Antimicrobial Resistance in Campylobacter jejuni Isolated from Brazilian Poultry Slaughterhouses

Campylobacteriosis is one of the most common foodborne diseases in the world. It is considered the most frequently reported foodborne illness in the European Union (EU) and one of the most important in the United States (US) (EFSA & ECDC, 2018; CDC, 2019a; WHO, 2019). Poultry is known to be the major reservoir and an important source for pathogen transmission to humans (Kaakoush et al., 2015). Campylobacteriosis is most often associated with the consumption of raw and undercooked poultry or the cross-contamination of other foods by these items (CDC, 2019a). Although Brazil is a leading supplier of the world’s poultry meat (ABPA, 2018), Brazil’s official data does not report Campylobacter infections. Resistance in foodborne pathogens presents the potential for their transmission to humans through the food chain (Wang et al., 2013). Campylobacteriosis is generally a self-limiting disease. However, in some patients, Campylobacter infection can result in a systemic disease requiring the use of antimicrobials (CDC, 2019b). Erythromycin is considered the first-line treatment, but fluoroquinolones are also frequently used due to their broad-spectrum activity against enteric pathogens (Engberg et al., 2001). Recently, however, multidrugresistant Campylobacter strains have been detected in poultry and several other sources around the world (Szczepanska et al., 2017; Du et al., 2018; Montgomery et al., 2018). In the EU, Campylobacter isolated from human and poultry sources have shown high to extremely high resistance to ciprofloxacin and tetracycline (EFSA & ECDC 2018), and both substances have been widely used in Brazilian poultry production in recent decades (Machinski Júnior et al., 2005). Ciprofloxacin resistance in Campylobacter strains is usually related to the Tre-86-Ile mutation in the quinolone resistancedetermining region (QRDR) of the gyrA gene, which results in the replacement of the amino acid threonine by isoleucine (Frasao et al., 2015a). Resistance to tetracycline is usually related to the presence of the tetO gene (Pratt & Korolik, 2005). Our aim was to assess the minimum inhibitory concentrations (MICs) for Campylobacter jejuni strains and determine their molecular resistance profiles to tetracycline and ciprofloxacin.


INTRODUCTION
Campylobacteriosis is one of the most common foodborne diseases in the world. It is considered the most frequently reported foodborne illness in the European Union (EU) and one of the most important in the United States (US) (EFSA & ECDC, 2018;CDC, 2019a;WHO, 2019). Poultry is known to be the major reservoir and an important source for pathogen transmission to humans (Kaakoush et al., 2015). Campylobacteriosis is most often associated with the consumption of raw and undercooked poultry or the cross-contamination of other foods by these items (CDC, 2019a). Although Brazil is a leading supplier of the world's poultry meat (ABPA, 2018), Brazil's official data does not report Campylobacter infections.
Resistance in foodborne pathogens presents the potential for their transmission to humans through the food chain (Wang et al., 2013). Campylobacteriosis is generally a self-limiting disease. However, in some patients, Campylobacter infection can result in a systemic disease requiring the use of antimicrobials (CDC, 2019b). Erythromycin is considered the first-line treatment, but fluoroquinolones are also frequently used due to their broad-spectrum activity against enteric pathogens (Engberg et al., 2001). Recently, however, multidrugresistant Campylobacter strains have been detected in poultry and several other sources around the world (Szczepanska et al., 2017;Du et al., 2018;Montgomery et al., 2018).
In the EU, Campylobacter isolated from human and poultry sources have shown high to extremely high resistance to ciprofloxacin and tetracycline (EFSA & ECDC 2018), and both substances have been widely used in Brazilian poultry production in recent decades (Machinski Júnior et al., 2005). Ciprofloxacin resistance in Campylobacter strains is usually related to the Tre-86-Ile mutation in the quinolone resistancedetermining region (QRDR) of the gyrA gene, which results in the replacement of the amino acid threonine by isoleucine (Frasao et al., 2015a). Resistance to tetracycline is usually related to the presence of the tetO gene (Pratt & Korolik, 2005). Our aim was to assess the minimum inhibitory concentrations (MICs) for Campylobacter jejuni strains and determine their molecular resistance profiles to tetracycline and ciprofloxacin.

