Lysates of Methylococcus capsulatus Bath induce a lean-like microbiota, intestinal FoxP3+RORγt+IL-17+ Tregs and improve metabolism

Abstract

Interactions between host and gut microbial communities are modulated by diets and play pivotal roles in immunological homeostasis and health. We show that exchanging the protein source in a high fat, high sugar, westernized diet from casein to whole-cell lysates of the non-commensal bacterium Methylococcus capsulatus Bath is sufficient to reverse western diet-induced changes in the gut microbiota to a state resembling that of lean, low fat diet-fed mice, both under mild thermal stress (T22 °C) and at thermoneutrality (T30 °C). Concomitant with microbiota changes, mice fed the Methylococcus-based western diet exhibit improved glucose regulation, reduced body and liver fat, and diminished hepatic immune infiltration. Intake of the Methylococcu-based diet markedly boosts Parabacteroides abundances in a manner depending on adaptive immunity, and upregulates triple positive (Foxp3⁺RORγt⁺IL-17⁺) regulatory T cells in the small and large intestine. Collectively, these data point to the potential for leveraging the use of McB lysates to improve immunometabolic homeostasis.

Fig. 1. McB feeding reverses WD-induced gut microbiota changes and increases cecal SCFA levels.

A Principle coordinate analysis (PCoA) using Weighted UniFrac distances of fecal microbiota sampled from first and second experiment biweekly during the dietary intervention period, as indicated by numbers post dietary intervention in centroids. The WD_CNTL and WD_McB groups were similar in microbiota composition prior to dietary intervention week 12 + 0 (PERMANOVA p = 0.88 and 0.43 in Exp1 and Exp2, respectively). At the end of each experiment, the microbiota composition was significantly different between these groups (PERMANOVA p = 0.001 and 0.002 in Exp1 and Exp2, respectively). B Taxasummary of most abundant bacterial genera showing mean relative abundance in % of indicated family and genera in each group at indicated time points. C Deseq analysis of fecal bacterial genera abundances significantly regulated by McB intervention compared to the WD_CNTL (p.adj. < 0.05). Relative abundance in % in each group and variation are shown for each regulated genus at the sampled time points. Fold-change and adjusted p values of individual genera are indicated in Supplementary Table 3. D Relative Firmicutes/Bacteroidetes ratio of fecal samples of individual mice before (12 + 0) and after 6 weeks of dietary intervention (12 + 6). Statistical significance is indicated by p value on Wilcoxon matched-pairs signed rank test. E Weighted UniFrac distance (instability test) between paired samples from indicated 2-weeks’ interval post dietary intervention. Statistical significance is indicated by p value on RM two-way ANOVA with multiple comparison and Bonferroni’s post-hoc. D, E LFD n = 6, WD_CNTL and WD_McB n = 15 per group. F Major short-chained fatty acids (SCFAs) in cecal content as fold-change pmol per cecum. Propionate data was tested by one-way ANOVA and Dunnett’s multiple comparisons test. Acetate and butyrate data was tested by Kruskal–Wallis test and Dunn’s multiple comparisons test. LFD n = 6, WD_CNTL n = 11, and WD_McB n = 10. G Minor SCFAs in cecal content as fold-change pmol per cecum. Isobutyrate and isovalerate data were tested by Kruskal–Wallis test and Dunn’s multiple comparisons test. Valerate data were tested by one-way ANOVA and Dunnett’s multiple comparisons test. LFD n = 6. WD_CNTL and WD_McB n = 10 per group. D–G Bars indicate group mean ± SEM and individuals data points in Exp1 (squares) and Exp2 (circles) with p-values < 1 × 10⁻¹ between WD_CNTL and indicated group using the specified statistical test.

Fig. 2. WD_McB feeding stimulates induction of gut-specific regulatory T cells.

