Interactions between Gut Microbiota, Host Genetics and Diet Modulate the Predisposition to Obesity and Metabolic Syndrome

Abstract

Obesity, diabetes, and metabolic syndrome result from complex interactions between genetic and environmental factors, including the gut microbiota. To dissect these interactions, we utilized three commonly used inbred strains of mice-obesity/diabetes-prone C57Bl/6J mice, obesity/diabetes-resistant 129S1/SvImJ from Jackson Laboratory, and obesity-prone but diabetes-resistant 129S6/SvEvTac from Taconic-plus three derivative lines generated by breeding these strains in a new, common environment. Analysis of metabolic parameters and gut microbiota in all strains and their environmentally normalized derivatives revealed strong interactions between microbiota, diet, breeding site, and metabolic phenotype. Strain-dependent and strain-independent correlations were found between specific microbiota and phenotypes, some of which could be transferred to germ-free recipient animals by fecal transplantation. Environmental reprogramming of microbiota resulted in 129S6/SvEvTac becoming obesity resistant. Thus, development of obesity/metabolic syndrome is the result of interactions between gut microbiota, host genetics, and diet. In permissive genetic backgrounds, environmental reprograming of microbiota can ameliorate development of metabolic syndrome.

Figure 1. Differences in weight gain and glucose tolerance between vendor-bred mouse strains

(A) Weight gain curves of B6J, 129J and 129T mice on CD and HFD (n=19-20 per group). (B) Intraperitoneal GTT of vendor-bred strains at 10 weeks on the indicated diets (n=6). Repeated measure two-way ANOVA (time and strain) with Bonferroni post-hoc test for A&B. (C) Insulin levels measured in non-fasted animals and (D) calculated HOMA-IR after 22 weeks of CD and HFD [n=6 per treatment group; two-way ANOVA (strain and diet) with Sidak's post-hoc test]. (E) Liver triglyceride (TG) content (n=4) and (F) liver weight (n=6) in mice after 22 weeks of consumption of CD and HFD. Two-way ANOVA [diet (p<0.0001) and strain (p=0.0004); interaction p=0.0054] with Bonferroni post-hoc test. (G) qPCR for inflammatory markers expressed in perigonadal fat (n=8) or liver (n=4) harvested from mice fed CD or HFD for 22 weeks. Data are normalized to TBP. (one-way ANOVA with Tukey's post-hoc test). ***, P<0.001, **, P<0.01, *, P<0.05. All values presented as Mean ±SEM. See also Figure S1.

Figure 2. Phenotypic changes associated with breeding at Joslin

(A) Schematic overview of the experimental design, indicating study groups and time points of fecal sampling. (B) Body weight gain of Joslin-bred mice (B6J/Jos, 129J/Jos and 129T/Jos) on either CD or (C) HFD (n=19-20 per group). Symbols show statistically significant differences between the indicated groups as judged by a repeated measures two-way ANOVA (time and strain) with Bonferroni post-hoc test. (D) Intraperitoneal GTT of Joslin-bred mice tested after 10 weeks on either CD or HFD [n=6 per group; symbols show significant differences between the indicated groups as judged by a repeated measures two-way ANOVA (time and strain) with Bonferroni post-hoc test]. (E) Liver weight of Joslin-bred strains at 22 weeks, [two-way ANOVA (diet and strain) with Tukey's post-hoc test; n=6]. (F) Heat map of gene expression in perigonadal fat (PGF) (n=8) and liver (n=4) after 22 weeks on either CD or HFD. (G) Insulin levels measured after 22 weeks of CD and HFD in non-fasted animals and (H) calculated HOMA-IR [n=6 per treatment group, two-way ANOVA (diet and strain) with Tukey's post-hoc test). ***, P<0.001, **, P<0.01, *, P<0.05. All values presented as mean ± SEM. See also Figure S2.

Figure 3. Food consumption, activity and metabolic rates in vendor-bred and Joslin-bred strains

(A) Daily food intake per gram lean body mass, (B) total activity (beam breaks/30 min), (C) oxygen consumption (VO₂) (D) carbon dioxide production (VCO₂) and (E) respiratory exchange ratio after 8 weeks on the indicated diets as measured in metabolic cages [n=5-8 per group; two-way ANOVA (diet and strain) with Bonferroni post-hoc test for panels A,B and E, one-way ANOVA with Tukey's post-hoc test for C and D ***, P<0.001, **, P<0.01, *, P<0.05]. All values presented as Mean ± SEM.