Bacterial strains and growth conditions
A total of 54 C. jejuni strains were selected for this study (Table 1). Isolates were obtained from broiler carcass samples collected between 2011 and 2012 from different Brazilian poultry slaughterhouses according to criteria described by the International Organization for Standardization (ISO 10272-1:2017 Phenotypic characterization of antimicrobial resistance As described by the Clinical and Laboratory Standards Institute (CLSI) (CLSI, 2013b), a broth microdilution test was performed to determine the MIC for six clinically relevant antibiotics (Sigma-Aldrich): chloramphenicol (0.25-128 mg/L), ciprofloxacin (0.007-16 mg/L), erythromycin (0.064-128 mg/L), gentamicin (0.064-32 mg/L), nalidixic acid (1-256 mg/L), and tetracycline (0.064-64 mg/L). The strains were classified as susceptible or non-susceptible (including intermediate strains) according to the breakpoints described in the CLSI standards (CLSI, 2013a;El-Adawy et al., 2015). The strains were also classified as wild type or nonwild type (nWT) based on their epidemiological MIC cut-off (ECOFFs), which were determined according to the EUCAST guidelines available at the time of data analysis (January, 2019) (EUCAST, 2019). A C. jejuni reference strain (ATCC 33560) was selected to ensure the validity of the tests. Strains that were resistant to three or more classes of antimicrobials were classified as multidrug resistant (MDR) strains (Schwarz et al., 2010). The multiple antibiotic resistance (MAR) index was determined as previously described (Krumperman, 1983).

Molecular characterization of antimicrobial resistance
Thermal extraction of DNA was performed as described (Sambrook & Russel, 2012). The strains with tetracycline MICs ≥2 mg/L were selected for PCR detection of the tetO gene. The primers were designed by Bacon et al. (2000). All PCR reactions (25 μL) contained 10X PCR buffer, 2.5 mM dNTPs, 10 pmol primer, 2 mM MgCl 2 , 5 U Taq DNA polymerase, and 2 μL template DNA. The cycling program consisted of 30 cycles of 94 °C for 30 s, 54 °C for 30 s, and 72 °C for 1 min. The amplified products (559 bp) were separated by electrophoresis in a 1% agarose gel stained with ethidium bromide, which was photographed under UV illumination. A total of 31 strains with ciprofloxacin MICs ≥4 mg/L were selected to characterize their molecular resistance. First, the QRDR in the gyrA gene was detected by PCR with primers designed by Price et al. (2005). All PCR reactions (25 μL) contained 10X PCR buffer, 1 mM dNTPs, 10 pmol primer, 2 mM MgCl 2 , 1 U Taq DNA polymerase, and 5 μL template DNA. The cycling program consisted of 35 cycles of 94 °C for 30 s, 55 °C for 30 s, and 72 °C for 1 min. The amplified products (454 bp) were separated by electrophoresis in a 1% agarose gel, stained with ethidium bromide, and photographed under UV illumination. All reactions were repeated three times. A PCR control containing the PCR mixture without the addition of the template DNA was included with all reactions.
The amplified products of gyrA were then sequenced in triplicate in an automated sequencer (ABI-PRISM 3500® Genetic Analyzer; Applied Biosystems) with 50 cm capillaries and polymer POP7 (Applied Biosystems). The sequences obtained in the chromatograms were processed using Chromas Lite (Technelysium) and Geneious (Biomatters) software. To confirm the presence of the mutation, the sequence of strain C. jejuni (L04566.1), obtained from GenBank, was used as a ciprofloxacin-sensitive strain standard.

Statistical analysis
The data were subjected to a descriptive statistical analysis using PASW Statistics software. The kappa index (Landis & Koch, 1977) was determined to evaluate the concordance between the classifications based on the CLSI breakpoints and ECOFF values.

