Comprehensive functional core microbiome comparison in genetically obese and lean hosts under the same environment

Abstract

Our study provides an exhaustive comparison of the microbiome core functionalities (captured by 3,936 microbial gene abundances) between hosts with divergent genotypes for intramuscular lipid deposition. After 10 generations of divergent selection for intramuscular fat in rabbits and 4.14 phenotypic standard deviations (SD) of selection response, we applied a combination of compositional and multivariate statistical techniques to identify 122 cecum microbial genes with differential abundances between the lines (ranging from -0.75 to +0.73 SD). This work elucidates that microbial biosynthesis lipopolysaccharides, peptidoglycans, lipoproteins, mucin components, and NADH reductases, amongst others, are influenced by the host genetic determination for lipid accretion in muscle. We also differentiated between host-genetically influenced microbial mechanisms regulating lipid deposition in body or intramuscular reservoirs, with only 28 out of 122 MGs commonly contributing to both. Importantly, the results of this study are of relevant interest for the efficient development of strategies fighting obesity.

Fig. 1. Comprehensive identification of cecal core microbial gene (MG) additive log-ratio (alr) transformed abundances evidencing a correlated response to selection for intramuscular fat content.

Individual plots (n = 89) and observed vs. cross-validation predicted values (with seven validation groups) obtained from a. Partial least squares discriminant analysis (PLS-DA) (each line is represented by a different color, contoured by 95% confidence ellipses) and b. Partial Least Squares (PLS) analysis (colored by increased intramuscular fat content, g/100 g of muscle (IMF)). The final models were built based on 4 latent components in both cases and 240 or 230 MG alrs, respectively. PLS-DA model showed a misclassification rate of 0%, and a cross-validation misclassification rate of 5%. PLS model showed a goodness of fit (R²) of 91% and a goodness of prediction after cross-validation (Q²) of 79%. c Venn diagram displays 122 MG alrs selected in both PLS-DA and PLS analysis, while 118 were selected exclusively in PLS-DA, and 108 exclusively in PLS model.

Fig. 2. Differences between obese (n = 47) and lean (n = 42) lines in core microbial gene (MG) additive log-ratio (alr)-transformed abundances presenting a correlated response to selection for intramuscular fat content.

Bars represent differences in units of standard deviation (SD) of MG alrs and highest posterior-density intervals at 95% probability. Colors represent the classification of MGs according to Cluster of Orthologous Groups of proteins (COG). Full information of the MGs can be found in Supplementary Data 4. The 122 MG alrs commonly identified in partial least squares discriminant analysis and partial least squares analysis were classified into 20 different COGs, but only the most represented are displayed in the graph: cell wall/membrane/envelope biogenesis (21 MGs), energy production and conversion (13 MGs), amino acid metabolism (11 MGs), lipid transport and metabolism (8 MGs), translation, ribosomal structure, and biogenesis (7 MGs), coenzyme transport and metabolism (6 MGs), post-translational modification protein turnover, chaperones (6 MGs), and signal-transduction mechanisms (5 MGs).

Fig. 3. Top 10 microbial taxa highly enriched with unique proteins classified within KEGG Orthologue groups (or microbial genes) presenting a correlated response to selection for intramuscular fat content.

Color scale represents the number of unique proteins mapping into each KEGG Orthologue group (or microbial gene) identified within each taxa. Full names of microbial genes are given in Supplementary Data 4.