SIGIRR, a Negative Regulator of TLR/IL-1R Signalling Promotes Microbiota Dependent Resistance to Colonization by Enteric Bacterial Pathogens

Enteric bacterial pathogens such as enterohemorrhagic E. coli (EHEC) and Salmonella Typhimurium target the intestinal epithelial cells (IEC) lining the mammalian gastrointestinal tract. Despite expressing innate Toll-like receptors (TLRs), IEC are innately hypo-responsive to most bacterial products. This is thought to prevent maladaptive inflammatory responses against commensal bacteria, but it also limits antimicrobial responses by IEC to invading bacterial pathogens, potentially increasing host susceptibility to infection. One reason for the innate hypo-responsiveness of IEC is their expression of Single Ig IL-1 Related Receptor (SIGIRR), a negative regulator of interleukin (IL)-1 and TLR signaling. To address whether SIGIRR expression and the innate hypo-responsiveness of IEC impacts on enteric host defense, Sigirr deficient (−/−) mice were infected with the EHEC related pathogen Citrobacter rodentium. Sigirr −/− mice responded with accelerated IEC proliferation and strong pro-inflammatory and antimicrobial responses but surprisingly, Sigirr −/− mice proved dramatically more susceptible to infection than wildtype mice. Through haematopoietic transplantation studies, it was determined that SIGIRR expression by non-haematopoietic cells (putative IEC) regulated these responses. Moreover, the exaggerated responses were found to be primarily dependent on IL-1R signaling. Whilst exploring the basis for their susceptibility, Sigirr −/− mice were found to be unusually susceptible to intestinal Salmonella Typhimurium colonization, developing enterocolitis without the typical requirement for antibiotic based removal of competing commensal microbes. Strikingly, the exaggerated antimicrobial responses seen in Sigirr −/− mice were found to cause a rapid and dramatic loss of commensal microbes from the infected intestine. This depletion appears to reduce the ability of the microbiota to compete for space and nutrients (colonization resistance) with the invading pathogens, leaving the intestine highly susceptible to pathogen colonization. Thus, SIGIRR expression by IEC reflects a strategy that sacrifices maximal innate responsiveness by IEC in order to promote commensal microbe based colonization resistance against bacterial pathogens.


Introduction
Citrobacter rodentium is a mouse-specific attaching/effacing (A/E) bacterial pathogen related to the clinically important enteropathogenic Escherichia coli (EPEC) and enterohemorrhagic E. coli (EHEC). C. rodentium has been widely used to define the in vivo virulence strategies employed by A/E pathogens [1,2]. It has also proven a popular model to assess host immune responses against mucosal bacterial pathogens as well as explore how these pathogens compete with intestinal commensal microbes for colonization niches and nutrients [3,4]. We have shown that Myeloid differentiation factor (MyD) 88 dependent innate receptors play a critical role in driving the host response to infection as well as controlling C. rodentium burdens carried by mice [5]. Moreover, we and others have shown that infected Myd88 deficient (2/2) mice suffer severe intestinal epithelial cell (IEC) damage leading to widespread ulcers and necrosis of their colonic tissues. MyD88 signaling thus appears to limit tissue damage during infection by this pathogen, potentially by promoting increased IEC proliferation [5,6].
While the full array of receptors involved in the protective actions of MyD88 in this model is still unclear, TLR4 appears to drive the inflammatory response as well as promote IEC proliferation. We previously showed that infected Tlr4 2/2 mice displayed only modest increases in colonic crypt heights and were attenuated in the induction of the chemokines MCP-1 and MIP2a, as well as the recruitment of neutrophils and macrophages to the infected colon [7]. Despite these defects in their inflammatory responses, Tlr4 2/2 mice were still able to clear C. rodentium infection. In contrast, TLR2 and interleukin (IL)-1R both protect IEC integrity and drive IEC repair [2,7,8,9]. Infected Tlr2 2/2 mice suffer severe ulceration in their mid and distal colons. Surprisingly, their exaggerated pathology was not due to higher C. rodentium burdens, as colonization levels were similar to wildtype mice [10], but instead involved impaired IEC barrier function and reduced production of the cytokine IL-6, leading to exaggerated IEC apoptosis during infection. Similarly, mice lacking the IL-1R were found to be more susceptible to C. rodentium infection, developing exaggerated mucosal damage following challenge [8].
Since many of the MyD88 dependent responses to C. rodentium infection involve changes in IEC function or proliferation, we asked whether they reflected MyD88 dependent signalling within the IEC themselves. While it is clear that MyD88 signalling can occur within IEC, in general IEC are hypo-responsive to most bacterial products and other aspects of inflammatory signalling [11]. Since IEC exist in close proximity to trillions of commensal bacteria, their hypo-responsiveness may be necessary to maintain a mutualistic relationship with the resident microbiota and prevent spontaneous intestinal inflammation [2]. The commensal microbiota, aside from providing nutrients to the host, are important contributors to host defense by promoting resistance to colonization by enteric pathogens [12]. Disruption of the commensal microbiota (through antibiotic treatment, for example) enhances enteric pathogen colonization and subsequent disease in mice [13,14,15]. However, the molecular mechanisms that underlie commensal microbe based resistance to colonization in the mammalian intestine remain poorly defined.
Recent studies suggest that the innate hypo-responsiveness of IEC may in part reflect the actions of Single Ig IL-1 related receptor (SIGIRR) (also called TIR8), a negative regulator of IL-1R and TLR signaling [16,17]. SIGIRR suppresses MyD88 dependent signaling by sequestering IL-1 receptor-associated kinase (IRAK)1 and TNF receptor associated factor (TRAF) 6, both downstream targets of MyD88 [17,18,19]. Our laboratory recently demonstrated that stable overexpression of SIGIRR within IEC lines diminished NF-kB-mediated IL-8 responses to TLR ligands and to IL-1b. In contrast, suppression of SIGIRR by small interfering ribonucleic acid (siRNA) significantly enhanced Caco-2 and HT-29 cell line production of IL-8 in response to these ligands [20]. We have also shown that Sigirr 2/2 mice suffer extensive mucosal damage and increased mortality during dextran sodium sulfate (DSS) induced colitis [21]. The protective role of SIGIRR in this model was shown to reflect the suppression of inflammatory signalling within IEC, such that in the absence of SIGIRR, IEC exhibited exaggerated inflammatory responses driven by the presence of the commensal microbiota. Despite these findings, the specific receptors regulated by SIGIRR were not defined [21]. To address what impact increased innate responsiveness by IEC would have on intestinal host defense against pathogenic bacteria, we tested the effects of SIGIRR on the host response to C. rodentium infection.
As previously reported, the intestines of uninfected Sigirr 2/2 mice were histologically similar to those of WT mice [21,22] but following C. rodentium infection, Sigirr 2/2 mice developed exaggerated inflammatory and antimicrobial responses, along with increased mucosal damage and IEC proliferation. Bone marrow (BM) transplantation experiments confirmed that loss of SIGIRR in the non-BM compartment (putative IEC) controlled the exaggerated colitic and IEC responses. These responses appeared largely independent of TLR2 or TLR4 but were instead dependent on IL-1R signalling. Despite their exaggerated colitic and antimicrobial responses, Sigirr 2/2 mice showed dramatically increased susceptibility to C. rodentium infection, in association with the rapid depletion of their intestinal commensal microbes. This loss of commensals reduced the host's resistance to colonization by C. rodentium as well as to other enteric bacterial pathogens. Thus SIGIRR plays an unexpected role in enteric host defense, promoting commensal microbe-based colonization resistance to pathogens at the expense of limiting antimicrobial, colitic and IEC homeostatic responses.

