Extrathymically Generated Regulatory T Cells Establish a Niche for Intestinal Border-Dwelling Bacteria and Affect Physiologic Metabolite Balance

Abstract

The mammalian gut microbiota provides essential metabolites to the host and promotes the differentiation and accumulation of extrathymically generated regulatory T (pTreg) cells. To explore the impact of these cells on intestinal microbial communities, we assessed the composition of the microbiota in pTreg cell-deficient and -sufficient mice. pTreg cell deficiency led to heightened type 2 immune responses triggered by microbial exposure, which disrupted the niche of border-dwelling bacteria early during colonization. Moreover, impaired pTreg cell generation led to pervasive changes in metabolite profiles, altered features of the intestinal epithelium, and reduced body weight in the presence of commensal microbes. Absence of a single species of bacteria depleted in pTreg cell-deficient animals, Mucispirillum schaedleri, partially accounted for the sequelae of pTreg cell deficiency. These observations suggest that pTreg cells modulate the metabolic function of the intestinal microbiota by restraining immune defense mechanisms that may disrupt a particular bacterial niche.

Keywords: T reg cells; extrathymic; host-microbe interactions; microbiota.

Figure 1.. Properties of microbial communities selected in hosts with impaired pTreg cell generation.

(A) Schematic of experimental design. Multiple cohorts of Foxp3^GFP∆CNS1 and Foxp3^GFP WT littermate controls were co-housed post-weaning until analysis at 8 weeks of age. (B-F) Shotgun metagenomic sequencing of cecal contents. (B) Abundance of significantly altered microbial pathways. (C) Comparison of overall coverage of microbial pathways. (D) Shannon index for α-diversity. (E) β-diversity of intestinal microbial communities. Nonmetric Multidimensional Scaling (NMDS) representation of the Bray-Curtis index. (F) LEfSe analysis of bacterial abundance. Data in all panels are pooled from 3 independent experiments (n = 2–3 mice per group in each experiment). Error bars represent the median with min–max range. (B, F) Showing only significantly altered pathways/bacteria (p<0.05 KW test/MW U-test). See also Figure S1.

Figure 2.. Foxp3^GFP∆CNS1 mice mount exacerbated goblet cell and type 2 cytokine responses early after exposure to microbes.

(A) Schematic of experimental design. GF Foxp3^GFP and Foxp3^GFP∆CNS1 mice were gavaged with freshly collected fecal contents from SPF mice and analyzed on D7 post-conventionalization. (B) Fecal DNA was extracted and subjected to 16S rDNA sequencing. Plot shows median and range of normalized read counts associated with the indicated species and are representative of 3 independent experiments (n= 4 mice per genotype). (C-D) RNA-sequencing of the bulk epithelium of the LI of Foxp3^GFP∆CNS1 and Foxp3^GFP mice (n = 4 per genotype). (C) Volcano plots showing log₂ fold-change Foxp3^GFP∆CNS1/Foxp3^GFP expression (x axis) and –log₁₀ of P-value (y axis). Significantly altered transcripts (P adj < 0.05) with a log₂ fold-change > 0.5 are shown in red. (D) CDF plot showing cumulative log₂ fold-change across all genes (black) and genes associated with goblet cells (red). Goblet cells genes do not follow the same distribution as all genes (P-value < 1e-10, one-sided ks-test). (E-H) FACS analysis of intracellular cytokine staining after ex-vivo re-stimulation. Quantification of the proportion of LI lamina propria cells producing the indicated cytokines in D7 (E, F) and GF (G, H) mice. (E, G) CD3ε^–CD90⁺ ILCs. (F, H) Foxp3^–CD4⁺ T cells. Data in (E-H) are pooled from 2 independent experiments (n = 2–3 mice per group in each experiment). Mean ± SD are plotted. Statistical significance was determined by a two-way ANOVA with Sidak’s multiple comparisons correction. *P < 0.05. See also Figure S2.

Figure 3.. pTreg cells preserve the niche of border dwellers by suppressing type 2 responses during the process of community assembly.

