SF6015 - Validation of a Method for Determining the Species of Origin of Contaminant E. coli

Researcher:

Dr. Carlton Gyles, Dept. of Pathobiology, University of Guelph

Objectives:

• To develop a procedure for rapidly determining the animal or human source of E. coli recovered as a contaminant from environmental samples, with special emphasis on water.

Expected Benefits:

• To enable the source of contamination (human, livestock, wildlife) of water or foods to be rapidly identified, thereby permitting corrective measures to be taken very quickly.

Summary of Research Results:

The project was a continuation of our previous project, which showed that the nucleic acid-based method Amplified Fragment Length Polymorphism (AFLP) was highly effective and was the best of four methods that had been applied to a limited number (105) of E. coli isolates.

The database of AFLP determinations on 105 E. coli isolates from the previous study was expanded to a total of 300 E. coli isolates, so that the final number of isolates from each of the categories (human, livestock, wildlife) was 100. The additional testing for purposes of enhancing the database involved E. coli cultured from feces of humans, sewage, livestock, and wildlife. Within the database, the AFLP results showed that 94% of the livestock isolates, 86% of the wildlife isolates, and 93% of the human isolates were correctly classified when the primer pair EcoRI-A/MseI-G was employed. The resolving power of primer pair EcoRI-C/MseI-CA was slightly lower than that of EcoRI-A/MseI-G.

The database was then used to determine the origins of E. coli cultured from various human, sewage, livestock, and wildlife sources. The evaluations were done in a blinded manner - the individual doing the analysis and assignment of E. coli to categories did not know the sources of the organisms. A total of 321 E. coli isolates were obtained, and AFLP data were produced for each of these E. coli isolates using the two primer sets. The results showed that about 70% of the human isolates, 56% of the livestock isolates, and 58% of the wildlife isolates were correctly classified when the primer pair EcoRI-A/ MseI-G was employed. Detailed analysis also showed that 74.7% of the human isolates (n=79) from individuals, 85% of the chicken isolates (n=27), 70% of the pig isolates (n=57), and 70.3% of the deer isolates (n=37) were correctly classified when the primer set EcoRI-A/MseI-G was used. The primer pair EcoRI-C /MseI-CA was not as informative as EcoRI-A /MseI-G and the data from 8 isolates were not suitable for analysis. However, when the non-database E. coli isolates were analyzed using the primer pair EcoRI-C /MseI-CA, 79.5% of the livestock isolates (n=146) were correctly classified, especially for chicken (100%, n=26) and turkey (89.7%, n=29) isolates.

A recently published PCR method that was reported to be effective in detecting porcine and bovine E. coli was evaluated. Although, our preliminary results showed that the specific method was unsatisfactory, this remains a very attractive approach.

It is concluded that the database of 300 isolates is inadequate for satisfactory differentiation of the origins of E. coli isolates. There are two alternatives for future developments - either the building of large databases or the development of a non-library-based method.