Metagenomics-Based, Strain-Level Analysis of Escherichia coli From a Time-Series of Microbiome Samples From a Crohn's Disease Patient

Abstract

Dysbiosis of the gut microbiome, including elevated abundance of putative leading bacterial triggers such as E. coli in inflammatory bowel disease (IBD) patients, is of great interest. To date, most E. coli studies in IBD patients are focused on clinical isolates, overlooking their relative abundances and turnover over time. Metagenomics-based studies, on the other hand, are less focused on strain-level investigations. Here, using recently developed bioinformatic tools, we analyzed the abundance and properties of specific E. coli strains in a Crohns disease (CD) patient longitudinally, while also considering the composition of the entire community over time. In this report, we conducted a pilot study on metagenomic-based, strain-level analysis of a time-series of E. coli strains in a left-sided CD patient, who exhibited sustained levels of E. coli greater than 100X healthy controls. We: (1) mapped out the composition of the gut microbiome over time, particularly the presence of E. coli strains, and found that the abundance and dominance of specific E. coli strains in the community varied over time; (2) performed strain-level de novo assemblies of seven dominant E. coli strains, and illustrated disparity between these strains in both phylogenetic origin and genomic content; (3) observed that strain ST1 (recovered during peak inflammation) is highly similar to known pathogenic AIEC strains NC101 and LF82 in both virulence factors and metabolic functions, while other strains (ST2-ST7) that were collected during more stable states displayed diverse characteristics; (4) isolated, sequenced, experimentally characterized ST1, and confirmed the accuracy of the de novo assembly; and (5) assessed growth capability of ST1 with a newly reconstructed genome-scale metabolic model of the strain, and showed its potential to use substrates found abundantly in the human gut to outcompete other microbes. In conclusion, inflammation status (assessed by the blood C-reactive protein and stool calprotectin) is likely correlated with the abundance of a subgroup of E. coli strains with specific traits. Therefore, strain-level time-series analysis of dominant E. coli strains in a CD patient is highly informative, and motivates a study of a larger cohort of IBD patients.

Strain-level epidemiology of microbial communities and the human microbiome

“…While many studies have linked overall microbiome structure or microbial species enrichments to gastrointestinal (GI) or systemic disease, relatively few have identified strain-specific microbial variants associated with these diseases. The inflammatory bowel diseases (IBD) are among the best-studied chronic gastrointestinal conditions with respect to the microbiome, and in IBD, subspecies of E. coli and Ruminococcus gnavus have each been associated with disease severity [61,62]. Hall et al [13] noted a particular subpopulation of R. gnavus strains more abundant in the IBD gut, enriched for adaptations to oxidative stress response, adhesion, and the utilization of iron and mucus.…”

A workflow for generating multi-strain genome-scale metabolic models of prokaryotes

“…The models enable accurate and rapid computational prediction of auxotrophies and nutrient utilization capability across strains from only genome sequences without the need for experiments. The results then allow us to calculate correlations between strain-specific metabolic variations and attributes of the strain's lifestyle (such as host specificity) or health outcomes such as strain-specific implications in inflammatory bowel disease 7,8,12,15 . The reconstruction of multi-strain GEMs is much faster than reconstructing a reference model from scratch, yet still highly informative.…”

Metagenomic approaches in microbial ecology: an update on whole-genome and marker gene sequencing analyses

“…Further, when using an appropriate sequencing depth, it is possible to assemble full genomes from metagenome data to gain insights into the ‘genomic diversity’ of microbial ecosystems and to obtain draft genomes of uncultured organisms [5–7]. Although recent approaches have been developed to classify marker gene sequences at lower taxonomic levels than the genus [8–10], it is still not possible to distinguish between genomes with similar marker gene regions, while WGS metagenomics allows us to assign taxonomy at the species and strain levels [11–13]. Moreover, in comparison to the marker gene approach, WGS metagenomics is generally less affected by the biases associated with the PCR necessary for amplifying the marker genes, such as the number of cycles used or the primers and hyper-variable regions chosen [14–16].…”