN6

0.4

Body weight (%)

Axis.2 [33.4%]

E

90 ex-GF SPF-1 ex-GF SPF-2 ex-GF DysN6

80 70

0

2

4 6 Time (d)

8

10

100 50 0

ex-GF SPF-1 ex-GF SPF-2 ex-GF DysN6

0

2

4 6 Time (d)

****

****

8

10

Figure 1. Differences in Microbiota Composition Regulate the Severity of Acute DSS Colitis (A) DSS colitis was induced in SPF WT (SPF-1–SPF-6) and in-house bred dysbiotic Nlrp6 / (DysN6) mice by administering 2% DSS (w/v) for 7 days. Body weight and survival of mice were examined daily for 10 days. (B) Body weight and survival of the mice described in (A). DSS severity is depicted as ‘‘o’’ being mild and ‘‘+’’ being severe. n = 9–21 mice/group. (C and D) Analysis of fecal microbiota composition of the mice described in (A) before DSS colitis induction using 16S rRNA sequencing. Shown is analysis of b-diversity (PCoA) (C) and the ratio of relative abundances between Firmicutes to Bacteroides and Firmicutes to Proteobacteria (D). n = 15–33 mice/group. (E) Germ-free C57BL/6N mice were cohoused with donor SPF WT (SPF-1, SPF-2) and Nlrp6 / (DysN6) mice, followed by induction of DSS colitis. Shown is analysis of b-diversity (PCoA) before and disease severity (body weight and survival) upon induction of DSS colitis. n = 7–8 mice/group. Data are displayed as mean ± SEM from at least two independent experiments. The indicated p values represent unpaired Student’s t test: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. See also Figure S1.

during DSS colitis in SPF-1 mice as well as in cSPF-2 and cDysN6 mice, we compared their fecal microbial communities before and after induction of DSS colitis (day 5). Strikingly, b-diversity analysis (PCoA) as well as an analysis of relative abundances of different bacterial families revealed minor differences between the two time points for each community, respectively (Figure 2E; Figure S2D). Minor alterations included an increase in Verrucomicrobiaceae in cSPF-2 and an increase in abundance of some Bacteroidaceae in cDysN6 (Figure 2F; Figure S2D), but no unified changes were observed between the cSPF-2 and cDysN6 communities despite a similar induction of colitis at this time point. Hence, we hypothesized that colitogenic communities already modulate host immunity before disease induc-

996 Cell Reports 21, 994–1008, October 24, 2017

tion, which, in turn, results in enhancement of colitis severity. Thus, global gene expression in colonic tissues of mice harboring either SPF-1, cSPF-2, or cDysN6 was compared using RNA sequencing (RNA-seq). Interestingly, SPF-1 and cSPF-2 mice clustered together and separately from cDysN6 mice with a distinct gene expression signature (Figure 2G; Figure S2E). Specifically, pathway enrichment analysis showed that many upregulated genes in cDysN6 were involved in T cell and B cell signaling as well as cytokine and chemokine signaling (Figure 2H). In contrast, despite the fact that a similar colitis severity outcome was observed in cDysN6 mice, SPF-2 colonization of SPF-1 mice did not result in significant alterations in the host transcriptome (Figure 2G; Figure S2E). These data together

0.25

a

a

0.00

Cohousing None SPF-2 DysN6

−0.25 −0.25 0.00 0.25 Axis.1 [36.9%]

110 S1 vs S1+S2

C

S1 vs S1+Dys

**** **** **** ** **** **** **** ****

10

100

Colon length (cm)

Axis.2 [33.4%]

Microbiota SPF-1 SPF-2 DysN6

Body weight (%)

B

A

90 SPF-1 SPF-1+SPF-2 SPF-1+DysN6

80 70

0

2

4 6 Time (d)

8

10

cDysN6

SPF-1

8 7 6 SPF-1 cDysN6 cSPF-2

cDysN6 cSPF-2 SPF-1

D

9

cSPF-2 (Histology score)

15 10 5 0 SPF-1 cDysN6 cSPF-2

F

E

d0

0.4 cDysN6

SPF-1 cSPF-2 cDysN6

0.2 0.0

Treatment

-0.2

cSPF-2

Axis.2 [23.5%]

Microbiota

None DSS

d5 Clost ridiales Rum inococcaceae Clostridiales/z_ot hers Bact eroidaceae Pept ococcaceae Ent erococcaceae

Verrucom icrobiae Verrucom icrobiaceae Verrucom icrobia Verrucom icrobiales Odoribact eraceae

-0.4 -0.4

-0.2 0.0 0.2 Axis.1 [34.1%]

Clost ridiales/ot her Paraprevot ellaceae

Rikenellaceae Eubact eriaceae

6.0 4.8 3.6 2.4 1.2 0.0 1.2 2.4 3.6 4.8 6.0 LDA SCORE (log 10)

H

G -2 -1 0 1 2 3

B Cell Receptor Signaling Pathway

Node size (# genes) Node color (p value)

EBV LMP1 signaling

IL-9 Signaling Pathway

Toll-like receptor signaling pathway IL-4 signaling Pathway

IL-2 Signaling Pathway

Cytokines and Inflammatory Response Inflammatory Response Pathway

p < 10

43 genes

p < 10 7

185 genes

p = 1.0

Edge width (% shared genes) 1%

Type II interferon signaling (IFNG)

SPF-1

cSPF-2

cDysN6

Chemokine signaling pathway

T Cell Receptor Signaling Pathway

15

10 genes

Edge color (genes from input) 6

50%

3

100%

0

Macrophage markers

Figure 2. Alteration of Colitis Susceptibility and Distinct Host Responses by Colitogenic Microbiota (A–C) SPF-1 WT mice were cohoused with either SPF-2 WT or DysN6 Nlrp6 / (SPF-1 + DysN6) mice, resulting in SPF-1 + SPF-2 and SPF-1 + DysN6 mice, respectively. (A) Analysis of b-diversity (PCoA) of donor and recipient mice before induction of DSS colitis. n = 5–16 mice/group. (B–D) Acute DSS colitis was induced, and the weight of microbiota recipient mice was monitored for 10 days (B). Colon length was measured 5 days after induction of DSS colitis. Shown is a representative image of excised colons (C). Histological analysis of distal colon was performed 5 days after induction of DSS colitis (D). Representative pictures of H&E-stained colon sections are shown. The scale bars represent 50 mm. n = 5–16 mice/group. (E and F) 16S rRNA sequencing of fecal microbiota from WT SPF-1, cSPF-2, and cDysN6 on day 0 and day 5 of DSS colitis. Shown are analysis of b-diversity (PCoA) (E) and analysis of differentially abundant microbial families in cDysN6 and cSPF-2 mice on day 0 and day 5 of DSS by LEfSe (Kruskal-Wallis test, p < 0.05, LDA 4.0) (F). n = 8–12 mice/group. (G and H) RNA-seq analysis from total colonic tissue of WT mice colonized with SPF-1, cSPF-2, or cDysN6. The heatmap shows quantification of RNA reads (G). Also shown is a pathway analysis based on gene ontology (GO) terms of genes significantly upregulated (2-fold) in cDysN6 mice compared with SPF-1 (H). n = 4 mice/group. Data are displayed as mean ± SEM from at least two independent experiments. The indicated p values represent unpaired Student’s t test (B) and nonparametric Kruskal-Wallis test (C and D): *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. See also Figure S2.


suggest that alteration of the SPF-1 community by colonizing it with colitogenic SPF-2 or DysN6 triggers a very different response at the host transcriptional level. DysN6, but Not SPF-2, Microbiota Depends on Adaptive Immune Cells to Develop Colitis Because transfer of the colitogenic SPF-2 community, unlike the DysN6 community, did not trigger large changes in the host transcriptome in the intestine, we hypothesized that the mere presence of the SPF-2 community may be sufficient to trigger more severe colitis upon damage to the intestinal barrier. Therefore, we assessed disease severity in SPF-1 mice that received FT of the SPF-2 or DysN6 community 2 or 28 days prior to disease induction, respectively. Despite minor but detectable differences in communities of mice receiving FT for 2 or 28 days (Figures 3A and 3C; Figure S3A), brief colonization with the SPF-2 microbiota was sufficient to transfer exacerbated disease severity that was comparable with the result following extended colonization (Figure 3B). In contrast, brief colonization with the DysN6 microbiota did not transfer heightened disease susceptibility (Figure 3D). This inability of the DysN6 microbiota to transfer colitis severity potentially results from incomplete microbiota transfer, a requirement for extended immunomodulation or priming of adaptive immune responses. Comparison of the communities in mice receiving the DysN6 FT for 2 or 28 days by linear discriminant analysis (LDA) effect size (LEfSe) analysis revealed very minute differences (Figure 3E), including a higher abundance of SFB as well as Odoribacteriaceae 28 days after the transfer. Notably, despite successful transfer of colitis severity, communities differed stronger in the case of SPF-2 FT (Figure 3F). Interestingly, similar to the DysN6 FT, SFB and Odoribacteriaceae displayed higher abundances 28 days after SPF-2 transfer. This suggests that these bacteria may not be involved in modulating DSS colitis severity. Next, to test whether DysN6 requires priming of adaptive immunity, we compared the severity of DSS colitis between Rag2 / mice harboring either the SPF-1, cSPF-2, or cDysN6 communities. Strikingly, unlike in WT mice, cDysN6 could not enhance colitis severity in Rag2 / mice, as indicated by similar weight loss (Figure 3G) and colon length (Figure S3D) between Rag2 / mice with SPF-1 and cDysN6. In contrast, cSPF-2 also induced severe colitis in Rag2 / mice, as indicated by increased weight loss and mortality (Figure 3H; Figure S3E). Importantly, we confirmed comparable transfer of the donor communities into WT and Rag2 / mice (Figures S3B and S3C). We used permutational multivariate analysis of variance (ADONIS) (Anderson, 2001), considering the variables ‘‘genotype,’’ ‘‘microbiota,’’ and ‘‘cage’’ to evaluate their relative contribution to variability within the groups (Figures S3B and S3C). This analysis revealed that genotype contributed only 3% of variability, whereas microbiota contributed around 60%. Together, these data demonstrate that extended immunomodulation and priming of adaptive immunity by DysN6, but not SPF-2, are required to exacerbate colitis severity. Colitis Development Is Characterized by the Presence of Distinct Immune Signatures in DysN6 and SPF-2 Mice Despite similar disease severity in DysN6 and SPF-2 mice upon DSS colitis induction, our initial results corroborated the hypoth-


esis that distinct colitogenic communities contribute to disease development via different pathways. To further compare intestinal inflammation induced in cDysN6 compared with cSPF-2 mice, the presence of cytokines and chemokines was measured in tissue homogenates on day 7 of DSS colitis. The levels of the pro-inflammatory cytokines interleukin-6 (IL-6) and IL-17A were significantly higher in the distal colon of both cDysN6 and cSPF-2 mice compared with SPF-1 mice (Figure S4A). Compared with SPF-1 and cDysN6, colitis induced in cSPF-2 mice was distinctively characterized by higher levels of interferon g (IFN-g), IL-22, and tumor necrosis factor alpha (TNF-a) as well as lower levels of IL-18, mainly in the distal colon (Figure S4A). No changes were observed in IL-2, IL-4, IL-5, IL-10, and IL-13 between the three microbiota communities (data not shown). In line with our previous observations (Elinav et al., 2011), higher levels of the chemokine CCL5 were detected in the proximal colon of cDysN6 mice compared with SPF-1 and cSPF-2 mice (Figure S4B). In contrast, several other chemokines, including LIX and KC, which recruit and activate neutrophils, along with MIP-1a and MIP-1b, were significantly increased during colitis induced by cSPF-2 (Figure S4B). In parallel, we analyzed lamina propria leukocytes (LPLs) from colonic tissue by flow cytometry to identify whether distinct immune cell subsets are associated with disease induced by SPF-1, cDysN6, and cSPF-2 communities. Indeed, 2-fold increased numbers of CD45+ cells were observed in cDysN6 WT mice compared with SPF-1 and cSPF-2 WT mice both before and 5 days after induction of DSS colitis (Figure 4A). In line with the enhanced levels of neutrophil-attracting chemokines, colitis in cSPF-2 mice was associated with a specific increase in the relative abundance and total number of neutrophils (Figures 4B–4D). However, all SPF-1-, cSPF-2-, and cDysN6-colonized mice did not demonstrate any significant difference in disease outcome while being treated with antibody against Ly6G compared with the isotype control (data not shown). This might indicate a complex interaction among different components of the innate immune system to enhance microbiota-mediated colitis severity. Despite similar frequencies of immune cell subsets of the adaptive immune system (Figures S4C and S4D), significant increases in the numbers of B220+ B cells and CD3+ T cells were observed before and after induction of DSS colitis in cDysN6 mice (Figures 4E and 4F). Increases in the numbers of CD4+ and CD8+ T cells, but not gd T cells, contributed to this difference (Figure 4F). During, but not before DSS colitis, a higher frequency of CD4+ T cells in the colon of cDysN6 and cSPF-2 mice displayed an activated phenotype (Figure S4D). Notably, the absolute numbers of activated CD4+ T cells were only increased in cDysN6 mice, both before and after induction of DSS colitis (Figure 4F). These analyses show that two colitogenic communities trigger distinct inflammatory immune pathways—i.e., enhanced neutrophil recruitment and pathogenic adaptive immune cell responses— during DSS colitis. ab T Cells Trigger DysN6-Mediated, but Not SPF-2Mediated, Colitis Development To investigate which type of pathogenic adaptive immune responses contribute to disease exacerbation after colonization with the DysN6 community, we decided to compare the severity

B

0.0

Flora transfer No transfer 2d transfer 28d transfer

−0.3

110 S1 vs S2 (2d)

Microbiota SPF-1 DysN6

0.1

Body weight (%)

NMDS2

70

0

2

4 6 Time (d)

Flora transfer No transfer 2d transfer 28d transfer

0.0 −0.1

−0.2

28d

S1 vs Dys (28d)

90

0.0 NMDS1

80

2

4 6 Time (d)

