Std fimbriae-fucose interaction increases Salmonella-induced intestinal inflammation and prolongs colonization

Expression of ABO and Lewis histo-blood group antigens by the gastrointestinal epithelium is governed by an α-1,2-fucosyltransferase enzyme encoded by the Fut2 gene. Alterations in mucin glycosylation have been associated with susceptibility to various bacterial and viral infections. Salmonella enterica serovar Typhimurium is a food-borne pathogen and a major cause of gastroenteritis. In order to determine the role of Fut2-dependent glycans in Salmonella-triggered intestinal inflammation, Fut2+/+ and Fut2-/- mice were orally infected with S. Typhimurium and bacterial colonization and intestinal inflammation were analyzed. Bacterial load in the intestine of Fut2-/- mice was significantly lower compared to Fut2+/+ mice. Analysis of histopathological changes revealed significantly lower levels of intestinal inflammation in Fut2-/- mice compared to Fut2+/+ mice and measurement of lipocalin-2 level in feces corroborated histopathological findings. Salmonella express fimbriae that assist in adherence of bacteria to host cells thereby facilitating their invasion. The std fimbrial operon of S. Typhimurium encodes the π-class Std fimbriae which bind terminal α(1,2)-fucose residues. An isogenic mutant of S. Typhimurium lacking Std fimbriae colonized Fut2+/+ and Fut2-/- mice to similar levels and resulted in similar intestinal inflammation. In vitro adhesion assays revealed that bacteria possessing Std fimbriae adhered significantly more to fucosylated cell lines or primary epithelial cells in comparison to cells lacking α(1,2)-fucose. Overall, these results indicate that Salmonella-triggered intestinal inflammation and colonization are dependent on Std-fucose interaction.


Introduction
Glycosylation is an important type of post-translational modification of proteins and lipids and is involved in the regulation of a wide range of processes at the cellular and molecular level. The gastrointestinal tract is home to a vast array of glycan structures and glycoconjugates [1], where the mucosal surface is the site of complex interactions between the intestinal microbiota, intestinal barrier, and immune system. The mucosal surface is characterized by a heavily glycosylated mucus layer produced by goblet cells as well as membrane-bound glycosylated proteins and lipids that form them [2]. These glycoconjugates can be utilized by intestinal commensal bacteria and pathogens as molecular attachment sites or as nutrients [3]. Importantly, host-derived glycans can help foster beneficial relationships with symbiotic microbes, such as Bacteroides thetaiotaomicron, by providing an energy source in the absence of dietary polysaccharides [4].
The FUT2 gene encodes the α-1,2-fucosyltransferase, a glycosyltransferase well known for its role in the expression of ABH and Lewis histo-blood group antigens on the gastrointestinal epithelium and in bodily secretions. Individuals expressing a functional allele are commonly described as 'secretors' whereas those homozygous for loss-of-function mutations display a 'non-secretor' phenotype. Variation in host glycosylation may directly influence susceptibility to enteric pathogens such as enterotoxigenic Escherichia coli [5], Helicobacter pylori [6], and norovirus [7]. Recent studies have shown the importance of host glycans in supporting a beneficial relationship with the endogenous microbiota by nourishing the microbiota during the stress of systemic infection [8] or by controlling opportunistic pathogens within the microbiota in the context of infection (e.g. Enterococcus faecalis) [9]. The non-secretor phenotype is also associated with an increased risk to develop chronic inflammatory bowel diseases [10]. This is possibly due to the altered composition of the intestinal microbiota, which may in turn influence the capacity of pathogenic bacteria to bind to host mucosal surface structures [11].
Salmonella enterica serovar Typhimurium (S. Typhimurium) is one of the most successful mucosal pathogens, colonizing the human gastrointestinal tract and causing severe inflammatory diarrhea [12]. S. Typhimurium carries several virulence genes including fimbrial adhesins, which are hair-like appendages on the outer membrane and are involved in adherence to host epithelial cells. Adhesion to host tissues is critical for invasion and pathogenicity of S. TP8 (Projektnummer 158989968). Alibek Galeev and Samriti Sharma were supported by the Center for Infection Biology (ZIB) at the Hannover Biomedical Research School (HBRS). JFB was supported by DFG under Germany's Excellence Strategy -EXC 22167-390884018. MH was supported by DFG SFB 944, project P4 and BMBF Infect-ERA consortium grant 031L0093A. TS was funded by the DFG through grant TS 1921/3-1. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Typhimurium [13]. Type 1 fimbriae are one of the best characterized fimbrial adhesins and are encoded by the fim operon. FimH, a lectin-like protein, directly binds to high mannose oligosaccharides conjugated to surface glycoproteins of eukaryotic cells [14,15]. Another fimbrial operon, std, encodes the π-class Std fimbriae, which have been described to bind terminal α-1,2 fucose residues [16].
The expression of bacterial adhesins possibly involved in binding fucosylated host proteins suggests that these fimbriae may facilitate Salmonella to establish or maintain infection in the highly fucosylated large intestine. Here, we investigated the role of host fucosylation in disease development during Salmonella infection using mice with and without expression of the Fut2 gene (Fut2 +/+ and Fut2 -/-). Taken together, our results demonstrate that Std-fucose interaction contributes to S. Typhimurium persistence and inflammation.