Figure 4. Phylogenetic similarity and phylogenetic diversity of fecal microbiota

(A) PCoA of unweighted UniFrac distances for all 287 fecal samples, with fitted vectors indicating the effects of diet and birthplace calculated using weighted mean positions along principal coordinates 1 and 2 (PC1, PC2). (B) PCoA of unweighted UniFrac distances for all 287 fecal samples, separated by week. (C) Percent phylogenetic similarity [calculated as 100×(1 - unweighted UniFrac distance)] within each experimental group of mice and between the three vendor-bred strains and the three Joslin-bred strains at week 0 of the experiment. Comparisons of selected mean distances within and between experimental treatment groups were performed using permutation tests with 10,000 randomizations, followed by Holm's corrections for multiple tests. Shown are means and bootstrapped 95% confidence intervals. (D) Pie charts showing relative abundances of bacterial phyla in the fecal microbiota of mice at the indicated time points and diets. (E) Phylogenetic diversity of the fecal microbiota of all strains at week 0 on CD and at week 22 on both CD and HFD. Differences between means were tested for significance using two ANOVAs, one for week 0 and one for week 22, followed by planned contrasts of means with p-values adjusted using the single-step method. (F) Relative abundances of Joslin specific OTUs that colonize 129J, 129T and B6J mice over time (±SEM). Differences between the mean relative abundances were tested within each mouse strain for CD-fed mice at weeks 0 and 22, and for CD-fed versus HFD-fed mice at week 22, using t-tests and Holm's corrections for p-values. ***, P<0.001, **, P<0.01, *, P<0.05. See also Figure S3.

Figure 5. Representation of bacterial taxa in the fecal microbiota

(A) Heat map showing the phi-coefficients of OTUs that have at least one significant phi-coefficient more extreme than −0.8 or 0.8 or mean relative abundance of at least 1% in at least one group of mice at week 0 of the experimental period. (B) Abundance of OTUs assigned to the phylum Bacteroidetes in vendor-bred and Joslin-bred mouse strains. All Bacteroidetes OTUs with mean relative abundances of 0.1% or greater in at least one mouse group are included in the heatmap. Abundance values were log-transformed according to the formula log10 (abundance/10000 reads +1). Also see Figure S4.

Figure 6. Correlations between OTUs’ relative abundances and phenotypes, including all mouse groups

(A) A heat map shows Spearman's rank correlations between phenotypes and the relative abundances of selected OTUs. OTUs with at least four correlations more extreme than −0.5 or 0.5 or one correlation more extreme than −0.75 or 0.75 are shown. All phenotypes with at least one correlation more extreme than −0.5 or 0.5 are presented. Hierarchical clustering is presented for both OTUs and phenotypes, based on the Euclidean distances between their presented Spearman's rank correlations. (B) Linear correlations between the mean relative abundances (per 10,000 reads) of select OTUs and mouse phenotypes for each of experimental treatment groups. (C) Heatmap of Spearman's correlations between phenotypes and the relative abundances of OTUs using all HFD-fed 129 mice (129T, 129T/Jos, 129J, and 129J/Jos). P-values were corrected for false detection rate using the Benjamini-Hochberg method and significant correlations after correction are highlighted with a yellow box. See also Figure S5.

Figure 7. Gnotobiotic mouse experiments identify microbiota-transmissible phenotypes in HFD fed recipients

(A) Weight gain of germ free (GF) C57Bl/6 mice (n=5) and GF mice colonized with fecal microbiota from B6J, 129J and 129T (n=9-10) mice fed a HFD [repeated measures two-way ANOVA (time and donor strain), with Tukey's post-hoc tests]. (B) Random fed blood glucose of mice from A, after 6 weeks on HFD (one-way ANOVA with Tukey's post-hoc test). (C) AUC values calculated from data obtained during an oral glucose tolerance tests at 8 weeks post colonization. One exceptionally high outlier value was removed from the mice colonized with the B6J microbiota prior to analysis. (D) Liver weights for mice described in A after 8 weeks of high fat diet feeding (one-way ANOVA with Tukey's post-hoc test) *, P<0.05; **, P<0.01; ***, P<0.0001. All values presented as Mean ± SEM