0.5

F

2d

WT (SPF-1 vs cDysN6)

90 WT SPF-1 WT cDysN6

0

2

4 6 Time (d)

8

*

***

8

28d

100

***

n.s. **

50

0

10

WT SPF-1 WT DysN6 (2d) WT DysN6 (28d)

0

2

4 6 Time (d)

8

10

2d

4

Tenericut es Anaeroplasm at aceae Anaeroplasm at ales Mollicut es Clost ridiaceae_SFB YS2 YS2_Ot hers 4C0d_2 Cyanobact eria Clost ridiaceae Odoribact eraceae Clostridiales/ot her

5

10

Rag2-/- (SPF-1 vs cDysN6) 110 100 90 80 70

Rag2-/- SPF-1 Rag2-/- cDysN6

0

2

4 6 Time (d)

8

10

4

H Body weight (%)

* **** ********

100

70

2

cSPF-2

2

Odoribact eraceae Clost ridiaceae_SFB Clost ridiaceae 1 0 1 2 3 LDA SCORE (log 10)

110

80

0

**** **** 4 6 8 10 Time (d)

Porphyrom onadaceae

Body weight (%)

Body weight (%)

*

WT SPF-1 WT DysN6 (2d) WT DysN6 (28d)

5

G

10

100

Lachnospiraceae Rum inococcaceae Porphyrom onadaceae Eubact eriaceae

3

0

WT SPF-1 WT SPF-2 (2d) WT SPF-2 (28d)

ns ns ns ns

110 S1 vs Dys (2d)

0 −0.5

4

8

50

D

0.2

cDysN6

WT SPF-1 WT SPF-2 (2d) WT SPF-2 (28d)

80

1.0

0.3

E

90

3

2

1 0 1 LDA SCORE (log 10)

WT (SPF-1 vs cSPF-2)

110

* * *

100

Body weight (%)

C

0.0 0.5 NMDS1

100

**** **** ***

100

−0.6 −0.5

*** **** **** ***

S1 vs S2 (28d) **

Survival (%)

NMDS2

0.3

Body weight (%)

Microbiota SPF-1 SPF-2

Survival (%)

0.6

A

90 80 70

WT SPF-1 WT cSPF-2

0

2

4 6 Time (d)

8

10

2

3

4

Rag2-/- (SPF-1 vs cSPF-2)

110 100 90 80 70

Rag2-/- SPF-1 Rag2-/- cSPF-2

0

2

4 6 Time (d)

8

10

Figure 3. The Adaptive Immune System Is Important for DysN6-Mediated, but Not SPF-2-Mediated, Colitis (A–F) SPF-1 mice were mock-transferred or received a fecal transplant from Nlrp6 / DysN6 or WT SPF-2 donor mice 2 days or 28 days prior to colitis induction, respectively. (A and C) PCoA plot of fecal microbiota composition at steady state of WT SPF-1 mice receiving SPF-2 (A) or DysN6 (C) microbiota for different time periods. n = 3–11 mice/group. (B and D) Body weight and survival of WT SPF-1 mice receiving SPF-2 (B) or DysN6 (D) microbiota for different time periods during DSS colitis. n = 12–15 mice/ group. (E and F) Analysis of differentially abundant microbial families in mice with short (2 days) and prolonged (28 days) exposure to DysN6 (E) and SPF-2 (F) were analyzed by LEfSe (Kruskal-Wallis test, p < 0.05, LDA 2.0) before induction of DSS colitis. n = 3–11 mice/group. (G and H) SPF-1 WT and SPF-1 Rag2 / recipients were cohoused with donor SPF-2 or DysN6. Body weight was monitored upon colitis induction. n = 8–12 mice/ group. Data are displayed as mean ± SEM from at least two independent experiments. The indicated p values represent unpaired Student’s t test: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. See also Figure S3.


SPF-1

SPF-1 cDysN6 cSPF-2

4.0x105

cDysN6

5.56

d0

E

24.7

1.75

cSPF-2 22.6

F

B220+ cell numbers

2.76

6.0x105 21.7

57.0

11.0

62.5

12.3

62.3

53.0

6.12

CD44

35.0

5.0x104

5 0

36.0

3.70

38.0

3.34

6.63

53.7

4.07

d0

d5

CD4+ cell numbers 2.0x105 1.0x105

1.0x10

5

5.0x104 0

0

0

d0

d5

activated CD4+ cell numbers

54.6

0

d5

CD3+ cell numbers

4.0x105

d0 5.90

1.0x105

10

2.0x105

2.0x105

CD4

neutrophil numbers

15

d0 cDysN6

2.27

D

neutrophil 20

d5

SPF-1 19.0

CD8

10.8

6.02

Ly6G

0

CD62L

cSPF-2

in total CD45+ (%)

8.0x10

C

B

CD45+ cell numbers 5

Ly6C

A

CD8+ cell numbers 4.0x10

9.0x104

d0

d5

d5

TCRgd+ cell numbers 2.0x104

4

1.0x104

6.0x104

2.0x104

5.0x103

3.0x104 0

0

0 d0

d5

d0

d5

d0

d5

Figure 4. Colitis Driven by DysN6 and SPF-2 Is Characterized by Distinct Infiltration of Innate and Adaptive Immune Cells (A–F) Colonic lamina propria leukocytes (cLPLs) were isolated from WT mice harboring SPF-1, cDysN6, or cSPF-2 microbiota during the steady state (day 0) and on day 5 after DSS induction and analyzed by fluorescence-activated cell sorting (FACS). (A) Total number of CD45+ cells in cLPLs. (B–D) Analysis of neutrophil infiltration upon DSS induction. Representative FACS plots show frequencies of neutrophils (B). Also shown are frequencies (C) and total numbers (D) of neutrophils on day 0 and day 5 of DSS. (E and F) Analysis of adaptive immune cells upon DSS induction. Representative FACS plots show CD4 and CD8 frequencies gated on CD3+ cells and frequencies of naive and activated CD4+ T cells during the steady state (E). Also shown are total numbers of the indicated immune cell subsets on day 0 and day 5 of DSS (F). Data represent 5–17 mice/group as mean ± SEM from at least two independent experiments. The indicated p values represent nonparametric Kruskal-Wallis test: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. See also Figure S4.

of DSS colitis in WT as well as B or T cell-deficient mice under SPF-1 and cDysN6 conditions. To assure comparable microbiota composition in WT and gene-deficient mice at baseline, all gene-deficient mouse lines were initially rederived into SPF-1 conditions using embryo transfer. To then generate experimental cohorts of WT and gene-deficient mice, the DysN6 microbiota was transferred into SPF-1 recipients using FT or cohousing, and the composition of the fecal microbiota was recorded before induction of disease. To investigate an involvement of T and B cells, we studied SPF-1 and cDysN6 Tcrbd / and muMT / mice, respectively. Comparison of microbiota composition and multi-variate analysis before the start of DSS colitis revealed that mice clustered according to SPF-1 and cDysN6 communities, with genotype contributing only little (i.e., 3%) to differences in microbiome composition (Figures S5A and S5B). Despite a similar transfer of DysN6 into WT and Tcrbd / mice, strikingly, no difference in the severity of DSS colitis was observed between SPF-1 and cDysN6 Tcrbd / mice, as indicated by similar weight loss, unlike in WT mice, which showed microbiota-modulated disease severity (Figure 5A). An involvement of T cells in transferring exacerbated disease severity was further corroborated by analyzing intestinal inflam-


mation using histology (Figure 5B) and endoscopy (Figure S5C) as well as quantifying colon shortening (Figure S5D) of WT and Tcrbd / mice. In contrast to WT mice, deficiency in T cells resulted in no detectable differences in these parameters between SPF-1 and cDysN6 Tcrbd / mice. Transfer of the DysN6 community into SPF-1 muMT / mice resulted in exacerbation of DSS colitis severity, as indicated by significantly enhanced weight loss, colon shortening, and heightened intestinal inflammation compared with SPF-1 muMT / mice, suggesting limited involvement of B cells in colitis exacerbation (Figures 5C and 5D; Figure S5D). To investigate whether T cells are also required for disease exacerbation by the colitogenic SPF-2 microbiota, we introduced the SPF-2 community into SPF-1 WT, Tcrbd / , and muMT / mice. After confirming that the fecal microbiota of mice clustered according to their microbial communities and not by genotype (Figure S5E and S5F), we induced DSS colitis. As expected from the results with Rag2 / mice, deficiency in B or T cells alone did not affect the transfer of heightened disease severity by cSPF-2 (Figure S5G). gd T cells have been implicated in colonic tissue repair (Chen et al., 2002); hence, we next characterized DSS colitis severity in SPF-1 and cDysN6 Tcrd / mice. Notably, characterization of fecal microbiota

A

B

70

0

2

4 6 Time (d)

8

10

90 80 70

Tcrbd-/- SPF-1 Tcrbd-/- cDysN6

0

2

4 6 Time (d)

8

10

C

D

0

2

4 6 Time (d)

8

10

SPF-1

80 70

muMT-/- SPF-1 muMT-/- cDysN6

0

90 80 70

WT SPF-1 WT cDysN6

0

2

4 6 Time (d)

8

10

2

4 6 Time (d)

8

10

F

Tcrd-/- (SPF-1 vs cDysN6) 110

SPF-1

100

muMT-/-

10

90

WT (SPF-1 vs cDysN6) 110

WT

cDysN6

70

WT SPF-1 WT cDysN6

100

100 90 80 70

Tcrd-/- SPF-1 Tcrd-/- cDysN6

0

2

4 6 Time (d)

8

10

5

15

cDysN6

80

E Body weight (%)

Body weight (%)

90

Body weight (%)

Body weight (%)

100

110

10

0 SPF-1 cDysN6 SPF-1 cDysN6 WT Tcrbd-/-

muMT-/- (SPF-1 vs cDysN6)


Histology score

WT SPF-1 WT cDysN6

100

Histology score

80

15

5

0 SPF-1 cDysN6 SPF-1 cDysN6 WT muMT-/WT

Tcrd-/15 Histology score

90

110

SPF-1

100

Tcrbd-/-

cDysN6

110

WT

Tcrbd-/- (SPF-1 vs cDysN6) Body weight (%)

Body weight (%)


10 5

0 SPF-1 cDysN6 SPF-1 cDysN6 WT Tcrd-/-

Figure 5. ab T Cells Are Required for DysN6-Mediated Colitis (A–F) SPF-1 WT and SPF-1 gene-deficient mice were cohoused with a DysN6 donor. Body weight was monitored upon induction of DSS colitis (A, C, and E). Histological analysis of the distal colon was performed 5 days after induction of DSS colitis (B, D, and F). Shown are representative pictures of H&E-stained colon sections. Scale bars represent 50 mm (B, D, and F). (A and B) Body weight (A) and histological analysis (B) of SPF-1 and cDysN6 WT and Tcrbd / mice. n = 6–16 mice/group. (C and D) Body weight (C) and histological analysis (D) of SPF-1 and cDysN6 WT and muMT / mice. n = 5–18 mice/group. (E and F) Body weight (E) and histological analysis (F) of SPF-1 and cDysN6 WT and Tcrd / mice. n = 5–26 mice/group. Data are displayed as mean ± SEM from at least two independent experiments. The indicated p values represent unpaired Student’s t test: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. See also Figure S5.

demonstrated that mice clustered according to SPF-1 and cDysN6 microbiota (Figure S5H). Similar to what we observed in WT mice, transfer of DysN6 microbiota induced in Tcrd / mice enhanced weight loss, colon shortening, and heightened intestinal inflammation (Figures 5E and 5F; Figure S5I). From these results we concluded that T cells are essential for DysN6- but not SPF-2-induced exacerbation of disease. Specifically, our data suggest that modulation of ab T cells by members of the DysN6 community is important. Finally, we exclude a major contribution of B cells to DysN6-mediated colitis. Pathogenic CD4+ T Cells Are Crucial to Induce DysN6Mediated Colitis CD4+ but also CD8+ T cells contribute to different aspects of intestinal homeostasis and inflammation (Honda and Littman, 2012). Hence, we compared DSS colitis severity in SPF-1 and

cDysN6 CD4 / and CD8 / mice. Fecal microbiota of mice clustered again according to SPF-1 and cDysN6 but not according to genotype (Figures S6A and S6B). After DSS induction, CD8 / but not CD4 / mice showed enhanced weight loss and colitis severity after DysN6 transfer, comparable with WT mice (Figures 6A and 6B; Figure S6C). Furthermore, analysis of intestinal inflammation using histology and quantification of colon shortening (Figures 6C and 6D) in SPF-1 and cDysN6 WT and CD4 / mice corroborated that CD4+ but not CD8+ T cells are required for DysN6-induced exacerbation of disease. To evaluate whether CD4+ T cells contribute to enhanced colitis severity by the colitogenic SPF-2 microbiota, we introduced the SPF-2 community into SPF-1 WT and CD4 / mice. Analysis of fecal microbiota of mice confirmed clustering according to microbiota and not by genotype (Figure S6D). Deficiency in CD4+ T cells did not affect the transfer of heightened disease severity


90 80 70

WT SPF-1 WT cDysN6

0

2

4 6 Time (d)

8

100 90 80 70

10

CD8 SPF-1 CD8-/- cDysN6 -/-

0

2

4 6 Time (d)

8

100 90 80 70

10

WT

D

C


WT SPF-1 WT cDysN6

0

2

4 6 Time (d)

100 90 80 70

10

CD4-/- SPF-1 CD4-/- cDysN6

0

2

4 6 Time (d)

8

10

15 Histology score

SPF-1

9 7

cDysN6

5 SPF-1 cDysN6 SPF-1 cDysN6 WT CD4-/-

E

Body weight (%)

100 90 80 70

WT SPF-1 WT cDysN6

0

2

4 6 Time (d)

8

5

α-CD4 Ab (SPF-1 vs cDysN6)

α-CD8 Ab (SPF-1 vs cDysN6)

isotype (SPF-1 vs cDysN6) 110

10

0 SPF-1 cDysN6 SPF-1 cDysN6 WT CD4-/-

10

110

Body weight (%)

