A carbohydrate-active enzyme (CAZy) profile links successful metabolic specialization of Prevotella to its abundance in gut microbiota

Abstract

Gut microbiota participates in diverse metabolic and homeostatic functions related to health and well-being. Its composition varies between individuals, and depends on factors related to host and microbial communities, which need to adapt to utilize various nutrients present in gut environment. We profiled fecal microbiota in 63 healthy adult individuals using metaproteomics, and focused on microbial CAZy (carbohydrate-active) enzymes involved in glycan foraging. We identified two distinct CAZy profiles, one with many Bacteroides-derived CAZy in more than one-third of subjects (n = 25), and it associated with high abundance of Bacteroides in most subjects. In a smaller subset of donors (n = 8) with dietary parameters similar to others, microbiota showed intense expression of Prevotella-derived CAZy including exo-beta-(1,4)-xylanase, xylan-1,4-beta-xylosidase, alpha-L-arabinofuranosidase and several other CAZy belonging to glycosyl hydrolase families involved in digestion of complex plant-derived polysaccharides. This associated invariably with high abundance of Prevotella in gut microbiota, while in subjects with lower abundance of Prevotella, microbiota showed no Prevotella-derived CAZy. Identification of Bacteroides- and Prevotella-derived CAZy in microbiota proteome and their association with differences in microbiota composition are in evidence of individual variation in metabolic specialization of gut microbes affecting their colonizing competence.

Figure 1

Numbers of functional (A) and taxonomical (B) annotation of the identified peptides. (A) Most of the identified peptides belong to a single orthologous group. (B) At genus level, appr. 20–25% of peptides were unanimously annotated to one bacterial genus, while appr. 40–60% of peptides was shared by two or more genera. Peptides remaining without taxonomic annotation constituted 13.8–43.2% of peptides. (For a particular genus, see color codes).

Figure 2

Co-occurrence network of CAZy families of particular bacterial genera in the samples. Bacteroides- and Prevotella-derived glycosyl hydrolases (GH) cluster together in distinct sets of samples. Clusters containing fewer GH enzymes can be identified for Faecalibacterium, Eubacterium and Bifidobacterium- derived enzymes. Each node represents a genus–CAZy family pair, and the edges represent a significant co-occurrence relationship (see text for details). The top 500 correlating pairs were included in the graph.

Figure 3

Donors with high abundancy of Bacteroides and donors with high abundance of Prevotella differ in their expression of CAZy. (A) Relative abundance of the most prominent bacterial groups on genus-level according to 16S-RNA parallel sequencing. (B) A heatmap of CAZy enzymes identifies two distinct profiles of CAZy enzyme expression among donors (in vertical columns). CAZy enzymes (horizontal lines) annotated as Bacteroides- or Prevotella-derived (or ambiguous) are expressed at a higher level in two distinct groups of donors (consisting of 25 and 8 donors, demarcated with blue or red frames, respectively). In (A), each dot represents the value of abundance for one donor. Enzymes with a FDR < 0.01 are included in the heatmap.

Figure 4

(A) A list of glycosyl hydrolase (GH) and other CAZy families with unambiguous Prevotella or Bacteroides-annotation in the form of a heatmap. Of the 88 families found in this heatmap (horizontal lines), 21 are shared (framed with black for their Prevotella-annotated identifications) and appear in both clusters and groups of donors. Inclusion of CAZy families is based purely on peptide annotations without any profiling of donors.

Figure 5

Prevotella representation in microbiota of study subjects and its relationship to Prevotella-CAZy proteome. (A) Principal component analysis (PCA) of microbiota composition based on 16S-RNA sequencing at genus level and (B) based on intensity-weighed CAZy enzyme expression. In both PCA-plots, donors with high abundance of Prevotella (red) differ significantly from all others (black) according to permutational multivariate analysis of variance (PERMANOVA) (p < 0.001 and p < 0.01, for (A) and (B), respectively). Numbers refer to the number of each donor in heatmaps (Figs. 3B, 4).