Fut2 expression affects susceptibility to Salmonella-induced colitis
To test the hypothesis that expression of Fut2 influences host susceptibility to enteric pathogens, a model of S. Typhimurium-induced colitis was utilized. Fut2 +/+ and Fut2 -/littermates were treated with streptomycin, and 24 hours later, infected with wild-type S. Typhimurium. One day post infection (p.i.), the cecal tissue of Fut2 -/mice contained more S. Typhimurium than Fut2 +/+ mice (S1A Fig) in agreement with the observations of Goto and colleagues [17]. However, in contrast to their results, we found the total cecum weight and histopathology scores (S1B-S1D Fig) were comparable between Fut2 +/+ and Fut2 -/mice.
Under most conditions wild-type Salmonella kill C57BL/6 mice within approximately one week. Therefore, in order to follow the infection to later time points mice were infected with the S. Typhimurium ΔaroA mutant strain which is attenuated for systemic disease but causes extensive intestinal inflammation [18]. There was no significant difference in bacterial colonization or resulting inflammation of Fut2 +/+ and Fut2 -/mice on day 1 p.i. (S1 Fig) or on day 3 p.i. (S2A-S2C Fig). However, on day 7 and day 14 p.i., a significantly reduced Salmonella burden in the intestine of Fut2 -/compared to Fut2 +/+ mice was detected (Fig 1A, S2A-S2C Fig). Furthermore, 7 days p.i., the histopathological changes in the colon were significantly less severe in Fut2 -/mice compared to Fut2 +/+ animals (Fig 1B and 1C). Notably, the colons of infected Fut2 +/+ mice were characterized by a higher number of detached epithelial cells within the colon lumen, increased inflammatory cell infiltration within the mucosa, and stronger submucosal edema. Additionally, the levels of the inflammation-associated marker lipocalin-2 were quantified in the large intestine after S. Typhimurium infection. The concentration of lipocalin-2 in the colon and cecum 7 days p.i. were significantly higher in Fut2 +/+ mice in comparison to Fut2 -/- (Fig 1D, S2D Fig).
Next, colon tissue sections were analyzed by immunohistochemical staining and subsequent quantification of CD68-and MPO-positive cells, which represent macrophages and neutrophils, respectively. Consistent with elevated histopathological scores, significantly higher numbers of recruited neutrophils and macrophages were detected in the colon tissue of Fut2 +/+ mice compared to Fut2 -/mice (Fig 2A-2C). In addition, a significantly stronger infiltration of CD4 + T lymphocytes in the colonic lamina propria of Fut2 +/+ mice compared to Fut2 -/mice was detected by immunofluorescence staining and by flow cytometry (Fig 2A and  2D). No statistically significant differences were found in the numbers or infiltrate composition with respect to cytotoxic T lymphocytes, B lymphocytes, or dendritic cells by flow cytometric quantification of CD8 + , CD19 + , and CD11c hi cells, respectively (S3 Fig). To summarize, Fut2-expressing mice exhibited higher bacterial load in the intestine at later time points, which was also associated with an increase in inflammation assessed by Salmonella fimbriae interaction with host glycans histopathology and lipocalin-2 levels, demonstrating that Fut2-mediated fucosylation in the intestine plays an important role in Salmonella-triggered inflammation and colonization of the intestine.