Results
MyD88 signaling in IEC plays little role in the host response to C. rodentium infection MyD88 signaling plays a critical role in the development of protective host responses to C. rodentium infection, controlling pathogen burdens, driving inflammation and promoting IEC integrity/proliferation [5,6]. As a result, Myd88 2/2 mice are highly susceptible to C. rodentium infection, carrying heavy pathogen burdens and suffering severe mucosal damage and loss of IEC barrier function ( Figure 1A-1D). Since many of these MyD88 dependent protective responses involve changes in IEC function, we tested whether they reflected MyD88 signaling within the IEC. Crossing Myd88 flox mice with mice expressing the cre enzyme under the villin (IEC specific) promoter, we generated IEC-Myd88 2/2 mice that were then infected with C. rodentium along with control MyD88 flox mice. Interestingly, MyD88 signaling within IEC had little impact on the host response to infection. Pathogen burdens in the IEC-Myd88 2/2 mice were similar to those in control mice ( Figure 1A), as were crypt heights

Author Summary
Despite being in close contact with billions of commensal bacteria, the epithelial cells that line the intestine develop very weak innate inflammatory responses to bacterial products. The goal of this study was to explore why these cells respond so poorly, and how increasing their innate responsiveness would impact on host defense against invading bacterial pathogens. We show that a negative regulator of innate signaling called SIGIRR, limits the inflammatory responses of the intestine to bacteria. Following infection by the bacterial pathogen Citrobacter rodentium, the intestines of mice lacking SIGIRR showed exaggerated inflammatory, antimicrobial and proliferative responses. Through transplantation studies, we showed it was SIGIRR expression by intestinal epithelial cells that limits these responses, and that the exaggerated responses were driven by cytokine signaling through the interleukin-1 receptor. Despite their exaggerated responses, SIGIRR deficient mice proved extremely susceptible to infection by C. rodentium and other intestinal bacterial pathogens. We found the exaggerated inflammatory responses rapidly depleted intestinal commensal microbes, reducing their ability to outcompete invading pathogens for space and nutrients (colonization resistance). Our study thus clarifies that the hypo-responsiveness of epithelial cells plays an unexpected but critical role in host defense, by promoting commensal microbe based competition against enteric pathogens. and tissue integrity. Moreover, we noted no overt difference in IEC barrier function or inflammatory cell recruitment to the infected cecum between these two mouse strains ( Figure 1B-1D). These data indicate that MyD88 signaling within IEC plays little role in driving the innate host response to C. rodentium infection.
Sigirr 2/2 mice develop exaggerated colitis during C. rodentium infection While it is unclear why IEC play such a limited role in driving MyD88 dependent responses to C. rodentium infection and other forms of colitis [23,24,25], we recently showed that SIGIRR, a negative regulator of TLR and IL-1R signaling, is expressed by IEC and limits their responses to IL-1b and to most bacterial PAMPs [20]. To test the impact of SIGIRR in vivo, Sigirr 2/2 mice were infected with C. rodentium under co-housing conditions (WT and Sigirr 2/2 mice in same cage) or with WT and Sigirr 2/2 mice housed in different cages, and similar results were obtained. As shown in Figure 2A, infected WT mice displayed modest weight loss at day (D) 2 post-infection (pi) followed by recovery and progressive weight gain. In contrast, infected Sigirr 2/2 mice exhibited significantly greater (P,0.05) weight loss (,10%) Figure 2. Sigirr 2/2 mice suffer more severe colitis during C. rodentium infection. Sigirr 2/2 mice exhibited (A) rapid weight loss by D4 pi. At both D6 and D10 pi, (B-C) their ceca displayed severe damage, with loss of stool contents and focal ulcers. (D) Cecal tissues from Sigirr 2/2 mice had significantly higher pathology scores at D6 and D10 pi compared to WT mice. Plating revealed Sigirr 2/2 mice carried significantly higher pathogen burdens than WT mice in (E) cecal and colonic tissues, but their burdens were similar in (F) liver or spleens. Pathogen counts represent mucosal associated bacteria. Results are pooled from 2 independent infections with n = 3-4 per group. Error bars = SEM, (Two-way ANOVA ( starting from D4 until D10 pi (Figure 2A). Following their euthanization, the Sigirr 2/2 mice showed greater macroscopic signs of infection than WT mice, with overt loss of stool content and edema seen throughout their large intestines ( Figure 2B). Moreover, pus and blood were often found within the ceca of infected Sigirr 2/2 mice and between 60 and 80% of these mice suffered macroscopic cecal ulcers at D6 and D10 pi, a phenotype not observed in WT mice.
Based on the observed pathology, as well as the fact that baseline SIGIRR gene transcript levels in WT mice were highest in the cecum (not shown), we focused our subsequent analysis on the cecum. At both D6 and D10 pi, Sigirr 2/2 mice showed significantly higher cecal pathology scores compared to WT mice, reflecting increased crypt hyperplasia, edema, and greater inflammatory cell infiltration ( Figure 2C and 2D, P,0.05). Specifically, we noted increased macrophage and neutrophil infiltration into the cecal mucosa of Sigirr 2/2 mice. Similarly, gene transcript levels for chemokines (MCP-1 and MIP-2a), antimicrobial factors (mCRAMP, b-defensin III, and Reg-3c) and cytokines (IL-17A, TNF-a and IFN-c) were all significantly elevated at D6 and D10 pi in Sigirr 2/2 mice compared to WT mice ( Figure S1). Based on their exaggerated colitic responses, we expected Sigirr 2/2 mice would show some level of protection against infection. Instead, they carried significantly higher (10-100 fold) C. rodentium burdens in their ceca and colons compared to WT mice at both D6 and D10 pi ( Figure 2E, P,0.01). Interestingly, these higher burdens reflected increased numbers of C. rodentium infecting the intestinal mucosal surface, as well as within the intestinal lumen. Notably, despite their high intestinal pathogen burdens, Sigirr 2/2 mice did not show increased susceptibility at systemic sites, as C. rodentium burdens in the spleens and livers of Sigirr 2/2 mice were similar to those in WT mice ( Figure 2F).
