Antimicrobial Resistance Determinants Circulating among Thermophilic Campylobacter Isolates Recovered from Broilers in Ireland Over a One-Year Period

Campylobacteriosis is the leading cause of human bacterial gastroenteritis, very often associated with poultry consumption. Thermophilic Campylobacter (Campylobacter jejuni and Campylobacter coli) isolates (n = 158) recovered from broiler neck skin and caecal contents in Ireland over a one-year period, resistant to at least one of three clinically relevant antimicrobial classes, were screened for resistance determinants. All ciprofloxacin-resistant isolates (n = 99) harboured the C257T nucleotide mutation (conferring the Thr-86-Ile substitution) in conjunction with other synonymous and nonsynonymous mutations, which may have epidemiological value. The A2075G nucleotide mutation and amino acid substitutions in L4 and L22 were detected in all erythromycin-resistant isolates (n = 5). The tetO gene was detected in 100% (n = 119) of tetracycline-resistant isolates and three of which were found to harbour the mosaic tetracycline resistance gene tetO/32/O. Two streptomycin-resistant C. jejuni isolates (isolated from the same flock) harboured ant(6)-Ib, located in a multidrug resistance genomic island, containing aminoglycoside, streptothricin (satA) and tetracycline resistance genes (truncated tetO and mosaic tetO/32/O). The ant(6)-Ie gene was identified in two streptomycin-resistant C. coli isolates. This study highlights the widespread acquisition of antimicrobial resistance determinants among chicken-associated Campylobacter isolates, through horizontal gene transfer or clonal expansion of resistant lineages. The stability of such resistance determinants is compounded by the fluidity of mobile genetic element.