RESULTS
The phenotypic antimicrobial resistance profiles and MIC results are described in Tables 1 and 2. Only one strain was susceptible to all substances and all strains were clinically susceptible to gentamicin and chloramphenicol, regardless of the breakpoint (CLSI or EUCAST) evaluated. Resistance for tetracycline and erythromycin was higher when EUCAST parameters were applied. 46.3% (25/54) of the strains were classified as non-susceptible and 51.8% (28/54) as nWT for tetracycline, and 42.6% (23/54) of the strains were classified as non-susceptible and 48.1% (26/54) as nWT for erythromycin. Resistance to ciprofloxacin was equal for both parameters, and 94.4% (51/54) of the strains were classified as non-susceptible or nWT. Regarding resistance for nalidixic acid, 94.4% (51/54) of the strains were non-susceptible according to the CLSI breakpoints and 83.3% (45/54) were nWT according to EUCAST breakpoint. CLSI also classifies the strains as "intermediate", which were considered as non-susceptible in the present study (Borges et al., 2019) due to their uncertain therapeutic effect in vivo (CLSI, 2013b).

DISCUSSION
Antimicrobial resistance is a complex challenge and a major problem for global public health. Each year, about 25,000 patients in the EU and 23,000 in the US die from infections caused by multiresistant bacteria, with annual treatment costs of approximately 1.5 billion euros and 20 billion dollars, respectively (WHO, 2014). The Brazilian government does not have an integrated program for monitoring antimicrobial resistance in the primary human and production animal pathogens such as Salmonella spp. and C. jejuni, making the adoption of new measures to control and restrict the use of antimicrobials difficult (Borges et al., 2019). In addition, unlike European countries, Brazil has no specific legislation mandating the analysis of campylobacteriosis. Therefore, studies addressing antimicrobial resistance are essential for characterizing the circulating C. jejuni strains in the Brazilian poultry production chain.
Although similar, the results based on the ECOFF values showed a great number of nWT strains (nonsusceptible). Considering that MIC determinations depend on breakpoints and that MIC results affect clini-cal decisions and official data reports (Kassim et al., 2016), changes in the breakpoint parameters can result in significant changes in the final MIC. The breakpoints are set by three international agencies: the U.S. Food and Drug Administration Center for Drug Evaluation and Research (USDA-CDER), the CLSI, and the European Committee on Antimicrobial Susceptibility Testing (EUCAST) (Humphries et al., 2019). The guidelines proposed by CLSI are some of the most used worldwide and are based on the drugs' properties and mechanisms of resistance (Kassim et al., 2016), whereas EUCAST bases its breakpoints on the drugs' properties and ECOFFs (EUCAST, 2019). We compared the results for both interpretative criteria through kappa analysis. It showed perfect agreement for ciprofloxacin, gentamicin and chloramphenicol, almost perfect agreement for tetracycline (κ = 0.889) and erythromycin (κ = 0.888) and fair agreement for nalidixic acid (κ = 0.400). Comparisons among studies is challenging due to the wide variation in interpretative techniques and parameters. The agreement seen between the EUCAST and CLSI guidelines not only provides important information about antimicrobial susceptibility, it indicates that international data on C. jejuni resistance could be compared, thus allowing the establishment of more specific control measures for this species in the poultry production chain.
The use of chloramphenicol in production animals has been banned in Brazil since 2003(MAPA, 2003 and the use of gentamicin in poultry production is restricted (Giacomelli et al., 2014). These are probably the reasons for the absence of non-susceptible strains in our study.
Our results indicate that almost 50% of the strains were resistant to erythromycin, which is higher than the previously published data (Bolinger & Kathariou, 2017;Szczepanska et al., 2017). These results are a public health concern, because this agent is the treatment of choice for Campylobacteriosis (Engberg et al., 2001). These high rates may be associated with the wide use of this drug in animal production up until 2012, when erythromycin was banned as a food additive in Brazil (MAPA, 2012). Higher erythromycin resistance rates can also be related to the several mechanisms by which Campylobacter acquires resistance to these antimicrobial agents (Bolinger & Kathariou, 2017). We also observed a high level of resistance to tetracycline. Tetracycline resistance in Campylobacter has been previously reported in strains isolated from animal products (Abdi-Hachesoo et al., 2014;Giacomelli et al., 2014;Hungaro et al., 2015;Sierra-Arguello et al., 2015). Over the past decade, the tetracycline compound class has been used in farm animal production as a growth promoter and for the treatment of diseases. The high resistance levels suggest that the overuse of tetracycline may have selected resistant strains. The majority of tetracycline resistance determinants confer increased resistance to the other compounds from the same class, though it is also possible that the use of oxitetracycline and doxycycline has also caused tetracycline resistance (Fairchild et al., 2005). A high level of tetracycline resistance in Campylobacter is usually associated with the presence of the tetO gene. This gene encodes the TetO protein, which protects ribosomes from the inhibitory effects