Sigirr 2/2 mice exhibit a hyper-proliferative IEC response against C. rodentium infection Despite carrying heavy C. rodentium burdens and suffering exaggerated colitic damage, Sigirr 2/2 mice survived infection and healed their mucosal ulcers by D14 pi ( Figure S2), suggesting that intestinal mucosal repair was enhanced in these mice. We assessed IEC proliferation in Sigirr 2/2 and WT mice (by Ki-67 staining) to examine their ability to repair mucosal damage and also measured IEC barrier integrity. Increased IEC proliferation, (as measured by the percentage of IEC in a crypt showing Ki-67 positive staining) was evident by D4 pi in the Sigirr 2/2 mice, but not in WT mice ( Figure 3A and 3B, 4761% vs 2163%, P,0.01). By D6 pi, IEC proliferation in Sigirr 2/2 mice was significantly greater than in WT mice ( Figure 3A and 3B, 7567% vs 4164%, P,0.01) and similar results were observed at D10 pi ( Figure 3A and 3B, 6065% vs 3464%, P,0.01). We also assessed IEC barrier function by oral fluorescein isothiocyanate dextran (FD4) gavage, and while infection did cause an increase in serum FD4 levels in both strains, we noted no differences in FD4 levels between infected Sigirr 2/2 mice and WT mice ( Figure 3C).
The negative regulator SIGIRR suppresses the host response to C. rodentium infection We next examined whether SIGIRR impacted on very early host responses to C. rodentium by employing the cecal ligation model of acute C. rodentium infection, allowing us to inject equal pathogen numbers into the ceca of WT and Sigirr 2/2 mice. As expected, similar numbers of C. rodentium were found infecting the cecal tissues of WT and Sigirr 2/2 mice after 10 hr of infection (WT: 1.960.7610 8 CFU/g, Sigirr 2/2: 1.260.5610 8 CFU/g). Cecal tissues were collected and qPCR was used to determine transcript levels for genes encoding several antimicrobial peptides and other IEC derived factors. As shown in Figure 3D, infection of WT mice led to a modest increase in the transcription of genes encoding antimicrobial peptides (mCRAMP, Reg3c, b-defensin III) and chemokines (MCP1 and MIP2a). In contrast, we observed strikingly elevated transcription of most genes in infected Sigirr 2/2 mice (P,0.05), despite the two mouse strains carrying similar pathogen burdens. These results indicate that in the absence of SIGIRR, the host intestine responds very rapidly to C. rodentium infection, and that the exaggerated colitis suffered by orally infected Sigirr 2/2 mice likely reflects both their lack of SIGIRR, as well as their increased pathogen burdens.
SIGIRR's actions reflect its expression by non-BM derived cells SIGIRR thus plays a critical role in suppressing C. rodentium induced colitis, but the cellular source of SIGIRR is unclear. SIGIRR is expressed by IEC, as well as by several BM-derived cell types including dendritic cells and T cells [17,18]. To better define the cell type(s) regulated by SIGIRR during C. rodentium infection, we generated haematopoietic chimeras by BM transplantation between Sigirr 2/2 mice and WT mice expressing Ly5.1 on their BM-derived cells. Following infection by C. rodentium, Sigirr 2/2 + WT BM chimeric mice developed severe colitis, similar to that seen in Sigirr 2/2 mice. Moreover the ceca of infected WT + Sigirr 2/2 BM mice were macroscopically similar to those of WT mice, although histologically, they displayed a greater increase in crypt lengths ( Figure 4B-4D) than WT mice. Pathology scoring also revealed Sigirr 2/2 + WT BM mice had exaggerated crypt hyperplasia, edema and immune cell infiltration when compared to WT + Sigirr 2/2 BM mice ( Figure 4D, P,0.05). Similar to Sigirr 2/2 mice, Sigirr 2/2 + WT BM chimeras carried significantly higher intestinal C. rodentium burdens, compared to WT mice and WT + Sigirr 2/2 BM mice at D10 pi ( Figure 4A). Furthermore, Sigirr 2/2 + WT BM mice exhibited hyperproliferative IEC responses (6565%) similar to those of Sigirr 2/2 mice (7266%), as revealed by Ki-67 staining while the IEC proliferative response was similar between WT mice (3964%) and WT + Sigirr 2/2 BM mice (3363%) ( Figure 4E and 4F). Thus expression of SIGIRR by non-BM derived cells plays the major role in suppressing the host inflammatory response and controlling C. rodentium burdens. Moreover, loss of SIGIRR in this compartment leads to the exaggerated colitis suffered by C. rodentium infected Sigirr 2/2 mice.
Next, we sought to identify the specific TLRs responsible for the exaggerated colitis seen in Sigirr 2/2 mice. Previous studies have shown that the host response to C. rodentium infection is largely dependent on TLR2 and TLR4 [7,10]. Therefore, we crossbred Tlr2 2/2 mice and Tlr4 2/2 mice with Sigirr 2/2 mice to generate Tlr2/Sigirr 2/2 mice and Tlr4/Sigirr 2/2 mice. Upon infecting these mice along with WT and Sigirr 2/2 mice, and Tlr2 2/2 and Tlr4 2/2 mice, all mouse strains were found to survive the infection ( Figure 5A and Figure S3). As expected, all readouts were dramatically greater in the Sigirr 2/2 mice, as compared to WT mice. Interestingly, Tlr2 2/2 and Tlr4 2/2 mice developed modestly elevated pathology compared to WT mice but much less than that suffered by Sigirr 2/2 mice ( Figure S3). Moreover, the responses in Tlr2/Sigirr 2/2 mice and Tlr4/Sigirr 2/2 mice were similar to those in the Sigirr 2/2 mice including pathogen burdens and pathological damage ( Figure 5B and 5C). Hence, the data suggest that the exaggerated damage and increased pathogen burden observed in the Sigirr 2/2 mice are largely independent of TLR2 and TLR4.
We next tested IEC barrier function in these mice (and Tlr2 2/2 mice). At D6 pi, FD4 serum levels within Tlr2 2/2 mice were significantly elevated compared to WT mice, confirming their impaired barrier function (Figures 5F and S3, P,0.05). Similar to WT mice and Tlr4 2/2 mice, Sigirr 2/2 mice showed no overt barrier dysfunction, but strikingly, the concurrent loss of SIGIRR (Tlr2/Sigirr 2/2 mice) compensated for the barrier dysfunction seen in Tlr2 2/2 mice. Furthermore, the impaired phospho STAT-3 staining seen in Tlr2 2/2 mice was restored in Tlr2/Sigirr 2/2 mice ( Figure 5E). These results indicate that loss of SIGIRR leads to IEC responses that promote mucosal integrity in a manner largely able to compensate for the defects caused by TLR2 or TLR4 deficiency, suggesting another SIGIRR regulated receptor is at play.
To further explore the potential for IL-1R signaling to contribute to the exaggerated responses seen in infected Sigirr 2/2 mice, we assessed the expression of its ligands IL-1a and IL-1b. Interestingly, IL-1a gene expression showed a dramatic increase in the cecal tissues of Sigirr 2/2 mice during infection. Moreover protein analysis (ELISA) revealed significantly increased IL-1b levels in Sigirr 2/2 mice compared to WT mice under both baseline and infected conditions ( Figure 7A and 7B, P,0.05). Similarly, cecal crypts isolated from Sigirr 2/2 mice showed elevated IL-1b levels under baseline and infected conditions as measured by Western blot analysis ( Figure 7C).

