Strains and plasmids are listed in Table EV1. The EHEC O157:H7 strain EDL933 (O'Brien et al., 1983) was used in this study. B. thetaiotamicron strain VPI-5482 was grown anaerobically at 37°C in a complex medium containing clarified rumen fluid (Leedle and Hespell, 1980). The medium was prepared, dispensed and inoculated by using strictly anaerobic techniques in Balch tubes. The EDL933 ΔnagC and ΔnanR mutants were generated by allelic exchange using a suicide vector as described in EV Methods. When required, the growth medium was supplemented with kanamycin (25 mg/ml), ampicillin (50 mg/ml), NANA (0.1 mM or 1 mM), or NAG (0.1 mM or 1 mM) (Sigma Aldrich).

The entire intergenic region between ler (LEE1) and espG (bp −1,225 to +19) containing two ler promoters (Sperandio et al., 2002; Porter et al., 2005) was inserted upstream of lacZ in pRS551 (see EV Methods). The resulting plasmid pGLB was introduced into EDL933 or its isogenic mutants. After growth in DMEM with or without NANA or NAG until OD₆₀₀ of 0.6, β-galactosidase assays were performed as described previously (Miller, 1972). Student t-tests were performed to determine statistical significance.

Bacteria were harvested at OD₆₀₀ of 0.6 and RNA was extracted as previously described (Le Bihan et al., 2015). cDNAs were synthesized from 10 μg RNA using reverse transcriptase. The concentration of cDNA samples was then adjusted to 25 ng/μL. A standard curve was performed for genes of interest to determine the copy number of targeted transcripts in 50 ng of cDNA. Primers used in this study are listed in Table EV2. Results are presented as the ratios between the cDNA copy number of the gene of interest and the cDNA copy number of the housekeeping gene. Student t-tests were performed to calculate p-values.

Western blot analyses were performed with slight modifications to those previously described (Chekabab et al., 2014). Culture supernatants (8 ml) were harvested and supplemented with 1 μg Bovine Serum Albumin (BSA). Proteins were precipitated overnight at 4°C using 10% trichloroacetic acid and sodium deoxycholate, pelleted by centrifugation, washed with acetone, resuspended in SDS sample buffer and boiled. Proteins were then run on 14% SDS-PAGE gels and transferred to nitrocellulose membranes. Protein transfer including BSA was assessed using Ponceau S dye. The EspB protein was revealed using a rabbit EspB specific polyclonal antiserum (1:2,000) and a goat anti-rabbit HRP-linked secondary antibody (Bio-Rad Laboratories, Hercules, CA).

Epithelial cell lines HeLa, HCT-8, and HCT-116 were maintained in MEM with 10% FBS, 100 U/mL penicillin and 100 mg/mL streptomycin at 37°C under 5% CO₂(Branchu et al., 2014). Cells were seeded into 6-well plates (5 × 10⁵ cells/well) and grown for 24 h without antibiotics. Bacteria were pre-grown in DMEM in the presence or absence of NANA or NAG before the infection assays. Cells were then washed and infected with bacteria with an MOI of 10 for 90 min, in the presence or absence of NANA or NAG. Following infection, cells were washed 3 times with DPBS, trypsinized for 5 min at 37°C, pelleted by a low-speed centrifugation (100 g; 3 min). The pellet was washed once to ensure the plating of adherent bacteria only. Results are presented as the percentage of adherent bacteria as compared to the wild-type strain incubated without NANA or NAG. The EDL933 ΔescN strain that does not produce the T3SS (Deng et al., 2004) was used as a negative control. Student t-tests were performed to determine the significance.

The EMSA was adapted from previous report (Chekabab et al., 2014). The EMSA reaction mix consisted of purified NagC at the desired concentration (0.5–2.5 μM), 50 nM of 5′ 6-FAM labeled pLEE1 probe, 0.1 mg/mL calf thymus DNA and 0.1 mg/mL BSA in EMSA buffer (50 mM NaCl, 20 mM Tris, pH 7.4, 0.02% v/v sodium azide). Reactions were incubated for 30 min at 25°C, and then loaded onto a 12% native polyacrylamide gel running at 120 V in 1x TBE buffer. The forward primer contained a 5′ 6-FAM fluorescein tag. Competitive EMSA assay was done with 50 nM of 6-FAM labeled pLEE1 probe and unlabeled probes corresponding to P_LEE1, P_kan (negative control) or P_nagB−nagE (positive control). The ratio “cold probe/labeled probe” was 10/1.