Introduction
Campylobacter is the most commonly reported foodborne bacterial pathogen causing human gastroenteritis in the European Union (EU) and Ireland, most often associated with the broiler reservoir [1]. Ireland was found to have a 98% prevalence of campylobacter-contaminated broiler carcasses in 2008 [2]. Frequent isolation of antimicrobial-resistant Campylobacter spp. of food animal origin continues to limit the spectrum of clinically useful antimicrobials and is internationally recognised as a major societal challenge. Veterinary antimicrobials used therapeutically and prophylactically are often the same as, or belong to the same class as those used clinically [3].
Macrolides, fluoroquinolones and aminoglycosides are classified as critically important antimicrobials, while tetracycline is considered a highly important antimicrobial [3]. Resistance to (fluoro)quinolones and tetracyclines is highly prevalent in clinical and broiler-associated Campylobacter spp. isolates, while resistance to erythromycin is typically low to moderate across Europe [4][5][6]. Macrolides are the first line antibiotic for the treatment of enteric gastroenteritis, while fluoroquinolones and tetracyclines remain as alternatives [7][8][9]. Systemic infections are routinely treated with aminoglycosides [9,10] and resistance to aminoglycosides is low in clinical and broiler isolates across Europe [5]. Cross-resistance between aminoglycoside antimicrobials is incomplete and although streptomycin is not used clinically to treat campylobacteriosis, resistance can be used as an indicator for acquired aminoglycoside resistance genes.
In Gram negative bacteria, DNA gyrase is the primary target of fluoroquinolones [11]. DNA gyrase is a heterotetrameric type IIA topoisomerase, consisting of two polypeptide subunits (GyrA and GyrB, encoded by gyrA and gyrB, respectively), catalysing ATP-dependent negative supercoiling of DNA to regulate replication, repair and gene expression [12][13][14]. Resistance to (fluoro)quinolones among Campylobacter spp. is largely mediated by chromosomal mutations in the quinolone resistance-determining region (QRDR) of gyrA, typically conferred by the C257T nucleotide mutation (Thr-86-Ile) [15]. The QRDR is located near the Tyr-125 active site, involved in DNA-protein bridge formation during DNA strand passage [15].
Macrolides act by binding to the 50S bacterial ribosomal subunit and inhibit translational elongation, and interfere with protein synthesis and subsequent ribosomal subunit assembly [16,17]. Polymorphisms in the 23S ribosomal RNA (rRNA), mutations in 50S ribosomal proteins L4 and L22 (encoded by rplD and rplV, respectively) or the presence of the emerging ermB gene contribute to macrolide resistance [8,18]. Beta-hairpin extensions from 50S ribosomal proteins L4 and L22 are involved in the regulation of nascent peptide exit from the large ribosomal subunit [19,20]. Mutations in L4 and L22, combined with the overexpression of antimicrobial efflux genes have been reported to contribute to high-level macrolide resistance [21]. The ribosomal methylase encoded by ermB was reported recently for the first time in thermophilic Campylobacter spp., located on a chromosomal multidrug resistance genomic island (MDRGI) (likely originating from a Gram positive species) in a high-level erythromycin-resistant Campylobacter coli (C. coli) isolate (ZTC113) of swine origin in China [8,22]. ErmB dimethylates adenine at position 2074 of the 23S rRNA gene, reducing the binding affinity of macrolides [23].
Tetracycline resistance in Campylobacter spp. is largely conferred by a ribosomal protection protein (RPP), TetO, capable of displacing tetracycline from its primary binding site on the 30S ribosomal subunit [17,31]. Bacterial resistance to tetracycline is also associated with ATP-dependent efflux or enzymatic inactivation of tetracycline [32][33][34]. Campylobacter tetO can be located chromosomally but is often plasmid-mediated [31,33,[35][36][37][38][39]. Tetracycline RPPs are widely distributed among bacterial genera and it has been reported that the tetO gene exists in at least eleven bacterial genera, including four Gram negative and seven Gram positive genera [33]. The conserved acquisition of tetO between members of different bacterial genera indicates that conjugative plasmids, transposons, or recombination events contribute to the dissemination and maintenance of the tetO gene [37].
Aminoglycosides are broad spectrum antimicrobials and inhibit protein synthesis by binding to 16S rRNA of the 30S ribosome [44,45]. Campylobacter spp. resistance to aminoglycosides is mediated by reduced antimicrobial binding affinity for target sites due to enzymatic modification, via acetylation, phosphorylation or adenylation of amino or hydroxyl groups of the aminocyclitol nucleus or sugar moieties [44][45][46]. Although there are two main nomenclature systems used to identify aminoglycoside modifying enzymes [47][48][49], we followed the system proposed by Shaw et al. (1993), later extended to include an expanded panel of aminoglycoside 6-nucleotidyltransferases (also known as adenyltransferases) [50,51]. The designation proposed by Shaw et al. (1993) is as follows: the type of modification (nucleotidyltransferase/adenyltransferases (ANT)); the modification site (6'); a roman numeral to denote unique resistance profiles (I), and a letter to represent unique protein sequences (b) [48]. Genes for ANT enzymes are found on transposons, plasmids or chromosome, often in associated with other resistance genes and very often as part of the transposon-associated aminoglycoside-streptothricin resistance gene cluster (ant(6)-I-sat4-aphA3), first isolated from Staphylococci [52,53]. ANT(6)-I encoding genes are widely distributed among clinical and animal streptomycin-resistant thermophilic Campylobacter spp. isolates [51].