Std fimbriae mediate adhesion to intestinal epithelial cells in a fucosedependent manner
The Fut2 protein facilitates intestinal epithelial fucosylation by catalyzing the addition of Lfucose residues via an α(1,2) linkage to the terminal β-D-galactose residue of glycans. The std operon of S. Typhimurium encodes a fimbrial adhesin known to be important for the attachment to fucosylated structures on intestinal epithelial cells [16]. To investigate the role of fucose-Std fimbriae interaction during Salmonella adherence, human intestinal epithelial cell lines HT29-MTX-E12 and Caco-2 were utilized. HT29-MTX-E12 are colon epithelial cells that differentiate into goblet-like cells and produce mucus after three weeks of in vitro culture [19]. Ulex europaeus agglutinin I (UEA-I) lectin staining revealed extensive fucosylation of cell surface and mucus in the differentiated HT29-MTX-E12 cells in contrast to undifferentiated HT29-MTX-E12 cells. Wheat germ agglutinin (WGA) lectin staining for the ubiquitously expressed N-acetylglucosamine was positive in both differentiated and undifferentiated HT29-MTX-E12 cells (Fig 3A). Previous studies have shown that bacterial expression of Std fimbriae is a subject of complex and tight regulation, both in vivo and in vitro [20,21]. Only a very small proportion of the Salmonella population express std fimbriae in vitro [22] and the std operon is completely absent in E. coli. Therefore, to analyze the role of Std fimbriae in vitro, an inducible expression plasmid containing the Salmonella stdABCD operon encoding the structural genes of Std fimbriae was transformed into a common laboratory E. coli K-12 and afimbriated E. coli ORN172. Upon induction with anhydrotetracycline, Std fimbriae were expressed by E. coli (referred as E. coli StdON) as confirmed by flow cytometry and Western blotting (Fig 3C and 3D). HT29-MTX-E12 cells were infected with either std-expressing (E. coli StdON) or non-expressing bacteria (E. coli StdOFF). In contrast to E. coli StdOFF, only E. coli StdON strain showed increased adherence to differentiated HT29-MTX-E12 cells ( Fig 3B). Importantly, expression of Std fimbriae had no effect on adhesion of E. coli StdON to undifferentiated HT29-MTX-E12 cells which do not contain fucosylated glycoproteins (Fig 3B). Similarly, E. coli ORN172 In conclusion, Std fimbriae are important for binding α(1,2)-fucosylated residues on cell lines corroborating the results by Chessa and colleagues [16].
Next, Std-dependent bacterial adherence to primary epithelial cells was investigated. To this end, primary intestinal epithelial cells were isolated from Fut2 +/+ mice and cultivated as three-dimensional organoids in matrigel. These enteroids were expanded and seeded onto transwell filters resulting in the formation of a 2D monolayer consisting of various primary epithelial cell types. Monolayer barrier integrity and the degree of differentiation were evaluated by measuring transepithelial electrical resistance. Polarized monolayers were infected with E. coli StdON and StdOFF bacteria and adherence was analyzed by immunofluorescence. We counted the number of E. coli bacteria attached to UEA-1-positive and -negative cells. E. coli StdON bacteria were primarily associated with fucosylated cells, while E. coli StdOFF adhered equally to fucosylated and non-fucosylated cells ( Fig 4A). Furthermore, a significantly higher number of the α(1,2)-fucose-associated E. coli StdON cells compared to the α(1,2)fucose-associated E. coli StdOFF bacteria was detected ( Fig 4B). Overall, this data demonstrate that std-expressing bacteria preferentially bind to fucosylated cells.