Colon length (cm)

8

CD4-/- (SPF-1 vs cDysN6)

110

CD4-/-

11

Body weight (%)

Body weight (%)

100

B

CD8-/- (SPF-1 vs cDysN6)

110

Body weight (%)


110

Body weight (%)

Body weight (%)

A

100 90 80 70

α-CD8 Ab SPF-1 α-CD8 Ab cDysN6 0

2

4 6 Time (d)

8

110 100 90 80 70

10

α-CD4 Ab SPF-1 α-CD4 Ab cDysN6 0

2

4 6 Time (d)

8

10

SPF-1 cDysN6 cSPF-2

F

CD4+ IL-17A+

CD4+ IFN-γ+

CD4+ IL-17A+IFN-γ+

1.0x103

1.0x104

0 d0

0

d5

H

pg/ml

2.0x103

d5

d0

I

95 SPF-1 cDysN6 cSPF-2

90 85

0 3

7 11 Time (d)

15

Colonoscopy score

100

10 5 0 SPF-1 cDysN6 cSPF-2

400

4000

200 0

0 d0

d5

J 15

105

1000

0

0 d0

Colon weight/length(g/cm)

2.0x10

Body weight (%)

8000

2.0x103 4

TNF-α 600

2000

3.0x104 4.0x103

IFN-γ

IL-17A

G

d0

d5

0.08

K

d5

d0

d5

CD4+ cells in colon 3.0x106 2.0x106

0.04 1.0x106 0.00 SPF-1 cDysN6 cSPF-2

0 SPF-1 cDysN6 cSPF-2

Figure 6. Pathogenic CD4+ T Cells Are Crucial to Develop DysN6-Mediated Colitis (A) SPF-1 WT and CD8 / mice were cohoused with DysN6 donor mice, and DSS colitis was induced. (B–D) SPF-1 WT and CD4 / mice were cohoused with DysN6 donor mice, and DSS colitis was induced. Body weight (B) during DSS colitis as well as colon shortening (C) and intestinal inflammation (D) on day 5 of DSS colitis were monitored. Shown are representative pictures of H&E-stained colon sections. Scale bars represent 50 mm (D). n = 5–20 mice/group. (E) SPF-1 and cDysN6 WT mice were injected with isotype control, anti-CD8, or anti- CD4 antibodies on day 1, day 3, and day 7 of DSS colitis, and body weight was compared. n = 10–12 mice/group. (F) Total numbers of CD4+ T cells producing IFN-g and/or IL-17A from isolated cLPLs from IL-17AGFP IFN-gKatushka FoxP3RFP triple reporter mice with different microbiota. n = 6–14 mice/group.

(legend continued on next page)


(Figure S6E), further supporting that SPF-2 drives colitis severity irrespective of T cells. To investigate whether pathogenic CD4+ T cells are required during DysN6-enhanced DSS colitis, we treated SPF-1 and cDysN6 WT mice during DSS colitis with an isotype control antibody or depleting antibodies against CD4 or CD8, respectively. Depletion of CD4-expressing cells, but not CD8-expressing cells, resulted in failure of the DysN6 community to exacerbate DSS colitis (Figure 6E), highlighting that CD4+ T cells are required during the development of DysN6-enhanced colitis. Consequently, we extended our immunophenotyping and analyzed the production of proinflammatory cytokines in CD4+ T cells before and during DSS colitis. We initially focused on IFN-g and IL-17 and, hence, isolated colonic LPLs (cLPLs) from SPF-1, cSPF-2, and cDysN6 IL-17AGFP IFN-gKatushka FoxP3RFP triple reporter mice allowing the in situ monitoring of cytokine production (Gagliani et al., 2015). Transfer of the DysN6 but not SPF-2 community resulted in enhanced numbers of IL-17A and IFN-g single and IL-17A/IFN-g double cytokineproducing CD4+ T cells already before induction of DSS colitis (Figure 6F). After induction of DSS colitis, enhanced numbers of cytokine-producing CD4+ T cells were observed in mice with both colitogenic communities (Figure 6F). In addition to monitoring cytokine production in situ, we isolated cLPLs from SPF-1, cSPF-2, and cDysN6 mice before and after induction of DSS colitis and stimulated them with aCD3 and aCD28 to quantify cytokine production from T cells. Strikingly, T cells from cDysN6 mice produced larger amounts of IL-17A and IFN-g than T cells from SPF-1 and cSPF-2 mice (Figure 6G), both during the steady state and colitis. Notably, TNF-a production after restimulation of T cells was highest during colitis in mice colonized with SPF-2 (Figure 6G). To further investigate the ability of the DysN6 to drive T cellmediated intestinal inflammation, we transferred CD45RB(high) Foxp3-CD4+ T cells from IL-17AGFPIFN-gKatushka FoxP3RFP triple reporter mice into SPF-1, cSPF-2, and cDysN6 Rag2 / mice. After 2 weeks, when mice differed only mildly in their weight loss (Figure 6H), we already observed higher intestinal inflammation, as quantified by colonoscopy in cDysN6 compared with SPF-1 and cSPF-2 recipients (Figure 6I). cDysN6 mice displayed an enhanced colon weight to length ratio and cellular infiltration (Figures 6J and 6K). Specifically, IFN-g+ CD4+ T cells numbers were significantly increased (Figure S6G). Although the numbers of IL-17A+ and double cytokine-producing T cells were also significantly enhanced in cDysN6-colonized mice, their total numbers were much lower than those of IFN-g+ CD4+ T cells (Figure S6G). Taken together, this demonstrates that DysN6 induces pathogenic CD4+ T cells producing high levels of proinflammatory cytokines. Moreover, these microbiota-induced cells are essential to drive disease in two distinct colitis models.

Recognition of Antigens from Dominant Microbial Members by CD4+ T Cells Drives DSS Colitis Severity in DysN6 Mice To investigate whether recognition of microbial antigens by CD4+ T cells is required for exacerbation of colitis in DysN6 mice, DSS colitis was induced in OTII transgenic mice colonized with the SPF-1, cSPF-2, or cDysN6 communities (Figure S7A). Strikingly, cDysN6 OTII mice did not display exacerbation of DSS colitis severity, as indicated by the lack of DysN6 transferinduced changes in body weight loss, intestinal inflammation, and colon shortening (Figures 7A and 7B; Figure S7B). In contrast, cSPF-2 OTII mice were characterized by similar weight loss compared with cSPF-2 WT mice (Figures S7C and S7D). This shows that antigen specificity of CD4+ T cells is required for modulation of disease severity by the DysN6 but not SPF-2 community. Although the DysN6 and SPF-2 communities both trigger severe colitis in the host, the mechanisms of pathogenesis are completely opposing. Consequently, we wanted to understand whether triggering of innate or adaptive immunity by the SPF-2 or DysN6 community dominate over each other when cotransferring them into SPF-1 recipients. Analysis of microbiota composition in recipient mice after cohousing of SPF-1 recipient as well as SPF-2 and DysN6 donor mice showed that the resulting community largely resembled the cDysN6 community (Figure 7C). Accordingly, the host gene expression signatures in cDysN6+SPF-2 mice were similar to the ones observed in cDysN6 mice, including upregulation of genes associated with T cell, B cell, cytokine, and chemokine signaling as well as upregulation of Cd4 (Figure 7D; Figures S7E and S7F). Moreover, mice with cDysN6+SPF-2 displayed high weight loss, intestinal inflammation, colon shortening, and mortality compared with SPF-1 mice (Figure 7E; Figures S7G and S7H). Strikingly, the cDysN6+SPF-2 community failed to induce severe colitis in CD4 / mice (Figure 7F; data not shown). These results demonstrate that the DysN6 community and its pathogenesis mechanism (i.e., the induction of pathogenic antigen-specific CD4+ T cell responses) dominate over SPF-2-induced changes during colitis induction. DISCUSSION Alterations in the microbiome have been hypothesized to contribute to the development of IBD, and patient data suggest the existence of microbial signatures associated with specific disease entities such as CD (Frank et al., 2007; Gevers et al., 2014). However, it remains in question whether those changes are causal and can be potentially used to improve the selection of IBD therapy or, rather, are the result of ongoing inflammation that alters the intestinal microenvironment (Bo¨rnigen et al.,

(G) cLPLs were isolated from SPF-1, cDysN6, and cSPF-2 mice during the steady state and on day 5 after DSS colitis induction and restimulated with a-CD3/ CD28 for 3 days. Cytokine levels were measured from supernatant. n = 5 mice/group. (H–K) T cell transfer colitis was induced by injecting CD4+Foxp3 CD45RB(high) T cells into SPF-1, cDysN6, or cSPF-2 Rag2 / recipients. Body weight was measured after T cell transfer (H). Shown is the colonoscopy severity score on day 14 after transfer (I). Colon weight/length ratio (J) and total numbers of CD4+ cells in cLPLs on day 16 after injection were monitored by FACS (K). n = 7–14 mice/group. Data are displayed as mean ± SEM from at least two independent experiments. The indicated p values represent unpaired Student’s t test (A–E) and nonparametric Kruskal-Wallis test (F–K): *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. See also Figure S6.


A

B

70

WT SPF-1 WT cDysN6

0

2

4 6 Time (d)

8

100

Histology score

90

SPF-1

100

15

110

90 80 70

10

OTII SPF-1 OTII cDysN6

0

2

4 6 Time (d)

8

10

C

cDysN6

110

80

OTII

WT

OTII (SPF-1 vs cDysN6) Body weight (%)

Body weight (%)


D

Macrophage markers Node size (# genes) Node color (p value)

SPF-1

cDysN6

-0.2

cSPF-2

Chemokine signaling pathway

cDysN6+SPF-2

-0.4 -0.6 -1.2

Inflammatory Response Pathway

-0.8

-0.4 0.0 NMDS1

0.4

F 100 Survival (%)

100 90 SPF-1 cDysN6+SPF-2

0

2

4 6 Time (d)

8

10

10 genes

p < 10

91 genes

p < 10 6

831 genes

p = 1.0

Edge width (% shared genes)

13

Edge color (genes from input)

1%

5

50%

2

100%

0

T Cell Receptor Signaling Pathway

WT (SPF-1 vs cDysN6+SPF-2) 110

80

EBV LMP1 signaling

B Cell Receptor Signaling Pathway

Body weight (%)

NMDS2

XPodNet - protein-protein interactions in the podocyte expanded by STRING

-

0.0

Body weight (%)

Type II interferon signaling (IFNG)

Microbiota

0.2

70

5 0 SPF-1 cDysN6 SPF-1 cDysN6 WT OTII

0.4

E

10

50 0

SPF-1 cDysN6+SPF-2

0

2

4 6 8 Time (d)

10

CD4-/- (SPF-1 vs cDysN6+SPF-2) 110 100 90 80

CD4-/- SPF-1 CD4-/- cDysN6+SPF-2

70 0

2

4 6 Time (d)

8

10

Figure 7. CD4+ T Cells Drive DSS Colitis Severity in DysN6 Mice by Recognizing Antigens from Dominant Microbial Members (A and B) SPF-1 WT and OTII transgenic mice were cohoused with DysN6 donor mice, and DSS colitis was induced. Body weight (A) and intestinal inflammation (B) on day 5 of DSS colitis were monitored. Shown are representative pictures of H&E-stained colon sections. Scale bars represent 50 mm. n = 5–10 mice/group. (C–E) SPF-1 WT mice were cohoused with DysN6, SPF-2, or both DysN6 and SPF-2 donor mice, respectively. Shown is b-diversity analysis (PCoA) of fecal microbiota (C) and pathway analysis based on GO terms of genes significantly upregulated (2-fold), as determined by RNA-seq in cDysN6+SPF-2 compared with SPF-1 mice (D). Also shown are body weight loss and survival after induction of DSS colitis (E). (F) SPF-1 CD4 / mice were cohoused with DysN6 and SPF-2 donor mice, and DSS colitis was induced. n = 5–12 mice/group. Data are displayed as mean ± SEM from at least two independent experiments. The indicated p values represent unpaired Student’s t test: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. See also Figure S6.

2013). Here we identified and then characterized distinct types of microbial communities that directly affect the severity of intestinal inflammation in an immunocompetent host. We showed that these communities alter disease susceptibility via opposing mechanisms, one requiring antigen-specific CD4+ T cell responses and the other being mediated by innate immune cells. Healthy human individuals can differ greatly in the composition of their intestinal microbiota, but despite this variability, alterations in the microbiota of patients have been associated with different types of human diseases (Clemente et al., 2012; Falony et al., 2016). Germ-free mice have been fundamental to address the causal role of alterations of the microbiome in mouse models of human disease. Likewise, conventionally housed laboratory mice that feature tremendous differences in the microbiome represent a valuable resource to study the contribution of diverse microbial ecosystems to disease development (Stappenbeck


and Virgin, 2016). In a first effort to reduce experimental variability, the concept of SPF housing conditions was introduced to exclude unwanted influences imposed by the presence of potential pathogens, such as Helicobacter spp. or mouse norovirus, commonly present in wild and conventionally housed mice (Stappenbeck and Virgin, 2016). However, microbiota composition differs greatly between SPF mice from different commercial breeders and academic institutions (Rausch et al., 2016), and those differences influence host responses; e.g., the presence of Th17 cells in SFB-colonized mice (Ivanov et al., 2009) or lowered susceptibility to malaria infection as a consequence of an increased abundance of Lactobacillaceae and Bifidobacterium spp. (Villarino et al., 2016). These observations also make genetically identical SPF mice a versatile experimental model to explore diverse microbial communities and to study host-microbiota interactions in health and disease.

