Subgrouping of ESBL-producing Escherichia coli from animal and human sources: an approach to quantify the distribution of ESBL types between different reservoirs.

Escherichia (E.) coli producing extended-spectrum beta-lactamases (ESBLs) are an increasing problem for public health. The success of ESBLs may be due to spread of ESBL-producing bacterial clones, transfer of ESBL gene-carrying plasmids or exchange of ESBL encoding genes on mobile elements. This makes it difficult to identify transmission routes and sources for ESBL-producing bacteria. The objectives of this study were to compare the distribution of genotypic and phenotypic properties of E. coli isolates from different animal and human sources collected in studies in the scope of the national research project RESET. ESBL-producing E. coli from two longitudinal and four cross-sectional studies in broiler, swine and cattle farms, a cross-sectional and a case-control study in humans and diagnostic isolates from humans and animals were used. In the RESET consortium, all laboratories followed harmonized methodologies for antimicrobial susceptibility testing, confirmation of the ESBL phenotype, specific PCR assays for the detection of bla(TEM), bla(CTX), and bla(SHV) genes and sequence analysis of the complete ESBL gene as well as a multiplex PCR for the detection of the four major phylogenetic groups of E. coli. Most ESBL genes were found in both, human and non-human populations but quantitative differences for distinct ESBL-types were detectable. The enzymes CTX-M-1 (63.3% of all animal isolates, 29.3% of all human isolates), CTX-M-15 (17.7% vs. 48.0%) and CTX-M-14 (5.3% vs. 8.7%) were the most common ones. More than 70% of the animal isolates and more than 50% of the human isolates contained the broadly distributed ESBL genes bla(CTX-M-1), bla(CTX-M-15), or the combinations bla(SHV-12)+bla(TEM) or bla(CTX-M-1)+bla(TEM). While the majority of animal isolates carried bla(CTX-M-1) (37.5%) or the combination bla(CTX-M-1)+bla(TEM) (25.8%), this was the case for only 16.7% and 12.6%, respectively, of the human isolates. In contrast, 28.2% of the human isolates carried bla(CTX-M-15) compared to 10.8% of the animal isolates. When grouping data by ESBL types and phylogroups bla(CTX-M-1) genes, mostly combined with phylogroup A or B1, were detected frequently in all settings. In contrast, bla(CTX-M-15) genes common in human and animal populations were mainly combined with phylogroup A, but not with the more virulent phylogroup B2 with the exception of companion animals, where a few isolates were detectable. When E. coli subtype definition included ESBL types, phylogenetic grouping and antimicrobial susceptibility data, the proportion of isolates allocated to common clusters was markedly reduced. Nevertheless, relevant proportions of same subtypes were detected in isolates from the human and livestock and companion animal populations included in this study, suggesting exchange of bacteria or bacterial genes between these populations or a common reservoir. In addition, these results clearly showed that there is some similarity between ESBL genes, and bacterial properties in isolates from the different populations. Finally, our current approach provides good insight into common and population-specific clusters, which can be used as a basis for the selection of ESBL-producing isolates from interesting clusters for further detailed characterizations, e.g. by whole genome sequencing.