Lack of Evidence for erm (B) Infiltration Into Erythromycin-Resistant Campylobacter coli and Campylobacter jejuni from Commercial Turkey Production in Eastern North Carolina: A Major Turkey-Growing Region in the United States

Introduction

Erythromycin and other macrolides constitute drugs of choice for treatment of human campylobacteriosis, and macrolide resistance determinants in Campylobacter are, therefore, of major public health significance. Macrolide resistance is markedly more common in Campylobacter coli than in Campylobacter jejuni, and has been largely mediated by substitutions in the 23S rRNA, especially A2075G (Bolinger and Kathariou, 2017). Until 2014, evidence was lacking for involvement of erm (erythromycin ribosome methylation) determinants in macrolide resistance in Campylobacter, although such determinants, especially erm(B), are widely disseminated through horizontal gene transfer among bacteria. In 2014, however, erm(B) was shown to mediate high-level macrolide resistance in Campylobacter coli ZC113, a swine-derived isolate from China (Qin et al., 2014). Subsequent analyses placed the first erm(B)-positive Campylobacter isolates in 1994 (Zhou et al., 2016). Analysis of multiple isolates, primarily C. coli, from humans and food animals in China revealed that erm(B) is localized on multidrug resistance genomic islands (MDRGIs) (Qin et al., 2014; Wang et al., 2014; Zhou et al., 2016; Bolinger and Kathariou, 2017).

Reports of erm(B)-harboring Campylobacter from humans or the food supply outside of China have remained scarce with only four such erm(B)-harboring isolates reported to date. Three were C. coli isolated in Spain from poultry, including two from turkeys in 2014 (Florez-Cuadrado et al., 2016, 2017), whereas the first erm(B)-harboring C. jejuni strain in the United States was recently identified through whole genome sequencing (WGS) of human clinical isolates collected through the National Antimicrobial Resistance Surveillance system. This strain was isolated in 2016 from a patient who had recently returned from international travel, and remains the only erm(B)-harboring Campylobacter strain reported in the United States (Chen et al., 2018). Previously, WGS of macrolide-resistant C. coli and C. jejuni from humans, retail meat, and poultry intestinal samples in the United States revealed that all harbored substitutions in the 23S rRNA (mostly A2075G), whereas erm(B) was not detected (Zhao et al., 2016; Whitehouse et al., 2018).

Turkeys in the United States are frequently colonized by multidrug-resistant Campylobacter strains, with macrolide resistance being highly prevalent in C. coli (Luangtongkum et al., 2006; Bolinger, 2017). The objective of this study was to determine whether erm(B) has infiltrated turkey-derived Campylobacter spp. in eastern North Carolina, one of the major turkey-growing regions in the United States. We hypothesized that this would be detected by targeted PCR screening of erythromycin-resistant isolates using primers derived from erm(B) sequences conserved in C. coli, C. jejuni, and Gram-positive bacteria (Wang et al., 2014), similarly to the approach used to detect erm(B)-harboring isolates from turkeys in Spain (Florez-Cuadrado et al., 2017).

Materials and Methods

The panel of 174 C. coli and 4 C. jejuni isolates was chosen based on their previously determined resistance to erythromycin, with erythromycin minimum inhibitory concentration (MIC) breakpoint for resistance set at ≥8 μg/mL, as described (Luangtongkum et al., 2006). Erythromycin MIC was typically >128 μg/mL. The erythromycin-resistant isolates were collected between July 2013 and June 2016 for a larger project assessing antimicrobial resistance and risk factors for Campylobacter colonization of turkeys grown conventionally in North Carolina (Bolinger, 2017). Most isolates (n = 159) were from ceca which were kindly provided throughout the study period by a vertically integrated commercial turkey company, from flocks grown on different farms in eastern North Carolina. The remaining isolates were derived from flies in different turkey houses (n = 14) and from turkey feces (n = 5) in the same region (Bolinger, 2017). The preponderance of C. coli in the panel reflects the fact that although turkeys in North Carolina are extensively colonized with both C. coli and C. jejuni, erythromycin resistance is primarily encountered in C. coli (Bolinger, 2017). Erythromycin-resistant C. jejuni was detected only in four farms, and one isolate from each farm was included in the panel. Multilocus sequence typing revealed several sequence types (STs) among C. coli (STs 1067, 1101, 8086, 8212, and 8551) and two (ST-1839, ST-7729) among the four erythromycin-resistant C. jejuni (Bolinger, 2017).

PCR for erm(B) used primers ermB-F: 5′-GGGCATTTAACGACGAAACTGG-3′ and ermB-R: 5′-CTGTGGTATGGCGGGTAAGT-3′ (Wang et al., 2014). All isolates were additionally tested using PCR for the aminoglycoside-resistance cluster (ARC) from the type I MDRGI with primers cluster-F: 5′-GGATGGATTCCTATGAAAACAT-3′ and cluster-R: 5′-GGCTTTGTTCATCTTCATACTCT-3′ (Qin et al., 2012). Genomic DNAs of Campylobacter coli ZC113 and Campylobacter coli NADC2A were included each time as positive controls for erm(B) and the ARC, respectively.

The 23S rRNA gene region typically associated with erythromycin resistance in Campylobacter was amplified from each strain and sequenced using Sanger sequencing. Sequenced amplicons from the erythromycin-resistant C. coli strain 14983A (Miller et al., 2016) and the erythromycin-sensitive C. coli type strain ATCC 33559 were used as positive and negative controls, respectively. All sequences were assembled in SeqMan (Lasergene v. 8.0; DNASTAR, Madison, WI).

Results and Discussion

All 178 erythromycin-resistant Campylobacter isolates were PCR-negative for erm(B), as well as for the ARC from the type I MDRGI. On the other hand, analysis of the 23S rRNA gene sequences of a subset of isolates (45 C. coli, 4 C. jejuni) selected to include different farms and STs revealed that all harbored the A2075G transition in the 23S rRNA typically associated with macrolide resistance in Campylobacter (Bolinger and Kathariou, 2017). Thus, our findings suggest that, as of 2016, erm(B) had not yet infiltrated the genomes of the macrolide-resistant C. coli or C. jejuni from conventionally grown turkeys in this region.

As erm(B) has already been identified in Campylobacter from food production animals in at least two locations, China and Spain, we consider it likely that it will be eventually encountered in Campylobacter from other regions as well, including the United States. Although not detected in our survey, erm(B) may emerge in the U.S. turkey production in response to yet unidentified selective pressures. Our investigation focused on Campylobacter from turkeys grown in eastern North Carolina, one of the leading turkey-growing regions in the United States, and needs to be complemented by similar investigations from other major turkey-producing regions. Nonetheless, our findings and those from other surveys (Zhao et al., 2016; Whitehouse et al., 2018) suggest a relatively slow pace of acquisition and spread of the erm(B) determinant in Campylobacter in the United States, where, with the sole exception of the one clinical C. jejuni strain already discussed (Chen et al., 2018), macrolide resistance continues to involve 23S rRNA substitutions. The current findings will contribute to worldwide efforts to monitor macrolide resistance and the presence of erm(B) in Campylobacter from food animal production and other sources.