A comparison of AFLP and ERIC-PCR analyses for discriminating Escherichia coli from cattle, pig and human sources.

Amplified fragment length polymorphism (AFLP) and enterobacterial repetitive intergenic consensus polymerase chain reaction (ERIC-PCR) genomic fingerprinting assays were compared for their ability to differentiate Escherichia coli isolates obtained from various host sources, and with respect to their pathogenicity. One hundred and ten verotoxigenic, enterotoxigenic and non-pathogenic E. coli isolates obtained from cattle, humans and pigs were used in this study. The AFLP assay was shown to be highly effective in predicting both the host source and pathogenicity of the E. coli isolates. A stepwise discriminant function analysis showed that 91.4, 90.6 and 97.7% of the human, bovine and pig isolates were classified into the correct host types, respectively. The analysis also distinguished the non-pathogenic E. coli from the verocytotoxigenic and enterotoxigenic virulence phenotypes at 100, 100 and 90.9% accuracy, respectively. Sixty-two E. coli strains from the collection were subjected to the ERIC-PCR fingerprinting analysis. Using this method, only 28.6, 0 and 75.0% of the human, bovine and pig isolates were classified into the correct host types, respectively. Overall, the AFLP method was able to ascribe host source with a high level of confidence and readily discriminate pathogenic from non-clinical isolates of E. coli.