DNase I footprinting of free DNA and DNA–protein complexes was performed as described (El Qaidi et al., 2009; Graveline et al., 2011). The DNA fragment corresponding to the ler regulatory region (259 bp) was amplified using primers listed in Table EV2, alternately end labeled with ³²P (140,000 cpm, 0.6 nM). Each end-labeled amplicon was subsequently incubated in a total volume of 80 μl in binding buffer (25 mM Hepes (pH 8.0), 100 mM K glutamate (pH 8.0), 0.5 mg/ml BSA). After incubation with purified NagC for 10 min at room temperature, 2 μl of DNase I (1.3 U/ml; New England BioLabs) containing 5 mM CaCl₂ and 25 mM MgCl₂ was added for 5 min. The reaction was stopped by the addition of 150 μL of phenol/chloroform/isoamyl alcohol and 350 μl of stop buffer (0.5 M Na acetate pH 5.0, 2.5 mM EDTA, 10 μg/ml Salmon sperm DNA) to each sample. DNA fragments were precipitated in ethanol, and amounts with equivalent cpm (5.10⁴) from each reaction were loaded onto 6% polyacrylamide–7 M urea gels. Maxam-Gilbert A+G reactions were carried out on the appropriate 32 P-labeled DNA fragments, and the products loaded alongside the DNase I footprinting reaction mixtures. The gels were dried and analyzed by autoradiography. A control footprinting experiment realized with nagE-nagB regulatory region and with NagC (Figure EV5).

BALB/c mice were purchased from Janvier Labs (Le-Genest-St-Isle, France). Sets of 5 mice aged 5 weeks were given drinking water containing streptomycin sulfate (5 g/l) throughout the experiment. On day 1 following the addition of streptomycin, each mouse was infected intragastrically with 100 μl of a mix containing 10⁷ each of EDL933 Sm^R and EDL933 Sm^R ΔnagC strains. Mice treated with B. thetaiotaomicron were gavaged daily with 5 × 10⁹ cells of B. thetaiotaomicron strain VPI-5482, starting from day 1 before EHEC infection to day 7. At indicated time points, fecal or tissue samples were collected, homogenized in PBS and subsequently diluted before plating on LB-streptomycin agar plates and LB-streptomycin-kanamycin agar plates. Output ratios were calculated for each time point and competitive indices were obtained by dividing the output ratio by the input ratio. A One-way ANOVA was performed to determine the significance. For NANA and NAG quantification in intestinal contents, see EV Methods.

All animal experiments were reviewed and approved by the Auvergne Committee for Animal Experimentation (C2E2A). All procedures were carried out according to the European directives for the protection of animals used for scientific purposes, 2010/63/EU, and to the guidelines of the local ethics committee.

The effect of NANA and NAG on EHEC adhesion was examined by measuring the ability of the EHEC O157:H7 EDL933 strain to adhere to cultured epithelial cells in the presence or absence of NANA or NAG. Our data showed that EDL933 adhesion to HeLa cells was significantly decreased by 40 ± 21 or 23 ± 11% in presence of 1 mM NAG or NANA, respectively (Figure 1A; Figure EV1). At 0.1 mM, only NAG significantly decreased the number of cell-attached bacteria (53 ± 22%).

The adhesion of EHEC O157:H7 to HeLa cells is mainly driven by the production of a T3SS, as previously demonstrated (Branchu et al., 2014) and as verified using the ΔescN strain which is defective in the production of the T3SS (Figure EV1). Thus, we investigated the effect of NANA and NAG on LEE gene expression. As a first step, strain EDL933 carrying a P_LEE1:lacZ fusion was cultured in the presence of different concentrations of the sugars. The expression of LEE1 was significantly repressed in the presence of either 1 mM NANA or 0.1 and 1 mM NAG (Figure 1B), but not at 0.01 mM of either sugar. Next, we examined the expression of genes from the five LEE operons and observed that expression of ler (LEE1), sepZ (LEE2), escV (LEE3), espB (LEE4), tir, and eae (LEE5) was significantly lower in presence of 1 mM NANA or NAG (Figure 1C). Consistent with the decreased expression of the LEE4 gene, the secretion of the effector EspB, encoded by the LEE4, was dose dependently reduced when EDL933 was incubated with 0.1 and 1 mM of either sugar (Figure 1D). Taken together, these data indicate that NANA and NAG inhibit the adhesion of EDL933 to epithelial cells and repress T3SS encoding genes. In addition no significant difference was observed upon addition of other mucin sugars, such as mannose, galactose, N-acetylgalactosamine and glucuronate on the expression of ler (Figure EV2).