Std fimbriae-fucose interaction is critical for Salmonella-induced inflammation and colonization
Using cut sections of the cecum of CBA/J mice, it was previously demonstrated that purified Std fimbriae of S. Typhimurium are able to bind terminal α(1,2)-fucose residues in the mucosa [16]. However, the functional consequences of this interaction for disease development, as well as the extent of Std fimbriae expression in vivo are not known.
To assess Std fimbriae production in vivo, stdA gene expression was first examined using RT-qPCR. Similar levels of stdA gene expression were detected in the colon of both Fut2 +/+ and Fut2 -/mice infected with S. Typhimurium ΔaroA strain (S5A Fig). In order to determine whether the absence of std or the presence of intestinal fucosylated glycans affects expression of fucose or 1,2-propanediol utilization pathways we quantified levels of fucI and pduBC by RT-qPCR. We saw comparable levels of these genes expressed in either mouse strain in stdcontaining and std-deficient bacteria (S5B and S5C Fig). In order to look more closely at the spatial regulation of Std expression, we used reporter strains of S. Typhimurium containing a stdAstop::gfp fusion [22] and staining of tissue sections with anti-Std serum. Std was observed to be specifically expressed in the lumen of the large intestine of both Fut2 +/+ and Fut2 -/mice on day 1 (S6 Fig) and day 7 p.i. (Fig 5 and S7 Fig). In contrast, Std-expressing Salmonella were not observed after invasion of the mucosa (Fig 5, S6 and S7 Figs). These data demonstrate that there is a tight spatial regulation of Std-expression whereby Std fimbriae are expressed prior to invasion of the large intestine.
Fut2 -/mice lack terminal fucose on intestinal epithelium [23]. To test whether epithelial fucosylation directly affects Salmonella colonization via interaction with Std fimbriae, the stdA mutation was transferred into the S. Typhimurium ΔaroA background strain. This mutant strain lacked functional Std fimbriae (S. Typhimurium ΔaroAΔstdA) yet had the same growth rate and motility as the parental S. Typhimurium ΔaroA strain (S8 Fig). Fut2 +/+ and Fut2 -/mice were then infected with S. Typhimurium ΔaroAΔstdA via oral gavage for 7 days. In contrast to S. Typhimurium ΔaroA (Fig 1), the S. Typhimurium ΔaroA ΔstdA strain colonized the colon of Fut2 +/+ and Fut2 -/mice to similar levels ( Fig 6A). In addition, histopathology scores showed similar intestinal inflammation of Fut2 +/+ and Fut2 -/mice. H&E staining of colonic tissue showed moderate numbers of necrotic epithelial cells, mild inflammatory cell infiltration within intestinal mucosa and mild submucosal edema (Fig 6B and 6C), which was  In order to further explore whether Std fimbriae play a role in the colonization of S. Typhimurium in the presence or absence of host intestinal fucosylation, competitive index (CI) experiments were performed by orogastrically infecting both Fut2 +/+ and Fut2 -/mice with equal numbers of S. Typhimurium ΔaroA and S. Typhimurium ΔaroA ΔstdA. Fecal pellets Salmonella fimbriae interaction with host glycans were collected at 1, 3, and 5 days p.i. and bacterial counts of both strains were determined. After 7 days, the bacterial load in intestinal tissues and luminal content was determined and the CI ratio of the two strains was calculated. Interestingly, in cecum and colon, S. Typhimurium ΔaroA significantly outcompeted the isogenic Std-deficient strain in Fut2 +/+ mice, but not in Fut2 -/mice (Fig 7). Accordingly, CI results from feces at day 1, 3, and 5 showed that S. Typhimurium ΔaroA outcompeted the isogenic ΔstdA mutant in Fut2 +/+ mice (S9 Fig). Taken together, our in vivo data demonstrate that Std fimbriae are important for Salmonella colonization, persistence, and induction of inflammation in a fucosylated host environment.