SIGIRR deficiency increases susceptibility to C. rodentium colonization/infection
While Sigirr 2/2 mice develop exaggerated colitic and IEC proliferative/reparative responses during infection, these mice also appear more susceptible to C. rodentium colonization, carrying significantly higher pathogen burdens than mice expressing SIGIRR. To test whether Sigirr 2/2 mice are truly more susceptible to C. rodentium infection, we infected WT and Sigirr 2/2   mice with a 100 fold lower dose (LD) of C. rodentium and followed their course of infection by plating feces. Previous studies have found this low dose is unable to effectively infect WT mice [28], and indeed, LD WT mice shed only low numbers of C. rodentium (10 4 -10 6 CFU/g of feces). In contrast, LD Sigirr 2/2 mice showed significantly (1006) higher C. rodentium burdens by D2 pi Figure 6. The exaggerated colitis seen in the Sigirr 2/2 mice is attenuated by treatment with the IL-1R antagonist anakinra. WT and Sigirr 2/2 mice were intra-peritoneally injected twice with anakinra (1 mg/mouse) each day for 6 days, or with PBS during C. rodentium infection. Anakinra treatment did not impact on pathogen burdens in infected mice (A) however treatment with anakinra ameliorated macroscopic ulceration in infected Sigirr  and their pathogen burdens continued to increase until D6 pi when they reached ,2.6610 9 CFU/g of feces, similar to the burdens carried by Sigirr 2/2 mice given a full infectious dose ( Figure 8A). At D10 pi, following tissue collection, Sigirr 2/2 mice were found to carry 1000 fold more C. rodentium in both cecal and colonic tissues compared to WT mice ( Figure 8B, P,0.01). As expected from their pathogen burdens, Sigirr 2/2 mice developed severe inflammation and cecal pathology similar to that seen with a full dose infection whereas WT mice did not display any overt or histological signs of infection or pathology ( Figure 8B).
Sigirr 2/2 mice are also highly susceptible to S.

Typhimurium colonization/infection
Thus Sigirr 2/2 mice show unusual susceptibility to C. rodentium colonization and infection. We next tested whether they showed heightened susceptibility to infection by another enteric bacterial pathogen. We orally inoculated WT and Sigirr 2/2 mice with the enteric bacterial pathogen Salmonella enterica serovar Typhimurium. S. Typhimurium is known to poorly colonize the intestines of mice, in large part because it is unable to displace competing commensal microbes [15,29,30]. As a result, pretreating mice with the antibiotic streptomycin is commonly used to remove competing commensals and facilitate S. Typhimurium colonization and infection of the cecum [15]. While we found Sigirr 2/2 mice were more susceptible than WT mice to S. Typhimurium infection following streptomycin pretreatment (not shown), we also tested their susceptibility to S. Typhimurium in the absence of antibiotic pretreatment. As expected, oral inoculation of WT mice with a dose of 5610 6 CFU of DaroA S. Typhimurium did not lead to any significant cecal pathology or colonization by this pathogen. In contrast, the same dose led to dramatically heavier (1000 fold) S. Typhimurium burdens in the cecum and severe inflammation and pathology in the ceca of Sigirr 2/2 mice ( Figure 8C), similar to that seen in S. Typhimurium infected mice pretreated with streptomycin. Taken together, these results show that loss of SIGIRR expression leads to dramatically enhanced susceptibility to oral infection by enteric bacterial pathogens, in a manner able to overcome the typical commensal microbe dependent resistance to such colonization.

Sigirr 2/2 mice undergo rapid commensal microbe depletion following infection
Previous studies by our group and others found that differences in the makeup of the intestinal microbiota can affect host susceptibility to C. rodentium infection [31]. Previous sequencing studies found no overt differences between the microbiota of Sigirr 2/2 mice and WT mice [32], but considering the susceptibility of Sigirr 2/2 mice to enteric pathogens, we speculated they might instead carry either fewer total commensal microbes, or harbour a different microbial composition than WT mice. We enumerated the microbiota in the feces of these mice by fluorescence microscopy under baseline conditions, but noted no significant differences in commensal numbers between WT mice and Sigirr 2/2 mice ( Figure 9A). Through qPCR analysis of fecal pellets, we found members of the phylum Bacteroides to be the dominant taxonomic group present within the stool of both Sigirr 2/2 and WT mice (,50-60%), with smaller numbers of Firmicutes (,20%), and other microbes making up minor components of the microbiota. We also looked at representative members of these phyla such as Bifidobacteria, Lactobacillus and Clostridium [33] and did not note any large or overt differences between the WT and Sigirr 2/2 mice (data not shown).
We also analyzed the above mentioned stool samples using fluorescence in situ hybridization (FISH). The FISH analysis relied on staining microbes with probes for Eubacteria as well as probes specific for c-Proteobacteria or probes that recognize only three genera (Cytophaga, Flexibacter and Bacteroides) from the phylum Bacteroides. Microbes that stained with the Eubacteria probe, but did not stain with the other probes were placed in the ''Other'' category, with these microbes primarily encompassing the Firmicutes as well as minor components of the microbiota. While the makeup of the microbiota as determined by the FISH staining may be less accurate when compared to the more thorough qPCR approach ( Figure 9B and 9C), both techniques were consistent in showing no statistically significant differences between uninfected Sigirr 2/2 and WT mice, confirming previous reports by Chan et al. [32] Interestingly, Lupp et al. (2007), showed that C. rodentium infection causes a dramatic shift in the makeup of the gut microbiota as well as a reduction in total commensal numbers. These commensal changes were host mediated and took approximately 6 days to occur, corresponding with the time taken for C. rodentium to heavily colonize its host. We, and others have speculated these changes in the microbiota may facilitate C. rodentium colonization, as antibiotic based removal of commensals accelerates the course of C. rodentium infection [31,33]. We quantified commensal populations within the feces of infected WT mice, and confirmed it takes approximately 6 days for the aforementioned commensal depletion (80%) to occur. In contrast, total commensal numbers in the fecal samples collected from the Sigirr 2/2 mice dropped dramatically by D1 pi, and this reduction remained evident until D6 pi and beyond ( Figure 9D). The depletion of commensal microbes affected all relevant phyla to a roughly similar degree as measured by qPCR (data not shown). The corresponding C. rodentium burdens increased in concert (albeit slightly delayed) with the loss of commensals. Rapid depletion of total commensal microbe numbers was also detected in Sigirr 2/2 mice following infection with LD C. rodentium (not shown). Similarly, we also observed rapid depletion of total commensal microbe numbers by D2 post-S. Typhimurium infection (in the absence of antibiotics) ( Figure 9E, *P,0.05). In contrast, this reduction in commensals was not observed in S. Typhimurium infected WT mice. Interestingly, we did not observe commensal depletion when Sigirr 2/2 mice were infected with an avirulent strain (DescN) of C. rodentium (not shown). These results suggest the impaired resistance to pathogen colonization we observed in Sigirr 2/2 mice reflects the ability of C. rodentium and S. Typhimurium to rapidly trigger commensal depletion in these mice, thereby opening niches within the intestine for rapid pathogen colonization.

