Genetic variance in the murine defensin locus modulates glucose homeostasis

Abstract

Insulin resistance is a heritable risk factor for many chronic diseases; however, the genetic drivers remain elusive. In seeking these, we performed genetic mapping of insulin sensitivity in 670 chow-fed Diversity Outbred in Australia (DOz) mice and identified a genome-wide significant locus (QTL) on chromosome 8 encompassing 17 defensin genes. By taking a systems genetics approach, we identified alpha-defensin 26 (Defa26) as the causal gene in this region. To validate these findings, we synthesized Defa26 and performed diet supplementation experiments in two mouse strains with distinct endogenous Defa26 expression levels. In the strain with relatively lower endogenous expression (C57BL/6J) supplementation improved insulin sensitivity and reduced gut permeability, while in the strain with higher endogenous expression (A/J) it caused hypoinsulinemia, glucose intolerance and muscle wasting. Based on gut microbiome and plasma bile acid profiling this appeared to be the result of disrupted microbial bile acid metabolism. These data illustrate the danger of single strain over-reliance and provide the first evidence of a link between host-genetics and insulin sensitivity which is mediated by the microbiome.

Keywords: Bile Acids; Defensin; Diversity Outbred; Insulin Sensitivity; Microbiome.

Figure 1. Genetic mapping of insulin sensitivity in 670 chow-fed Diversity Outbred in Australia mice.

(A) Glucose and insulin concentrations during a glucose tolerance test in 670 DOz mice. (B) The equation used to calculate the Matsuda Index and genetic mapping of the Matsuda Index in DOz mice, including schematic of the defensin locus. (C) Schematic of murine defensin secretion from Paneth cells into the small intestine. Biological replicates are shown as individual data points. Source data are available online for this figure.

Figure 2. Analysis of microbial composition and co-housing effects in mice carrying a putative insulin sensitivity allele within the defensin locus.

(A) Single nucleotide polymorphism (SNP) mapping of the Matsuda Index on chromosome 8. (B) Blood glucose concentrations during a glucose tolerance test of mice carrying the minor allele of rs23754102 (AB) and non-carrier (AA) controls. (C) Blood insulin concentrations during a glucose tolerance test of mice carrying the minor allele of rs23754102 (AB) and non-carrier (AA) controls. (D) Insulin sensitivity (Matsuda Index) of mice carrying the minor allele of rs23754102 (AB), non-carrier (AA) controls, and cage mates of AB mice. (E) Alpha-diversity (Inverse Simpson) of microbiomes from AA, Cage mate, and AB mice. (F) PCoA visualisation of beta-diversity between AA, Cage mate, and AB mice calculated by Bray-Curtis dissimilarity. (G) Relative abundance of Akkermansia muciniphila, (H) Bifidobacterium pseudolongum, (I) Ligilactobacillus spp. (J) Limosilactobacillus spp. in AA, Cage mate, and AB mice. Data are mean with biological replicates shown as individual data points. **P < 0.01, *P < 0.05 compared to AA mice. Metabolic phenotypes were compared using two-way RM ANOVA, with Tukey’s LSD (B, C), one-way ANOVA (D). Alpha-diversity was compared by one-way ANOVA (E). Difference in microbial relative abundance (G–J) was identified by Analysis of Compositions of Microbiomes with Bias Correction (ANCOM-BC). Source data are available online for this figure.

Figure 3. Analysis of defensin protein expression in Diversity Outbred founder strains.

(A) Schematic of study design to investigate insulin sensitivity and defensin protein expression in small intestines of inbred mouse strains. (B) Matsuda Index in Diversity Outbred founder strains. (C) Adiposity in Diversity Outbred founder strains. (D) Correlations of all quantified defensin peptides (Defa) with Matsuda Index across Diversity Outbred founder strains. (E) Correlation of mean normalised alpha-defensin 26 abundances with Matsuda Index in Diversity Outbred founder strains. (F) Founder strain contribution estimates for the DOz Matsuda Index QTL (top) and QTL LOD score (bottom) on chromosome eight. (G) Correlation of founder strain contribution estimates for the DOz Matsuda Index QTL with mean normalised alpha-defensin 26 abundance. Dashed lines denote 95% confidence intervals. Data are mean with biological replicates shown as individual data points. **P < 0.01, *P < 0.05 Metabolic phenotypes were compared by one-way ANOVA (B, C). Pearson’s correlations were performed without P value adjustment (D, E, G). Source data are available online for this figure.

Figure 4. Metabolic phenotyping of western-diet fed C57BL/6J mice supplemented with alpha-defensin 26.