A common feature in IBD, particularly in UC, is impairment of the intestinal barrier, resulting in enhanced exposure to luminal microbes. By employing a mouse model of damage to the intestinal barrier, DSS colitis, we demonstrate that isogenic SPF mice with differences in microbiome composition feature altered susceptibility to intestinal inflammation. Specifically, we noted that transfer of colitogenic communities into mice relatively resistant to induction of DSS colitis is sufficient to alter disease susceptibility even in immunocompetent mice. Upon induction of disease, the DysN6 community as well as the SPF-2 community induced severe colitis compared with the relatively resistant SPF-1 community, but the mechanisms of pathogenesis differed strongly. Inflammation in SPF-2 mice was characterized by high levels of TNF-a and neutrophil-attracting chemokines coinciding with significant higher infiltration of neutrophils into the inflamed tissue. In line with previous findings, DysN6 mice featured higher levels of the chemokine CCL5, known to attract innate and adaptive immune cells carrying CCR1, CCR3, CCR4, and CCR5 (Elinav et al., 2011). Here we identified high infiltration of activated CD4+ T cells in DysN6 mice, hinting toward a potential involvement of these cells in intestinal pathogenesis. The hypothesis of adaptive immune cells being involved in DysN6 mice was further corroborated by the observation that extended colonization with the DysN6 but not SPF-2 community was required to transfer disease susceptibility. Subsequently, we evaluated the effect of the transfer of the two colitogenic communities in mice lacking specific subsets of adaptive immune cells. For these comparisons we employed WT and gene-deficient mice that were embryo-transferred into our vivarium using SPF-1 foster mothers, resulting in a standardized microbiota (E.J.C.G, unpublished data). Moreover, we included cohousing of WT and gene-deficient mice to further reduce microbiota variability within experiments and documented, for all experiments, microbiota composition using 16S rRNA gene sequencing. Using this carefully controlled approach, we observed significant increases in IL-17A and IFN-g secretion by CD4+ T cells during DysN6- and SPF-2 driven colitis. Notably, this is in line with an association of CD4+ T cells and proinflammatory cytokines, including IL-17, IFN-g, and IL-23, with human IBD (Kaser et al., 2010). Strikingly, our experiments demonstrated that CD4+ T cells are only essential to mediate the exacerbation of DSS colitis in DysN6 but not SPF-2 mice. In contrast, despite measurable CD4+ T cell activation during DSS colitis, SPF-2 modulated disease severity independent of adaptive immune cells. T cell receptor (TCR)-mediated recognition of cognate antigens is required for proper T cell function, and recognition of microbial antigens has been suggested to significantly contribute to the development of colitis (Feng et al., 2010). Using OTII transgenic mice, we could show that DysN6-driven but not SPF-2-driven colitis development strongly depended on the presence of antigen-specific CD4+ T cells. The presence of in vivo cytokine-secreting CD4+ T cells before induction of DSS colitis in DysN6 mice suggests that colonic CD4+ T cells already recognize cognate microbial antigens during this phase, similar to what has been observed for SFB-specific CD4+ T cells in the small intestine (Yang et al., 2014). Importantly, antibody-mediated depletion of CD4+ T cells during colitis resulted in failure to transfer enhanced colitis susceptibility. This demonstrated that, to enhance colitis, modu-

lation of the mucosal barrier by CD4+ T cells in the steady state was not sufficient and, rather, required the presence and, presumably, the effector functions of CD4+ T cells during colitis. The distinct property of the DysN6 community to prime and activate pathogenic CD4+ T cell responses was further corroborated using a model for CD4+ T cell-mediated colitis. Specifically, transfer of CD4+ T cells in Rag2 / mice harboring the DysN6 but not the SPF-2 microbiota enhanced intestinal inflammation and cytokine production by CD4+ T cells. Whether these different communities also cause different disease susceptibility or pathogenesis via shared or distinct pathways in other inbred mouse strains or IBD models such as the Il10 / model of colitis and the TNFdeltaARE model of ileitis, remains to be tested (Keubler et al., 2015; Schaubeck et al., 2016). This shows that colitogenic communities exert their pathogenic effects in the same disease model by opposing mechanisms. Detailed characterization of the colitogenic communities using 16S rRNA gene sequencing revealed the varying presence of potential pathobionts such as SFB, Prevotella spp., Helicobacter spp., Enterobacteriaceae, and Verrucomicrobiaceae in DysN6 and SPF-2 mice. SFB have been shown to modulate intestinal T cell immunity and systemic autoimmunity (Ivanov et al., 2009). However, based on their presence in both SPF-2 and DysN6 mice, a role in driving the differential requirement for CD4+ T cells can be excluded. Similarly, members of the genus Prevotella, previously found to be enriched in the colitogenic microbiota of Nlrp6 / mice (Elinav et al., 2011), were present in both colitogenic communities, indicating that they are not involved in regulating the different pathogenicity modes. Helicobacteraceae have been demonstrated to induce the development of colitis in Il10 / mice in cooperation with other members of the microbiota (Keubler et al., 2015). Although Helicobacteraceae were absent in SPF-2 mice, DysN6 mice harbored different members of this family, including H. typhlonius, H. rodentium, and H. muridarum, but did not harbor H. hepaticus. Finally, both Enterobacteriaceae and Verrucomicrobiaceae, specifically Akkermansia muciphilia, bloomed during induction of DSS colitis in SPF-2 mice, but it is being debated whether expansion during disease suggests a contribution to disease development or, rather, a consequence of the ability to utilize inflammation-induced metabolites. In contrast to the ‘‘one microbe one disease’’ model, the concept of dysbiosis, an imbalance of the community, has been proposed for microbiome-mediated modulation of diseases (Petersen and Round, 2014). One characteristic of dysbiotic communities, including those in IBD patients, has been suggested to be an imbalance between Bacteroides, Firmicutes, and Proteobacteria, with an overexpansion of Bacteroides and Proteobacteria over Firmicutes (Frank et al., 2007). Lowered Firmicutes/Bacteroides ratios were noted in all colitogenic communities, including SPF-2 and DysN6, whereas the ratios between Firmicutes and Proteobacteria (F/P) was not consistently different between susceptible and resistant groups. Notably, the F/P ratio was the lowest in DysN6 mice, and according to our data, this is associated with a distinct mode of pathogenicity. Whether, in the cases of the SPF-2 and DysN6 community, specific pathobionts or a general dysbiosis are responsible for driving distinct


pathogenicity requires further investigation because dysbiotic communities have also been reported in other gene-deficient mouse lines (Couturier-Maillard et al., 2013; Hu et al., 2015; Roberts et al., 2014). Finally, it remains debated how these dysbiotic communities arise in gene-deficient mice and how similar they are in regard to their composition in different vivariums, taking into account that a large variability in the composition and function of microbial ecosystems in WT mice already exist. In this study, we employed Nlrp6 / mice with a similar microbiome compared with what has been reported previously (Elinav et al., 2011). However, it remains to be tested whether the pathological mechanism of dysbiotic communities occurring in unrelated lines of Nlrp6 inflammasome-deficient mice causes exacerbated pathology via the same or different mechanisms or does not cause any pathology at all. Along these lines, a recent study has suggested that the intestinal microbiomes of WT and Nlrp6 / mice raised under SPF conditions did not differ in their composition, suggesting that the development of dysbiotic communities reflects a complex interplay between genetic and environmental factors (Mamantopoulos et al., 2017). In summary, our data show how distinct microbial communities drive the development of intestinal inflammation in immunocompetent hosts by modulating opposing arms of the immune system. Our study suggests the concept that triggering of different immune pathways by microbial communities can alter disease susceptibility, eventually resulting in similar host pathophysiology. This implies that personalized immunomodulatory treatment according to distinct microbial signatures may be beneficial for IBD patients.

(Caporaso et al., 2011). Samples were sequenced on an Illumina MiSeq platform (PE250). Filtering of sequences for low-quality reads (q > = 30) and barcode-based binning were performed by using QIIME v1.8.0 (Caporaso et al., 2010). Reads were clustered into operational taxonomical units (OTUs) based on 97% nucleotide identity of the amplicon sequences using UCLUST reference OTU picking, followed by taxonomic classification using the Ribosomal Database Project (RDP) classifier executed at 80% bootstrap confidence cutoff (Edgar, 2010; Wang et al., 2007). Sequences without a matching reference dataset were grouped as de novo using UCLUST. The OTU absolute abundance table and mapping file were used for statistical analyses and data visualization in the R statistical programming environment package phyloseq (McMurdie and Holmes, 2013). To determine bacterial OTUs that explained differences between microbiota settings, the LEfSe method was used (Segata et al., 2011). OTUs with Kruskal-Wallis test < 0.05 and LDA scores > 3.5 were considered informative. Raw data are available in the Sequence Read Archive (SRA): PRJNA407363. DSS-Induced Colitis To induce acute colitis, mice were provided 2% (w/v) DSS (molecular mass = 36–50 kDa, MP Biomedicals) in drinking water for 7 days, followed by 7 days of access to regular drinking water. Daily clinical assessment of DSS-treated animals included body weight loss measurement, stool consistency, and detection of blood in the stool. Experimental samples were collected on days 0, 5, and 7 of DSS treatment. CD45Rbhi Colitis CD4+Foxp3 CD45RB(high) cells were transferred adoptively into Rag2 / mice according to the protocol described by Ostanin et al. (2009). Briefly, splenic lymphocytes were isolated from IL-17AGFP IFN-gKatushka FoxP3RFP triple reporter mice. CD4 enrichment was performed according to the manufacturer’s instructions using CD4 (L3T4) microbeads (Miltenyi Biotec). CD4-enriched cells were then stained with antibodies against CD45RB and CD4. Cells were sorted in a BD FACSAria II cell sorter by gating CD45RB(high), CD4+Foxp3 cells. Antibodies used for staining were anti-CD45RB (C36316A) and anti-CD4 (GK1.5). A total of 500,000 cells were injected intraperitoneally (i.p.) per mouse. Disease development was monitored by weighing animals 3 times a week and performing colonoscopies.

EXPERIMENTAL PROCEDURES Mice Wild-type and all transgenic mice, Rag2 / , Tcrbd / , muMT / , Tcrd / , OTII, CD4 / , CD8 / , and IL-17AGFP IFN-gKatushka FoxP3RFP reporter mice used in the study were on the C57BL/6N background, rederived into SPF-1 microbiota by embryo transfer, and bred at the SPF animal facilities of the Helmholtz Centre for Infection Research (HZI). Nlrp6 / mice were obtained from Yale University and subsequently bred under conventional housing conditions at the HZI without rederivation. Other donor microbiota for different composition were purchased from different commercial vendors (Janvier, Taconic, and Harlan) (Table S1). Germ-free C57BL/6NTac mice were bred in isolators (Getinge) in the germ-free facility of the HZI. All experiments were performed with 10- to 14-week-old age-matched and gender-matched animals. Both male and female animals were used for every experiment to exclude influence of gender.

Statistical Analyses Statistical analysis was performed using the GraphPad Prism program (GraphPad). Data are expressed as mean ± SEM. Differences were analyzed by Student’s t test and ANOVA. The indicated p values represent non-parametric Mann-Whitney U test or Kruskal-Wallis test comparison between groups. p Values % 0.05 were considered significant: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. DATA AND SOFTWARE AVAILABILITY The accession number for the RNA-seq data reported in this paper is Sequence Read Archive (SRA): SRP118483. The accession number for the 16S rRNA gene sequencing data reported in this paper is SRA: SRP119278. SUPPLEMENTAL INFORMATION

DNA Isolation and 16S rRNA Microbial Community Analysis Fresh stool samples of mice were collected and immediately stored at 20 C. DNA was extracted according to established protocols using a method combining mechanical disruption (bead-beating) and phenol/chloroformbased purification (Turnbaugh et al., 2009). Briefly, a sample was suspended in a solution containing 500 mL of extraction buffer (200 mM Tris, 20 mM EDTA, and 200 mM NaCl [pH 8.0]), 200 mL of 20% SDS, 500 mL of phenol: chloroform:isoamyl alcohol (24:24:1), and 100 mL of 0.1 mM zirconia/silica. Samples were homogenized twice with a bead beater (BioSpec) for 2 min. After precipitation of DNA, crude DNA extracts were resuspended in Tris, EDTA (TE) buffer with 100 mg/mL RNase and column-purified to remove PCR inhibitors (BioBasic). Amplification of the V4 region (F515/R806) of the 16S rRNA gene was performed according to previously described protocols


Supplemental Information includes Supplemental Experimental Procedures, seven figures, and two tables and can be found with this article online at https://doi.org/10.1016/j.celrep.2017.09.097. AUTHOR CONTRIBUTIONS U.R. and T.S. designed the experiments and wrote the manuscript with input from co-authors. U.R., A.I., and A.J.B. performed and analyzed the experiments. E.J.C.G. analyzed the RNA sequencing data. E.J.C.G. and T.R.L. designed and supported the analysis of the 16S rRNA sequencing data. M.C.P. and U.H. performed histological evaluation and analysis. R.A.F. contributed essential reagents. S.H., R.A.F., and T.S. supervised the study.

