Antibiotic resistance in the lower intestinal microbiota of dairy cattle: longitudinal analysis of phenotypic and genotypic resistance.

University of Minnesota Ph.D. dissertation. February 2012. Major: Environmental Health. Advisor: Dr. Randall Singer. 1 computer file (PDF); viii, 148 pages.This research focused on methods of measuring antibiotic resistance and analysis of antibiotic resistance data in dairy cattle that were sampled repeatedly over time. Specific objectives included: characterization and longitudinal analysis of phenotypic antibiotic resistance of commensal Escherichia coli, development of a statistical model for the analysis of low quantity resistance genes measured by quantitative real-time polymerase chain reaction (qPCR), measurement of antibiotic resistance genes in the lower intestinal bacterial communities of dairy cattle that received a short-term therapeutic dose of antibiotic and untreated cattle, and measurement and longitudinal analysis of the quantities of six antibiotic resistance genes in the lower intestinal bacterial communities of dairy cattle. Enteric E. coli collected from dairy cattle over 1.5 years were tested for phenotypic resistance to 17 antimicrobials. A total of 93 phenotypic patterns were observed among 3,402 isolates tested, with a majority (67%) susceptible to all 17 antimicrobials. The most prevalent resistances were to tetracycline, sulfamethoxazole, and streptomycin. Latent class and latent transition analyses were carried out to group the animals into classes according to their resistance phenotypes and to estimate the probabilities of transitioning into and out of classes over time. Probabilities of transitioning to a pan-susceptible class were high, as were the probabilities of remaining in the pan-susceptible class. Probabilities of transitioning form a pan-susceptible class to a resistant class were very low. Measurement of antibiotic resistance genes by qPCR presents challenges for genes that are present in very low quantities. A statistical model was developed to analyze qPCR data made up of a significant proportion of observations that fall below the limit of quantification of a qPCR assay. Computer simulations showed that the statistical model produced less biased estimates of regression parameters than common methods of handling low quantity qPCR data. qPCR was applied to a cohort of dairy cattle that received a five day course of ceftiofur and matched untreated cattle. Quantities of a gene (blaCMY-2) that confers resistance to ceftiofur were measured and analyzed using the statistical model developed for low quantity genes. Treated animals had significantly higher quantities of blaCMY-2 during treatment than untreated animals. By the first day post-treatment, gene quantities had returned to pre-treatment levels. The quantities of six different antibiotic resistance genes were measured by qPCR in the fecal community bacterial DNA of a cattle population that was sampled repeatedly over 2.5 years. Significantly increasing trends over time were observed for three of the six genes conferring resistance to tetracyclines, macrolides, and cephalosporins. Comparison of phenotypic resistance and qPCR data showed that qPCR performed on community DNA is a more sensitive method of detection that phenotypic testing of cultured isolates