Total Lipopolysaccharide from the Human Gut Microbiome Silences Toll-Like Receptor Signaling

Abstract

Cohabitation of microbial communities with the host enables the formation of a symbiotic relationship that maintains homeostasis in the gut and beyond. One prevailing model suggests that this relationship relies on the capacity of host cells and tissues to remain tolerant to the strong immune stimulation generated by the microbiota such as the activation of Toll-like receptor 4 (TLR4) pathways by lipopolysaccharide (LPS). Indeed, gut microbial LPS is thought to be one of the most potent activators of innate immune signaling and an important mediator of the microbiome's influence on host physiology. In this study, we performed computational and experimental analyses of healthy human fecal samples to examine the TLR4 signaling capacity of the gut microbiota. These analyses revealed that an immunoinhibitory activity of LPS, conserved across the members of the order Bacteroidales and derived from an underacylated structural feature, silences TLR4 signaling for the entire consortium of organisms inhabiting the human gut. Comparative analysis of metagenomic data from the Human Microbiome Project and healthy-donor samples indicates that immune silencing via LPS is a microbe-intrinsic feature in all healthy adults. These findings challenge the current belief that robust TLR4 signaling is a feature of the microbiome and demonstrate that microbiome-derived LPS has the ability to facilitate host tolerance of gut microbes. These findings have broad implications for how we model host-microbe interactions and for our understanding of microbiome-linked disease. IMPORTANCE While the ability for humans to host a complex microbial ecosystem is an essential property of life, the mechanisms allowing for immune tolerance of such a large microbial load are not completely understood and are currently the focus of intense research. This study shows that an important proinflammatory pathway that is commonly triggered by pathogenic bacteria upon interaction with the host is, in fact, actively repressed by the bacteria of the gut microbiome, supporting the idea that beneficial microbes themselves contribute to the immune tolerance in support of homeostasis. These findings are important for two reasons. First, many currently assume that proinflammatory signaling by lipopolysaccharide is a fundamental feature of the gut flora. This assumption influences greatly how host-microbiome interactions are theoretically modeled but also how they are experimentally studied, by using robust TLR signaling conditions to simulate commensals. Second, elucidation of the mechanisms that support host-microbe tolerance is key to the development of therapeutics for both intestinal and systemic inflammatory disorders.

Keywords: Toll-like receptors; lipopolysaccharide; microbial communities; microbiome; symbiosis; tolerance.

FIG 1

Total LPS from the human gut microbiome is immunoinhibitory. (A) Schematic diagram of sample processing and analysis in this study. (B) Principal-component analysis plot comparing the HMP1 sample set (blue) and a new set of nine healthy-donor (Novartis [NVS]) samples (black). (C) Human PBMCs were stimulated in the presence of increasing amounts of fecal LPS. IL-6 and IL-1β concentrations in supernatants were measured after 20 h of culture. E. coli LPS was included as a control. (D and E) HEK-293 cells expressing an NF-κB–luciferase reporter and either hTLR2 and hCD14 (D) or hTLR4, hMD2, and hCD14 (E) were stimulated in the presence of increasing doses of fecal LPS from each donor, B. dorei LPS, or E. coli PGN. A sterile-water negative control was included in each experiment (not shown). Luciferase activity was measured after 6 h of activation. Data shown are the mean ± the standard deviation of triplicate evaluations from one out of three independent experiments. (F) PBMCs were cotreated with 1 ng/ml E. coli LPS and increasing doses of fecal LPS from the donors indicated. IL-6 and IL-1β concentrations in supernatants were measured after 20 h of culture and compared to those obtained with LPS treatment alone (0:1). Data shown are the mean ± the standard deviation of triplicates in one representative experiment. *, corrected P value of <0.05.

FIG 2

Bacteroidetes bacteria are the main contributors to LPS biosynthesis in the human gut microbiome. (A) Average abundance per sample of genes related to the three main LPS biosynthesis-related functions. (B) Relative contributions of the different phyla to the total LPS-encoding capacity of the gut microbiome determined in both cohorts. The three functions of LPS biosynthesis were pooled. The darker-color set is HMP samples, and the lighter-color set is NVS samples. (C) Contribution of individual species to LPS biosynthesis functions. The average relative abundances of genes related to any of the three LPS-related GO functions are shown for individual species within each phylum. Only species detected in >50% of the donors in each data set are shown. The darker-color set is HMP samples, and the lighter-color set is NVS samples. (D) Correlation of individual-species abundance with stimulation of IL-6 production as determined in panel A. (E) Correlation of individual-species abundance with inhibition of IL-6 production as determined in panel B. Each data point represents one independent experiment. Rs, Spearman rho.

FIG 3

Bacteroidales LPS inhibits TLR stimulation. (A to C) LPS was isolated from the species indicated. (A) Human PBMCs were stimulated in the presence of increasing amounts of LPS from the Bacteroidales species indicated. IL-6, TNF-α, and IL-1β concentrations in supernatants were measured after 20 h of culture. E. coli LPS was included as a control. Data shown are the mean ± the standard deviation of triplicates in one representative experiment. (B and C) Human PBMCs were cotreated with E. coli LPS and increasing doses of LPS from Bacteroidales (B) and Alistipes (C) spp. TNF-α and IL-1β concentrations in supernatants were measured after 20 h of culture. E. coli LPS is shown as a control (dotted line). Data shown are the mean + the standard deviation of triplicate measurements in one representative of three or more independent experiments. *, corrected P value of <0.05. (C) MALDI-TOF MS analysis of the lipid A moiety of individual Bacteroidales and E. coli LPSs. m/z values are indicated for major peaks. On the right are the lipid A structures associated with the main m/z values.