A–D Number of indicated cells in colon. E–G Representative plots of colonic TCRβ⁺CD4⁺ FoxP3⁺ RORγt⁺ _pT_regs (left) and IL-17⁺ _pT_regs (right) in LFD (E), WD_CNTL (F), and WD_McB (G) group. H–K Number of indicated cells in small intestine. L–N TCRβ⁺CD4⁺ FoxP3⁺ RORγt⁺ _pT_regs (left) and IL-17⁺ _pT_regs (right) in LFD (L), WD_CNTL (M), and WD_McB (N) group. O–R Percentage of indicated cells in SI- and LI-LP from ‘weight-matched’ mice housed at thermoneutrality. A–N Dots indicate individuals data points in Exp1 (squares) and Exp2 (circles) with n = 6 (LFD) or 10 (WD_McB and WD_CNTL). O–R SI-LP n = 5 per group and LI-LP n = 9 (LFD and WD_REF groups) or 11 (WD_McB). A–D, H–K, O–R Bars indicate group mean ± SEM and dots indicate individual data points. All p-values < 1 × 10⁻¹ between WD_CNTL and indicated group by one-way ANOVA with multiple comparisons and Dunnett post-hoc are depicted.

Fig. 3. Dietary intervention with McB blunts progression of insulin resistance and fat mass accumulation.

A–C Oral glucose tolerance test (OGTT) and 5 h fasting blood glucose prior to dietary intervention (Week 11) and 5 weeks post intervention (Week 12 + 5) of LFD (A), WD_CNTL (B), and WD_McB (C) groups. D–F Glucose-stimulated insulin concentration during OGTT and 5 h fasting insulin levels prior to dietary intervention (Week 11) and 5 weeks post intervention (Week 12 + 5) of LFD (D), WD_CNTL (E), and WD_McB (F) groups. G Intraperitoneal insulin tolerance test (ITT) 3 weeks post dietary intervention. H Body weight development. Mice were fed either LFD or WD_REF the first 12 weeks followed by 6 week dietary intervention period. Dotted vertical line depicts intervention start. I Fat mass in gram through the dietary intervention period (Week 12 + 0 to 12 + 6) measured by MR scan. J Feed intake per cage as average grams per mouse during 48 h. K Fecal content of fat and protein. L NAFLD activity score based on hepatic steatosis grade (0–3), inflammation (0–2), and hepatocellular ballooning (0–3) graded by blinded histological assessment of liver tissue. M One representative H&E stained picture per group out of 6 (LFD) or 13 (WD_McB and WD_CNTL) liver tissue sections. A, F n = 6 (LFD) and 15 (WD_McB and WD_CNTL) except for F timepoint 60–120 min at Week 11 where n = 14, 12, and 12, respectively due to insufficient sample material. Statistical significance within each timepoint is indicated by p values at paired two-way ANOVA-RM with Bonferroni post-hoc test. Fasting glucose and insulin levels were evaluated by paired t-tests. Lines indicate group means and dots represents individual data points. G n = 8 (WD_McB) and 9 (WD_CNTL). WD_McB was compared to WD_CNTL by two-way ANOVA-RM with Bonferroni post-hoc test. Lines indicate group means and dots represents individual data points. H–I n = 6 (LFD) or 15 (WD_McB and WD_CNTL). LFD and WD_McB were compared to WD_CNTL by two-way ANOVA-RM, adjusted for multiple comparisons by Dunnett’s post-hoc. Lines indicate group means and dots represents individual data points. J, K Each data point represents the average of one cage. n = 2 (LFD) to 5 (WD_McB and WD_CNTL). LFD and WD_McB were compared to WD_CNTL by one-way ANOVA, adjusted for multiple comparisons by Dunnett’s post-hoc. Dot shapes indicate individuals data points in Exp1 (squares) and Exp2 (circles). L Bars represent median and interquartile range and dots represents individual data points. All p-values < 1 × 10⁻¹ between WD_CNTL and indicated group are depicted. Dot shapes indicate individuals data points in Exp1 (squares) and Exp2 (circles). LFD and WD_McB were compared to WD_CNTL by Kruskal–Wallis, adjusted for multiple comparisons by Dunn’s post-hoc. n = 6 (LFD) or 13 (WD_McB and WD_CNTL). A–L All p-values < 1 × 10⁻¹ between WD_CNTL and indicated group are depicted.

Fig. 4. Dietary McB intervention improves diet-induced metabolic and hepatic phenotype after prolonged WD feeding.