ACKNOWLEDGMENTS We thank the members of the Strowig, Huber, and Flavell laboratories, as well as Nicola Gagliani, for valuable discussions. We thank Achim Gronow, Annett Kluge, the staff of the animal unit, and the genome analytics core facility at the Helmholtz Centre for Infection Research for excellent technical support. The project was supported by the Helmholtz Association (VH-NG-933 to T.S.), by the DFG (STR-1343/1 and STR-1343/2 to T.S.), and the EU (MCCIG618925 to T.S. and StG337251 to S.H.). Received: May 23, 2017 Revised: August 15, 2017 Accepted: September 28, 2017 Published: October 24, 2017 REFERENCES Anderson, M.J. (2001). A new method for non-parametric multivariate analysis of variance. Austral. Ecol. 26, 32–46. Bloom, S.M., Bijanki, V.N., Nava, G.M., Sun, L., Malvin, N.P., Donermeyer, D.L., Dunne, W.M., Jr., Allen, P.M., and Stappenbeck, T.S. (2011). Commensal Bacteroides species induce colitis in host-genotype-specific fashion in a mouse model of inflammatory bowel disease. Cell Host Microbe 9, 390–403. Bo¨rnigen, D., Morgan, X.C., Franzosa, E.A., Ren, B., Xavier, R.J., Garrett, W.S., and Huttenhower, C. (2013). Functional profiling of the gut microbiome in disease-associated inflammation. Genome Med. 5, 65. Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello, E.K., Fierer, N., Penã, A.G., Goodrich, J.K., Gordon, J.I., et al. (2010). QIIME allows analysis of high-throughput community sequencing data. Nat. Methods 7, 335–336. Caporaso, J.G., Lauber, C.L., Walters, W.A., Berg-Lyons, D., Lozupone, C.A., Turnbaugh, P.J., Fierer, N., and Knight, R. (2011). Global patterns of 16S rRNA diversity at a depth of millions of sequences per sample. Proc. Natl. Acad. Sci. USA 108 (Suppl 1), 4516–4522. Chassaing, B., Aitken, J.D., Malleshappa, M., and Vijay-Kumar, M. (2014). Dextran sulfate sodium (DSS)-induced colitis in mice. Curr. Protoc. Immunol. 104, Unit 15.25. Chen, Y., Chou, K., Fuchs, E., Havran, W.L., and Boismenu, R. (2002). Protection of the intestinal mucosa by intraepithelial gamma delta T cells. Proc. Natl. Acad. Sci. USA 99, 14338–14343. Cho, J.H. (2008). The genetics and immunopathogenesis of inflammatory bowel disease. Nat. Rev. Immunol. 8, 458–466. Clemente, J.C., Ursell, L.K., Parfrey, L.W., and Knight, R. (2012). The impact of the gut microbiota on human health: an integrative view. Cell 148, 1258–1270. Couturier-Maillard, A., Secher, T., Rehman, A., Normand, S., De Arcangelis, A., Haesler, R., Huot, L., Grandjean, T., Bressenot, A., Delanoye-Crespin, A., et al. (2013). NOD2-mediated dysbiosis predisposes mice to transmissible colitis and colorectal cancer. J. Clin. Invest. 123, 700–711. Devkota, S., Wang, Y., Musch, M.W., Leone, V., Fehlner-Peach, H., Nadimpalli, A., Antonopoulos, D.A., Jabri, B., and Chang, E.B. (2012). Dietary-fatinduced taurocholic acid promotes pathobiont expansion and colitis in Il10-/- mice. Nature 487, 104–108. Edgar, R.C. (2010). Search and clustering orders of magnitude faster than BLAST. Bioinformatics 26, 2460–2461. Elinav, E., Strowig, T., Kau, A.L., Henao-Mejia, J., Thaiss, C.A., Booth, C.J., Peaper, D.R., Bertin, J., Eisenbarth, S.C., Gordon, J.I., and Flavell, R.A. (2011). NLRP6 inflammasome regulates colonic microbial ecology and risk for colitis. Cell 145, 745–757.

Fox, J.G., Ge, Z., Whary, M.T., Erdman, S.E., and Horwitz, B.H. (2011). Helicobacter hepaticus infection in mice: models for understanding lower bowel inflammation and cancer. Mucosal Immunol. 4, 22–30. Frank, D.N., St Amand, A.L., Feldman, R.A., Boedeker, E.C., Harpaz, N., and Pace, N.R. (2007). Molecular-phylogenetic characterization of microbial community imbalances in human inflammatory bowel diseases. Proc. Natl. Acad. Sci. USA 104, 13780–13785. Gagliani, N., Amezcua Vesely, M.C., Iseppon, A., Brockmann, L., Xu, H., Palm, N.W., de Zoete, M.R., Licona-Limoń, P., Paiva, R.S., Ching, T., et al. (2015). Th17 cells transdifferentiate into regulatory T cells during resolution of inflammation. Nature 523, 221–225. Garrett, W.S., Gallini, C.A., Yatsunenko, T., Michaud, M., DuBois, A., Delaney, M.L., Punit, S., Karlsson, M., Bry, L., Glickman, J.N., et al. (2010). Enterobacteriaceae act in concert with the gut microbiota to induce spontaneous and maternally transmitted colitis. Cell Host Microbe 8, 292–300. Gevers, D., Kugathasan, S., Denson, L.A., Va´zquez-Baeza, Y., Van Treuren, W., Ren, B., Schwager, E., Knights, D., Song, S.J., Yassour, M., et al. (2014). The treatment-naive microbiome in new-onset Crohn’s disease. Cell Host Microbe 15, 382–392. Honda, K., and Littman, D.R. (2012). The microbiome in infectious disease and inflammation. Annu. Rev. Immunol. 30, 759–795. Hu, S., Peng, L., Kwak, Y.T., Tekippe, E.M., Pasare, C., Malter, J.S., Hooper, L.V., and Zaki, M.H. (2015). The DNA Sensor AIM2 Maintains Intestinal Homeostasis via Regulation of Epithelial Antimicrobial Host Defense. Cell Rep. 13, 1922–1936. Imhann, F., Vich Vila, A., Bonder, M.J., Fu, J., Gevers, D., Visschedijk, M.C., Spekhorst, L.M., Alberts, R., Franke, L., van Dullemen, H.M., et al. (2016). Interplay of host genetics and gut microbiota underlying the onset and clinical presentation of inflammatory bowel disease. Gut, Published online October 8, 2016. https://doi.org/10.1136/gutjnl-2016-312135. Ivanov, I.I.I.I., Atarashi, K., Manel, N., Brodie, E.L., Shima, T., Karaoz, U., Wei, D., Goldfarb, K.C., Santee, C.A., Lynch, S.V.S., et al. (2009). Induction of intestinal Th17 cells by segmented filamentous bacteria. Cell 139, 485–498. Jostins, L., Ripke, S., Weersma, R.K., Duerr, R.H., McGovern, D.P., Hui, K.Y., Lee, J.C., Schumm, L.P., Sharma, Y., Anderson, C.A., et al.; International IBD Genetics Consortium (IIBDGC) (2012). Host-microbe interactions have shaped the genetic architecture of inflammatory bowel disease. Nature 491, 119–124. Kamada, N., Seo, S.-U., Chen, G.Y., and Nuń˜ez, G. (2013). Role of the gut microbiota in immunity and inflammatory disease. Nat. Rev. Immunol. 13, 321–335. Kaser, A., Zeissig, S., and Blumberg, R.S. (2010). Inflammatory bowel disease. Annu. Rev. Immunol. 28, 573–621. Keubler, L.M., Buettner, M., Ha¨ger, C., and Bleich, A. (2015). A Multihit Model: Colitis Lessons from the Interleukin-10-deficient Mouse. Inflamm. Bowel Dis. 21, 1967–1975. Laukens, D., Brinkman, B.M., Raes, J., De Vos, M., and Vandenabeele, P. (2016). Heterogeneity of the gut microbiome in mice: guidelines for optimizing experimental design. FEMS Microbiol. Rev. 40, 117–132. Levy, M., Thaiss, C.A., Zeevi, D., Dohnalova´, L., Zilberman-Schapira, G., Mahdi, J.A., David, E., Savidor, A., Korem, T., Herzig, Y., et al. (2015). Microbiota-Modulated Metabolites Shape the Intestinal Microenvironment by Regulating NLRP6 Inflammasome Signaling. Cell 163, 1428–1443. Ley, R.E., Ba¨ckhed, F., Turnbaugh, P., Lozupone, C.A., Knight, R.D., and Gordon, J.I. (2005). Obesity alters gut microbial ecology. Proc. Natl. Acad. Sci. USA 102, 11070–11075.

Falony, G., Joossens, M., Vieira-Silva, S., Wang, J., Darzi, Y., Faust, K., Kurilshikov, A., Bonder, M.J., Valles-Colomer, M., Vandeputte, D., et al. (2016). Population-level analysis of gut microbiome variation. Science 352, 560–564.

Mamantopoulos, M., Ronchi, F., Van Hauwermeiren, F., Vieira-Silva, S., Yilmaz, B., Martens, L., Saeys, Y., Drexler, S.K., Yazdi, A.S., Raes, J., et al. (2017). Nlrp6- and ASC-Dependent Inflammasomes Do Not Shape the Commensal Gut Microbiota Composition. Immunity 47, 339–348.e4.

Feng, T., Wang, L., Schoeb, T.R., Elson, C.O., and Cong, Y. (2010). Microbiota innate stimulation is a prerequisite for T cell spontaneous proliferation and induction of experimental colitis. J. Exp. Med. 207, 1321–1332.

McMurdie, P.J., and Holmes, S. (2013). phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PLoS ONE 8, e61217.


Neurath, M.F. (2014). Cytokines in inflammatory bowel disease. Nat. Rev. Immunol. 14, 329–342.

of murine microbiota activity and interactions with the host during acute inflammation and recovery. ISME J. 8, 1101–1114.

Ostanin, D.V., Bao, J., Koboziev, I., Gray, L., Robinson-Jackson, S.A., Kosloski-Davidson, M., Price, V.H., and Grisham, M.B. (2009). T cell transfer model of chronic colitis: concepts, considerations, and tricks of the trade. Am. J. Physiol. Gastrointest. Liver Physiol. 296, G135–G146.

Segata, N., Izard, J., Waldron, L., Gevers, D., Miropolsky, L., Garrett, W.S., and Huttenhower, C. (2011). Metagenomic biomarker discovery and explanation. Genome Biol. 12, R60.

Petersen, C., and Round, J.L. (2014). Defining dysbiosis and its influence on host immunity and disease. Cell. Microbiol. 16, 1024–1033. Rausch, P., Basic, M., Batra, A., Bischoff, S.C., Blaut, M., Clavel, T., Gla¨sner, €nther, C., et al. (2016). Analysis of facJ., Gopalakrishnan, S., Grassl, G.A., Gu tors contributing to variation in the C57BL/6J fecal microbiota across German animal facilities. Int. J. Med. Microbiol. 306, 343–355. Roberts, M.E., Bishop, J.L., Fan, X., Beer, J.L., Kum, W.W.S., Krebs, D.L., Huang, M., Gill, N., Priatel, J.J., Finlay, B.B., and Harder, K.W. (2014). Lyn deficiency leads to increased microbiota-dependent intestinal inflammation and susceptibility to enteric pathogens. J. Immunol. 193, 5249–5263. Saleh, M., and Elson, C.O. (2011). Experimental inflammatory bowel disease: insights into the host-microbiota dialog. Immunity 34, 293–302. Schaubeck, M., Clavel, T., Calasan, J., Lagkouvardos, I., Haange, S.B., Jehmlich, N., Basic, M., Dupont, A., Hornef, M., Von Bergen, M., et al. (2016). Dysbiotic gut microbiota causes transmissible Crohn’s disease-like ileitis independent of failure in antimicrobial defence. Gut 65, 225–237. Schwab, C., Berry, D., Rauch, I., Rennisch, I., Ramesmayer, J., Hainzl, E., €ller, M., et al. (2014). Longitudinal study Heider, S., Decker, T., Kenner, L., Mu


Stappenbeck, T.S., and Virgin, H.W. (2016). Accounting for reciprocal hostmicrobiome interactions in experimental science. Nature 534, 191–199. Turnbaugh, P.J., Ba¨ckhed, F., Fulton, L., and Gordon, J.I. (2008). Diet-induced obesity is linked to marked but reversible alterations in the mouse distal gut microbiome. Cell Host Microbe 3, 213–223. Turnbaugh, P.J., Ridaura, V.K., Faith, J.J., Rey, F.E., Knight, R., and Gordon, J.I. (2009). The effect of diet on the human gut microbiome: a metagenomic analysis in humanized gnotobiotic mice. Sci. Transl. Med. 1, 6ra14. Villarino, N.F., LeCleir, G.R., Denny, J.E., Dearth, S.P., Harding, C.L., Sloan, S.S., Gribble, J.L., Campagna, S.R., Wilhelm, S.W., and Schmidt, N.W. (2016). Composition of the gut microbiota modulates the severity of malaria. Proc. Natl. Acad. Sci. USA 113, 2235–2240. Wang, Q., Garrity, G.M., Tiedje, J.M., and Cole, J.R. (2007). Naive Bayesian classifier for rapid assignment of rRNA sequences into the new bacterial taxonomy. Appl. Environ. Microbiol. 73, 5261–5267. Yang, Y., Torchinsky, M.B., Gobert, M., Xiong, H., Xu, M., Linehan, J.L., Alonzo, F., Ng, C., Chen, A., Lin, X., et al. (2014). Focused specificity of intestinal TH17 cells towards commensal bacterial antigens. Nature 510, 152–156.

