Sublethal systemic LPS in mice enables gut-luminal pathogens to bloom through oxygen species-mediated microbiota inhibition

Abstract

Endotoxin-driven systemic immune activation is a common hallmark across various clinical conditions. During acute critical illness, elevated plasma lipopolysaccharide triggers non-specific systemic immune activation. In addition, a compositional shift in the gut microbiota, including an increase in gut-luminal opportunistic pathogens, is observed. Whether a causal link exists between acute endotoxemia and abundance of gut-luminal opportunistic pathogens is incompletely understood. Here, we model acute, pathophysiological lipopolysaccharide concentrations in mice and show that systemic exposure promotes a 100-10'000-fold expansion of Klebsiella pneumoniae, Escherichia coli, Enterococcus faecium and Salmonella Typhimurium in the gut within one day, without overt enteropathy. Mechanistically, this is driven by a Toll-like receptor 4-dependent increase in gut-luminal oxygen species levels, which transiently halts microbiota fermentation and fuels growth of gut-luminal facultative anaerobic pathogens through oxidative respiration. Thus, systemic immune activation transiently perturbs microbiota homeostasis and favours opportunistic pathogens, potentially increasing the risk of infection in critically ill patients.

Fig. 1. Sublethal systemic LPS exposure promotes gut-luminal pathogens to bloom in a TLR4-dependent manner.

a C57BL/6 specific pathogen-free (SPF) mice were intravenously (i.v.) injected with 5 µg lipopolysaccharide (LPS) and orally infected with 5 × 10⁷c.f.u. Salmonella Typhimurium (S. Tm), faeces were collected 24 h post-infection (h.p.i), unless indicated otherwise (b) Faecal S. Tm loads in mice systemically exposed to PBS or LPS or orally pre-treated with streptomycin (strep.)(minimum mice n = 8, at least two independent replicates, ^***P = 0.0002). c Number of S. Tm replications in mice systemically exposed to PBS or LPS or orally pre-treated with strep. (minimum mice n = 6, at least two independent replicates, ^*P = 0.0478, ^***P = 0.0008). d Faecal ΔsopB ΔsipA ΔsopE ΔsopE2 (Δ4) S. Tm loads in mice systemically exposed to PBS or LPS (minimum mice n = 6, at least two independent replicates, ^**P < 0.0012). e Percentage caecal S. Tm expressing SicA in mice orally pre-treated with strep. or systemically exposed to LPS (minimum mice n = 13, at least two independent replicates, ^****P < 0.0001). f–h Faecal Escherichia coli, Klebsiella pneumoniae, and Enterococcus faecium loads in mice systemically exposed to PBS or LPS (minimum mice n = 5, at least two independent replicates, ^**P = 0.0079, ^**P = 0.0022^{, **}P = 0.0022). (i) Faecal S. Tm loads at 24 h and 48 h.p.i. in Oligo-MM¹² mice orally infected with avirulent S. Tm at 0 h.p.i. and systemically exposed to PBS or LPS at 24 h.p.i. (minimum mice n = 5, at least two independent replicates, ^**P = 0.0087). j Faecal S. Tm loads in Tlr4^+/- and Tlr4^-/- littermates systemically exposed to PBS or LPS (minimum mice n = 6, at least two independent replicates). k Faecal S. Tm loads in mice systemically exposed to 1 or 5 µg LPS (minimum mice n = 6, at least two independent replicates, ^**P = 0.0043). (I, m) Faecal S. Tm loads in mice systemically exposed to PBS or 12.5 µg flagellin or 1.8 μg ODN 2395 (minimum mice n = 6, two independent replicates). Bars indicate median values. Dotted lines indicate conservative average limit of detection. P values were calculated using the two-sided Mann-Whitney U test (b, d–o) or two-sided Kruskal-Wallis test with Dunn’s multiple test correction (c). ns, not significant. Source data are provided in the Source Data file.

Fig. 2. Systemic LPS exposure triggers a transient inflammatory response that elevates gut-luminal oxygen species levels.

a Caecal lipocalin-2 levels at 24 h.p.i. in mice systemically exposed to PBS or LPS or orally pre-treated with streptomycin (minimum mice n = 5, at least two independent replicates, ^***P = 0.0002). b Representative images of haematoxylin and eosin-stained caecum tissue at 0, 6, 12 or 24 h post-injection (h.p.inj) or 24 h.p.i. in mice systemically exposed LPS or orally pre-treated with streptomycin. Scale bar=80 µm (mice n = 6, at least two independent replicates). c Caecal lipocalin-2 levels at 0, 3, 6, 9 or 12 h.p.inj. in mice systemically exposed to LPS (minimum mice n = 6, at least two independent replicates, ^**P = 0.0013, ^****P < 0.0001). d Caecum tissue RNAseq at 6 h.p.inj. in mice systemically exposed to LPS compared to PBS. Over-representation analysis of biological processes based on significantly upregulated genes (minimum mice n = 6). e Caecum tissue RNAseq log2 fold change of classical bacterial response genes at 0 and 6 h.p.inj. in mice systemically exposed to LPS compared to PBS (minimum mice n = 6). f Caecal oxygen species levels at 3 h.p.inj. in mice systemically exposed to PBS or LPS, treated with or without streptomycin (minimum mice n = 6, at least two independent replicates, ^**P = 0.0022)^. Bars indicate median values. Dotted lines indicate conservative average limit of detection. P values were calculated using the two-sided Kruskal-Wallis test with Dunn’s multiple test correction (a,c) or two-sided Mann-Whitney U test (f). ns, not significant. Source data are provided in the Source Data file.

