PLoS ONE (Jan 2018)

Absence of genetic selection in a pathogenic Escherichia coli strain exposed to the manure-amended soil environment.

  • Allison M Truhlar,
  • Thomas G Denes,
  • Keiran K Cantilina,
  • Selene K Leung,
  • M Todd Walter,
  • Anthony G Hay

DOI
https://doi.org/10.1371/journal.pone.0208346
Journal volume & issue
Vol. 13, no. 12
p. e0208346

Abstract

Read online

Escherichia coli that express curli are more common in subsurface soil drainage when manure is surface applied. However, it is unknown whether this arises from mutations in individual strains leading to curli expression or by selection for individuals already expressing higher levels of curli. To test this, we examined curli production in pathogenic E. coli O157:H7 EDL933 as a function of manure management. Five treatments were investigated: (1) soil only, (2) soil with surface-applied E. coli O157:H7 EDL933 Δstx1-2 (EcO157), (3) soil with incorporated EcO157, (4) soil with surface-applied EcO157-inoculated manure, and (5) soil with incorporated EcO157-inoculated manure. EcO157 was reisolated from soils immediately after application and weekly thereafter for 8 weeks. EcO157 in the surface-applied treatments died faster than their incorporated treatment counterparts. Phenotypic assays revealed differences between treatments as well. Half of surface-applied manure reisolates from week 6 developed a mixed red and white colony morphology on Congo Red plates, indicating changes in curli production that were not seen in other treatments or times. In 37°C growth tests, week 6 reisolates from all treatments except soil surface-applied EcO157 left the lag phase at a significantly greater rate than week 0 isolates. We applied whole genome sequencing technology to interrogate the genetic underpinnings of these phenotypes. Surprisingly, we only found single-nucleotide polymorphisms in two of the 94 resequenced isolates from the different treatments, neither of which correlated with curli phenotype. Likewise, we found no differences in other genomic characteristics that might account for phenotypic differences including the count of gaps and the origin of discarded reads that failed to map to the parental strain. These results suggest there were no systematic genomic differences (i.e. individual-level selection) that correlated with time or treatment. We recommend future research focus on population-level selection of E. coli strains in the manure-amended soil environment.