Cell Reports, Volume 21

Supplemental Information

Distinct Microbial Communities Trigger Colitis Development upon Intestinal Barrier Damage via Innate or Adaptive Immune Cells Urmi Roy, Eric J.C. Gálvez, Aida Iljazovic, Till Robin Lesker, Adrian J. B1a_zejewski, Marina C. Pils, Ulrike Heise, Samuel Huber, Richard A. Flavell, and Till Strowig

Supplemental information Supplemental experimental procedures: Microbiota manipulation: Alteration of microbiota composition was conducted by cohousing or fecal transplantation (Thiemann et al., 2017). For cohousing both recipient and donor mice were cohoused at least 4 weeks prior starting experiment. For fecal transplantation donor mice were euthanized, intestinal content was collected in BBL thioglycollate media (BD Bioscience) and homogenized by vortexing. To remove course particle under anaerobic conditions the content was filtered through 70um sterile filter. After centrifugation (10min, 500g, 40C), the pellet containing fecal bacteria was resuspended in BHI medium (Sigma-Alrich). After 2-hr of starving recipient mice were orally gavaged with a total 200ul of fecal bacterial content. Again a 4 week time period was given for a successful establishment of fecal transplanted bacteria (unless mentioned otherwise). Every microbiota manipulation was further confirmed by 16S rRNA sequencing of fecal bacteria. Histology: At day 5 of DSS-treatment colon samples were collected, rolled up to “swiss roles”, fixed in 4% neutrally buffered formaldehyde and embedded in paraffin according to standard histological procedures. Sections of 3µm thickness were stained with hematoxylin-eosin (HE) and evaluated by light microscopy blinded to the experimental groups. The histological scoring used to evaluate the severity of colitis in DSS treated mice microscopically, was adapted from the TJL-score, which was developed for scoring colitis in mice by The Jackson Laboratory (Mähler et al., 1998). The alteration of the score has been previously described (Pils et al., 2010). The colon was divided into a proximal (oral), middle and distal (aboral) section, each of about the same size. The three sections were scored for the general criteria: severity (0-3), ulceration (0-3), oedema (0-3), goblet cell metaplasia (0-3), and area involved (0-3) where score 0 depicted no alteration to score 3 massive alteration in the given parameters. The scores were added up to a total of up to 15 per section and the scores of the three sections to a total of up to 45 per colon sample. Colonoscopy: Colonoscopy was performed using a high-resolution mouse video endoscopic system (‘Coloview’, Carl Storz, Tuttlingen, Germany). The severity of colitis was blindly scored using MEICS (Murine Endoscopic Index of Colitis Severity), which is based on five parameters: granularity of mucosal surface (0-3); vascular pattern (0-3); translucency of the colon mucosa (0-3); visible fibrin (03); and stool consistency (0-3) (Becker et al., 2007). Isolation of colonic lamina propria leukocytes (cLPL) and flow cytometry: To isolate cLPL, density gradient centrifugation using Percoll was done as previously described (Weigmann et al., 2007). In brief, colons were collected during steady state and at d5 of DSS treatment. Fecal content was removed, tissues were opened longitudinally, washed with PBS and then shaken in HBSS containing 2 mM EDTA for 20 min at 37°C. Tissues were cut into small pieces and incubated with digestion solution (DMEM containing 1% fetal bovine serum (FBS), 0.25 mg/ml collagenase D, 0.5 U/ml dispase and 5 µg/ml DNase I) in a shaker for 20 min at 37°C. Digested tissues were filtered through 70uM cell strainer (Falcon) and DMEM + 5% FBS was added to inactivate enzymes. The last two steps were repeated until all tissue was digested. After centrifugation, cells were resuspended in 4 ml of 40% Percoll (GE Healthcare) and overlaid on 4 ml of 80% Percoll. Percoll gradient separation was performed by centrifugation at 450 g for 25 min at 25°C. Cells in the interphase were collected and used as LPL. The collected cells were then suspended in staining buffer containing PBS, 1% FBS and 2 mM EDTA. The following antibodies were used: anti-CD45 (30-F11), anti-CD3 (17A2), anti-B220 (RA3-6B2), anti-CD4 (RM4-5, GK1.5), anti-CD8a (53-6.7), anti-TCR-gamma/delta (GL3), anti-CD44 (IM7), anti-CD62L (MEL-14), anti-MHC class II (M5/114.15.2), anti-CD11b (M1/70), anti-CD11c (N418), anti-Ly6G (IA8), anti-Ly6C (HK1.4) (Biolegend). To distinguish live dead cells AlexaFluor350 NHS Ester (Life Technologies) was used. Flow cytometry analysis was performed using a BD LSR (BD Biosciences) and data were analyzed with FlowJo software (TreeStar Inc.). In-vitro T cell activation of cLPL: For T cell activation freshly isolated cLPL (200,000/well) were cultured in 96-well round-bottom plates in complete culture medium containing soluble, plate-bound, anti-CD3 (1mg/ml) and soluble anti CD28 (5mg/ml) (Biolegend) for 3 days. Supernatant were collected after 3 days for cytokines and chemokines measurement.

1

Collection of colonic tissue homogenate: Colonic tissue homogenate was collected from euthanized mice during steady state and d7 colitis. Colons were excised into proximal and distal colon. Each part was cut longitudinally and cleaned by washing with autoclaved 1x PBS. Parts of proximal or distal colon were weighed and homogenized mechanically using Mini-Beadbeater-96 (Biospec) in NP-40 lysis buffer containing protease inhibitors (Complete Mini EDTA-free, Roche). Protein extracts were centrifuged (10,000 r.p.m. for 5 min at 4°C) and the supernatants were collected as tissue homogenate samples and stored at -80°C. Cytokine detection- ELISA, multiplex and Legendplex: Concentration of IL-18 in the tissue homogenates was measured using the following commercial ELISA kits: IL-18 (MBL) according to manufacturer’s instruction. Different other cytokines and chemokines were measured by using the ProcartaPlex Multiplex Immunoassay (eBioscience) and FACS based Legendplex kit (Biolegend) according to the manufacturer’s instructions. RNA isolation and quantitative PCR: Tissues were preserved in RNAlater solution (Ambion) and subsequently homogenized in Trizol reagent (Invitrogen). One microgram of total RNA was used to generate cDNA by the protocol for first strand cDNA synthesis using RevertAid RT (Thermo Scientific). RealTime-PCR was performed using gene-specific primer sets (Applied Biosystems) of Cd4 primer (F: 5’-TAGCAACTCTAAGGTCTCTAAC and R: 5’-GATAGCTGTGCTCTGAAAAC) and Kapa Sybr Fast qPCR kit (Kapa Biosystems) on a LightCycler 480 instrument (Roche). PCR conditions were 950C for 60 s, followed by 40 cycles of 950C for 3 s and 600C for 30 s. Data were analyzed using the the deltaCt method with hprt (F: CTGGTGAAAAGGACCTCTCG and R: TGAAGTACTCATTATAGTCAAGGGCA) serving as the reference housekeeping gene. RNA-Seq Analysis: Total RNA isolation from distal colonic tissue was performed as described at the RNA isolation section. RNA integrity was measured in a Bioanalyzer (Agilent Technologies, USA) and samples were selected according to RNA Integrity Number (RIN) > 9. Isolation of mRNA was performed with Dynabeads mRNA DIRECT Micro Kit (Ambion, USA) using 1ug of total RNA. Furthermore, cDNA synthesis, fragmentation and sequencing library preparation were done using ScriptSeq v2 RNA-Seq Kit (PCR 15 cycles) (Illumina, USA). Sequencing was performed through Illumia Hi-seq 2000 platform in single end mode for 50bp. Raw data is available in the SRA under accession number: PRJNA407361 We obtained and average of 52,2Mio of reads per sample (n=16). Reads were quality filtered using Trimmomatic with as follow parameters (LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15 MINLEN:35 HEADCROP:3). After quality control reads were aligned to the mouse reference genome (mm10) using STAR. Reads count to each gene was evaluated using HTseq. Normalization and differential expression were quantified using the DEseq2 package. Differential expressed gene networks were analyzed with Consensus Path DB-mouse webserver. Data was visualized using ggplot2 R library. Antibody-mediated depletion: Anti-CD4 (clone GK1.5) and anti-CD8a (2.43) neutralizing antibodies and an isotype control mAb (clone LTF-2) (Bio X Cell) were used. For each antibody, 200ug injections were given intraperitoneally (i.p.) at day -1, day 3 and day 7 of DSS treatment. Isolating cLPL and splenic lymphocytes followed by flow cytometry assessed depletion efficiency.

2

3.0 100

5

0

DysN6

15

SPF-6

SPF-4

SPF-4

D

SPF-2 SPF-1

DysN6 SPF-6 SPF-5 SPF-4 SPF-3 SPF-2 SPF-1

SPF-6

DysN6

SPF-5 SPF-4 SPF-3 SPF-2 SPF-1

DysN6 SPF-4 SPF-3 SPF-2 SPF-6 SPF-5 SPF-1

DysN6 SPF-6 SPF-4 SPF-2 SPF-1

Figure S1

Colon length (cm) 150

Body weight (%)

80

10

Histology score

3.5 90

Alpha Diversity Measure

200

100

DysN6 SPF-6

6

4.0

110

SPF-2

7

250

70

SPF-1

5 2.5

8

4.5

C Shannon Chao1

B

9 300 d8 DSS

A

Figure S1. Differences in microbiota composition of SPF mice regulate severity of acute inflammation upon barrier breakdown. Related to Figure 1 (A) Body weight on d8 of acute DSS colitis of SPF WT (SPF-1 – SPF-6) and dysbiotic Nlrp6-/(DysN6) mice described in Figure 1A . (B) Analysis of fecal microbiota composition in SPF WT (SPF-1 – SPF-6) and dysbiotic Nlrp6-/(DysN6) mice before induction of DSS colitis using 16S rRNA sequencing. Analysis of α-diversity (Chao1 and Shannon). (C-D) Colon length was measured 5 days after induction of DSS colitis of SPF-1, SPF-2, SPF-4, SPF-6 and DysN6 (C). Histological analysis of distal colon was performed 5 days after induction of DSS colitis (D). Representative pictures of H&E-stained colon sections. Bar represents approx. 50µm. Data represent n=6-18 mice/group as mean ± SEM from at least two independent experiments. P values indicated represent a unpaired Student’s t test. *p < 0,05; **p < 0,01; ***p< 0,001; ****p< 0,0001.

3

A

B

50 0

SPF-1

cDysN6

15

cSPF-2 Colonoscopy score

Survival (%)

100 **** SPF-1 **** SPF-1+SPF-2 SPF-1+DysN6

0

2

4 6 me (d)

8

10

Body weight (%)

Axis.2 [13.5%]

Microbiota SPF-6 SPF-2 a

0.0

Cohousing None SPF-2 SPF-6

-0.2

5 0 SPF-1 cDysN6 cSPF-2

C 0.2

10

-0.50 -0.25 0.00 0.25 Axis.1 [53.3%]

110 100 90 SPF-6 SPF-6 (coh SPF-2) SPF-2 (coh SPF-6)

80 70

0

2

4

6 me (d)

8

10

D d0 SPF-1

d5 SPF-1

d0 cDysN6

d5 cDysN6

Family

d5 cSPF-2

d0 cSPF-2

Actinobacteria (A)

80

Coriobacteriaceae

Bacteroidetes (B)

Relative Abundance %

Bacteroidaceae Odoribacteraceae

60

Paraprevotellaceae Porphyromonadaceae Prevotellaceae Rikenellaceae

40

S24-7

Cyanobacteria (C) Deferribacteres (D) Deferribacteraceae

20

Firmicutes (F) Clostridiaceae Clostridiaceae-SFB Dehalobacteriaceae Eubacteriaceae Lachnospiraceae Lactobacillaceae Mogibacteriaceae Ruminococcaceae eraceae z-Others

Proteobacteria (P) Alcaligenaceae Desulfovibrionaceae Helicobacteraceae

Anaeroplasmataceae

Verrucomicrobia (V)