Discussion
Variation in human glycosylation influences various metabolic diseases, cancers, inflammatory diseases, and susceptibility to infectious pathogens. Genome-wide association studies show that FUT2 nonsense polymorphisms are associated with increased risk for Crohn's disease [10] and primary sclerosing cholangitis [24]. Genetic variation in FUT2 is also linked to susceptibility to infections with bacterial and viral pathogens including Helicobacter pylori [25], norovirus [26,27], Enterotoxigenic E. coli [5], and progression of HIV [28]. In this study, we investigated the role of Fut2 expression for S. Typhimurium infection and found that Std fimbriae-fucose interaction was important for Salmonella-induced inflammation and colonization.
Host mucosal glycans can influence the susceptibility to infection indirectly or directly. Indirectly, glycan-dependent differences in microbiota composition may contribute to the susceptibility to infection with a particular pathogen. For example, we previously reported an influence of the histo-blood group related glycosyltransferase gene B4galnt2 on host susceptibility to S. Typhimurium infection. The expression of B4galnt2 in the gut results in differences Salmonella fimbriae interaction with host glycans in microbial composition which in turn affect the extent of Salmonella colonization, and hence, disease pathology [29]. Many complex carbohydrates degraded by the intestinal microbiota produce metabolites that can be utilized by Salmonella Typhimurium and Clostridium difficile thereby facilitating their expansion within the gut [4]. Host glycans can also directly influence a host's susceptibility to infection by modulating bacterial attachment to host tissues. Many bacteria produce specific adhesins which bind to host glycans. For example, the H. pylori adhesin BabA mediates adherence to the gastric mucosa of individuals with fucosylated ABO (H)/Lewis b blood group antigens [30,31]. Norovirus (strain GII.4) and rotavirus (strains with spike protein VP8) encode adhesins which bind α(1,2)-fucosylated glycans resulting in increased susceptibility in individuals with a secretor phenotype [32,33].
While Fut2 expression in the small intestine is inducible by ILC3-derived IL-22 [17], the large intestine is constitutively fucosylated [8]. Mice lacking Fut2 are more susceptible to S. Typhimurium infection at an early time point post infection, as demonstrated by Goto and colleagues [17] and confirmed in the present study. In contrast, we demonstrate that at later time points, a lack of Fut2 expression is associated with decreased intestinal colonization, pathology, and inflammatory responses. It has been hypothesized that at one day post S. Typhimurium infection, α(1,2)-fucose-containing glycans secreted from goblet cells may interfere with the attachment of Salmonella to intestinal epithelial cells, although this has not been proven so far [34]. Our study suggests that at later time points post infection, Salmonella exploits α(1,2)-fucose-containing glycans present in the intestine of Fut2 +/+ mice to their advantage. Fucose and its metabolic products such as 1,2-propanediol can be utilized by S. Typhimurium as carbon and energy sources [35]. In vivo, during intestinal inflammation, 1,2-propanediol is generated and serves as a nutrient source for Salmonella [36]. Our data do not show any differences in the expression of fucI and pduBC in Fut2 +/+ and Fut2 -/mice, even in the absence of stdABCD genes, suggesting that these pathways are not regulated by Std fimbriae.