A 5 h fasting plasma insulin levels before (Week 20) and after (Week 21 + 5) dietary intervention. Dots represents individual data point and bars represents mean ± SEM. n = 7 (LFD) or 9 (WD_REF and WD_McB). LFD and WD_McB were compared to WD_CNTL by paired t-test. B Fat mass in grams measured by MR scan at indicated time points from dietary intervention. Dots represents individual data point and lines depicts mean. n = 10 (LFD and WD_McB) or 9 (WD_REF) per group. C NAFLD activity score separated in hepatic steatosis grade (0–3), inflammation (0–2) and hepatocellular ballooning (0–3) graded by blinded histological assessment of H&E stained liver tissue. Dots represents individual data point and bars represent median and interquartile range. n = 10 (LFD) or 9 (WD_REF and WD_McB) per group. D One representative H&E stained picture per group out of 10 (LFD) or 9 (WD_REF and WD_McB) liver tissue sections. E Number of lipid droplets in each liver sample per experimental group quantified by Oil Red O staining. Dots represents individual data point of 3 (LFD and WD_REF) and 4 (WD_McB) randomely chosen samples; bars represent median and interquartile range. F Average size of lipid droplets in each of 3-4 liver samples per experimental group quantified by Oil Red O staining in (E). Dots represents individual data point of 3 (LFD and WD_REF) and 4 (WD_McB) randomely chosen samples; bars represent median and interquartile range. G Distribution of lipid droplet size in % of all lipid droplets within each experimental group from E. H Plasma adiponectin concentration in 5 h fasted mice before (Week 20) and after (Week 21 + 5) dietary intervention. Dots represents individual data point and bars represents mean ± SEM. n = 8 (LFD), 7 (WD_REF), or 6 (WD_McB). I ITT after four weeks of dietary intervention (Week 21 + 4). Dots represents individual data point and lines depicts mean. n = 7 (LFD), 9 (WD_REF), or 10 (WD_McB). J Expression of key metabolic enzymes in liver tissue after 5 weeks of dietary intervention (Week 21 + 5) by RT-qPCR. Dots represents individual data point and bars represents mean ± SEM. n = 10 (LFD and WD_McB) or 8 (WD_REF) per group. A, H Statistical significance compared by paired t-test. B, I Statistical significance compared to WD_REF by two-way ANOVA-RM, adjusted for multiple comparisons by Dunnett post-hoc. C, E, F, J Statistical significance compared to WD_REF by Kruskal–Wallis test, adjusted for multiple comparisons by Dunn’s post-hoc. A–J All p-values < 1 × 10⁻¹ between WD_REF and indicated group are depicted.

Fig. 5. WD_McB feeding resets the hepatic lipidome and decreases hepatic immune infiltration alleviating NAFLD.

A Principle coordinate analysis (PCoA) of hepatic lipid species identified in negative ionization mode. B As in A but depicted in a volcano plot with significantly regulated lipid species (FDR < 0.05, adjusted for multiple comparisons by Benjamin–Hochberg) presented on a log2 scale in either green (WD_McB > WD_REF) or blue (WD_McB < WD_REF). C, D Identified lipid species differentially expressed (FDR < 0.01) between WD_McB and WD_REF as indicated. Adjusted p values of individual lipid species are indicated in Supplementary Table 4. n = 10 (LFD and WD_McB) or 9 (WD_REF). E Lipid pathways identified by fold-change analysis (FDR < 0.05, adjusted for multiple comparisons by Benjamin–Hochberg) seperating LFD (gray) and WD_McB (green) groups from WD_REF in negative ionization mode. F Hepatic cytokine levels of TNF-α and IL-6 as indicated. n = 9 (LFD and WD_McB) and 8 (WD_REF). G Ly6G⁺ and CD3⁺ immune cells revealed by immunohistochemistry^. n = 3 per group (randomly selected samples). H Cytokine levels in plasma of IL-22, IL-18, and IL-17. n = 10 (LFD and WD_REF) or 8 (WD_McB). I Amount of Tim4⁺ Kupffer cells. n = 8 (LFD), 9 (WD_REF), and 11 (WD_McB)^. J Mean Fluorescence intensity (MFI) of Ly6C⁺ liver monocytes. n = 8 (LFD), 9 (WD_REF), and 11 (WD_McB⁾. K Amount of hepatic IL-17⁺ γδ T cells. n = 8 (LFD), 9 (WD_REF), and 11 (WD_McB). C, D, F–K Bars represent group mean ± SEM and dots indicate individual data points. F, H Statistical significance compared to WD_REF group by Kruskal–Wallis test, adjusted for multiple comparisons by Dunn’s post-hoc. G, I–K Statistical significance compared to WD_REF by one-way ANOVA, adjusted for multiple comparisons by Dunnett post-hoc. All p-values < 1 × 10⁻¹ between WD_REF and indicated group are depicted.