Sigirr 2/2 crypt epithelial cells show heightened antimicrobial activity
We next addressed the basis for the rapid depletion of commensal microbes seen in infected Sigirr 2/2 mice. Notably, the massive commensal depletion occurs prior to overt intestinal increased (D) mortality rates, (E) elevated pathogen burdens and (F) severe mucosal damage. The severe damage suffered by the Il-1r/Sigirr 2/2 mice was accompanied by impaired IEC proliferation as shown by (G) immunostaining for Ki-67 and by higher IEC permeability as quantified by (H) FD4 in serum. Pathogen counts represent mucosal associated bacteria. Results are pooled from 2-4 independent infections with n = 3-4 per group. Error bars = SEM, (Student t test ( Figure A, B, E), one-way ANOVA ( Figure G, H), *P,0.05, **P,0.01). Images were taken at 2006 magnification. doi:10.1371/journal.ppat.1003539.g007 inflammation. As outlined in Figure 3D, acute exposure to C. rodentium rapidly increased antimicrobial gene expression, likely reflecting responses by IEC. We thus examined whether the IEC of Sigirr 2/2 mice harbored greater bactericidal activity than WT mice towards commensal bacteria, using commensal E. coli and Lactobacilli caesi as examples. We isolated intestinal crypts under baseline condition from WT and Sigirr 2/2 mice; collected their crypt supernatants and tested their bactericidal activity against the aforementioned commensal microbes. After a 1 hr incubation, isolates from Sigirr 2/2 mice showed a killing activity of 29.963.5% against commensal E. coli while crypt isolates from WT mice showed only 10.669.9%. Similarly, isolates from Sigirr 2/2 mice displayed greater killing activity than WT mice against Lactobacilli caesi (38.462.4% versus 23.465.4%) a gram-positive commensal microbe (P,0.05). Therefore, crypt isolates from Sigirr 2/2 mice possess significantly higher bactericidal activity than those from WT mice.
We next we sought to determine whether the bactericidal activity within the crypt isolates of Sigirr 2/2 mice increased during infection, and whether the enteric pathogens C. rodentium and S. Typhimurium were more resistant to this bactericidal activity compared to commensal microbes. We isolated intestinal crypts from WT and Sigirr 2/2 mice after 24 hrs of C. rodentium infection; collected their crypt supernatants and tested their bactericidal activity against both commensal microbes and pathogenic microbes. As shown in Figure 9F, while the killing of L. casei by Sigirr 2/2 crypt isolates was relatively unchanged from baseline, the killing of the commensal E. coli strain significantly increased to 41.463.9% at day 1 pi (P,0.05). Moreover, both C. rodentium and S. Typhimurium were found to be significantly more resistant to the bactericidal activity of the crypt isolates from both WT and Sigirr 2/2 mice compared to the commensal microbes (P,0.05). Considering the rapid induction of antimicrobial genes seen in the cecal loop model, the heightened antimicrobial activity of Sigirr 2/2 mice likely explains the rapid commensal depletion seen following their infection.

