Home-site advantage for host species-specific gut microbiota

Abstract

Mammalian species harbor compositionally distinct gut microbial communities, but the mechanisms that maintain specificity of symbionts to host species remain unclear. Here, we show that natural selection within house mice (Mus musculus domesticus) drives deterministic assembly of the house-mouse gut microbiota from mixtures of native and non-native microbiotas. Competing microbiotas from wild-derived lines of house mice and other mouse species (Mus and Peromyscus spp.) within germ-free wild-type (WT) and Rag1-knockout (Rag1^-/-) house mice revealed widespread fitness advantages for native gut bacteria. Native bacterial lineages significantly outcompeted non-native lineages in both WT and Rag1^-/- mice, indicating home-site advantage for native microbiota independent of host adaptive immunity. However, a minority of native Bacteriodetes and Firmicutes favored by selection in WT hosts were not favored or disfavored in Rag1^-/- hosts, indicating that Rag1 mediates fitness advantages of these strains. This study demonstrates home-site advantage for native gut bacteria, consistent with local adaptation of gut microbiota to their mammalian species.

Fig. 1.. Deterministic assembly of house-mouse microbiota from mixtures of native and non-native microbiotas.

(A) Phylogeny shows evolutionary relationships among wild-derived laboratory mice from which microbiotas were obtained. (B) Scatterplot shows negative association between microbiota similarity (Jaccard) and evolutionary divergence [millions of years ago (Ma ago)] among rodent hosts. Divergence times were estimated using data from TimeTree.org. Curve shows exponential decay regression (P < 0.001, R² = 0.54). Inset shows boxplots of microbiota similarity between pairs of samples. Wilcoxon test, FDR-adjusted ****P < 1 × 10⁻⁴. (C) Cartoon shows experimental design. Fecal microbiotas from three native M. m. domesticus lines and three non-domesticus mouse lines were mixed in pairwise combinations and inoculated into weaned germ-free mice, from which fecal samples were collected weekly for 4 weeks. (D) PCoA plots show similarities among microbiotas from donors and ex-germ-free recipients based on the Jaccard similarity index. Colors indicate the mouse line from which the samples were collected corresponding to (A). Sizes of gray circles indicate time points (weeks 1 to 4), and shades of gray delineate host individuals. The PCoA plot is faceted by inoculum but plotted along common axes showing microbiota similarity among all donors and recipient ex-germ-free mice. (E) Boxplots show positive LAI values of ex-germ-free mice throughout the 4-week experiment; (F) positive differences between observed LAI values and LAI values expected under neutral assembly; and (G) log₁₀-transformed ratios of native ASVs to non-native ASVs identified as sources by SourceTracker. In (E) to (G), shapes and colors denote identities of native and non-native donors, respectively, and lines connect samples from the same mouse. For each boxplot in (B) and (E) to (G), center lines denote medians, and lower and upper hinges correspond to first and third quartiles, respectively. FDR-adjusted P values were derived from Wilcoxon tests for non-zero mean, *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Fig. 2.. Competitive advantages of native microbiota in both WT and Rag1^−/− mice.

(A) Cartoons show experimental design. Fecal samples from two M. m. domesticus lines and two non-domesticus mouse strains were mixed in equal ratios and inoculated into germ-free mouse pups at 10 days of age reared in sterile, multicage gnotobiotic isolators. Fecal samples were collected at 4 and 6 weeks. (B) PCoA plots show microbiota similarities (Jaccard) among donors and ex-germ-free recipients. Colors indicate mouse lines from which samples were collected. Sizes of gray circles indicate time points (weeks 4 and 6). As in Fig. 1, the PCoA plot is faceted by inoculum but plotted along common axes showing microbiota similarity among all donors and recipient ex-germ-free mice. (C to E) Boxplots show LAI values (C), the differences between observed and expected LAI values (D), and log₁₀-transformed ratios of native ASVs to non-native ASVs identified as sources by SourceTracker (E). For (C) to (E), shapes and colors denote identities of native and non-native donors, respectively. Wilcoxon test for non-zero mean, FDR-adjusted ****P < 0.0001. For each boxplot in (B) to (D), center lines denote medians, and lower and upper hinges correspond to first and third quartiles, respectively.

Fig. 3.. Selective advantages for a subset of native ASVs depended on Rag1.

(A) Phylogeny shows relationships among M. m. domesticus–specific ASVs detected in ex-germ-free mice that received the NY4 + PMAN or FL1 + PAH microbiota mixtures. Colors of branches denote bacterial phyla. Rings correspond to ex-germ-free mouse groups (innermost: FL1 + PAH WT; second from innermost: FL1 + PAH Rag1^−/−; second from outermost: NY4 + PMAN WT; and outermost: NY4 + PMAN Rag1^−/−) and indicate significantly positive selection on ASVs (filled squares) within ex-germ-free mice based on binomial tests. Unfilled squares mark ASVs that were detected in ex-germ-free mice but not significantly positively selected. Absence of squares indicates that the ASV was not detected in the mouse group. (B) Phylogeny from (A) pruned to only ASVs displaying significant selective advantages in WT ex-germ-free mice but not in Rag1^−/− mice. Rows correspond to individual ex-germ-free mice, and columns correspond to the tips of the phylogeny. Filled squares indicate ex-germ-free mice in which the observed relative abundance of the ASV exceeded the relative abundance expected under neutrality. (C) Boxplots display differences between observed and expected ASV relative abundances in WT and Rag1^−/− mice that received the FL1 + PAH mixture. All ASVs from both inocula in (B) for which cage-mean differences between host genotypes remained significant after FDR correction are shown. FDR-adjusted *P < 0.05 and **P < 0.01.