Fig. 3. S. Tm employs aerobic respiration and the oxidative TCA cycle to bloom in the gut of LPS-exposed mice.

a Competitive index of ΔcyxA, ΔcydB, ΔcyxA ΔcydB, and ΔcyoA S. Tm in 1:1 competition with WT S. Tm at 24 h.p.i. in mice systemically exposed to LPS (minimum mice n = 5, at least two independent replicates, ^**P = 0.0087). b Competitive index of Δsdh and Δfrd S. Tm in 1:1 competition with WT S. Tm at 24 h.p.i. in mice systemically exposed to LPS (minimum mice n = 8, at least two independent replicates, ^****P < 0.0001). c Competitive index of ΔgltA, ΔacnAB, ΔicdA, ΔsucAB, ΔsucCD, Δfrd, Δsdh, ΔfumABC and Δmdh S. Tm in 1:1 competition with WT S. Tm at 24 h.p.i. in mice systemically exposed to LPS (minimum mice n = 4, at least two independent replicates). Bars indicate median values. Dotted lines indicate competitive index of 1. P values were calculated using the two-sided Kruskal-Wallis test with Dunn’s multiple test correction (a,b). ns, not significant. Source data are provided in the Source Data file.

Fig. 4. Systemic LPS exposure triggers a microbiota stress response that inhibits microbiota fermentation.

a Caecal content anaerobic microbiota loads at 6 h.p.inj. in mice systemically exposed to PBS or LPS (mice n = 6, at least two independent replicates). b 16S rRNA gene sequencing Shannon diversity index at 0, 6, 12 or 24 h.p.inj. in mice systemically exposed to LPS (minimum mice n = 5 at least two independent replicates). c Metatranscriptome analysis of the caecal microbiota at 6 h.p.inj. in Oligo-MM¹² mice systemically exposed to LPS compared to PBS, representative volcano plots of gene distribution of Blautia pseudococcoides YL58, Enterocloster clostridioformis YL32 and Bacteroides caecimuris I48 (minimum mice n = 6, at least two independent replicates). d Caecal L-glutamate levels in germ-free and SPF mice systemically exposed to PBS or LPS at 6 h.p.inj. (minimum mice n = 6, at least two independent replicates, ^***P = 0.0010). e–g Caecal acetate, propionate, and butyrate levels at 0, 3, 6, 9 or 12 h.p.inj. in mice systemically exposed to LPS (minimum mice n = 6, at least two independent replicates, acetate: ^**P = 0.0034, propionate: 0 h vs 9 h ^*P = 0.0446 and 0 h vs 12 h ^*P = 0. 0372, butyrate: ^*P = 0.0309 and ^**P = 0.0095). h Hydrogen levels of mice systemically exposed to PBS or LPS at 32 h, curves obtained by smoothing function of data obtained every 24 min per mouse (mice n = 4, at least two independent replicates, ^*P = 0.0286). Bars indicate median values. Dotted lines indicate conservative average limit of detection. Dashed lines indicate time of injection. Grey rectangles indicate dark phase. P values were calculated using the two-sided Mann-Whitney U test (a,h), two-sided Kruskal-Wallis test with Dunn’s multiple test correction (b,d-g) or two-sided Wald test with Benjamini-Hochberg multiple test correction (c). ns, not significant. Source data are provided in the Source Data file.

Fig. 5. Facultative anaerobic pathogens exhibit a higher tolerance to oxidative stress than common commensal microbiota strains.

a–c Bacterial strains grown twice overnight and adjusted to a starting OD of 0.05 were exposed to H₂O₂ concentrations ranging from 0.375 to 12 mM. Growth curves were measured and inhibitory concentrations 25 and 75 (IC25, IC75) were calculated (wells n = 4). Source data are provided in the Source Data file.

Fig. 6. Oxidative stress inhibits human gut microbiota fermentation and promotes facultative anaerobic pathogens to bloom.

a–d An inoculation reactor containing immobilised faecal microbiota of an adult donor inoculated test reactors (TR1-4). TRs were inoculated with S. Tm at −12h to reach a density of 1 × 10⁸c.f.u. ml^-1. 12 h post S. Tm inoculation, TRs were exposed to varying H₂O₂ concentrations 0 mM (TR1), 1.5 mM (TR2), 3 mM (TR3), 8 mM (TR4) (reactors n = 4). TR effluent was collected every 3 h to 12 h. L-glutamate, acetate, propionate, and butyrate concentrations in TR effluent post-H₂O₂ exposure (reactors n = 4). e S. Tm loads in TR effluent post-H₂O₂ exposure (reactors n = 4). f 16S rRNA gene sequencing family-level microbiota composition in TR effluent post-H₂O₂ exposure (reactors n = 4). Dashed lines indicate time of H₂O₂ exposure. Source data are provided in the Source Data file.