0

Verrucomicrobiaceae A B C DF P

E

V

A B C DF P

2 0

4

Duoxa2 Mzb1

AI747448 Ang4

Igj Nuggc

Duox2

2310034C09Rik Apol9a Mef2b Serpina3f Cd3d Icos Cxcl1 Ly6c2 Cd3g Cd6

Pou2af1 Zbp1 Cd177 Irf4 Gm12250 Cxcl9 SpnGpr110 Gcsam Gm5431 Retnlb Nlrc5 Cxcl10 AI662270 Sh2d1a Havcr1 Ccl8 Cd247 Lat Eaf2 Rhoh A530032D15Rik 5830418P13RikCd5 Gm5547 Ifi44 Vpreb3 Dynlt1b Gpr55 Psmb8 Glycam1Itgal Slamf1 Ptprcap Ly6a Zc3h12a Cd28Gimap7 Folr4 Dhrs9 Mfsd2a Serpina3g Ceacam10 Ccl20Slamf7 C130026I21Rik Ifit1 Lax1 Ceacam12 Klhl6 Il18bp Slamf6 Il21 Ifi47 Zap70Cd8a Oasl2 Ms4a4b Cd79b Slfn1 Isg15Ddah1 Fcrla Cecr2 2310069B03Rik Ptpn22 Gpr18 Upp1 1810065E05Rik Ddx60 rim40 Pou2f2 ap1 Pla2g2a Batf Rsad2 Gm14446 Btla Itgb7 Oas3 Gm6289 Prg2 Foxp3 Apol9b AicdaI830012O16Rik Ikzf1 Ces2f Igtp Ikzf3 Cd2 Map4k1 Il1rl1 Socs3 Gzmb Cma2 Lck Cacna1s Sash3 Rac2 St8sia1 Chst3 Pdcd1 Aqp4 Cd74 Lat2 Herc6 Sept1 Bhlha15 Ido1 Irgm2 Apol10a Prdm1 9330175E14Rik Psmb9 Socs1 AI504432 Cd8b1 Fcrl5Dmp1 Cd69 Gbp2 Coro1a Fcgr4 Ly9 Acap1 Cd40lg Il2rb raf3ip3 Ptk6 rim15 Mir8112 Cd3e Plet1 Casp4 Rltpr Reg3g Ltb Ccl5 Scimp Ccr5 Olfr56 Coch Srgn Btnl2 Cd79a BC147527 Cd19 Gimap1 Cd52 Sp140 Stac2 Siglecg Myo1g Mcpt2 Ms4a1 Pax5 Cd22 Fam78a Usp18 Rasal3 Hvcn1 Ciita Cpa3 Gimap3 Arhgdib Cxcl11 Slc9a3 Cd274 Def6 Sh2d2a Saa3 Il10 Ubash3a Parvg Selplg Stat1 Clca6 Rnase6 Gbp6 Ptpn7Il21r Pla2g5 Erh Gm12185 rat1 Il12rb1Bfsp2 Cd96 Gbp8 Cxcr5 Gpr171 Cst7 Cd300lf reml2 Iigp1 Stat4 Nod2 B2m Mgat4c Gda Gzma Il2rg Bin2 Irgm1 Cd53 Slc13a2 Slc51b ItkGbp2b Rtp4 2010002M12Rik Dok3 Gm14137 BlkMcpt1 2610206C17Rik Pim1Pik3cd Sh2d1b1 Bhlhe41 Ubd Mcpt4 Art2b Gbp5Hmha1 Hcls1 Gbp3 Cyp4f18 Cd37 Laptm5 Cd180Oas2 Il1r2 Slc13a2os Was E330020D12Rik Mx2 Bcl2a1b Ipcef1 Arhgap30 Fer1l4 Ccl12 Il1b Il27ra Plcg2 Pfkfb3 Sla Pglyrp2 Cd72 AW112010 Lta Lipc Clec4e Samsn1 Rgs14Sprr2h Nrg1 Lgals9 Pglyrp1 Fabp6 Gimap4 Gbp7 Fpr2 Mov10 Irf1 Gadd45b Ly6c1 rim10 Ocstamp Fcrl1 Rps21 Grm6 Ccdc88b Lcp2 Btk Dock2 Nxpe3 Rasgrp1 Xcl1 Sit1 March1 A2m Steap4 Spib Scn4a Ptprc Batf2 Napsa Fam169b Fam26f Bst2 Gm4841 A630023P12Rik Ubald2 C2 Ube2l6 Sp110 Prss16 Ifng Cd7 Zfp831 Dnase1l3 Rgs13 Kcna3 Ripk3 Ctsw Cd226 Nfkbiz Spink4 Ly6e ap2 Gpr65 Gm1966 Fgr Snx20 Sell Pdcd1lg2ox2 Stap1Rinl Madcam1 Cxcr6 espa1 Gbp4 rim30a Pyhin1 Fmnl1 Fermt3 Serpina1b Rpl34 Ly86 Cxcr4 Fpr1 Gm15987 Gm15056 5430437J10Rik Ar Grap2 Smpdl3b 5031414D18Rik Cd27 St8sia6 2010109I03Rik Sh2b2 Itgb2 Mybl1 Arhgap25 Parp14 Phf11b Susd3 Cmah Mefv Kmo Gimap6 Lgals3bp Lpxn Ms4a6d Rps2 Lcp1 Pik3ap1 Pirb Il18rap Ost4 Limd2 Gm8369 Cd40 Rpl22 Skap1 Il16 Mx1 faip8l2 Bcat1 2210407C18Rik Ccl7 Serpina3n Oas1a Aif1 Xaf1 Pipox Plxnc1 Wdfy4 Faim3Fam103a1 5430427O19Rik Runx3 Xdh Clec4d Ifit2 Padi2 Cxcl13 ReltPhgdh Rftn1 3930402G23Rik Gpr114 Pif1P2ry10 ac1 Inpp5d Dhx58 Cyfip2 Spta1 Il2ra Dusp2 Phlda1 Syk Glipr1 Sgk2 Rdh9 Irf8 Arhgap4 Fkbp11 Capg Ccr6 C920009B18Rik Abca13 B4galnt1 Ifi27l2b Oas1b Cd14 Clec4n Pik3r5 Gpr132 Csf3r Kynu Cd244 Ccr7 Ppp1r16b Vav1 Dtx3l LpoCd48 Cr2 Lyn Itln1 Havcr2 Arhgap9 Dlk1 fec Stat2 Irf9 Psmb10 Sectm1a Pla2g16 Grap Bach2 Gm12216 Arhgap15 Creld2 Bst1 Parp9 Ncf4 BC021614 Arl5c Card11 Hmgb2 Zbtb16 Slamf8 Ms4a4c Slc5a9 Apobec3 GchfrSlc9a7 Slc17a8 Psme2 Sdf2l1 Basp1 Cdkn3 Apol7c BC035044 AB124611 Ckap4Irf7 Cybb Pde9a Klra2 Junb Ifit3 Phf11d Atp5l Zmynd15 Nckap1l Rilpl2 Hck Lst1Insl6 Klrd1 Oas1g 2010003K11Rik Clca4 Cd200r2 Prex1 Pdcd5 Slc7a11 Fdps Atp12a Haao Cks1bClca2 Clec4a1 Fcer2a Cmpk2 Ccr10Flt3Fcgr1 Cystm1 Rnf24 Gpr141 Slc15a3 rim31

2

Krt20

0 Myt1l Gpr83 Epha5 Snhg11 Cacna1aOlfr78 Gpr124 Itih2 Gucy1a3 Angptl1Nos1 Amy1Srcin1 Ankrd29 Gp6 Adamtsl3 4930412C18Rik Bmp3 Fbxw10 Greb1l Htra4 Cyp2c67 Kcna1 Wnt8b Col25a1 Stx1b Zfp551 Abca6 Rnf32 Abcg8 Zdbf2 Opcml Slc24a2 Adcyap1r1 Sned1 Klk15Iqcg Eci3 Abca8a Col14a1 Insrr Cdh19 Abca8b Pnpla3 Plk3 Nhs DcxSlc38a4 Scn2a1 Cd36 Mfap3l Klb Bco2 Chd5 Pbld1 Slc6a20a Necab1 Hepacam Ccdc11 Nek5 Ndst4 Esr2 Slc35e3 Pde6a Itpripl2 Ggt1 Kcna2 Meox2 Elavl2 Atp2b3 Sphkap Cd209b Rtn1 Agr3 Agbl2 Nudt10 Nr4a1 Rgs1 Adipoq Odf3b Cd209f Mme Cyp2c69 Atp6v0c Sfmbt2 Cyp2a12 Lep Gjc3 Gm6329 4930432K21Rik Nalcn Nr1i3 Limch1 Plcd4 Mettl7a2 Kcnn3 A930011G23Rik Kcnh6 Sec14l4 Flrt1 G6pc2 Asxl3 Ccl24 ymp Atp7b Wbscr17 Siglech Iqch Fxyd4

Car1

Cyp2c55

Mal

AF529169

1

Significant No Yes

Lum

Cyp2b10 Pcdhac2 Crisp1

Slc20a1

Mettl7a2Higd1c

100 10000 Mean abundance

Nos2

Derl3 Ctla4

Gpr15

Hs3st5

1

V A B C DF P

SPF-1 vs cDysN6 Log2 fold change

Log2 fold change

SPF-1 vs cSPF-2 4

V A B C DF P

V A B C DF P

V A B C DF P

Fndc5

100 10000 Mean abundance

Figure S2

V

Figure S2. Alteration of colitis susceptibility and distinct host response driven by colitogenic microbiota. Related to Figure 2 (A-B) Acute colitis was induced in SPF-1 WT mice cohoused with either SPF-2 WT or DysN6 Nlrp6-/(SPF-1 + DysN6) mice resulting in SPF-1 + SPF-2 and SPF-1 + DysN6 mice, respectively. Survival of mice was monitored (A). Representative pictures and colitis severity score by colonoscopy performed on day 6 after colitis induction (B). (C) SPF-6 WT mice were cohoused with SPF-2 WT resulting in SPF-6 (coh SPF-2) and SPF-2 (coh SPF-6) mice. Analysis of β-diversity (PCoA) of non-cohoused SPF-6 and cohoused SPF-6 (coh SPF-2) and SPF-2 (coh SPF-6) mice before induction of DSS colitis. Acute DSS colitis was induced and body weight of mice was monitored for 10 days. (D) 16S rRNA sequencing of fecal microbiota from WT SPF-1, cSPF-2 and cDysN6 at d0 and d5 of DSS colitis was performed. Relative abundances of different microbial families are displayed. (E) RNAseq analysis from total colonic tissue of WT mice colonized with SPF-1, cSPF-2 or cDysN6. DEseq analysis compares significant up/down-regulation of genes (fold change > 2) in different microbiota conditions. Data represent n=4-16 mice/group as mean ± SEM from at least two independent experiments. P values indicated represent a unpaired Student’s t test (A) and nonparametric Kruskal-Wallis test (B) *p < 0.05; **p < 0.01; ***p< 0.001; ****p< 0.0001.

4

A

Family Actinobacteria (A) Proteobacteria (P) Coriobacteriaceae Alcaligenaceae Desulfovibrionaceae Bacteroidetes (B) Helicobacteraceae Bacteroidaceae Odoribacteraceae Paraprevotellaceae Porphyromonadaceae Prevotellaceae Verrucomicrobia (V) Rikenellaceae

cDysN6 (2d)

Relative Abundance %

60

Verrucomicrobiaceae

40

Deferribacteres (D) Deferribacteraceae Firmicutes (F) Clostridiaceae Clostr Dehalobacteriaceae Erysipelotrichaceae Eubacteriaceae Lachnospiraceae Lactobacillaceae Mogibacteriaceae Ruminococcaceae Turicibacteraceae

20

0 A B D F P V

A B D F P V

A B D F P V

A B D F P V

B

A B D F P V

C Genotype B/6N Rag2

Genotype B/6N Rag2

Microbiota

Microbiota

a

a

cDysN6

a

Axis

Axis

a

Microbiota Genotype Cage

R2

P

Microbiota Genotype Cage Axis

Axis

E

Colon length (cm)

D

* * WT F WT cDysN6 Rag2- F Rag2- - cDysN6

6 0 cDysN6 WT

cDysN6 Rag2

0

2

4

6 Time (d)

Figure S3

WT F WT c F 2 Rag2 F Rag2 c F 2

** 0 0

2

4 6 Time (d)

R2

P

Figure S3. Adaptive immune system is important for DysN6 mediated colitis. Related to Figure 3 (A) SPF-1 WT recipient mice were transferred with fecal content from Nlrp6-/- DysN6 or SPF-2 WT donor mice. Relative abundance of different bacterial families in fecal microbiota in mice with short (2d) and prolonged (28d) exposure to DysN6 and SPF-2 before DSS induction . (B-C) SPF-1 WT and Rag2-/- recipients were cohoused with donor DysN6 or SPF-2 mice. Analysis of β-diversity (PCoA) of DysN6 recipients (B) and SPF-2 recipients (C) is shown along with multivariate analysis of variance (ADONIS test) of variables ‘microbiota composition’, ‘genotype’ and ‘cage’. (D-E) SPF-1 and cDysN6 WT and Rag2-/- were sacrificed on d5 of DSS colitis and colons were excised. Colon length was measured (D) and survival of mice with different microbiota are given (E). Data represent n=3-15 mice/group as mean ± SEM from at least two independent experiments. P values indicated represent a unpaired Student’s t test *p < 0,05; **p < 0,01; ***p< 0,001; ****p< 0,0001.

5

A pg/mg tissue

IL-6

IL-22

IL-17A

IFN8

150

3

2

50

1

1

8 4

0.4 0.2 0

0.4 0.2 0

20

PC

B

0

DC

PC

DC

CCL5 pg/mg tissue

20 10 0

PC

PC

LIX

30

DC

200 4

0.2

0

IL-18

TNF-

DC

0 PC

20

15

10

10

4 2 0

2

2

0

PC

0

DC

0

DC

MIP-1b

20

DC

PC

DC

MIP-1a

30

PC

100

PC

KC 60 40 20

PC

4 2 0

DC

SPF-1 cDysN6 cSPF-2

PC

DC

B220+

0 d0

60 40 20 0 d0

d5 CD8+

in total CD3+ (%)

80

30 20 10 0 d0

d5

d5

30 20 10 0 d0

d5

30

SPF-1 cDysN6

20

cSPF-2

10 0 d0

CD44

activated CD4+

CD4+

CD3+

TCR + 40

CD62L CD4

in total CD3+ (%)

in total CD45+ (%)

50

in total CD3+CD4-CD8- (%)

in total live cells (%)

CD45+ 100

CD3

d5

Figure S4

60

in total CD4+ (%)

CD45

in total CD45+ (%)

D

CD8

B220

FSC-H

FSC-H

C

Live/dead

40 20 0 d0

DC

d5

80 60 40 20 0

d0

d5

Figure S4. Colitis driven by DysN6 and SPF-2 are characterized by distinct local infiltration of innate and adaptive immune cells and cytokine profile. Related to Figure 4 (A-B) DSS colitis was induced in WT mice harboring SPF-1, cDysN6 or cSPF-2 communities. Colon samples were collected at day 7 of DSS and divided into two parts as proximal (PC) and distal (DC) colon. Cytokines (A) and chemokines (B) were measured using Multiplex/LEGENDplex kit from colonic tissue homogenates. (C-D) Colonic lamina propria leukocytes (cLPL) were isolated from WT mice harboring SPF-1, cDysN6 or cSPF-2 communities during steady state (d0) and d5 of DSS colitis and analyzed by FACS. Gating strategy of FACS data is displayed (C). Frequencies of different immune cells at d0 and d5 of DSS (D). Data represent n=5-17 mice/group as mean ± SEM from at least two independent experiments. P values indicated represent a nonparametric Kruskal-Wallis test *p < 0,05; **p < 0,01; ***p< 0,001; ****p< 0,0001.

6

A

B

Microbiota

Genotype WT muMT

Axis

Axis

Axis

Genotype WT Tcrbd


Microbiota

R

P


Axis

C

R

P

D Tcrbd engt (cm)

Colonoscopy score

WT

WT

E

Tcrbd

9

WT

Tcrbd

F

Microbiota

Microbiota

Axis

G

Genotype WT muMT R

Axis

Axis

Genotype WT Tcrbd


P

muMT

Tcrbd SPFTcrbd cSPF-

WT SPFWT SPF-

H

Time (d)

I engt (cm)

Microbiota Genotype WT Tcrd

Axis

muMT SPFmuMT cSPF-

Time (d)

Time (d)

Axis


Axis

Tcrbd


muMT

R

9

P

WT

Figure S5

Tcrd

R

P

Figure S5. αβ T cells are required for DysN6 but not SPF-2 mediated colitis. Related to Figure 5 (A-D) SPF-1 WT, Tcrbd-/- and muMT-/- mice were cohoused with DysN6 donor mice. Fecal microbiota composition was analyzed by 16S rRNA sequencing before induction of DSS colitis. Analysis of βdiversity (PCoA) of SPF-1 and cDysN6 WT and Tcrbd-/- mice (A) or muMT-/- mice (B) is shown along with multivariate analysis of variance (ADONIS test) of variables ‘microbiota composition’, ‘genotype’ and ‘cage’. Representative pictures and colitis severity score by colonoscopy performed on day 6 after colitis induction in SPF-1 and cDysN6 WT and Tcrbd-/- mice (C). Colon shortening was measured 5 days after colitis induction in SPF-1 and cDysN6 WT, Tcrbd-/- and muMT-/- mice (D). (E-F) SPF-1 WT, Tcrbd-/- and muMT-/- were cohoused with SPF-2 donor mice. Fecal microbiota composition was analyzed by 16S rRNA sequencing before induction of DSS colitis. Analysis of βdiversity (PCoA) of SPF-1 and cSPF-2 WT and Tcrbd-/- mice (E) or muMT-/- mice (F) is shown along with multivariate analysis of variance (ADONIS test) of variables ‘microbiota composition’, ‘genotype’ and ‘cage’. (G) DSS colitis was induced in SPF-1 and cSPF-2 WT, Tcrbd-/- and muMT-/- mice. Body weight was monitored over 10 days after DSS induction. (H-I) SPF-1 WT and Tcrd-/- mice were cohoused with DysN6 donor mice. Analysis of βdiversity (PCoA) of fecal microbiota from SPF-1 and cDysN6 WT and Tcrd-/- mice (H) is shown along with multivariate analysis of variance (ADONIS test) of variables ‘microbiota composition’, ‘genotype’ and ‘cage’. Colon lengths of WT and Tcrd-/- mice 5 days after colitis induction were measured (I). Data represent n=5-26 mice/group as mean ± SEM from at least two independent experiments. P values indicated represent a unpaired Student’s t test *p < 0,05; **p < 0,01; ***p< 0,001; ****p< 0,0001.

