Genomic evidence of Escherichia coli gut population diversity translocation in leukemia patients

Abstract

Escherichia coli, a commensal species of the human gut, is an opportunistic pathogen that can reach extra-intestinal compartments, including the bloodstream and the bladder, among others. In non-immunosuppressed patients, purifying or neutral evolution of E. coli populations has been reported in the gut. Conversely, it has been suggested that when migrating to extra-intestinal compartments, E. coli genomes undergo diversifying selection as supported by strong evidence for adaptation. The level of genomic polymorphism and the size of the populations translocating from gut to extra-intestinal compartments is largely unknown. To gain insights into the pathophysiology of these translocations, we investigated the level of polymorphism and the evolutionary forces acting on the genomes of 77 E. coli isolated from various compartments in three immunosuppressed patients. Each patient had a unique strain, which was a mutator in one case. In all instances, we observed that translocation encompasses much of the genomic diversity present in the gut. The same signature of selection, whether purifying or diversifying, and as anticipated, neutral for mutator isolates, was observed in both the gut and bloodstream. Additionally, we found a limited number of non-specific mutations among compartments for non-mutator isolates. In all cases, urine isolates were dominated by neutral selection. These findings indicate that substantial proportions of populations are undergoing translocation and that they present a complex compartment-specific pattern of selection at the patient level.IMPORTANCEIt has been suggested that intra and extra-intestinal compartments differentially constrain the evolution of E. coli strains. Whether host particular conditions, such as immunosuppression, could affect the strain evolutionary trajectories remains understudied. We found that, in immunosuppressed patients, large fractions of E. coli gut populations are translocating with variable modifications of the signature of selection for commensal and pathogenic isolates according to the compartment and/or the patient. Such multiple site sampling should be performed in large cohorts of patients to gain a better understanding of E. coli extra-intestinal diseases.

Keywords: Escherichia coli; adaptation; blood stream infection (BSI); evolution; genomics; immunosuppression; infection; whole-genome sequencing.

Fig 1

Patient follow-up with their main clinical characteristics and the sampling schemes. Absence (light gray) or presence (dark gray) of each criterion is shown. Patient A had four infectious episodes, and patients B and C had one infectious episode each.

Fig 2

Genomic diversity of E. coli isolates among samples and compartments for the three patients. (A–C) Heatmaps showing the number of SNPs between each isolate of patients (A–C). Samples are ordered by site horizontally and clustered according to their SNP similarities vertically (method: complete). All isolates of patient B are mutators (*). The isolate corresponding to the reference sequence is indicated (#). We used the same color scale for all patients. Note that the number of SNPs for patients A and C is between 0 and 14 and between 0 and 400 for patient B.

Fig 3

Unrooted trees of E. coli isolates from patients A, B, and C. Trees were built using neighbor joining from the substitution presence/absence matrix. The scale indicates the number of substitutions. All isolates of patient B are mutators (*). We zoomed in on a clade of the patient B to highlight the scale difference. The bicolor points (patients A and C) denote the presence of isolates sampled in different sites with identical sequences (zero SNP).

Fig 4

Venn diagrams showing the SNP distribution among compartments for patients A, B, and C. The ellipses are proportional to the number of SNPs. All isolates of patient B are mutators (*).

Fig 5

Presence/absence heatmaps of antibiotic resistance genes of E. coli isolates when compared to the pan-resistome (including the resistance genes of all isolates). We considered a gene as present when at least 80% of its length was covered by at least one read. Genes are ordered by synteny on contigs. All isolates of patient B are mutators (*). The prevailing predicted localization of genes by PlaScope (chromosomic or plasmidic) is indicated (full list in Table S8). Note that chromosomal genes are not mobile.

Fig 6

Presence/absence heatmaps of virulence-associated genes of E. coli isolates when compared to the pan-virulome (including the virulence genes of all isolates). We considered a gene as present when at least 80% of its length was covered by at least one read. Genes are ordered by synteny on contigs. All isolates of patient B are mutators (*). The prevailing predicted localization of genes by PlaScope (chromosomic or plasmidic) is indicated (full list in Table S9). Note that plasmidic genes are not mobile, at the opposite of resistance genes (see Fig. 5).

Fig 7

Action of selection on sequences (dN/dS) of E. coli isolates for patients A, B, and C. Significant results are framed in black (see Table S9). All isolates of patient B are mutators (*). Neutrality (dN/dS not significantly different from 0) is indicated in pale yellow, whereas purifying and diversifying selection are indicated in green and red, respectively.