Fig. 6. WD_McB treatment improves colonic mucus production and reverses obesity-induced gut microbiota changes to resemble the composition found in lean LFD-fed mice.

A Colon length in centimeters. Statistical significance compared to WD_REF group by Kruskal–Wallis test, adjusted for multiple comparisons by Dunn’s post-hoc. n = 10 (LFD and WD_McB) or 9 (WD_REF). B Area of neutral mucins determined by Periodic Acid Schiffs (PAS) staining in three discrete locations, i.e., proximal, middle, and distal colon segments. Three longitudinally sectioned crypts were measured in each sample (one data point) at each location. Statistical significance compared to WD_REF group by one-way ANOVA, adjusted for multiple comparisons by Dunnett post-hoc, for proximal and distal segments while middle segment was by Kruskal–Wallis test, adjusted for multiple comparisons by Dunn’s post-hoc. n = 9 (LFD and WD_REF) or 10 (WD_McB) except for distal segment where n = 9 for WD_McB. C Colon crypt depth (CD) in proximal, middle, and distal colon segments, measured in histological sections. Measurements are given in µm, and results are given as means of six crypt measurements per location per individual mice (3 longitudinally PAS stained and HID-AB stained crypts, respectively, per data point). Statistical significance compared to WD_REF group by one-way ANOVA, adjusted for multiple comparisons by Dunnett post-hoc test. n = 10 (LFD and WD_McB) or 9 (WD_REF). D PAS staining for mucus in colon middle segments. One representative picture per group, as indicated, out of 9 (LFD and WD_REF) or 10 (WD_McB) middle segment colon sections. E Area of sulfomucins in proximal, middle, and distal colon segments by HID staining. Three longitudinally sectioned crypts were measured in each sample and at each location. Statistical significance compared to WD_REF group by one-way ANOVA, adjusted for multiple comparisons by Dunnett post-hoc, for proximal and middle segments while distal segment was by Kruskal–Wallis test, adjusted for multiple comparisons by Dunn’s post-hoc. LFD n = 10, WD_REF n = 8 except proximal segment where n = 9, and WD_McB n = 10 except for proximal segment where n = 9. F Area of sialomucins in proximal, middle, and distal colon segments by AB staining. Three longitudinally sectioned crypts were measured in each sample and at each location. Statistical significance compared to WD_REF group by one-way ANOVA, adjusted for multiple comparisons by Dunnett post-hoc, for proximal and middle segments while distal segment was by Kruskal–Wallis test, adjusted for multiple comparisons by Dunn’s post-hoc. LFD n = 10, WD_REF n = 8 except proximal segment where n = 9, and WD_McB n = 9 except for middle segment where n = 10. G Representative HID-AB staining of middle segment of indicated experimental group. One representative picture per group out of 10 (LFD and WD_McB) or 8 (WD_REF) evaluated sections. H PCoA of fecal microbiota composition of indicated group before (Week 21 + 0) and after (21 + 5) dietary intervention with group mean indicated as centroids. Microbiota composition was significantly different between the WD_REF and WD_McB groups at the end of the experiment week 21 + 5 (PERMANOVA p = 0.001). I Taxasummary of most abundant bacterial genera showing mean relative abundance of indicated genera in each group at indicated timepoint. J Firmicutes/Bacteroidetes ratio of fecal samples of individual mice at termination (Week 21 + 5). Statistical significance compared to WD_REF group by one-way ANOVA, adjusted for multiple comparisons by Dunnett post-hoc test. n = 10 (LFD and WD_McB) or 9 (WD_REF). K Deseq analysis of fecal bacterial genera abundances significantly regulated by McB intervention (p.adj. < 0.05). Relative abundance in % in each group and variation are shown for each regulated genera at the sampled time points. Fold-change and adjusted p values of individual genera are indicated in Supplementary Table 5. A–K n = 8–10. A–C, E, F, J) Bars represent group mean ± SEM and dots indicate individual data points. All p-values < 1 × 10⁻¹ between WD_REF and indicated group are depicted.

Fig. 7. McB lysates rely on adaptive immunity to boost Parabacteroides.

