Driving gut microbiota enterotypes through host genetics

Abstract

Background: Population stratification based on interindividual variability in gut microbiota composition has revealed the existence of several ecotypes named enterotypes in humans and various animal species. Enterotypes are often associated with environmental factors including diet, but knowledge of the role of host genetics remains scarce. Moreover, enterotypes harbor functionalities likely associated with varying abilities and susceptibilities of their host. Previously, we showed that under controlled conditions, 60-day-old pig populations consistently split into two enterotypes with either Prevotella and Mitsuokella (PM enterotype) or Ruminococcus and Treponema (RT enterotype) as keystone taxa. Here, our aim was to rely on pig as a model to study the influence of host genetics to assemble enterotypes, and to provide clues on enterotype functional differences and their links with growth traits.

Results: We established two pig lines contrasted for abundances of the genera pairs specifying each enterotype at 60 days of age and assessed them for fecal microbiota composition and growth throughout three consecutive generations. Response to selection across three generations revealed, per line, an increase in the prevalence of the selected enterotype and in the average relative abundances of directly and indirectly selected bacterial genera. The PM enterotype was found less diverse than the RT enterotype but more efficient for piglet growth during the post-weaning period. Shotgun metagenomics revealed differentially abundant bacterial species between the two enterotypes. By using the KEGG Orthology database, we show that functions related to starch degradation and polysaccharide metabolism are enriched in the PM enterotype, whereas functions related to general nucleoside transport and peptide/nickel transport are enriched in the RT enterotype. Our results also suggest that the PM and RT enterotypes might differ in the metabolism of valine, leucin, and isoleucine, favoring their biosynthesis and degradation, respectively.

Conclusion: We experimentally demonstrated that enterotypes are functional ecosystems that can be selected as a whole by exerting pressure on the host genetics. We also highlight that holobionts should be considered as units of selection in breeding programs. These results pave the way for a holistic use of host genetics, microbiota diversity, and enterotype functionalities to understand holobiont shaping and adaptation. Video Abstract.

Keywords: Enterotype; Genetic selection; Genetics; Gut microbiota; Heritability; Holobiont; Metagenomics; Pig.

Fig. 1

Animal protocol and stratification of the Large White pigs according to their enterotypes PM (Prevotella-Mitsuokella) or RT (Ruminococcus-Treponema). A Timing for stool sampling (blue triangle) and body weight records (colored circles) from birth (D0) until the end of the post-weaning period (D70). Microbiota data was obtained in all animals by sequencing the 16S rRNA gene using DNA extracted from stools sampled at D60. For a subset of 30 G0 animals, whole-metagenome sequencing data was obtained at D60. B Selection strategy of the two pig lines HPM (High PM) and HRT (High RT) over three successive generations (G1 to G3). The G0 generation was not genetically selected and corresponds to the founding population obtained from 30 litters from 30 males crossed with 30 females. The generations G1 to G3 comprised 30 litters produced from 6 males and 30 females each. The number of piglets per generation and per pig line is indicated. The relative prevalence of the two enterotypes within each population is represented by pie charts, the blue and red sections representing the % of pigs with the PM or RT enterotype, respectively. The number of pigs for each enterotype is reported in the pie charts. C Enterotype distribution of the whole population of 1067 pigs (generation G0 to G3, two pig lines) into two groups that correspond to the enterotypes PM (blue) and RT (red). D Notched box plots showing the differences in alpha diversity (left: Shannon index, right: richness) according to generation and pig line

Fig. 2

Main taxonomic and functional differences between the two enterotypes that were characterized on animals from the G0 basal population. A Average relative abundances of the main bacterial genera for the two pig groups harboring either PM (142 piglets) or RT (87 piglets) enterotypes based on 16S rRNA sequencing data. This analysis was narrowed down to the piglets with an unchanged enterotype across multiple repeats of the clustering process. B Co-abundance networks based on the most abundant and differentially abundant genera between the two enterotypes defined on the G0 population at D60, using 16S rRNA sequencing data. C Most contrasted MetaGenomic species (MGS) identified based on shotgun metagenomics data between two subsets of 15 females representative of each enterotype at G0. Effect size was estimated using the Cliff’s Delta statistic. Blue and red bars correspond to MGS enriched in PM and RT animals, respectively

Fig. 3

Genetic parameters of fecal microbiota composition and body weight phenotypes at D60. A Heritability estimates (h² blue dots) and litter effect (c² pink dots) with their standard error (lines) for 64 gut microbiota genera and 2 diversity indexes (genera) based on 16S rRNA sequencing data. Selected genera are written in dark blue (B). Genetic correlation between genera under selection in the HPM line (red squares) or the HRT line (blue triangles), diversity indexes (green disk), and post-weaning growth rate (ADG_pw, yellow star). Negative correlations are marked with dashed lines. Black lines represent correlations for which the absolute value is higher than 0.9, purple lines represent correlations with absolute values between 0.6 and 0.9 and gray lines correlations with absolute values between 0.3 and 0.5

Fig. 4

Responses to genetic selection oriented towards a high abundance of Prevotella and Mitsuokella or a high abundance of Ruminococcus and Treponema at D60 across three generations. The differences between the HPM and HRT lines for 74 genera, 2 diversity indexes (light blue highlighting), and one growth feature (ADG_postweaning: average daily gain during the post-weaning period, light green highlighting) are expressed in standard deviation. They are ranked from the largest positive difference (on the left) to the largest negative difference (on the right) at generation G3 of the selection. The differences are in pale green for G1, green for G2, and dark green for G3. Positive differences are associated with higher values in the HPM line whereas negative values are associated with higher values in the HRT line. Microbiota results are based on 16S rRNA sequencing data