It was previously shown that purified Std fimbriae of S. Typhimurium can bind terminal α (1,2)-fucose residues [16]. Salmonella enterica encodes up to 13 fimbrial and at least 7 nonfimbrial adhesins depending on the serovar [13,37]. The production of adhesins involves complex and tight regulation since inappropriate expression could be detrimental for bacterial colonization and pathogenesis. While the majority of adhesins are expressed in temporal and spatially highly controlled manner during animal infections, they are often not produced under laboratory conditions [20,38]. In vitro expression of the S. Typhimurium std operon is bistable resulting in the emergence of a minor subpopulation of Std-positive cells [22]. Therefore, to investigate the role of Std fimbriae in vitro we took advantage of an Std-expressing E. coli strain. Our in vitro infection experiments revealed enhanced attachment of Std-expressing E. coli to fucosylated cell lines, which could be specifically blocked by the addition of the α (1,2)-fucose-binding lectin UEA-I. These observations confirm the findings of Chessa and colleagues [16] and were further corroborated by preferential binding of Std-expressing E. coli to fucosylated cells in primary epithelial monolayers. Altogether, bacteria expressing Std fimbriae exhibited increased adhesion to human cell lines and murine intestinal crypt organoids when terminal α(1,2)-fucose was present.
The extent and localization of Std fimbriae expression in vivo was not previously known. Notably, we observed the spatial expression pattern of Std fimbriae in vivo. In this work, we present, for the first time, evidence that Std fimbriae are differentially produced during intestinal infection. Std-positive Salmonella were found predominantly in the intestinal lumen, which corroborates the existence of a regulatory fimbrial switch.
Previous work has shown that Std fimbriae are required for the long-term S. Typhimurium colonization of CBA mice [39]. In agreement with these data, we show that Std fimbriae are important for persistence in C57Bl/6 mice and demonstrate that this is strictly dependent on the ability of Std fimbriae to adhere to fucosylated host glycoproteins or glycolipids in the large intestine. While the expression of α(1,2)-fucose on enterocytes and goblet cells is facilitated by Fut2, M cells in the follicle associated epithelium overlying Peyer's patches [40] and Paneth cells [41] are α(1,2)-fucosylated via Fut1. Thus, based on our CI data, we can conclude that Fut2-dependent fucosylation is important for Salmonella persistence only in the large intestine. Furthermore, we present evidence that Std fimbriae are also important for the induction of host intestinal pathology and inflammation in a Fut2-dependent manner.
In conclusion, our results demonstrate a substantial role for glycosylation of the intestinal mucosa in the susceptibility to S. Typhimurium infection. Std fimbriae binding of terminal α-1,2 fucose residues mediate bacterial adherence to host glycoproteins or glycolipids. Taken together, our results conclusively demonstrate that host fucosylation in the intestine is exploited by S. Typhimurium during the course of infection in a mechanism that requires Salmonella expression of Std fimbriae.