7

A

B

Microbiota

Genotype WT CD4 R

Axis

Axis

Genotype WT CD8

Microbiota Genotype

Microbiota

P

Microbiota Genotype

WT

CD4

CD8

WT

D

P

Axis

Axis

C

R

E

Microbiota

CD4

CD8

CD4 (

WT

Axis

Genotype WT CD4

Microbiota Genotype

R

P

CD4 SPFCD4 cSPF-

WT SPFWT SPF-

Time (d)

Time (d)

Axis

F

-

-

IFN-

SPFcDys

G

-

Figure S6

cSPF-

Figure S6. Pathogenic CD4+ T cells are crucial for DysN6 but not SPF-2 to enhance colitis severity. Related to Figure 6 (A-C) SPF-1 WT, CD8-/- and CD4-/- mice were cohoused with DysN6 donor mice. Fecal microbiota was analyzed by 16S rRNA sequencing before DSS induction and analysis of β-diversity (PCoA) of WT control, CD8-/- mice (A) and CD4-/- mice (B) is shown along with multivariate analysis of variance (ADONIS test) of variables ‘microbiota composition’, ‘genotype’ and ‘cage’. Representative pictures and total score after colonoscopy performed on day 6 after colitis induction in WT, CD8-/- and CD4-/mice harboring SPF-1 and cDysN6 communities (C). (D-E) SPF-1 WT and CD4-/- mice were cohoused with SPF-2 donor mice for 4 weeks. Fecal microbiota was analyzed by 16S rRNA before DSS induction and analysis of β-diversity (PCoA) of WT control, and CD4-/- mice is shown along with multivariate analysis of variance (ADONIS test) of variables ‘microbiota composition’, ‘genotype’ and ‘cage’ (D). Body weight was monitored for 10 days after inducing DSS (E). (F) Colonic lamina propria leukocytes (cLPL) were isolated from SPF-1, cSPF-2 and cDysN6 IL17AGFP IFN-γKatushka FoxP3RFP triple reporter mice during steady state (d0) and d5 of DSS colitis and analyzed by FACS. Representative FACS plots and percentage of different cytokine producing CD4+ T cells are displayed. (G) T cell transfer colitis was induced by injecting CD4+Foxp3-CD45RB(high) T cells into SPF-1, cDysN6 or cSPF-2 Rag2-/- recipients. Mice were sacrificed 16 days after T cell transfer. Lymphocytes were isolated from colon and analyzed by FACS. Infiltration of different cytokine producing cell population in colon is displayed. Data represent n=5-20 mice/group as mean ± SEM from at least two independent experiments. P values indicated represent a unpaired Student’s t test (C,E) and nonparametric Kruskal-Wallis test (F,G) *p < 0,05; **p < 0,01; ***p< 0,001; ****p< 0,0001.

8

B

A

C engt (cm)

Microbiota

Axis

Genotype WT OTII


Axis

R

8 WT SPFWT SPF-

P

8 Time (d) WT

D

OTII

E

Duo

Der Tigit

Igj

Nuggc

P

Tnf Gcsam

Ha

F

GzmbGzma Coch Lat L

Ser

Tnfrsf8 Cd8a

Duo

Spn

Icos Rhoh Ptprcap

Itgal Ubd P

Psmb8

Ccl8 Fcrla Ser Aicda Btla Rltpr T TrL r Igtp Ciita F Ltb P Siglecg T Par Tr Srgn Themis Fo LpoMy Arhgdib Selplg Gbp8 Ar ra Mo WasF Sla Rasgr Cmah Pglyr Sell Ccdc88b Scimp F Spib Gr Fcr Pappa Fgr Pirb Lipc Lta Kmo F Fer Ser Gr Myb Tr T P F TrT rap RinlLpxn L Fcamr To F raNapsa TrTrSer Pipox Tg Slpi Va F Grap T Cybb

L

Batf

Gda Ptprc Pglyr A

Par Xdh

L

f

Fcr Tnfsf8

OTII SPFOTII SPF-

Cnp

a Itgb8

Pb Meo

Itpr Iqch

Flr

Tppp

Mf Mal

Tymp

8

Siglech

Time (d) bundance

F

H

G

8

Figure S7

Significant No Yes

Figure S7. CD4+ T cells drives DSS colitis severity in DysN6 but not SPF-2 mice by recognizing antigens from dominant microbial members. Related to Figure 7 (A-B) SPF-1 WT and OTII transgenic mice were cohoused with DysN6 donor mice for 4 weeks. Analysis of β-diversity (PCoA) of fecal microbiota from SPF-1 and cDysN6 WT and OTII transgenic mice is shown along with multivariate analysis of variance (ADONIS test) of variables ‘microbiota composition’, ‘genotype’ and ‘cage’ (A). Colon length of SPF-1 and cDysN6 WT and OTII transgenic mice after 5 days of DSS induction were measured (B). (C-D) DSS colitis was induced in SPF-1 and cSPF-2 WT and OTII transgenic mice and body weight was monitored for 10 days. (E) RNAseq analysis from total colonic tissue of WT mice harboring different communities. Heatmap shows quantification of RNA reads and DEseq analysis to identify significant up/down-regulation (fold change >2) of genes in SPF-1 and cDysN6+SPF-2 conditions. (F) Relative Cd4 expression in colonic tissue of mice harboring different microbiota. (G-H) DSS colitis was induced in WT mice colonized with SPF-1 and cDysN6+SPF-2 communities. At d5 of DSS mice were sacrificed and colon length was measured (G). Histological analysis of distal colon was performed 5 days after induction of DSS colitis (H). Representative pictures of H&E-stained colon sections. Bar represents approx. 50µm. Data represent n=5-20 mice/group as mean ± SEM from at least two independent experiments. P values indicated represent a unpaired Student’s t test (B-D, G, H) and nonparametric Kruskal-Wallis test (F) *p < 0,05; **p < 0,01; ***p< 0,001; ****p< 0,0001.

9

Supplemental Tables: Table S1: Provider of mice. Related to Figure 1 Table S2: Microbial signatures of experimental mice. Related to Figure 1

10

Table&S1:&Provider&of&mice Mouse&line:

Genotype

SPF51 SPF52 SPF53 SPF54 SPF55 SPF56 DysN6

C57Bl/6NCrl C57Bl/6NRj C57Bl/6NRj C57Bl/6NHsd C57Bl/6NTac C57Bl/6NTac B6.Cg5Nlrp6tm1Flv

Source Breeder Barrier HZI T2 Janvier A1 Janvier C10 Envigo 2 Taconic 1C5HC Taconic EBU401 HZI T1

Commercial: N Y Y Y Y Y N

Table S2: Microbial signatures of experimental mice OTU numbers 681370 4393532 181719 835900 4449525 4372003 4378740 4449518 2212505 4405128 1136443 179575 376862 176868 4426641 170248 186043 135956 228140 263895 179547 180944 1105376 1108599 782953 1141335 4390755 4306262

SPF-1 0.000 0.240 0.000 0.000 0.000 9.702 0.000 13.321 0.000 0.000 1.493 0.848 0.000 0.171 0.000 0.138 24.499 15.271 0.268 0.000 10.430 17.893 0.000 0.000 0.000 0.000 0.366 0.000

SPF-2 0.000 0.000 2.015 1.270 5.752 0.097 13.268 2.520 26.084 0.255 0.463 0.547 0.000 0.281 0.462 0.141 11.461 11.556 0.091 0.000 3.060 5.999 0.268 1.032 0.000 0.000 0.000 1.025

SPF-3 0.000 0.000 2.527 0.742 5.325 0.000 7.778 2.502 27.303 0.569 0.000 1.050 0.423 0.295 1.143 0.060 18.245 8.692 0.000 0.000 4.588 10.792 0.340 0.761 0.000 0.000 0.000 0.373

SPF-4 0.000 0.000 1.269 0.986 3.660 0.000 1.973 1.403 37.901 0.168 0.208 0.907 0.113 0.211 0.375 1.251 11.438 19.410 0.087 0.000 3.145 5.217 0.457 0.280 0.066 0.000 0.000 0.205

SPF-5 0.096 0.156 0.793 0.412 0.000 0.359 8.634 2.806 23.072 0.095 0.111 0.672 0.388 0.243 0.358 1.317 13.987 13.736 0.078 0.000 5.668 10.968 0.328 0.250 1.830 0.000 0.000 2.025

SPF-6 0.000 0.000 2.102 0.000 0.000 0.442 0.082 0.121 20.639 0.000 1.210 0.438 0.000 0.296 0.000 1.216 22.135 7.097 0.136 0.385 9.615 16.115 0.111 0.000 0.000 0.000 0.681 0.086

DysM Kingdom 0.000 Bacteria 0.232 Bacteria 1.817 Bacteria 0.986 Bacteria 4.123 Bacteria 0.000 Bacteria 28.874 Bacteria 1.888 Bacteria 13.292 Bacteria 0.000 Bacteria 1.829 Bacteria 2.184 Bacteria 0.239 Bacteria 0.629 Bacteria 0.000 Bacteria 0.147 Bacteria 12.357 Bacteria 5.508 Bacteria 0.000 Bacteria 0.000 Bacteria 1.810 Bacteria 8.156 Bacteria 0.440 Bacteria 1.098 Bacteria 0.000 Bacteria 7.751 Bacteria 0.000 Bacteria 0.000 Bacteria

Phylum Class Actinobacteria Actinobacteria Actinobacteria Coriobacteriia Bacteroidetes Bacteroidia Bacteroidetes Bacteroidia Bacteroidetes Bacteroidia Bacteroidetes Bacteroidia Bacteroidetes Bacteroidia Bacteroidetes Bacteroidia Bacteroidetes Bacteroidia Cyanobacteria 4C0d-2 Deferribacteres Deferribacteres Firmicutes Clostridia Firmicutes Clostridia Firmicutes Clostridia Firmicutes Erysipelotrichi Firmicutes Clostridia Firmicutes Clostridia Firmicutes Bacilli Firmicutes Clostridia Firmicutes Clostridia Firmicutes Clostridia Firmicutes Clostridia Proteobacteria Betaproteobacteria Proteobacteria Deltaproteobacteria Proteobacteria Gammaproteobacteria Proteobacteria Epsilonproteobacteria Tenericutes Mollicutes Verrucomicrobia Verrucomicrobiae

Order Bifidobacteriales Coriobacteriales Bacteroidales Bacteroidales Bacteroidales Bacteroidales Bacteroidales Bacteroidales Bacteroidales YS2 Deferribacterales Clostridiales Clostridiales Clostridiales Erysipelotrichales Clostridiales Clostridiales Lactobacillales Clostridiales Clostridiales Clostridiales Clostridiales Burkholderiales Desulfovibrionales Enterobacteriales Campylobacterales Anaeroplasmatales Verrucomicrobiales

Family Bifidobacteriaceae Coriobacteriaceae Bacteroidaceae Odoribacteraceae Paraprevotellaceae Porphyromonadaceae Prevotellaceae Rikenellaceae S24-7 z-Others Deferribacteraceae Clostridiaceae Clostridiaceae-SFB Dehalobacteriaceae Erysipelotrichaceae Eubacteriaceae Lachnospiraceae Lactobacillaceae Mogibacteriaceae Peptococcaceae Ruminococcaceae z-Others Alcaligenaceae Desulfovibrionaceae Enterobacteriaceae Helicobacteraceae Anaeroplasmataceae Verrucomicrobiaceae

Supplemental references: Becker, C., Fantini, M.C., and Neurath, M.F. (2007). High resolution colonoscopy in live mice. Nat. Protoc. 1, 2900–2904. Mähler, M., Bristol, I.J., Leiter, E.H., Workman, A.E., Birkenmeier, E.H., Elson, C.O., and Sundberg, J.P. (1998). Differential susceptibility of inbred mouse strains to dextran sulfate sodium-induced colitis. Am. J. Physiol. 274, G544-51. Pils, M.C., Pisano, F., Fasnacht, N., Heinrich, J.-M., Groebe, L., Schippers, A., Rozell, B., Jack, R.S., and Müller, W. (2010). Monocytes/macrophages and/or neutrophils are the target of IL-10 in the LPS endotoxemia model. Eur. J. Immunol. 40, 443–448. Thiemann, S., Smit, N., Roy, U., Lesker, T.R., Gálvez, E.J.C., Helmecke, J., Basic, M., Bleich, A., Goodman, A.L., Kalinke, U., et al. (2017). Enhancement of IFNγ Production by Distinct Commensals Ameliorates Salmonella-Induced Disease. Cell Host Microbe 21, 682–694.e5. Weigmann, B., Tubbe, I., Seidel, D., Nicolaev, A., Becker, C., and Neurath, M.F. (2007). Isolation and subsequent analysis of murine lamina propria mononuclear cells from colonic tissue. Nat. Protoc. 2, 2307–2311.

11