Development of Methods for Genetic Assessment of Antibiotic Resistance In Animal Herds

Abstract

Antibiotic drugs are important in treating bacterial infectious diseases in humans and animals. There are severe consequences when infectious bacteria become resistant to antibiotics such as treatment failure and even death. Since antibiotics were discovered, their use has been associated with a parallel selection for resistant bacteria. Since the hazards related to antibiotic resistance development have been recognized, the prudent use of antibiotics has been in focus, especially concerning their use in animal production. For many years antibiotics have been, and still are, recklessly used in the animal production especially in the form of growth promoters. Due to the associated risks of resistant zoonotic bacteria transmission from animals to humans, it is of interest to keep antibiotic use and antibiotic resistance under strict surveillance.This PhD study was based on the development of real-time PCR (qPCR) assays that supply an easy and rapid method for quantifying antibiotic resistance levels in animal herds. The pig production is accountable for a large portion of the antibiotics used for food producing animals in Denmark. Therefore, the antibiotic resistance genes included in this study had previously been described in association with pig herds, and they encoded resistance to antibiotics used in the Danish pig production.The first objective had emphasis on the qPCR assays’ design and development. The goal was to design 10-15 qPCR assays representing different antibiotic classes that ultimately would be tested in a swine herd. A total of 14 assays were developed, representing the following antibiotic classes: Tetracycline (tet(A), tet(B), tet(C), tet(M), tet(O), tet(W)), β-lactam (blaSHV family, blaCTX-M-1 group, blaCMY-2), sulphonamide (sulI, sulII), macrolide, lincosamide, and streptogramin B (ermB, ermF), and glycopeptide (vanA).The glycopeptide vanA gene was included as a follow-up to the avoparcin growth promoter ban implemented in Denmark in 1995. Besides the 14 antibiotic resistance gene qPCR assays, a 16S rDNA assay was also included.Manuscript I was an investigation of the affects PCR conditions had on the diversity and prevalence of antibiotic resistance genes detected in swine manure. This work was carried out in Dr. Zhongtang Yu’s laboratory at The Department of Animal Science, The Ohio State University, Columbus, Ohio. At this point of the first objective, decisions were being made concerning qPCR chemistry (probe vs. DNA-binding dye) and mastermix composition. In this study, three cycle numbers and 4 MgCl2 concentrations were evaluated for their effect on the diversity and prevalence of ribosomal protection proteins (RPPs) in a 3 x 4 factorial design. Significant differences in genetic diversity and prevalence of tet genes were found amongst the cycle number and MgCl2 combinations, and suggested that 35 PCR cycles and 7 mM MgCl2 enabled optimal detection of RPP genes in swine manure using the Ribo2_new_FW/Ribo2_RV primer pair. The results emphasized the importance of the PCR conditions when performing studies involving tet gene prevalence, and when results are interpreted.Upon completion of the qPCR assay development and optimization the project progressed to the second objective . The second objective was to establish if the qPCR assays could quantify antibiotic resistance genes in swine herds by comparing this principle to culture dependent antibiotic resistance detection. In order to do so, fecal samples in a swine herd were collected using different sampling methods that were also pooled at different levels. The antibiotic resistance levels were then determined both by the qPCR assays and coliform colony forming unit (CFU) estimates. Furthermore, the different sampling and pooling methods were evaluated. This established the qPCR assays’ capacity to quantify antibiotic resistance genes in a swine herd (Manuscript II) where significant differences in qPCR gene copy number estimates between pens were found (p<0.05 for ermB, tet(C), tet(W); p<0.0001 for ermF and tet(O)), while other antibiotic resistance genes were not detected in any samples during sampling 1 and sampling 2 (blaCTX-M-1 group, blaCMY-2, blaSHV family, and vanA). On the other hand, the coliform CFU counts differed significantly between pens for ampicillin and not tetracycline nor erythromycin. This emphasizes the major shortcomings of both methods, namely that the CFU counts only represent the coliform bacteria while the qPCR assays only detect the pre-defined genes. However, the qPCR gene copy numbers had lower relative standard deviations compared to the coliform CFU counts meaning that there is less variation in the qPCR gene copy estimates compared to the coliform CFU counts (Manuscript II). The variation in the coliform CFU counts consequently complicated the comparison of different sampling methods using coliform CFU counts (Manuscript II). In order to compare the qPCR principle of antibiotic resistance quantification, 20 of the individually sampled animals were randomly selected and analyzed by qPCR, coliform CFU counts, and colony hybridization using probes that correspond to the fragments amplified by the qPCR assays’ primers (Manuscript III). This study showed that it is important to define which bacterial population is relevant in achieving the specific goal of the antibiotic resistance quantification, and the method chosen