Mice
B6.129X1-Fut2 tm1Sdo /J (Fut2 -/-) [42] mice were purchased from the Jackson Laboratory and intercrossed with wild-type C57BL/6J (Fut2 +/+ ) mice. Mice were backcrossed for 14 generations. Heterozygous breeding pairs produced litters of mixed genotypes. Slc11a1 (Nramp1) is an important resistance gene for S. Typhimurium. 129X1 mice harbor Slc11a1 resistant alleles while in C57Bl/6J mice a point mutation results in Slc11a1 sensitive alleles. The Slc11a1 genotype of mice was verified as described [ [43]] using a common reverse primer and forward primers (S2 Table) specific for the sensitive and resistant allele, respectively. All mice were homozygous for the Slc11a1 sensitive allele. Mice were housed together under specific pathogen-free conditions in individually ventilated cages (IVC). Standard chow and water were provided ad libitum. Experiments were conducted in the animal facilities of the University of Kiel and Hannover Medical School in Germany.

Ethics statement
Animal experiments were conducted in direct accordance with the German Animal Protection Law consistent with the ethical requirements and approval of the Animal Care Committee of the Ministry of Energy, Agriculture, the Environment and Rural Areas of Schleswig-Holstein, Germany (protocol # V244-7224.121.3 (99-10/10)) and by the Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit (protocol # 33.12-42502-04-16/2071).
For in vitro infection experiments, E. coli stdABCD was grown overnight in LB broth supplemented with carbenicillin 50 μg/ml at 37˚C and then grown to logarithmic phase in the presence (StdON) or absence (StdOFF) of AHT (IBA, 100 ng/ml). The non-fimbriated E. coli ORN172 strain [38] harboring the empty vector or a plasmid for expression of the Salmonella stdABCD operon under control of the tetR P tetA expression cassette were grown in LB broth. Expression was induced by addition of 100 ng/ml anhydrotetracycline (AHT). Imaging of bacteria by atomic force microscopy (AFM) was performed as previously described [45].

Competitive index (CI)
Fut2 +/+ and Fut2 -/mice were pre-treated with 20 mg of streptomycin and infected with a mixture of S. Tm ΔaroA and S. Tm ΔaroAΔstdA strains (1:1, 3x10 6 total bacteria per mice). During the infection, fecal pellets were collected at days 1, 3, and 5 p.i. At 7 days p.i., mice were sacrificed and bacterial loads in cecum tissue, cecum content, colon tissue, colon content, and ileum were enumerated. Feces and intestinal tissues were homogenized and plated on selective LB agar plates supplemented with either streptomycin (100 μg/ml) alone (to determine total Salmonella load) or with streptomycin (100 μg/ml) and kanamycin (50 μg/ml) (to enumerate S. Tm ΔaroAΔstdA). The number of S. Tm ΔaroA was calculated by subtracting the CFU counts of S. Tm ΔaroAΔstdA from the total Salmonella counts. Competitive index (CI) was calculated as the ratio of (ΔaroA / ΔaroAΔstdA) at the time of sampling divided by (ΔaroA / ΔaroAΔstdA) of the inoculum.

Motility assay
Motility was assessed by inoculating motility agar plates (10 g/l tryptone, 5 g/l NaCl, and 0.3% Bacto-agar) with saturated bacterial cultures grown overnight in LB broth. Motility halos were compared after incubation at 37˚C for 6 hours.

Pathology and histology
Organs were fixed in 10% formalin, dehydrated with ethanol, and embedded in paraffin. Paraffin sections were stained with hematoxylin-eosin (H&E) according to standard laboratory procedures. Histological scores of ceca and colons were determined as previously described [46]. Briefly, pathological changes were assessed by evaluating the presence of necrotic epithelial cells and neutrophils in lumen; desquamation and ulceration in surface epithelial cells; crypt abscesses; infiltrating inflammatory and immune cells in mucosa and submucosa layer; and the formation of edema in submucosa layer.

In vitro infection
Human colon epithelial clonal cell line, HT29-MTX-E12 [47,48] (a kind gift from Marguerite Clyne, University College Dublin), and the colorectal carcinoma cell line, Caco-2BBe1 (ATCC CRL-2102), were grown in DMEM supplemented with 10% fetal bovine serum (Biochrom) and 1% MEM non-essential amino acids solution (Gibco, Life Technologies). Cells were seeded in 24 well plates and incubated at 37˚C in a humidified 5% CO 2 atmosphere. Cells were grown for 7 days (Caco-2-BBe1) or 21 days (HT29-MTX-E12) to achieve differentiation of the monolayers. Cells were then infected with E. coli StdOFF or E. coli StdON, E. coli ORN172 StdOFF or E. coli ORN172 StdON for 30 min at an MOI of 50. For quantification of adherence, cells were washed four times with PBS and lysed in PBS containing 1% (v/v) Triton X-100. The number of adherent bacteria was determined by serial dilutions plating. Where indicated, 30 min prior to infection, cells were incubated with medium in presence of 0.3 mM UEA-I or DBA lectins (CosmoBio) at 37˚C in a humidified 5% CO 2 atmosphere.