Fluoroquinolone Resistance
Isolates resistant to ciprofloxacin/nalidixic acid were screened for mutations in the gyrA gene and 100% of isolates (n = 99) harboured a C257T point mutation, which is the dominant mutation conferring resistance among campylobacters. Resistant isolates were grouped into C. jejuni gyrA and C. coli gyrA (arbitrarily named GTJs and GTCs, respectively) sequence types based on the presence of synonymous and nonsynonymous mutations present in the portion of the gyrA gene sequenced (Table 1). Ciprofloxacin-resistant C. jejuni isolates (n = 85) were grouped into three GTJs (GTJ-I, -II, and -III). A large proportion (47.1%) carried the Thr-86-Ile substitution exclusively (GTJ1). Synonymous mutations T72C, C243T, T357C, C360T, C471T, T483C, and C622T were exclusively associated with C. jejuni and were present in both GTJ-II and GTJ-III. Nonsynonymous Ser-22-Gly (A64G) and Ala-206-Thr (G616A) mutations were present in 35.3% and 17.7% of isolates of ciprofloxacin-resistant C. jejuni isolates, respectively and were the basis of defining GTJ-II and GTJ-III, respectively. Both GTJ-II and GTJ-III were associated with the Asn-203-Ser (A608G) substitution. All CIP-resistant C. coli isolates tested (n = 14) harboured one nonsynonymous mutation only (Thr-86-Ile), but were grouped into seven GTCs based on the presence of various synonymous mutations (Table 1). Table 1. GyrA sequence types (GTs) and associated polymorphisms distributed among 85 ciprofloxacin-resistant C. jejuni isolates (GTJs) and 14 ciprofloxacin-resistant C. coli isolates (GTCs). Polymorphisms causing an amino acid substitution (nonsynonymous mutations) are highlighted in black. .

Overall Distribution of Antimicrobial Resistance
The antimicrobial resistance rates of this pool of broiler-associated thermophilic Campylobacter spp. isolates (resistant to at least one antimicrobial; n = 158) have been detailed previously [67], and are summarised below (Figure 2).

Discussion
This study reports the antimicrobial resistance determinants circulating among Irish broilerassociated Campylobacter isolates, collected throughout the Republic of Ireland, from the three largest poultry processors, over a one-year period (2017-2018).

