Genome Dynamics of Escherichia coli during Antibiotic Treatment: Transfer, Loss, and Persistence of Genetic Elements In situ of the Infant Gut

Abstract

Elucidating the adaptive strategies and plasticity of bacterial genomes in situ is crucial for understanding the epidemiology and evolution of pathogens threatening human health. While much is known about the evolution of Escherichia coli in controlled laboratory environments, less effort has been made to elucidate the genome dynamics of E. coli in its native settings. Here, we follow the genome dynamics of co-existing E. coli lineages in situ of the infant gut during the first year of life. One E. coli lineage causes a urinary tract infection (UTI) and experiences several alterations of its genomic content during subsequent antibiotic treatment. Interestingly, all isolates of this uropathogenic E. coli strain carried a highly stable plasmid implicated in virulence of diverse pathogenic strains from all over the world. While virulence elements are certainly beneficial during infection scenarios, their role in gut colonization and pathogen persistence is poorly understood. We performed in vivo competitive fitness experiments to assess the role of this highly disseminated virulence plasmid in gut colonization, but found no evidence for a direct benefit of plasmid carriage. Through plasmid stability assays, we demonstrate that this plasmid is maintained in a parasitic manner, by strong first-line inheritance mechanisms, acting on the single-cell level, rather than providing a direct survival advantage in the gut. Investigating the ecology of endemic accessory genetic elements, in their pathogenic hosts and native environment, is of vital importance if we want to understand the evolution and persistence of highly virulent and drug resistant bacterial isolates.

Keywords: Escherichia coli; antibiotic treatment; genome evolution; horizontal gene transfer; infant gut; plasmid persistence; urinary tract infections; virulence plasmid dynamics.

Figure 1

Resistance profiles of co-existing E. coli lineages during the course of a urinary tract infection and antibiotic treatment. E. coli isolates were obtained during the first 12 months of an infant's life. At 8 days of age, the infant was admitted to the hospital due to a urinary tract infection from which the lineage A clone was isolated at day 11. After 5 days of i.v. trimethoprim the treatment was switched to 5 days of i.v. ampicillin followed by an additional 8 days of oral (p.o.) amoxicillin. Trimethoprim was administered to prevent reoccurring infections for the following 7 months. Along the course of treatment, lineage B acquired a TEM-1b encoding IncX plasmid from lineage A; rendering both lineages resistant to the β-lactam treatment at day 32.

Figure 2

SNP tree and genomic events of lineage A. Branch-numbers show the amount of SNPs separating the isolates obtained from 2 days to 12 months after birth. The majority of samples were isolated during hospitalization and prophylactic treatment. No SNPs were detected between the isolates collected at days 2, 9, and 16, and only one SNP (in the blaTEM1b promoter) was detected between these isolates and the isolate collected at 32 days. Two isolates were included at the 2 months sampling point (2 m1 and 2 m2) and these shared 5 SNPs in common compared to the 32 days isolate. Two 6 month isolates were also included. The infant was subjected to antibiotic treatment from day 11 to 8 months after birth. Apart from SNP differences one sub-lineage had undergone major genomic changes by means of chromosomal deletions and acquisition of new plasmid DNA.

Figure 3

Overview of plasmids resembling pNK29-2 and their host phylogeny. Thirteen plasmids with high similarity to pNK29-2 were downloaded from Genbank (see Table S4 for accession numbers and references). Info on host strain, its disease associations and geographical origin was obtained from the literature. The “Year” column denotes the first mentioning of the host strain in the literature unless the isolation year was clearly stated. In silico MLST typing was performed and FimH types were added to the ST131 clade highlighting their internal diversity. A core-genome based maximum-likelihood tree was constructed to illustrate the diversity of the E. coli plasmid hosts. Node numbers are bootstrap confidence values.

Figure 4

Genetic variability and conservation within similar virulence plasmid backbones. six out of the Thirteen virulence plasmids similar to pNK29-2 showed signs of major restructuring events and are illustrated here. While these plasmids show some degree of variation, they also share a conserved core of transfer (green), stability (blue), and virulence genes (red). Mobile elements constituting inverted repeats (highlighted in orange) allow for instability of the cjrABC-containing region exemplified in pSaT040 and pECO-bc6. Additional genes involved in antibiotic resistance have been inserted upstream the fully conserved iron acquisition cluster of the genetic load region in p1ESCUM, pMVAST0167_1, pKPN7c3, and pECO-fce. The TEM-1b and tetA genes are highlighted in cyan and light green respectively. pECO-fce is significantly larger than the remaining plasmids due to a duplication of its transfer region which has been condensed for simplicity. Colored shades illustrate BLAST identity of pairwise comparisons and orange shades highlight inverted regions.

Figure 5

Competitive growth of lineage a strain against a pNK29-2 cured variant within streptomycin treated mice. Mice where inoculated by oral gavage with equal amounts of plasmid carrying and plasmid-free cells. The competitive index (top left equation) is the ratio of CFUs obtained from selective plating. This ratio was also quantified for the small intestine at the endpoint (day 12) of the experiment to assess the potential role of adhering cells. Replicates were excluded when the strains were no longer detected. Significance indicators illustrate pairwise comparisons of means (Mann-Whitney U-Test: ^*P = 0.014, ^**P = 0.0015, ^***P = 0.001)