Flow cytometry
Isolation of colonic lamina propria cells was achieved using the Lamina Propria Dissociation Kit (Miltenyi Biotec) according to the manufacturer's protocol. Leukocyte isolation was performed with 40% / 80% discontinuous Percoll gradient (GE Healthcare). Cells were incubated with FcγR blocking reagent (rat anti-mouse CD16/CD32, BD Biosciences) for 30 minutes on ice prior to incubation with the other antibodies. Antibodies used for flow cytometry analysis are listed in S1 Table. To detect expression of Std fimbriae, flow cytometry was performed as previously described [16] with modifications. In brief, approximately 5x10 8 bacteria were fixed with 10% formalin and incubated at room temperature for 20 minutes. After washing with PBS, cells were resuspended in 2% NGS diluted in PBS and incubated at room temperature for 30 minutes. Polyclonal rabbit anti-StdA serum [38] was added to the cell suspensions following incubation at room temperature for 30 min. After washing with PBS, fluorescently labeled secondary antibodies (Invitrogen) were added. Flow cytometry was performed using a MACS-Quant Analyzer 10 (Miltenyi Biotec). The data were analyzed using FlowJo v.10 software (TreeStar).

ELISA
Supernatants from the organ homogenates were collected and stored at −20˚C. Lipocalin-2 levels were detected using mouse lipocalin-2/NGAL DuoSet ELISA (R&D Systems) according to the manufacturer's protocol. Absorbance was measured using a Synergy HTX microplate reader and acquired using Gen5 software (Biotek).

Western blot
E. coli StdOFF and E. coli StdON were grown in LB broth supplemented with carbenicillin (50 μg/ml) and to induce std fimbrial expression anhydrotetracycline (100 ng/ml) at 37˚C until an OD 600 of 0.6 was reached. 10 8 bacteria were pelleted, resuspended in PBS, mixed with an equal volume of Laemmli buffer supplemented with 10% DTT, and boiled for 10 min. These whole-cell lysates were spun down and supernatants were loaded immediately onto a SDS-PAGE gel (15%). Proteins were transferred to Hybond-P 0.45 PVDF (Amersham) membranes using a Trans-Blot semi-dry transfer cell (Bio-Rad). After blocking with Roti-block (Carl Roth), membranes were incubated first with anti-StdA [38] serum (Humphries et al., 2003) diluted 1:500 in blocking buffer, and then with a goat anti-rabbit-HRP conjugate, and finally with Pierce ECL Western Blotting Substrate (Thermo Fisher Scientific). Images were obtained using the ImageQuant LAS 4000 system (GE Healthcare).

Salmonella gene expression in vivo
Feces of the infected Fut2 +/+ and Fut2 -/mice (day 7 p.i.) was immediately stored in RNAlater (Ambion). Total RNA was extracted using the High Pure RNA Tissue Kit (Roche) and reverse transcription was performed with the cDNA Synthesis Kit (Roche) in accordance with the manufacturer's instructions. Quantitative real-time PCR (qPCR) was performed on a CFX96 Real-Time PCR Detection System (Bio-Rad) using the Power SYBR Green PCR Master Mix (Applied Biosystems). Gene-specific primers are listed in S2 Table. Data were normalized to the house-keeping gene rpoD and analyzed by ΔΔCt method [50] with median values of Fut2 +/+ mice infected with S. Tm ΔaroA as calibrators.

Statistical analyses
All data were analyzed using GraphPad Prism V7.0d software. Statistical analyses were performed using one-way analysis of variance followed by Tukey's multiple comparison test or Wilcoxon-Mann-Whitney test as indicated. Graphs display the mean values ± SD, unless stated otherwise. Competitive index data were analyzed by the Wilcoxon signed-rank test by comparing medians with a hypothetical value of 1.
Supporting information S1