Discussion
MyD88 dependent signaling plays a critical protective role during C. rodentium infection, promoting inflammatory and antimicrobial responses that control C. rodentium burdens, as well as homeostatic responses that protect IEC barrier function and limit/repair IEC injury [5,6]. While many of these responses involve changes in IEC function, we clearly show that MyD88 signaling within IEC plays little role in driving these responses. Rather than IEC being unable to respond to C. rodentium infection, we instead show that the negative regulator SIGIRR suppresses their responsiveness. While Sigirr 2/2 mice react strongly to pathogenic insults to the GI tract, Sigirr 2/2 mice do not develop spontaneous colitis or other baseline pathologies, but instead show a modestly elevated inflammatory tone within their intestines [21]. When we infected Sigirr 2/2 mice, they developed exaggerated inflammatory and antimicrobial responses as well as increased IEC proliferation and repair. These responses began within the IEC layer, and through BM transplantation studies, we confirmed that it was SIGIRR expression by non-BM cells (putative IEC) that limits the host inflammatory and IEC responses to C. rodentium. This data thus expands our previous in vitro studies that the innate responsiveness of IEC to A/E bacterial pathogens is dramatically elevated in the absence of SIGIRR.
We previously showed Sigirr 2/2 mice develop exaggerated DSS colitis and IEC proliferation that depended on the presence of commensal microbes [21], but it was unclear which SIGIRR regulated receptors drove these responses. In the current study, we tested the role of MyD88, as well as individual receptors and as expected, loss of MyD88 left Sigirr 2/2 mice suffering severe mucosal damage and rapidly succumbing to infection, much like Myd88 2/2 mice that express SIGIRR. As previously mentioned, TLR4 signaling drives the majority of inflammation and IEC proliferation during C. rodentium infection of SIGIRR expressing mice [9]. Similarly TLR2 signaling in these mice is required to protect and repair IEC barrier function during infection [9,10]. Interestingly, neither TLR2 nor TLR4 were required for these roles in the absence of SIGIRR. In fact, SIGIRR deficiency led to increased IEC proliferation/barrier protection that was largely able to compensate for the defects caused by either TLR2 or TLR4 deficiency, suggesting another receptor was responsible for the protective IEC responses seen in Sigirr 2/2 mice.
Our data indicate that IL-1R signaling drives the exaggerated IEC responses seen in Sigirr 2/2 mice. While IL-1R signaling plays a protective role during C. rodentium infection of SIGIRR expressing hosts [8], it appears to be primarily responsible for the heightened IEC proliferation and barrier protection seen in Sigirr 2/2 mice, as these exaggerated IEC responses were attenuated following anakinra treatment, and lost in Il-1r/Sigirr 2/2 mice. While the increased pathogen burdens and heightened inflammation seen in the Sigirr 2/2 mice were not overtly IL-1R dependent, IL-1R signaling did prove crucial in limiting the severity of the resulting mucosal damage and ultimately for the survival of infected Sigirr 2/2 mice. Interestingly, we found elevated levels of both IL-1a and IL-1b in the intestinal tissues of Sigirr 2/2 mice under both baseline and infected conditions. Both IL-1a and IL-1b have been shown to induce IEC proliferation in vitro, as well as induce expression of chemokines and antimicrobial factors by IEC [34,35,36]. At present it is unclear which cytokine (or perhaps both) drives the exaggerated IL-1R dependent epithelial proliferative and repair responses seen in the Sigirr 2/2 mice.
Our data also indicates it is primarily SIGIRR expression by non-BM derived cells that limits the host response to C. rodentium infection. While several BM derived cell types (macrophages, dendritic cells, T cells) found in the intestine express SIGIRR, the only described non-BM derived cell type is the IEC [16,17]. It has been speculated that maintaining the innate hypo-responsiveness of IEC helps prevent spontaneous commensal microbe driven intestinal inflammation, as seen in patients with Inflammatory Bowel Disease (IBD) [37]. While this may be the case in some situations, our data suggests that SIGIRR instead controls the threshold at which IEC respond to invading bacterial pathogens.
Notably, despite their innate hyper-responsiveness, Sigirr 2/2 mice proved highly susceptible to infection, undergoing very rapid pathogen colonization and ultimately carrying significantly heavier C. rodentium burdens than WT mice. Unlike Myd88 2/2 mice where their immunodeficiency not only led to increased intestinal pathogen burdens, but also lethally high C. rodentium burdens in the liver and spleen, the Sigirr 2/2 mice only showed increased pathogen burdens in their ceca and colons. Heavy C. rodentium colonization of Sigirr 2/2 mice was evident between D2 and 4 pi, compared to the 6 days typically required for WT mice. Interestingly, we previously found that such rapid C. rodentium colonization was only seen in mice pretreated with the antibiotic streptomycin to displace commensal microbes. Taken together with the susceptibility of Sigirr 2/2 mice to S. Typhimurium infection without the typical requirement for streptomycin pretreatment, we focused on potential defects in commensals as underlying the susceptibility of Sigirr 2/2 mice to pathogen colonization of their intestines.
As outlined by Chan et al. [32], previous deep sequencing studies identified few differences between Sigirr 2/2 mice and WT mice in their baseline intestinal microbiota. In accordance with their results, we also found they carried similar numbers of intestinal commensal microbes, and showed only modest differences in their baseline microbiota makeup, as compared to WT mice. We therefore compared how the intestinal microbiota of Sigirr 2/2 and WT mice responded to C. rodentium infection. As shown by Lupp et al. [4], infection of WT mice leads to a host mediated depletion of the intestinal microbiota by D6 pi, concomitant with C. rodentium expansion and spread through the intestines of WT mice. We confirmed this timing in WT mice, whereas overt commensal depletion was found to occur by D1 pi in Sigirr 2/2 mice, and persisted thereafter. Though qPCR analysis, we found that the observed commensal depletion affected all commensal microbial phyla to a roughly similar degree. Interestingly, this rapid commensal depletion did not occur when Sigirr 2/2 mice were gavaged with avirulent C. rodentium, but it did occur when the mice were challenged with S. Typhimurium suggesting the commensal depletion results from a host response to active infection.
While infected Sigirr 2/2 mice do not display overt colitis at the time of commensal depletion, we showed that acute exposure to C. rodentium dramatically elevated mRNA transcript levels for a number of antimicrobial genes within the ceca of Sigirr 2/2 mice. This increased antimicrobial tone had a functional effect, as intestinal crypt IEC isolates from even uninfected Sigirr 2/2 mice displayed significantly greater killing activity against commensal E. coli and Lactobacilli bacteria than IEC from WT mice. We also demonstrated that the bactericidal activity of the Sigirr 2/2 crypt IEC selectively but significantly increased by as early as D1 pi, and that compared to the commensals, the pathogens C. rodentium and S. Typhimurium were comparatively resistant to killing by crypt IEC. Thus the rapid depletion of intestinal commensal microbes seen in infected Sigirr 2/2 mice likely reflects the heightened antimicrobial activity of their IEC and provides much of the basis for their reduced colonization resistance against enteric pathogens.
Colonization resistance was initially described by van der Waaij et al., [14] as the process whereby the intestinal microbiota protects itself as well as the host against incursion by new and often harmful microbes. While there is significant evidence for colonization resistance, such as the hyper-sensitivity of germfree mice to enteric infections, the exact mechanisms by which colonization resistance exerts its protective role are unclear [13,14,38]. Moreover, while such commensal mediated resistance to enteric pathogens is undoubtedly beneficial to the host, it is unclear whether host factors play a role in promoting colonization resistance. Our data suggests that it is in fact the innate hyporesponsiveness of IEC that is critical in promoting colonization resistance. While it may seem counter-intuitive that limiting host antimicrobial and inflammatory responses in the GI tract would be protective, as our LD infection data shows, in most cases when the host is exposed to small numbers of enteric bacterial pathogens, the resident microbiota are able to prevent the pathogens from successfully infecting the host. It appears that SIGIRR helps maintain host-commensal mutualism (and colonization resistance) in the face of noxious threats, such as the incursion by a bacterial pathogen, potentially by controlling the threshold at which the IEC respond.
Our results are intriguing since the exaggerated inflammatory/ antimicrobial responses in the intestines of Sigirr 2/2 mice appeared to deplete only commensals and not C. rodentium. While our crypt killing assays suggest the responses elicited in Sigirr 2/2 mice are poorly effective against C. rodentium and S. Typhimurium, we believe the effects are more complex. It is well known that enteric bacterial pathogens utilize a number of strategies to subvert host defenses, including suppressing the inflammatory signaling of infected IEC [2], as well as possessing greater inherent resistance to the effects of antimicrobial peptides [39]. Moreover, recent studies have shown that enteric bacterial pathogens can utilize nutrients and metabolites released within the inflamed intestine that are unusable to commensal species [3,40]. Therefore the success of bacterial pathogens at replacing the microbiota in Sigirr 2/2 mice may in fact also demonstrate their evolutionary success at withstanding host antimicrobial responses, and/or their ability to obtain nutrients within the intestinal lumen of the Sigirr 2/2 mice, in comparison to commensal microbes.
In conclusion, our data demonstrate that Sigirr 2/2 mice develop exaggerated intestinal inflammatory, antimicrobial and IEC proliferative responses to C. rodentium infection. While IL-1R signaling is key to many of the changes in IEC function and proliferation seen in these mice, there are other aspects of their exaggerated colitis (inflammation, commensal depletion etc) that are at least partially IL-1R independent but are still MyD88 dependent (not shown). We speculate these responses may reflect the combined actions of several TLRs since we were unable to identify a specific TLR responsible for these phenotypes. Moreover, although SIGIRR suppresses host inflammatory, antimicrobial and IEC reparative responses, it actually plays a critical role in enteric host defense by promoting commensal based colonization resistance against enteric bacterial pathogens. Thus SIGIRR's role in limiting the innate responsiveness of IEC is evidence of a complex defense strategy within the GI tract. This strategy delays the development of protective host responses to infection, by relying on cell types other than IEC to drive protective host (MyD88) responses to infection. However SIGIRR's key role in promoting and protecting host mutualism with the gut microbiota has obvious advantages as well, preventing many enteric infections by supporting the ability of the microbiota to resist invading microbes. While promoting colonization resistance to enteric infections is of great benefit, ultimately we believe that modulating SIGIRR expression within the intestine could also offer therapeutic potential for GI diseases, such as IBD. Increasing SIGIRR expression could potentially reduce inflammation in IBD patients, whereas suppressing SIGIRR could potentially help promote mucosal repair.

