Patient-Specific Bacteroides Genome Variants in Pouchitis

Abstract

A 2-year longitudinal microbiome study of 22 patients who underwent colectomy with an ileal pouch anal anastomosis detected significant increases in distinct populations of Bacteroides during 9 of 11 patient visits that coincided with inflammation (pouchitis). Oligotyping and metagenomic short-read annotation identified Bacteroides populations that occurred in early samples, bloomed during inflammation, and reappeared after antibiotic treatment. Targeted cultivation of Bacteroides isolates from the same individual at multiple time points and from several patients detected subtle genomic changes, including the identification of rapidly evolving genomic elements that differentiate isogenic strains of Bacteroides fragilis from the mucosa versus lumen. Each patient harbored Bacteroides spp. that are closely related to commonly occurring clinical isolates, including Bacteroides ovatus, B. thetaiotaomicron, B. vulgatus, and B. fragilis, which contained unique loci in different patients for synthesis of capsular polysaccharides. The presence of unique Bacteroides capsular polysaccharide loci within different hosts and between the lumen and mucosa may represent adaptations to stimulate, suppress, and evade host-specific immune responses at different microsites of the ileal pouch.

Importance: This longitudinal study provides an opportunity to describe shifts in the microbiomes of individual patients who suffer from ulcerative colitis (UC) prior to and following inflammation. Pouchitis serves as a model for UC with a predictable incidence of disease onset and enables prospective longitudinal investigations of UC etiology prior to inflammation. Because of insufficient criteria for predicting which patients will develop UC or pouchitis, the interpretation of cross-sectional study designs suffers from lack of information about the microbiome structure and host gene expression patterns that directly correlate with the onset of disease. Our unique longitudinal study design allows each patient to serve as their own control, providing information about the state of the microbiome and host prior to and during the course of disease. Of significance to the broader community, this study identifies microbial strains that may have genetic elements that trigger the onset of disease in susceptible hosts.

FIG 1

Longitudinal sampling of pouchitis and nonpouchitis patients. Samples from p_patients (patients who experienced inflamed ileal pouches at least once during the 2-year study) were collected during periods of inflammation (inflamed pouch), in the absence of inflammation, and 30 days after administration of antibiotics (post antibiotic). Samples were also collected from n_patients who never developed inflammation during the study period (non-pouchitis).

FIG 2

(A) Cluster dendrogram and heat map for the 20 most abundant oligotypes. Clustering employed average linkage of a Bray-Curtis dissimilarity matrix. The labels for terminal branches show the patient number first (preceded by “p” or “n” indicating whether the patient eventually develops pouchitis elsewhere) and then the number of days after initialization of the pouch; the colors of the terminal branches indicate the condition of the pouch at the time of sampling. The heat map represents scaled value for relative abundance of each oligotype per sample. The labels to the left of the heat map indicate the best match of the oligotype sequence in the NCBI RefSeq RNA database followed by the unique oligotype identifier (ID). Longitudinal patterns of nonpouchitis (B) and pouchitis (C) patients are displayed as bar plots. The patient ID is shown above each individual bar plot. The color of the bar corresponds with the oligotype ID of the heat map. The condition of the pouch and the number of days elapsed since pouch activation are displayed below each bar plot.

FIG 3

Bar plots show the temporal change in percent abundance of B. fragilis, B. vulgatus, B. ovatus, and B. thetaiotaomicron MAGs recovered from four patients that developed pouchitis. The letters “a” and “b” in B. theta-a and B. theta-b denote two different MAGs that match B. thetaiotaomicron. The condition of the pouch during each sampling is indicated by the color of the circle below each bar. Each MAG derived from the sample is displayed in chronological order as the number of days following pouch activation. The relative abundance of each B. fragilis cultivar isolated from patients p204, p207, and p214 is displayed as a black dot within the plot.

FIG 4

Read mapping to the p-214 B. fragilis genome cultivar isolated during inflammation. The top green histogram shows the percent GC for 2-kbp segments over 2.5 Mbp of the 5-Mbp draft genome from the B. fragilis cultivar isolated from the inflamed pouch of patient 214. The black histograms in the middle of the figure represent mean coverage over 2-kbp genomic segments of short metagenomic reads (luminal or mucosal samples collected during inflammation from patients p-204, p-207, p-214, and from the lumen of the nonpouchitis patient n-212) that mapped 100% over their length with 97% identity to the p-214 cultivar genome. Black bars on the purple background at the bottom of the figure display the positions of four genes [cpsM(V), wcbM, neuB, and upxZ] within capsular polysaccharide biosynthesis (CPS) loci highlighted in yellow. Two regions highlighted in red represent mobile elements within the genome.