for antibiotic resistance quantification has a large influence on the results obtained. This is exemplified by the higher level of tetracycline resistance observed in the coliform bacteria CFU counts compared to the colony hybridization (p<0.0001) (Manuscript III) despite the smaller bacterial fraction represented in the coliform estimates compared to the colony hybridization. This suggests that the majority of the tetracycline resistant bacteria were not detected by colony hybridization because they do not carry the specific genes that were used as probes.The first and second objectives established that the qPCR assays could be utilized in quantifying antibiotic resistance genes in total DNA extracted from swine feces. It was also confirmed that qPCR and culture dependent antibiotic resistance estimates represent two completely different populations, and cannot be compared directly. This was perceived by the lack of correlation between the total coliform CFU counts and the total number of bacteria in the population represented by the 16S copy number (R2=0.1) (Manuscript II), and by the differences in resistance estimates obtained by qPCR, coliform bacteria CFU counts, and colony hybridization (Manuscript III). Furthermore, pen floor sampling (pooled at stable level or not pooled), shoe cover samples (not pooled), and slurry tank samples were evaluated and are promising sampling methods when determining antibiotic resistance at herd level.The third objective involved the application of the qPCR assays in an animal population that was completely distant from the Danish pig production. Fecal samples from wildlife and Massai cattle in Tanzania were screened for the presence of the 14 antibiotic resistance genes using the qPCR assays. The wildlife and cattle samples were collected in the Ngorongoro Conservational Area (NCA) (wildlife and cattle interaction), and wildlife samples from the Mikumi National Park (MNP) (cattle are prohibited). Antibiotic resistant coliform bacteria estimates were also determined. This study constitutes Manuscript IV and the findings were surprising. The antibiotic resistance genes that were found in the cattle were also detected in the wildlife samples, regardless of the sampling site. Eight of the antibiotic resistance genes were detected in the samples, the most prevalent being tet(W) and blaCMY-2. Due to the nature of the blaCMY-2 antibiotic resistance spectrum, and the finding of this gene in 10 of 12 screened samples gives rise to concern. However, the finding of the blaCMY-2 gene in the wildlife further substantiates the qPCR assay as this gene was not detected in any of the pig samples collected and described in Manuscript II. Nevertheless, further studies should be conducted to study the antibiotic resistance gene pool among the wildlife in northern Tanzania.In conclusion, the 14 qPCR assays developed here successfully quantified antibiotic resistance in pig herds, where pen floor sampling (pooled at stable level or non-pooled), shoe cover sampling (non-pooled), and slurry tank sampling are promising sampling collection methods. To our knowledge, Manuscript II is the first study to describe sampling and pooling methods for qPCR quantification of antibiotic resistance genes in total DNA extracted from swine feces. The qPCR assays were also capable of detecting antibiotic resistance genes in Tanzanian wildlife and cattle samples representing a completely different population than the Danish pig production. Also, a gene not detected in the Danish pigs was detected in the Tanzanian wildlife and cattle samples further validating the qPCR assay. Generally, our results indicate that there is a large variation in the antibiotic gene abundance, regardless of animal species or sampling method. Furthermore, the present study highlights how different methods for antibiotic resistance detection (resistance coliform counts, colony hybridization on MacConkey and BA (anaerobic), and qPCR gene copy number estimates) reflect antibiotic resistance levels in different bacterial populations (Manuscript III). This emphasizes the importance of defining which bacterial population is relevant in the specific goal of antibiotic resistance quantification. When the aim is to monitor and quantify antibiotic resistance at herd level, using a method where few chosen indicator bacteria represent the resistance in the intestinal bacterial population let alone the herds´ is not optimal due to the large portion of neglected bacteria. When using qPCR for antibiotic resistance quantification the complete gene pool of the bacterial community is reflected. Besides being rapid and having minimal processing requirements, qPCR also enables precise measurement of small differences in DNA levels between samples, while detecting antibiotic resistance in viable and non-viable bacteria. However, qPCR only enables detection of the chosen genes, it is subject to inhibition, and results can be biased. It is therefore critical to assess qPCR assay parameters in order to establish the optimal conditions to accurately depict the antibiotic resistance levels. Additional studies evaluating the sampling methods in several animal herds should be tested in order to assist in understanding the antibiotic resistance gene variation. This study illustrates the immensity of the antibiotic resistance problem and the necessity for systematic surveillance of antibiotic consumption and resistance development at global, national, and local scales.