Mouse strains and infection of mice
Mouse strains used in this study: C57B/6 (WT), Tlr2 2/2, Tlr4 2/2, Myd88 2/2, Sigirr 2/2, Il-1r/Sigirr2/2, Tlr2/Sigirr 2/2, Tlr4/Sigirr 2/2 and Myd88/Sigirr 2/2 mice (8-12 weeks) were bred in house, and kept under specific pathogen free conditions at the Child and Family Research Institute. Il-1r12/2 and Myd88 flox (B6.129P2-Myd88 tm1Defr /J) and Villin cre mice were purchased from Jackson Laboratory. All double gene deficient mice were generated by crossbreeding single gene deficient mice. IEC-Myd88 2/2 mice were generated by crossbreeding Myd88 flox mice with Villin-Cre mice. Mice were orally gavaged with an overnight culture of ,2.5610 8 CFU of streptomycin resistant C. rodentium DBS100 or 1610 6 CFU of DaroA S. Typhimurium SL1344 (DaroA chosen because it does not kill Nramp1 sensitive mice) [41] and euthanized at specified time points pi. Infections were performed under both co-housing conditions (WT and mutant mice in same cage) and single housing conditions (WT and mutant in different cages). For IL-1R antagonist experiments, anakinra (Kineret) at 1 mg/mouse (or PBS as a control) was injected intraperitoneally twice daily (12 hrs apart) during the course of the infection (6 days) beginning on the day of the infection (day 0) just prior to oral gavage [42]. Body weight data are presented as the mean percentage of starting weight.

Ethics statement
All experiments were performed according to protocols approved by the University of British Columbia's Animal Care Committee and in direct accordance with The Canadian Council on Animal Care (CCAC) guidelines. The protocols were approved by the UBC Animal Care committee (Protocols: A11-0290 and A11-0253). Mice were monitored for mortality and morbidity throughout their infection and euthanized if they showed signs of extreme distress or .15% body weight loss. All surgeries were performed under anesthesia (isofluorane and xylazine), and all efforts were made to minimize suffering.

Tissue collection, bacterial counts and pathology scoring
Tissue collection and bacterial counts were performed as described previously [43]. Briefly, mice were anesthetized with isofluorane and euthanized over the course of infection, dissected and their large bowel was divided into cecum and colon to be collected in 10% neutral buffered formalin (Fisher) for histological analyses, or processed for tissue pathology assays. For viable cell counts, cecum, colon tissues and stool pellets were collected separately, and homogenized in PBS pH 7.4, with dilutions plated onto LB agar or streptomycin agar plates. Pathology was scored using a previously adapted scoring system [44]. In brief, paraffin-embedded cecal tissue sections (5 mm) that had been stained with haematoxylin and eosin were examined by two blinded observers. Tissue sections were assessed for i) submucosal edema (0-no change 1-mild 2-moderate 3-profound), ii) hyperplasia (0-no change 1-1-50% 2-51%-100% 3 -.100%), iii) goblet cell depletion (0 no change 1 mild depletion 2 severe depletion 3-absence of goblet cells), iv) epithelial damage/ integrity (0 no change 1-few cells sloughing, epithelial surface rippled 2-epithelial surface is rippled, damaged 3-epithelial surface is severely disrupted/damage, large amount of cells sloughing). For ulceration, an additional score of 1 was added for each 25% (or fraction of) of the tissue in the cross section affected ie.( a small ulcer would score as 4+1, whereas a tissue where 60% of the tissue cross section is ulcerated would be 4+3) v) mononuclear cells infiltration (per 406 field) (0-no change 1-,20 2-20 to 50 3-.50 cells. The maximum possible score was 18.

RNA extraction and quantitative real-time PCR
Following euthanization of mice, tissues were immediately transferred to RNA-later (Qiagen). Total RNA was purified using QiagenRNEasy kits (Qiagen) stored according to manufacturer's instructions. cDNA was synthesized with Omniscript RT kit (Qiagen) and OligodT (Applied Biological Material Inc) followed by quantitative real-time PCR techniques. qPCR was carried out on MJ Mini-Opticon Real-Time PCR System (Bio-Rad) using IQ SYBR Green Supermix (Bio-Rad) and MIP-2, MIP3-a, b-actin, mCRAMP, Reg 3-c, b-defensin III primers using sequences and conditions previously described [44]. Il-1a primers were generated by PrimerBank (http://pga.mgh.harvard.edu/primerbank/), 59 GAAGACTACAGTTCTGCCATT39, reverse 59 GACGTTT-CAGAGGTTCTCAGAG 39 Quantification was performed using the Gene Ex Macro OM 3.0 software (Bio-Rad) where PCR efficiencies for each primer set were incorporated into the final calculations.

Immunofluorescence staining
Immunofluorescence staining of control and infected tissues was performed using previously described procedures [43]. In brief, paraffin embedded tissues were cut (6 mm) and were deparaffinized by heating at 55-65uC for 20 mins, cleared with xylene, rehydrated through an ethanol gradient to water. Sections were blocked using blocking buffer (Goat serum in PBS containing 1% bovine serum albumin (BSA), 0.1% Triton-X100, and 0.05% Tween 20, and 0.05% sodium azide). The primary antibodies used were Ki-67 (Thermo) or phospho STAT-3 (Ab-Cam) while the secondary was AlexaFluor 568-conjugated goat anti-rabbit IgG. Tissues were mounted using ProLong Gold Antifade reagent (Invitrogen) containing DAPI for DNA staining. Sections were viewed on a Zeiss AxioImager microscope and images taken using an AxioCam HRm camera operating through AxioVision software.

Cecal loop surgery
Cecal loop infections were performed as recently described [43]. In brief, mice were anaesthetized (2.5% isofluorane and xylazine at 10 mg/kg) and following a midline abdominal incision, the cecum was exposed. The proximal colon close to the cecum was ligated twice, while the ileocecal valve prevented backflow into the ileum, thus isolating the cecum. C. rodentium cultures were diluted in PBS, and 300 ml of culture containing ,1610 7 CFU were injected into the cecal loop which was returned to the abdominal cavity and the incision closed with discontinuous sutures. The mice were euthanized at 10 hr pi and tissues and stool contents were collected for CFU counts and mRNA analysis as described above.

