Antibiotic resistant Escherichia coli and Salmonella enterica, and the characterization of E. coli O26 recovered from bovine feces

Escherichia coli and Salmonella, are commonly isolated from cattle and can harbor and transfer antimicrobial resistance (AMR) genes, however, it is uncertain of where variation of AMR occurs in feedlot production systems. Shiga toxin-positive E. coli O26:H11 has been extensively studied, however, there is a lack of available literature into non-Shiga toxin variants. The objectives of this research were to 1) partition unexplained model variance in AMR among E. coli and Salmonella into differing levels of aggregation in cattle prior to harvest; 2) molecularly investigate the genetic diversity and pathogenic potential of E. coli O26 strains recovered from bovine feces; and 3) assess the phylogenetic relatedness of Shiga toxin-negative and eae-positive E. coli O26 strains from cattle to highly pathogenic O26 clones isolated in humans. These objectives provide a better characterization and understanding of AMR and E. coli O26 to improve food safety and public health. To explore the variation of AMR in E. coli and Salmonella in cattle prior to harvest, a convenience sampling was performed where four feedlots were included, with each feedlot visited on four occasions every 14 to 28 days. At each visit, four pens of cattle were sampled. Of these pens, two pens each of cattle newly received within thirty days into the feedlot and two pens of cattle within six weeks of harvest were sampled. Twenty pen-floor fecal samples (N=1264) were collected within each pen, except in one pen where only four samples were collected. Three non-type specific E. coli and up to two Salmonella isolates per sample were isolated. Salmonella isolates were serotyped using serological agglutination and antimicrobial susceptibility was evaluated using microbroth dilution. Phenotypic tetracycline resistance was observed among 25.1 and 26.6% of Salmonella and E. coli isolates, respectively. Ceftiofur resistance was rare (1.21 and 4.69% for E. coli and Salmonella, respectively), except in one feedlot where it was observed among 16.5% of Salmonella isolates. To explore the extent of unexplained variation attributable to different levels of organization, multi-level, binomial models were constructed for E. coli and Salmonella isolates displaying tetracycline and ceftiofur resistance. In the multilevel models of tetracycline resistance in Salmonella, 35.7% of the variance occurred at the pen-level and 15.6% was observed at the feedlot-level. In contrast, in the Salmonella ceftiofur resistance multilevel model feedlot-level variation (44.0%) was greater than that observed at the pen-level (9.1%). In the four and five-level E. coli tetracycline resistance models, there was little evidence of feedlot-to-feedlot variation (0 and 2.5%, respectively). While the E. coli ceftiofur resistance model that treated animal as the constant, feedlot-level contributed to most of the observed variation (12.1%). These models inform the design of new sampling schemes to describe the burden and changes over time in antimicrobial resistance. For the monitoring of a phenotype where more variation occurs at the pen-level (e.g. tetracycline resistance), sampling should focus on more pens within fewer feedlots. However, for monitoring a phenotype in which more variation occurs at the feedlot-level (e.g. ceftiofur resistance), the primary sampling should include more feedlots and subsequently fewer pens within each feedlot. Escherichia coli O26 has been identified as the most common non-O157 Shiga toxin-producing (STEC) serogroup to cause human illnesses in the United States and has been implicated in outbreaks around the world. E. coli has high genomic plasticity, which facilitates loss or acquisition of virulence genes. Attaching and effacing E. coli (AEEC) O26 strains have frequently been isolated from bovine feces, and there is a need to better characterize the relatedness of these strains to defined molecular pathotypes and to describe the extent of their genetic diversity. High-throughput real-time PCR was used to screen 178 E. coli O26 isolates from a single U.S. cattle feedlot, collected from May to July 2011, for the presence or absence of 25 O26 serogroup-specific and virulence-associated markers. The selected markers were capable of distinguishing these strains into molecularly defined groups (yielding 18 unique marker combinations). Analysis of the clustered regularly interspaced short palindromic repeat 1 (CRISPR1) and CRISPR2a loci further discriminated isolates into 24 CRISPR types. The combination of molecular markers and CRISPR typing provided 20.8% diversity. The recent CRISPR PCR target SP_O26-E, which was previously only identified in stx2-positive O26:H11 human clinical strains, was identified in 96.4% (161/167, [95% confidence interval, 99.2 to 93.6%]) of the stx-negative AEEC O26:H11 bovine fecal strains. This supports that these stx-negative strains may have previously contained a prophage carrying stx or could acquire this prophage, thus possibly allowing them the potential to become pathogenic to humans. These results show that investigation of specific genetic markers may further elucidate our understanding of the genetic diversity of AEEC O26 in bovine feces. Enterohemorrhagic E. coli (EHEC) O26:H11, a serotype within STEC that causes severe human disease, has been considered to have evolved from AEEC O26:H11 through the acquisition of a Shiga toxin-encoding gene. Targeted amplicon sequencing using next-generation sequencing technology of forty-eight phylogenetically informative single nucleotide polymorphisms (SNPs) and three SNPs differentiating stx-positive from stx-negative strains were used to infer the phylogenetic relationship of 178 (6 stx-positive and 172 stx-negative AEEC) E. coli O26:H11 strains from cattle feces to 7 publically available genomes of human clinical strains. The AEEC cattle strains displayed synonymous SNP genotypes with stx2 positive sequence type 29 (ST29) human O26:H11 strains, while stx1 ST21 human and cattle strains clustered separately, demonstrating the close phylogenetic relatedness of these Shiga toxin-negative AEEC cattle strains and human clinical strains. With the exception of seven stx-negative strains, five of which contained espK, three stx-related SNPs differentiated the STEC from non-STEC strains, supporting that these AEEC cattle strains could serve as a potential reservoir for new or existing pathogenic human strains. Our results support that targeted amplicon sequencing for SNP genotyping expedites strain identification and the genetic characterization of E. coli O26:H11, which is important for food safety and public health.