(A) Experimental design scheme. GF Foxp3^GFP and Foxp3^GFP∆CNS1 mice were injected with neutralizing antibodies against IL-4, IL-5 and IL-13 (NAb) or isotype control, gavaged with freshly collected fecal contents from SPF mice and analyzed on D7 post-conventionalization. (B-C) Effect of NAb treatment on intestinal bacteria. DNA was extracted from fecal pellets and subjected to 16S rDNA sequencing. (B) Bar plots showing log₂ fold-change NAb/Isotype for bacteria significantly affected by treatment in either Foxp3^GFP (black bars) or Foxp3^GFP∆CNS1 (red bars). (C) Abundance of M. schaedleri. Displaying median with range. Statistical significance for (B-C) was determined by mathematical modeling (see Methods for details). (D) Gene expression analysis of GC-related transcripts by RT-qPCR. Levels of indicated genes in FACS-purified GC were normalized to Villin. Statistical significance was determined by a two-way ANOVA with Sidak’s multiple comparisons correction. *P < 0.05. See also Figure S3.

Figure 4.. Intestinal and circulating metabolites are affected in pTreg cell-deficient animals.

Serum and cecal contents from SPF Foxp3^GFP∆CNS1 and Foxp3^GFP control mice were subjected to unbiased analysis of polar and non-polar metabolites by LC/MS and DSM/MS, respectively. (A-B) Metabolite analysis in SPF mice. (A) Volcano plots showing log₂ fold-change Foxp3^GFP∆CNS1/Foxp3^GFP (x-axis) and –log₁₀ of p-value (y-axis) for abundance of metabolites detected in the cecum (left panel) and serum (right panel). Significantly different metabolites were colored according to biochemical groups. (B) Random forest analysis for metabolites in the cecal contents (left) and serum (right). (C-F) GF Foxp3^GFP and Foxp3^GFP∆CNS1 mice were gavaged with ASF or ASF lacking M. schaedleri (ASF^∆457) consortia and analyzed 14 days after colonization. (C) Experimental scheme of ASF vs. ASF^∆457 colonization experiments. (D-F) Metabolite analysis in colonized mice. (D) Volcano plots showing log₂ fold-change ASF^∆457/ASF (x-axis) and –log₁₀ P-value (y-axis) for metabolites detected in the cecum (left panel) and serum (right panel) of wild-type Foxp3^GFP mice. Significantly different metabolites were colored according to biochemical groups. (E) FC/FC plot showing log₂ fold-change ASF^∆457/ASF for Foxp3^GFP mice (x-axis) and log₂ fold-change ASF^∆457/ASF for Foxp3^GFP∆CNS1 mice (y-axis). (F) Metabolites significantly altered in the comparison between SPF Foxp3^GFP∆CNS1 vs. Foxp3^GFP mice (panel A) were colored according to biochemical groups and projected onto volcano plots showing log₂ fold-change ASF^∆457/ASF for Foxp3^GFP mice (x-axis) and –log₁₀ of p-value (y-axis) for compounds detected in the cecum (left panel) and serum (right panel). Displaying data pooled from 3 independent experiments (n = 2–3 mice per group). Significance was determined by a MW U-test. Data in panels (A-B) relate to the same animals shown in Figure 1. See also Figure S4.

Figure 5.. Impaired pTreg cell generation affects intestinal and organismal homeostasis

(A-B) RNA-sequencing of the bulk epithelium of the small (left panel) and large (right panel) intestines from SPF (top) and GF (bottom) Foxp3^GFP∆CNS1 and Foxp3^GFP control mice (n = 2–3 mice per group). (A) Volcano plots showing log₂ fold-change Foxp3^GFP∆CNS1/Foxp3^GFP (x axis) and –log₁₀ of p-value (y axis). Significantly altered transcripts (p<0.05) are shown in red. (B) Plot showing Gene Ontology (GO) terms enriched in genes more highly expressed in Foxp3^GFP∆CNS1 mice (red) or Foxp3^GFP mice (black). GO terms with an FDR q-value < 0.05 are shown. Enrichment and q-value were determined by GOrilla. (C-D) Comparison of food intake (C) and body weight (D) between SPF Foxp3^GFP∆CNS1 and Foxp3^GFP mice. Displaying amount of food consumed per 24 hours, normalized by body weight, averaged over 8 days of monitoring. Data pooled from 3 litters (C, n = 7–10) and 6 litters (D, n = 20–24). (E) GF Foxp3^GFP and Foxp3^GFP∆CNS1 mice were gavaged with freshly collected fecal contents from SPF mice and weights were monitored for 8 weeks. Plot shows weight gain of GF Foxp3^GFP∆CNS1 and Foxp3^GFP mice after conventionalization. Data pooled from 2 independent experiments (n = 4–5 mice per genotype in each experiment). Mean ± SD are plotted (C, D) or Median, quartiles and min/max (E). Statistical significance was determined by a two-way ANOVA with Sidak’s multiple comparisons correction. *P < 0.05, **P < 0.01, ***P < 0.001.