Bone marrow transplantation
Bone marrow was isolated from C57BL/6.Ly5.1 (WT), or Sigirr 2/2 mice. Four million cells were intravenously injected into lethally irradiated wild-type (WT), or Sigirr 2/2 recipient mice that were allowed to recover for 12 weeks. Reconstitution was confirmed by FACs analysis of peripheral blood leukocytes for Ly5.1 expression. Only mice exhibiting greater than 90% chimerism were infected. Similar methods were performed to obtain Sigirr 2/2 +WT BM mice.

Assessment of total microbes and FISH hybridization
Enumerating total microbes and FISH staining were performed as described previously [44].
At least 2 fecal pellets were collected from each animal at time points indicated. After homogenization, samples were placed in 10% Neutral Buffered Formalin to a final concentration of 3%. Samples were further diluted 1:10 in PBS, vortexed briefly, and stored at 4uC. 2-5 ml of the 1:10 diluted sample stored in PBS was diluted in 1 ml PBS and filtered onto Anodisc 25 filters (Whatman International Ltd) with a pore size of 0.2 mM and 2.5 cm diameter. The samples were allowed to thoroughly dry, and then were mounted on glass slides with ProLong Gold Antifade reagent containing DAPI (Molecular Probes) and viewed as above. The mean number of DAPI positive microbes counted in 3 to 6 randomly chosen fields per disc (10006) was determined by two scorers. The total number of commensal microbes was calculated based on the mean numbers of all the counted fields and the dilution factor. The total number of commensal microbes was presented as the percentage of uninfected controls. For FISH staining, filters were passed through an ethanol gradient following filtering and samples were incubated overnight at 37uC in the dark with Texas red-conjugated EUB338 general bacterial probe (59-GCT GCC TCC CGT AGG AGT-39) and an AlexaFluor 488 conjugated GAM42a probe (59-GCC TTC CCA CAT CGT TT-39) that recognizes bacteria that belong to the c-Proteobacter class [4,45]. Filters were then washed in the dark with hybridization solution (0.9 M NaCL, 0.1 M TRIS pH 7.2, 30% Formamide, 0.1% SDS) for 15 min with gentle shaking. This step was repeated once with wash buffer (0.9 M NaCl, 0.1 M TRIS pH 7.2), and sections were placed in dH 2 O, and then mounted using ProLong Gold Antifade reagent with DAPI (Molecular Probes) and imaged as described above.

Microbial composition analysis
Microbial composition was performed by qPCR as described previously [33]. Briefly, at least two fecal pellets were collected and DNA was extracted using a Qiagen DNA stool extraction kit. Following DNA extraction, 50 ng DNA/reaction was used for qPCR. qPCR was completed using 16 s rRNA group specific primers [33] to determine the relative abundance of selected bacterial phyla. The primers used were specific to the phyla Bacteroides (59-GAG AGG AAG GTC CCC CAC-39, 59-CGC TAC TTG GCT GGT TCA G-39) [46], and Firmicutes (59-GGA GYA TGT GGT TTA ATT CGA AGC A-39, 59-AGC TGA CGA CAA CCA TGC AC-39) [46], and c-Proteobacteria (59-TCG TCA GCT CGT GTY GTG A-39, 59-CGT AAG GGC CAT GAT G-39) [47]. The relative abundance of each taxonomic group was determined by calculating the average Ct value relative to the total bacterial 16 s rRNA, as determined using the universal Eubacteria primers (59-ACT CCT ACG GGA GGC AGC AGT-39, 59-ATT ACC GCG GCT GCT GGC -39) [48] and normalized to each primer's experimentally determined efficiency. Statistical significance was determined using a one-way ANOVA with a threshold of p,0.05.

Colonic crypt killing assay
The bactericidal capacity of cecal crypt secretions was performed as described previously [49,50]. Cecal and Colonic crypts were isolated from naïve WT and Sigirr 2/2 mice and mice infected with C. rodentium for 1 day. Ceca and colons were extracted from mice following euthanization and fecal contents were removed. Tissues were washed for 30 min at room temperature on an orbital shaker with HEPES buffered HBSS containing Pen/Strep and gentamycin. Tissues were then cut into small sections and placed in a petri dish containing 10 mL of BD cell recovery solution. Tissues were incubated for 2 hours at 4uC and crypts were dislodged from the tissues. Following centrifugation, crypts were resuspended in iPIPES buffer (10 mM PIPES; pH 7.4 and 137 mM NaCl), and counted. For experiments, 1000 crypts were incubated in 45 mL of iPIPES buffer at 37uC for 30 minutes. Following this period, 10 mL of a crypt-free isolate was added to 10 4 E. coli (commensal isolates, a gift from Dr. Ben Willing) or Lactobacili casei (a gift from Dr. Alan Stintzi) or C. rodentium or DaroA S. Typhimurium and incubated for 37uC for 60 minutes. The antimicrobial capacity of crypt-culture extracts was assessed by counting the overnight growth of WT and Sigirr 2/2 crypt-treated E. coli or Lactobacili casei, and expressed as a percentage of the growth observed in untreated cultures.

FD4 intestinal permeability assay
The assay was performed as previously described [9]. Uninfected or infected mice at D6 pi were gavaged with 150 ml of 80 mg/ml 4 kDa FITC-dextran (Sigma; FD4) in PBS 4 hrs prior to sacrifice. Mice were anaesthetized and blood (,500 ml) was collected by cardiac punctures, which was added immediately to a final concentration of 3% acid-citrate dextrose (20 mM citric acid, 100 nM sodium citrate, 5 mM dextrose). Plasma was collected and fluorescence was quantified using a VictorX3 (Perkin-Elmer Life Sciences) at excitation 485 nm, emission 530 nm for 1 sec.

Measurement of IL-1b in tissues
Tissues were collected as described above. After homogenization and centrifugation, tissue supernatants were collected at specified time points and IL-1b levels were assessed using an ELISA kit (BD Bioscience) following manufacturer's instructions.

Western blotting
Following isolation of intestinal crypts, cells were lysed in RIPA buffer on ice. IEC proteins (50 mg) were resolved by 12% SDS-PAGE and transferred to PVDF membranes. Blots were blocked for 1 h with 5% milk in TBST. Membranes were incubated with primary antibody in TBST overnight at 4uC and probed with the respective secondary antibody for 1 hr at room temperature. Rabbit polyclonal IL-1b (Genetex) and mouse polyclonal actin (SantaCruz) antibodies were used in this study. Membranes were then incubated with ECL Prime Western Blotting kit (GE health) and imaged using a Biorad ChemiDoc XRS (BioRad)

Statistical analysis
All results are expressed as the mean value 6 SEM. Unless specified otherwise, non-parametric Mann-Whiney t-tests, or Student t tests were performed (Prism version 4.00). A P value of 0.05 or less was considered significant.

Gene accession numbers
The following are the GeneIDs (Database: Entrez Gene) for each gene analyzed in this manuscript, given as gene name