Discussion
This study reports the antimicrobial resistance determinants circulating among Irish broiler-associated Campylobacter isolates, collected throughout the Republic of Ireland, from the three largest poultry processors, over a one-year period (2017-2018).
C. jejuni isolates harbouring multiple amino acid substitutions in the gyrA QRDR (GTJ-II and GTJ-III) ( Table 1) [80]. Moreover, the Ser-22-Gly, Asn-203-Ser and Ala-206-Thr mutations have been reported in fluoroquinolone-sensitive strains [81][82][83]. To confirm the apparent lack of involvement of these accessory gyrA mutations in the development of fluoroquinolone resistance, the introduction of these mutations in fluoroquinolone-susceptible strains could be investigated. Authors have previously reported that double mutations in gyrA (at amino acid positions 86 and 90) were necessary to produce high level moxifloxacin resistance [84,85]. Moxifloxacin is a potent fluoroquinolone with activity against fluoroquinolone-resistant campylobacters that harbour a single mutation in gyrA [86]. In this study, no high-level resistance to moxifloxacin was observed among the (fluoro)quinolone-resistant isolates. These data indicate that mutations outside the gyrA QRDR have a negligible effect on (fluoro)quinolone resistance.
Variation in gyrA alleles within a population of Campylobacter isolates have been identified as epidemiological markers and may serve as a supplementary approach to classical epidemiological typing methods [81,83,87,88]. In our study, the GTs detected were species specific, although Ragimbeau et al. (2014) reported the presence of a typical C. coli gyrA type in 0.23% (n = 1) of 430 C. jejuni isolates tested, and 1.4% (n = 4) C. coli isolates harboured a typical C. jejuni gyrA type. The amino acid substitutions present in each of the three GTJ (Table 1) lineages have been associated with poultry Campylobacter isolates previously [89][90][91]. Only one nonsynonymous mutation (Thr-86-Ile) was detected among the seven GTCs detected (Table 1), similar to previous studies reporting a single nonsynonymous mutation (Thr-86-Ile substitution) present in (fluoro)quinolone-resistant C. coli QRDR sequences [83,89,90]. It is likely that additional variants of Campylobacter spp. gyrA alleles exist, and may reflect ecological evolution [83].
Identical mutations in the 23S rRNA, rplD, and rplV genes were detected in erythromycin-resistant isolates (n = 5), while MICs ranged from 128 mg/L to ≥ 128 mg/L. Three of these isolates were collected from the same flock in north-central Ireland while one isolate was collected from a farm approximately 10 km away, the following week. The fifth erythromycin-resistant isolate was recovered from a farm in the mid-south-west of Ireland, two months previously, but all birds from these farms were processed in the same processing plant.
The A2075G point mutation in the 23S rRNA gene remains the most prevalent genetic event conferring macrolide resistance [8,92] and was detected in all erythromycin-resistant isolates in this study. The 23S rRNA A2074G and A2074C mutations were not detected. Mutations in the rplV gene, encoding the 50S ribosomal protein L22 were detected in the C-terminal region (amino acids 109-142) [18], including Thr-109-Ala, Ala-111-Glu, Ala-114-Thr, and Pro-120-Thr. Nonsynonymous mutations (T282A and C294T) were also observed in the region encoding the highly conserved β-hairpin loop at amino acids 78-98 [18]. Mutations in the RplD β-hairpin (spanning amino acid positions 55-77 [18,93]) are often associated with bacterial macrolide resistance, and such mutations were not observed among the Irish erythromycin-resistant isolates tested in this study. Moreover, polymorphisms were detected outside the cmeR regulatory IR. The effects of mutations detected in this study in rplD, rplV, and the intergenic region of cmeR-cmeA are unknown, but may contribute to erythromycin resistance.