Activation of the metabolic pathways required for the catabolism of NANA and NAG influence the activity of two transcriptional regulators, NanR and NagC (Figure 2A). Intracellular NANA inactivates NanR while NAG-6P derived from the catabolic conversion of both NANA and NAG inactivates NagC. To evaluate the role of NanR and NagC on LEE1 promoter activity, the P_LEE1:lacZ fusion was introduced into ΔnanR and ΔnagC mutants. The nagC deletion led to a significant decrease of the activity of P_LEE1 whereas the deletion of nanR had no significant effect (Figure 2B). Moreover, the addition of NANA or NAG still repressed the activity of P_LEE1 in the ΔnanR mutant whereas it did not in the ΔnagC mutant. These results indicate that LEE1 repression by NANA and NAG was NagC-dependent but NanR-independent. We further demonstrated that a nagC deletion also impairs the expression of escV, sepZ, espB, tir, and eae genes with a fold change ranging from −2.9 to −5.5 (Figure 2C), as well as secretion of EspB (Figure 2D). Consequently, the ΔnagC mutant adhered less to epithelial cells, using HeLa and the two human intestinal cell lines HCT-8 and HCT-116 (Figure 2E; Figures EV1, EV3). Importantly, LEE gene expression, EspB secretion and adhesion levels were restored to a wild-type status in a ΔnagC-complemented strain. Additionally, we assessed the potential involvement of NagC in the transcription of other genes encoding adhesins in EDL933. Neither fliC (flagellin), ycbQ (fimbriae) (Samadder et al., 2009), hcpA (hemorrhagic coli pilus), espP (autotransporter), lpfA (long polar fimbriae), csgA (curli) nor pgaB (poly-β-1,6-N-acetyl-D-glucosamine) were differentially expressed between wild type and ΔnagC strains (Supplementary Figure EV4). Since NagC does not affect the synthesis of these adhesins, it suggests that the defect in cell adhesion observed with the nagC mutant is mainly driven by a reduced production of the T3SS.

Using a NagC consensus DNA binding site generated from seven NagC binding sequences, we identified a single putative NagC binding site in the LEE1 promoter region (5′-GTATTTTACACATTAGAAAAAAG-3′) located at a position that overlaps the −10 box of the distal promoter (Figure 3A). The binding of NagC to the LEE1 promoter was first investigated by EMSA that showed that NagC forms a specific low-mobility complex with the LEE1 promoter as previously observed with NagC interacting with type 1 fimbriae fim intergenic region upstream of the fimB promoter (Sohanpal et al., 2004). Competition EMSA using an unrelated probe (P_kan) demonstrated that NagC binding to the LEE1 promoter is specific (Figure 3B). Further, DNase footprinting experiments confirmed that NagC bound to the predicted NagC binding site (Figure 3C). Consistent with the expected specificity, a single base substitution at position 18 of the putative NagC binding site prevented DNAse protection by NagC (Figure 3D). An additional DNase footprinting control experiment using nagBE intergenic region as expected showed clear protection zones indicating the functional activity of NagC (Figure EV5). These findings demonstrate that NagC interacts with the LEE1 promoter region in a sequence specific manner. Interestingly, the NagC-binding sequence in the promoter of ler is conserved among other EHEC O157:H7 strains and this is correlated with ler repression in the presence of 1 mM of NAG (Figure EV6). EHEC, EPEC, or C. rodentium strains with degenerated NagC binding site in the LEE1 promoter region were insensitive to NAG exposure.

To assess if NagC regulates the gut colonization process, we co-infected mice with an equal mixture of wild-type EDL933 and the ΔnagC mutant and followed the outcome of each strain overtime. We observed a marked increase in the wild-type strain over the ΔnagC mutant in the feces at days 6 and 8 post-infection (competitive index (CI) of 13 ± 5 and 20 ± 6, respectively), as well as in the cecal content at day 8 (CI of 270 ± 76) (Figures 4A,B). The competitive advantage of the wild-type strain was also recorded for bacteria adhering to cecal and colonic mucosa (Figure EV7). These data demonstrate that the deletion of nagC greatly impaired the ability of EDL933 to colonize the intestinal tract of mice.

We next sought to determine if the concentrations of NANA and NAG may alter EDL933 fitness in vivo through the modulation of NagC activity. For that, the drinking water of WT/ΔnagC-infected mice was supplemented with purified NANA or NAG. Supplementation led to increased sugar concentrations in the cecal content of uninfected mice but not in the cecal content of infected mice (Figure 4C). Interestingly, NANA and NAG concentrations also significantly decreased in supplemented mice upon infection, with fold-change of 4.9- and 1.9, respectively, suggesting that EDL933 consumes NANA and NAG in the intestine of mice. In these conditions, NAG supplementation significantly decreased the competitive advantage of the wild-type strain over the ΔnagC mutant by a factor 7.1 (Figure 4B). Co-infected mice were also subjected to a daily gavage with B. thetaiotaomicron to see if the behavior of EHEC is modulated by the population level of a mucin degrading bacterium. Gavage of mice did not change NANA concentration in the gut of infected mice but led to a 1.9-fold increase of NAG concentration (Figure 5). Importantly, we observed that the competitive index between the wild-type strain and the ΔnagC mutant significantly dropped from 297 in control mice to 54 in B. thetaiotaomicron–treated animals (Figure 5). Overall, our findings indicate that NAG concentration in the intestine, derived notably from activity of mucin degrading commensals, such as B. thetaiotaomicron, affects the fitness of EHEC in vivo in a NagC-dependent manner.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.