A, B Taxasummary of most abundant bacterial genera showing mean relative abundance of indicated genera in each group at indicated timepoint in WT (A) and RAG2^−/− (B) mice. C, E Relative abundance of Parabacteroides in fecal samples of indicated genotype at indicated timepoint. D, F Weighted UniFrac distance (instability test) between paired samples from indicated 2-weeks interval post dietary intervention in WT (D) and RAG2^−/− (F) mice. C–F Bars represent group mean ± SEM and dots indicate individual data points. All p-values < 1 × 10⁻¹ between WD_REF and indicated group are depicted. Statistical significance compared to WD_REF by two-way ANOVA-RM, adjusted for multiple comparisons by Dunnett post-hoc. G, H PCoA of fecal microbiota composition of indicated group at baseline (Week 0, all mice fed LFD) and after 2, 4, and 6 weeks of dietary intervention with group mean indicated as centroids.

Fig. 8. Altered gut microbiota by WD feeding affects glucose regulation.

A, C OGTT in 5 h fasted RAG2^-/- mice after 1 and 6 weeks of dietary intervention as indicated. Statistical significance compared to WD_REF by two-way ANOVA-RM, adjusted for multiple comparisons by Dunnett post-hoc. n = 12 mice per group. B, D Glucose stimulated Insulin Secretion during OGTT after 1 and 6 weeks of dietary intervention. Statistical significance compared to WD_REF by mixed effects analysis RM, adjusted for multiple comparisons by Dunnett post-hoc. n = 12 mice per group except for WD_McB group at 120 min in B, WD_McB group at 60 min in D, and WD_REF group at 15 min in D where n = 11 due to insufficient sample material. E–G Body weight, fat mass and lean mass, as indicated, at indicated time points post dietary intervention. Statistical significance of body weight compared to WD_REF by mixed effects analysis RM, adjusted for multiple comparisons by Dunnett post-hoc, and fat and lean mass by two-way ANOVA-RM, adjusted for multiple comparisons by Dunnett post-hoc. n = 12 mice per group except for WD_McB group Week 1 in E where n = 9. H Cecum weight at termination. Statistical significance compared to WD_REF group by one-way ANOVA, adjusted for multiple comparisons by Dunnett post-hoc test. n = 7 (LFD and WD_REF) or 11 (WD_McB). I Concentration of short chain fatty acids in cecum from H. Statistical significance compared to WD_REF group by Kruskal–Wallis test, adjusted for multiple comparisons by Dunn’s post-hoc. n = 12 (LFD and WD_McB) or 10 (WD_REF). J, K Body weight and fat mass development in ABX treated WT mice fed either LFD (grey bars) or WD_REF (blue bars), receiving cecal microbiota from indicated donor mice. Statistical significance compared to WD_REF by two-way ANOVA-RM, adjusted for multiple comparisons by Dunnett post-hoc. WD-fed mice n = 11 (LFD donor), 9 (WD_REF donor), and 12 (WD_McB donor). LFD-fed mice n = 12 (LFD donor), 9 (WD_REF donor), and 12 (WD_McB donor). L, N OGTT in 5 h fasted WD_REF-fed WT recipient mice 1 and 5 weeks after first cecal microbiota transfer, as indicated. Statistical significance compared to WD_REF by two-way ANOVA-RM, adjusted for multiple comparisons by Dunnett post-hoc. LFD donor n = 11, WD_REF donor n = 9, and WD_McB donor n = 12. M, O Glucose stimulated Insulin Secretion during OGTT (L + N) 1 and 5 weeks after first cecal microbiota transfer. Statistical significance compared to WD_REF by two-way ANOVA-RM, adjusted for multiple comparisons by Dunnett post-hoc. In M, LFD donor n = 11 except at 0 min where n = 4, WD_REF donor n = 9 except at 0 min where n = 6, and WD_McB donor n = 12 except at 0 min where n = 9. In O, LFD donor n = 10, WD_REF donor n = 8 except at 90 min where n = 9, and WD_McB donor n = 11 except at 90 min where n = 12. Reduced n-size reflects insufficient sample material at indicated timepoint. A–E, G, J, L–O Lines represent group mean and dots indicate individual data points. F, H, I, K Bars represents mean ± SEM. Dots indicate individual data points. A–O All p-values < 1 × 10⁻¹ between WD_REF and indicated group are depicted.