The ermB gene was not detected among the erythromycin-resistant isolates tested. Resistance mediated by ermB in Campylobacter spp. is largely confined to China, which may reflect the extensive use of antimicrobials in food producing animals in China [8]. Three reports of genetically distinct ermB-positive C. coli isolates recovered from poultry exist in Europe [94][95][96] and an ermB-positive isolate was detected for the first time in the United States in a C. jejuni isolate of clinical origin [97], while the ermB gene was recently detected in 18.3% of 240 thermophilic Campylobacter spp. retail meat associated-isolates in South Africa [98]. Mutations in C. jejuni 23S rRNA has been associated with a fitness cost and reduced doubling times [99][100][101], although tolerance to low temperatures may facilitate persistence in the environment and transmission of resistant strains through the food supply [100].
All tetracycline-resistant isolates harboured a portion of the tetO gene, while three isolates harboured the mosaic tetO/32/O type II gene were detected. Mosaic tetracycline resistance genes in Campylobacter spp. are typically derived from tetO and tet32 sequences in the type II conformation, with a shorter central tet32 segment [34], although there are limited reports of these resistance genes circulating among Campylobacter spp. The first mosaic tetracycline RPP gene was detected in Megasphera elsdenii, composed of a central tetW region flanked by two tetO regions [102]. However, the progenitors of these mosaic genes are based only on the order in which they were discovered, and the current classification system does not adequately reflect the variable nature of tetracycline RPPs [34,43]. The tetO primers [103] used in this study amplify a region at the beginning of the tetO gene and enable the detection of mosaic tet genes with a central portion flanked by an initial tetO region until position 228. The tetO/32/O type II gene reported here was associated exclusively with an MIC of 64 µg/L. However, 27 of 119 (22.7%) tetracycline-resistant Campylobacter spp. isolates tested had MICs of 64 µg/L or ≥ 64 µg/L, indicating that other factors contribute to enhanced tetracycline resistance. It should be noted that in this study, all three isolates harbouring mosaic tetracycline genes were also co-resistant to streptomycin, enabling co-selection and persistence of these antimicrobials. The burden of mosaic tetracycline resistance genes within the genus should be considered as part of the approach to elucidate developing and newly acquired antimicrobial resistance determinants within the genus.
Genomic sequencing of tetracycline-/streptomycin-resistant C. jejuni isolates (CITCj625-18 and CITCj727-18, isolated from the same flock, first and final thin) revealed identical genes in a multidrug resistance genomic island (Table 3, Figure 1A). Both isolates harboured a truncated tetO gene and a mosaic tetO/32/O type II gene, homologous that of Gram positive (GenBank accession number: KY994102.1) and Campylobacter spp. (GenBank accession number: WP_002823161.1). The presence of multiple tet genes (coding for similar or different mechanisms) in Gram negative isolates has also been documented [33]. However, the truncated form detected in this study (CITCj625-18 and CITCj727-18) is likely a remnant of a region of insertion or recombination. Truncated tetO genes have also been reported in C. coli MDGRI containing aadE (ant(6)-Ib) and ermB [94,111]. Aminoglycoside (ant(6)-Ib (867 bp)) and streptothricin resistance (satA) genes were also located within the multidrug resistance island ( Figure 1A). Streptothricin acetyltransferase A (satA) is frequently reported in Gram positive bacilli [112] and shares less than 40% identity with the streptothricin acetyltransferase A (sat4) reported in Campylobacter [52,111,113]. Plasmid replication proteins within the MDRGI suggest a plasmid as the insertion vehicle of the resistance genes.
CITCc3448-18 belonged to ST-828 clonal complex (ST-1096). ST-1096 has been isolated from a case of gastroenteritis in the United Kingdom (UK) in 2016 [114] and has also been previously reported in C. coli of swine origin in Spain, America, and Grenada [51,115,116]. CITCc1631 was ST-6543 (ST-828 clonal complex), which has been associated with clinical and chicken-associated isolates [117]. There are a total of fifteen depositions (all are UK associated) of C. coli ST-6543 on the PubMLST Campylobacter database at the time of writing [114], including eleven clinical isolates (faeces), two chicken-associated isolates, and two isolates with no source allocation.

