Propionate Fermentative Genes of the Gut Microbiome Decrease in Inflammatory Bowel Disease

Abstract

Changes in the gut microbiome have been associated with inflammatory bowel disease. A protective role of short chain fatty acids produced by the gut microbiota has been suggested as a causal mechanism. Nevertheless, multi-omic analyses have failed to identify a clear link between changes in specific taxa and disease states. Recently, metagenomic analyses unveiled that gut bacterial species have a previously unappreciated genomic diversity, implying that a geno-centric approach may be better suited to identifying the mechanisms involved. Here, we quantify the abundance of terminal genes in propionate-producing fermentative pathways in the microbiome of a large cohort of healthy subjects and patients with inflammatory bowel disease. The results show that propionate kinases responsible for propionate production in the gut are depleted in patients with Crohn's disease. Our results also indicate that changes in overall species abundances do not necessarily correlate with changes in the abundances of metabolic genes, suggesting that these genes are not part of the core genome. This, in turn, suggests that changes in strain composition may be as important as changes in species abundance in alterations of the gut microbiome associated with pathological conditions.

Keywords: Crohn’s disease; metagenomics; microbiota; short chain fatty acids.

Figure 1

Levels of propionate in healthy subjects (H) and patients with Crohn´s disease (CD) and ulcerative colitis (UC) in absolute concentration units. (a) A kernel density estimate plot represents the abundance of propionate from the metabolomic profiles measured in fecal samples from the cohort analyzed in Lloyd-Price [19]. (b) A boxplot represents the abundance of propionate measured in fecal samples. Horizontal black bars represent median levels, while boxes and whiskers represent the data from first to third quartiles and from the quartiles to the minimum and maximum, respectively. Black dots correspond to outlier values outside the interquartile range. Statistical significance of the differences between groups was evaluated using a pairwise Mann–Whitney test with the Benjamini–Hochberg false discovery rate correction. No significant differences among groups were obtained for CD and H, while significant differences between H and UC were observed with p = 0.029 (*).

Figure 2

Characterization of the terminal genes involved in the formation of bacterial propionate. (a) Scheme with the three terminal reactions involved in the formation of propionate. Enzymes, EC numbers, and coding genes are represented in purple, green, and blue, respectively. As seen, some of the participating enzymes have substrate broadness between acetate and propionate (depicted in light grey). For example, the enzyme encoded by the acs gene forms acetate (EC 6.2.1.1) and propionate (EC 6.2.1.17); (b) boxplot representing the relative abundance of the four terminal genes involved in propionate formation in all the GM metagenomic samples. Horizontal black bars represent median levels, while boxes and whiskers represent the data from first to third quartiles and from the quartiles to the minimum and maximum, respectively. Abundances were measured as indicated in the Materials in Methods and are represented in thousands of reads.

Figure 3

Abundance of genes involved in the terminal formation of propionate in healthy subjects (H) and patients with Crohn´s disease (CD) and ulcerative colitis (UC). (a,b) Propionate kinase genes tdcD and pduW. (c,d) Propionate CoA transferase gene pct and propionate CoA ligase gene prpE. Density plots represent the abundances of terminal microbial genes involved in the formation of propionate, from the metagenomic measures of the fecal samples from the cohort analyzed in [16]. Differences in abundances were evaluated using a pairwise Mann–Whitney test with the Benjamini–Hochberg false discovery rate correction. Significant differences among groups were obtained for tdcD and pduW between CD and H (*).

Figure 3

Abundance of genes involved in the terminal formation of propionate in healthy subjects (H) and patients with Crohn´s disease (CD) and ulcerative colitis (UC). (a,b) Propionate kinase genes tdcD and pduW. (c,d) Propionate CoA transferase gene pct and propionate CoA ligase gene prpE. Density plots represent the abundances of terminal microbial genes involved in the formation of propionate, from the metagenomic measures of the fecal samples from the cohort analyzed in [16]. Differences in abundances were evaluated using a pairwise Mann–Whitney test with the Benjamini–Hochberg false discovery rate correction. Significant differences among groups were obtained for tdcD and pduW between CD and H (*).

Figure 4

tdcD, pduW, and 16S average gene-abundance differences between healthy and CD conditions. (a,b) Differences in kinase abundances. Horizontal bars in the plots represent the difference in total average gene abundances between H and CD groups for tdcD and pduW. (c) Differences in average 16S abundance between the CD and H groups. Bars indicate the difference in log10 average abundances between H and CD groups, inferred from the 16S data in [16]. Only genera with the highest net variation (positive or negative) are represented.

Figure 5

Dot plots between (a) tdcD and 16S, and (b) tdcD and rpoB average gene abundance. Dots in the plot represent the average gene abundance of GM bacterial genera in CD, UC, and healthy conditions on a log10 scale. Only genera with consistent taxonomic annotation between data from [14,16] are presented. Labels indicate bacterial taxa with higher differences between the 16S- and rpoB-based quantifications. The regression lines with the correlation coefficient between the corresponding genes are presented in both plots.