Age-related changes in microbial composition and function in cynomolgus macaques

Abstract

Age can significantly affect human physiology and disease risk. Recent studies have shown that age may affect the composition and function of the gut microbiota, but the underlying mechanisms remain largely unknown. Non-human primates are an ideal model for uncovering how age shapes the gut microbiota, as their microbial composition is highly similar to that of humans and is not easily affected by confounding factors. Here, using the 16S rRNA and metagenomic sequencing methods, we characterized the microbial phenotypes of 16 female cynomolgus macaques from three age groups (young, adult and old). Our findings revealed significant differences in microbial composition among the three groups. With increased age, the relative abundances of Veillonellaceae, Coriobacteriaceae and Succinivibrionaceae were significantly increased, Ruminococcaceae and Rikenellaceae were significantly decreased at the family level. Functional enrichment showed that genes that differed among the three groups were mainly involved in arginine biosynthesis, purine metabolism and microbial polysaccharides metabolism. Moreover, CAZymes corresponding to polysaccharide degrading activities were also observed among the three groups. In conclusion, we characterized the composition and function of the gut microbiota at different ages, and our findings provide a new entry point for understanding the effects of age on the human body.

Keywords: age; cynomolgus macaques; gut microbiota; metagenomics.

Figure 1

Comparison of the microbial composition between the three groups. (A) Venn diagram depicting OTU richness and the overlap in microbial communities between the young (green), adult (red) and old (blue) monkeys. (B, C) Relative abundance of OTUs assigned at the phylum and family levels.

Figure 2

Comparison of the microbial composition between the young, adult and old groups. (A) 3-D Principle component analysis (PCA) plot of samples along principle component (PC) 1,2 and 3, which explained 14.73%, 9.67% and 9.18% of the total variance, respectively. (B) Partial least squares discriminant analysis (PLS-DA) plot of gut microbiota among three groups: young (n=5, 2–4 years, green dots), adult (n=6, 5–13 years, red dots) and old (n=5, 17–20 years, blue dots).

Figure 3

The most differentially expressed taxa among the three groups. (A) Heatmap of the 148 discriminative OTU abundances among the young, adult and old groups (LDA>2.0). OTUs (raw) were sorted by taxa and enriched groups, samples (column) were sorted by age. The intensity of color (blue to red) indicated the score normalized abundance for each OTU. (B) Venn diagram for different OTUs among the three groups. Blue designates enriched taxa between the young and adult groups; green designates enriched taxa between the adult and old groups; yellow designates the enriched taxa between the young and old groups. (C) Scatter diagram of the relative abundances of the age-related microbial families Ruminococcaceae and Veillonellaceae. The correlation was tested by Pearson’s correlation analysis and was adjusted by partial correlation analysis.

Figure 4

Age-related microbial functions in the KEGG pathway. Different KEGG enzymes identified by LEfSe analysis of the metagenomic sequences (LDA>2.0). (A) Heatmap of the abundances of different enzymes. Enzymes (raw) were sorted by taxa and enriched group, and samples (column) were sorted by age. The intensity of the color (blue to red) indicates the score normalized abundance for each enzyme. (B) Boxplot for the marked KEGG enzymes in different age groups. ECs were classified into pathways for arginine, purine and microbial polysaccharide metabolism. The abundances of different ECs were calculated by reads number. LEfSe was used to detect features with significantly different abundances using the Kruskal–Wallis rank sum test, and LDA was performed to evaluate the effect size of each feature. *P<0.05 **P<0.01. LPS, lipopolysaccharide; GAG, glycosaminoglycan; PGN, peptidoglycan.

Figure 5

Age-related microbial functions in CAZymes and food utilization. Different CAZymes identified using LEfSe analysis of the metagenomic sequences (LDA>2.0). (A) Heatmap of the abundances of different CAZymes. CAZymes (raw) were sorted by taxa and enriched group, and samples (column) were sorted by age. The intensity of color (blue to red) indicates the score normalized abundance for each enzyme. (B) The ratio of CAZymes represented within the metagenomes related to plant and animal carbohydrate utilization (left) or the ratio of mucin glycan to plant carbohydrate utilization (right) in the cynomolgus macaques. The boxplot distributions were tested using the nonparametric two-sided Wilcoxon rank sum test. (C) Boxplot for the marked CAZymes in different age groups. Representative CAZymes were classified into pathways for starch (left) and microbial peptidoglycan (right). The abundances of different CAZymes were calculated by the log RPKM. LEfSe detected the features with significantly different abundances using the Kruskal–Wallis rank sum test, and LDA was performed to evaluate the effect size of each feature. *P<0.05, **P<0.01.