Bacterial Isolate Culture Conditions and Susceptibility Testing
A total of 350 thermophilic Campylobacter isolates (296 C. jejuni and 54 C. coli) were recovered from free range and intensively-reared broiler carcasses (neck skin and caecal contents) using ISO 10272-2:2017 [2]. Isolates were collected between September 2017 and September 2018, from the three largest poultry processing plants in the Republic of Ireland. The collection of isolates was speciated using matrix-assisted laser desorption ionisation-time of flight (MALDI-TOF) mass spectrometry (MS) (Bruker, Billerica, MA, United States). Isolates were previously tested [67] for their MIC to six clinically relevant antimicrobials, namely ciprofloxacin, nalidixic acid, erythromycin, tetracycline, gentamicin, and streptomycin according to ISO 20776:2006 and EC Decision 2013/652/EU [118,119]. Overall, 158 (140 C. jejuni and 18 C. coli) isolates tested were resistant to at least one antimicrobial and were subsequently tested for resistance determinants.

Genotypic Characterisation of Antimicrobial Resistance-PCR Amplification and Sequencing
Primer sets, target genes and annealing temperatures are listed in Table 4. Primers to detect mosaic tetracycline resistance genes (tetO/32/O and tetO/W/O) were designed on SnapGene2.3.2 software (from Insightful Science; available at snapgene.com) and regions of primer complementarity were assessed on Primer-BLAST [120].
PCRs were performed with 2.5 U of Amplitaq polymerase (Applied Biosystems, Foster City, CA, USA), 1 × PCR buffer I and 2.5 mM magnesium chloride (Applied Biosystems), 0.2 mM of each deoxyribonucleotide (Sigma Aldrich, St Louis, MO, USA), 0.2 pmol/µL of each primer, and 1 µL (1-2 ng/µL) of DNA. Reaction conditions were denaturation at 94 • C for 2 min, 35 cycles of denaturation at 94 • C for 30 s, annealing (Table 4) for 30 s and extension at 72 • C for 1 min followed by a final extension at 72 • C for 5 min. PCR products were purified using the High Pure PCR Purification Kit (Sigma Aldrich). Purified PCR products were Sanger sequenced (forward and reverse reads) ( Table 4) by Eurofins Genomics (Eurofins Genomics, Ebersberg, Germany). Consensus sequences were aligned and assembled on SeqMan Pro (Lasergene) (DNAStar, Madison, WI, USA). Ciprofloxacin-resistant isolates (n = 99) were screened for mutations in the QRDR of the gyrA gene [15]. Products were purified and sequenced, as described above. Consensus sequences were aligned to the gyrA of the reference C. jejuni (GenBank accession number: L04566.1 and AL111168.1) and C. coli sequences (GenBank accession number: AF092101.1 and NZ_UIGM01000003.1) on SnapGene 2.3.2.
Tetracycline-resistant isolates (n = 119) were screened for the presence of tetO, according to Aminov et al. (2001) and products were visualised on 2% agarose gel electrophoresis. The tetO amplicon of a tetracycline-resistant C. jejuni isolate (CITCj382-18) from this study was purified and sequenced (as described above) as a positive control. Consensus sequences were aligned to the C. jejuni tetO gene (GenBank accession number: M18896.2).

Moxifloxacin Minimum Inhibitory Concentration Testing
All (fluoro)quinolone-resistant isolates (n = 99) were tested for moxifloxacin susceptibility. Briefly, isolates were recovered from −80 • C on CBA (Fannin Ltd) for 24 h at 42 • C under microaerobic conditions, and subcultured to CBA for 20 ± 2 h at 42 • C under microaerobic conditions. In microtiter plates, 100 µL serial dilutions of moxifloxacin (Sigma Aldrich) were prepared in of Mueller Hinton broth with lysed horse blood (Thermo Fisher Scientific, Waltham, MA, USA) ranging from 0.125-16 mg/L. A 0.5 McFarland inoculum was prepared in 5 mL demineralised water (Thermo Fisher Scientific) and 100 µL was transferred to 11 mL of Mueller Hinton broth with lysed horse blood (Thermo Fisher Scientific). Moxifloxacin serial dilutions were inoculated with 100 µL of cell suspension and incubated for 20 ± 2 hat 42 • C under microaerobic conditions.

Genome Sequencing and Genomic Analysis
The genomes of four streptomycin-resistant isolates (C. jejuni isolates CITCj625-18 and CITCj727-18 and C. coli isolates CITCc1631-18 and CITCc3448-18) were sequenced. DNA was quantified in triplicate with the Quant-iT dsDNA Assay Kit (Thermo Fisher Scientific). Genomic DNA libraries were prepared using the Nextera-XT protocol (Illumina, San Diego, CA, USA), with changes including 2 ng of input DNA and a minute PCR elongation time. DNA quantification and library preparation were performed on a Hamilton Microlab STAR system. Pooled libraries were quantified using the Kapa Biosystems library quantification kit on a Roche light cycler 96 qPCR machine. Libraries were sequenced on the Illumina HiSeq using a 250 bp paired-end protocol. Reads were adapter trimmed using Trimmomatic 0.30 with a sliding window quality cut-off of Q15 [124]. De novo assembly was performed using SPAdes version 3.7 [125] and assembly quality was assessed using QUAST [126].

Conclusions
Although non-poultry sources contribute to campylobacteriosis incidence, poultry are natural thermophilic Campylobacter spp. hosts. The broiler industry serves as a reservoir for the dissemination of resistant campylobacters. The enrichment and stability of Campylobacter spp. resistance determinants is noteworthy but the natural competence and potential of recombination or acquisition of mobile genetic elements contributes to the Campylobacter. Taken together, the data collected in this study point to the diversity of resistance determinants circulating among Irish broilers, contributing to the development of resistance to clinically relevant antimicrobials.