(A) Change in body weight (g) of western diet (WD) or WD + alpha-defensin 26 (WD + Defa26) fed C57BL/6J mice over 8 weeks. (B) Relative (fold-change) increase in adipose tissue mass of WD and WD + Defa26 fed C57BL/6J mice over 8 weeks. (C) Change in lean mass (g) of WD and WD + Defa26 fed C57BL/6J mice over 8 weeks. (D) Food intake (kcal/day) of WD and WD + Defa26 fed C57BL/6J mice over 8 weeks. (E) Post-dissection tissue weights of WD and WD + Defa26 fed C57BL/6J mice. (F) Blood glucose and (G) blood insulin concentrations during a glucose tolerance test of C57BL/6J mice fed either a WD or WD + Defa26 for eight weeks. (H) Matsuda Index of C57BL/6J mice fed either a WD or WD + Defa26 for 8 weeks. (I) Blood glucose concentrations during a insulin tolerance test of C57BL/6J mice fed either a WD or WD + Defa26 for 8 weeks. (J) Insulin secretion (K) and content from ex vivo islets collected from C57BL/6J mice fed either a WD or WD + Defa26 for 8 weeks. (L) Relative gut permeability (post – pre FITC fluorescence) of C57BL/6J mice fed either a WD or WD + Defa26 for 8 weeks. (M) Difference (log2 FC) in relative abundance of microbes identified by Analysis of Compositions of Microbiomes with Bias Correction (ANCOM-BC) between C57BL/6J mice fed either a WD or WD + Defa26 for 8 weeks. Data are mean with biological replicates shown as individual data points. For differentially abundant microbes, error bars represent SD of difference between groups. *P < 0.05 denotes statistical significance from WD control. Metabolic phenotypes were compared by two-way RM ANOVA with Tukey’s LSD (A, F, I), two-way ANOVA with Tukey’s LSD (E, G, J), or Student’s t test (B–D, H, K, L). Source data are available online for this figure.

Figure 5. Metabolic phenotyping of western-diet fed A/J mice supplemented with alpha-defensin 26.

(A) Change in body weight (g) of western diet (WD) or WD + alpha-defensin 26 (WD + Defa26) fed A/J mice over 8 weeks. (B) Relative (fold-change) increase in adipose tissue mass of WD and WD + Defa26 fed A/J mice over 8 weeks. (C) Change in lean mass (g) of WD and WD + Defa26 fed A/J mice over 8 weeks. (D) Combined mass of gastrocnemius, tibialis anterior and quadriceps muscles from A/J mice after WD or WD + Defa26 feeding for 8 weeks. (E) Food intake (kcal/day) of WD and WD + Defa26 fed A/J mice. (F) Blood glucose concentrations during a glucose tolerance test of A/J mice fed either a WD or WD + Defa26 for 8 weeks. (G) Glucose tolerance test ‘area-under-the-curve’ for A/J mice fed either a WD or WD + Defa26 for 8 weeks. (H) Blood insulin concentrations during a glucose tolerance test of A/J mice fed either a WD or WD + Defa26 for 8 weeks. (I) Blood glucose concentrations during an insulin tolerance test of A/J mice fed either a WD or WD + Defa26 for 8 weeks. (J) Insulin secretion from ex vivo islets collected from A/J mice fed either a WD or WD + Defa26 for 8 weeks. (K) Insulin content of islets collected from A/J mice fed either a WD or WD + Defa26 for 8 weeks. (L) Difference (log2 FC) in relative abundance of microbes identified by Analysis of Compositions of Microbiomes with Bias Correction (ANCOM-BC) between A/J mice fed either a WD or WD + Defa26 for 8 weeks. Data are mean with biological replicates shown as individual data points. ***P < 0.001, **P < 0.01, *P < 0.05 denotes statistical significance from WD control. Metabolic phenotypes were compared by two-way RM ANOVA with Tukey’s LSD (F, I), two-way ANOVA with Tukey’s LSD (A, H, J), or Student’s t test (B–E, G, K). Source data are available online for this figure.

Figure 6. Comparison of caecal microbiomes and circulating bile acids from C57BL/6J and A/J following alpha-defensin 26 supplementation.

(A) Visualisation of beta-diversity (ANCOM-BC) in WD and WD + Defa26 fed C57BL/6J and A/J mice. (B) Comparison of differentially abundant microbial taxa in mice fed either a WD or a WD + Defa26 assessed by ANCOM-BC. (C) Relative abundance of Akkermansia muciniphila (D) Alloprevotella sp. and (E) Erysipelatoclostridium spp. in C57BL/6J and A/J mice fed either a WD or a WD + Defa26. (F) Enrichment scores for statistically significant terms included in Taxon Set Enrichment Analysis. (G) Work-flow to quantify circulating bile acids by LC-MS and principal component visualisation of circulating bile acids in C57BL/6J and A/J mice fed either a WD or a WD + Defa26. (H) Heatmap visualisation of circulating bile acid concentrations in C57BL/6J and A/J mice fed either a WD or a WD + Defa26. Data are mean with biological replicates shown as individual data points. *P < 0.05 denotes statistical significance from WD control. Difference in microbial relative abundance (B–E) was identified by Analysis of Compositions of Microbiomes with Bias Correction (ANCOM-BC). Bile acids were compared by two-way ANOVA with Tukey’s LSD (H). Source data are available online for this figure.