Salmonella virulence factors induce amino acid malabsorption in the ileum to promote ecosystem invasion of the large intestine

Abstract

The gut microbiota produces high concentrations of antimicrobial short-chain fatty acids (SCFAs) that restrict the growth of invading microorganisms. The enteric pathogen Salmonella enterica serovar (S.) Typhimurium triggers inflammation in the large intestine to ultimately reduce microbiota density and bloom, but it is unclear how the pathogen gains a foothold in the homeostatic gut when SCFA-producing commensals are abundant. Here, we show that S. Typhimurium invasion of the ileal mucosa triggers malabsorption of dietary amino acids to produce downstream changes in nutrient availability in the large intestine. In gnotobiotic mice engrafted with a community of 17 human Clostridia isolates, S. Typhimurium virulence factors triggered marked changes in the cecal metabolome, including an elevated abundance of amino acids. In an ex vivo fecal culture model, we found that two of these amino acids, lysine and ornithine, countered SCFA-mediated growth inhibition by restoring S. Typhimurium pH homeostasis through the inducible amino acid decarboxylases CadA and SpeF, respectively. In a mouse model of gastrointestinal infection, S. Typhimurium CadA activity depleted dietary lysine to promote cecal ecosystem invasion in the presence of an intact microbiota. From these findings, we conclude that virulence factor-induced malabsorption of dietary amino acids in the small intestine changes the nutritional environment of the large intestine to provide S. Typhimurium with resources needed to counter growth inhibition by microbiota-derived SCFAs.

Keywords: Salmonella; colonization resistance; microbiota; short-chain fatty acids.

Fig. 1.

S. Typhimurium virulence factors change the cecal metabolome. Germ-free Swiss Webster mice were engrafted with a defined microbial consortium consisting of 17 human Clostridia isolates(29). One week later, mice were mock infected (mock) (N = 5), infected with the S. Typhimurium WT (N = 8), or with an avirulent S. Typhimurium invA spiB mutant (invA spiB) (N = 6). Cecal contents were collected 3 d after infection for untargeted metabolomics analysis. (A) Volcano blot showing metabolite abundance in the ceca of mock vs WT-infected mice. The Y axis shows the decadic logarithm of the false discovery rate (FDR)-corrected P value. The dashed line is set at an FDR-corrected P value of 0.05. Metabolites with a negative fold-change value decreased in mice infected with S. Typhimurium, while metabolites with a positive fold-change value increased. (B) Principal component analysis of the cecal metabolome in the indicated groups of mice. Ovals indicate the 95% CI. (C) The intestinal burden of S. Typhimurium was tracked for 3 d by enumerating colony-forming units (CFU) of S. Typhimurium in the feces (days 1 and 2) or cecal content (day 3). (B and C) Each dot represents data from one animal. (D) The graphs show the abundance of the indicated metabolites in mock-infected mice compared to mice infected with the S. Typhimurium WT. (C and D) Bars represent geometric means ± SE. ns, not significant (Mann–Whitney).

Fig. 2.

S. Typhimurium virulence factors trigger malabsorption of amino acids. (A–G) Groups of CBA/J mice (N = 8) were mock infected (mock) or infected with 10⁹ CFU of the S. Typhimurium WT or an isogenic S. Typhimurium invAspiB mutant (invA spiB). At the indicated timepoints, groups of 8 mice were killed to collect samples. (A–E) Ileal epithelial cells were isolated for extraction of host mRNA. Fold changes in transcript levels of Lcn2 (A), Slc6a19 (B), Slc7a9 (C), Slc1a1 (D), and Slc3a1 (E), were determined by qRT-PCR. Graphs show geometric mean ± SE. (A) The 3-d time point for Lcn2 mRNA elicited by infection with the S. Typhimurium WT has been published previously (42). (F and G) S. Typhimurium CFU were determined in cecal (F) or ileal (G) contents. Thick lines and symbols indicate the geometric mean. Thin lines indicate the SD. (H) Groups of CBA/J mice (N = 6) were mock infected or infected with WT or invA spiB. Four days later, mice were orally inoculated with a mixture of ¹³C-labeled alanine, ¹³C-labeled methionine, and ¹³C-labeled lysine, and serum samples were collected 30 min later. The graphs show the serum concentrations of the indicated ¹³C-labeled amino acids in the indicated groups. Bars represent geometric means ± SE. (A–E and H) Kruskal–Wallis; (F) Mann–Whitney; *P < 0.05; **P < 0.01; ***P < 0.001; ns, P > 0.05.

Fig. 3.

Decarboxylation of lysine and ornithine counters SCFA-mediated acidification of the bacterial cytosol. Filter-sterilized murine fecal homogenates were adjusted to the indicated pH and inoculated with the indicated S. Typhimurium strains. Each experiment was repeated three times. (A) Growth of S. Typhimurium at the indicated pH in the presence (+) or absence (–) of 50 mM acetate, 25 mM butyrate, and 6 mM propionate (SCFAs) was determined by measuring the optical density at 600 nm (OD₆₀₀). Bars represent geometric mean ± SD. (B–G) S. Typhimurium was transformed with pGFPR01 to measure cytosolic pH and bacterial growth (OD₆₀₀) in fecal homogenates supplemented with 50 mM acetate at pH 6.7 (B) or pH5.7 (C–G). S. Typhimurium growth and cytosolic pH were determined at pH 5.7 in the absence of amino acid supplementation (C) or after supplementation with lysine (F) or ornithine (G). (D) S. Typhimurium growth (OD₆₀₀) 48 h after inoculation of SCFA-supplemented fecal homogenates at pH 6.7 or 5.7 in the presence of the indicated metabolites or respiratory electron acceptors. lys, 10 mM lysine; orn, 10 mM ornithine; ser, 10 mM serine; asp, 10 mM aspartate; fum, 10 mM fumarate; mal, 10 mM malate; eth, 10 mM ethanolamine; O₂, 7% oxygen; NO₃⁻, 20 mM nitrate; S₄O₆²⁻, 40 mM tetrathionate. (E) Growth (OD₆₀₀) of the indicated S. Typhimurium strains in SCFA-supplemented fecal homogenates adjusted to pH 6.7 or 5.7 in the absence of amino acid supplementation or when supplemented with 20 mM lysine or 20 mM ornithine. *P < 0.05; ***P < 0.001; ****P < 0.0001; ns, not significant (one-way ANOVA with Dunnett’s multiple comparison test).

Fig. 4.

Malabsorption of dietary lysine promotes cadBA-mediated ecosystem invasion. (A and B) Groups of antibiotic-naïve CBA/J mice (N = 8) were infected with 10⁹ CFU of the S. Typhimurium WT or the indicated isogenic mutant strains by oral gavage. Thick lines and symbols indicate the geometric mean. Thin lines indicate the SD. (C) Groups of antibiotic-naïve CBA/J mice (N = 8) received regular drinking water (mock) or drinking water supplemented with 3% lysine. One day after the beginning of lysine supplementation, mice were infected with 10⁹ CFU of the S. Typhimurium WT or an isogenic cadBA mutant. The concentration of lysine was determined at the indicated time points after infection. Bars represent geometric mean ± SD. Each dot represents data from one animal. (A) Mann–Whitney; (H) Kruskal–Wallis; *P < 0.05; ***P < 0.001.