Antimicrobial Resistance in Campylobacter jejuni Isolated from Brazilian Poultry Slaughterhouses
of tetracycline (Connel et al., 2003). A total of 28 isolates had tetracycline MICs ≥2 mg/L, and 42.8% of them carried the gene. Reports from Brazil have shown lower frequencies of this gene than in other countries (Sierra-Arguello et al., 2015;Du et al., 2018). This gene can be present in conjugative plasmids containing resistance genes for other antimicrobials that continue to undergo selective pressure. The tetO gene can also be found as a chromosomal element. In this case, the stability of the chromosomal location ensures the gene replicates from generation to generation, even in the absence of the drug (Avrain et al., 2004;Crespo et al., 2012).
Fluoroquinolones are considered the second-line treatment against Campylobacter infection in humans (Engberg et al., 2001). Campylobacter strains showed high resistance to fluoroquinolones, with the CLSI breakpoints and ECOFF values indicating that 90.7% and 81.5% of the strains, respectively, were resistant to both ciprofloxacin and nalidixic acid. These high fluoroquinolone resistance rates have been previously described in Brazilian poultry sources (Sierra-Arguello et al., 2016) and are likely due to the large use of this antimicrobial class in poultry production (Iovine & Blaser, 2004). Although ciprofloxacin is more commonly used in humans, it is structurally related to enrofloxacin (WHO, 2011), which has been widely used in poultry production (Garcia-Migura et al., 2014), and previous studies have demonstrated that resistance for ciprofloxacin and enrofloxacin is developed through the same mechanisms (Frasao et al., 2015b). Campylobacter resistance to fluoroquinolones is usually related to a mutation in the QRDR region of the gyrA gene (Frasao et al., 2015b). This gene codes for the 'A' subunit of the enzyme DNA gyrase and confers a decreased susceptibility to fluoroquinolones (Wilson et al., 2000). All strains tested for the presence of mutations in the QRDR fragment of the gyrA gene showed a threonine to isoleucine (Thr-86-Ile) mutation at codon 86 (Table 1). This result confirms that this substitution is always related to high fluoroquinolone MICs. A second mutation at codon 149 (Val-149-Ile) was observed in 19.3% of the strains. As these amino acids belong to the same class, the replacement may not lead to any significant conformational modifications of the protein. Consequently, its function probably remains unmodified (Taylor, 1986). Other mutations associated with an intermediate level of resistance to quinolones such as Asp-90-Asn and Ala-70-Thr (Iovine, 2013) were not encountered in this study. Mutation in QRDR of gene gyrA is the main resistance mechanism to fluoroquinolones. However, chromosomal efflux pumps, especially those codified by cmeA, cmeB and cmeC genes, are important factors to antimicrobial in Campylobacter species (Wieczorek & Osek, 2013;Nascimento et al., 2019). Previous studies have demonstrated that almost all strains of Campylobacter jejuni isolated in Brazil presented these three genes .
Since 2000, several Latin American countries are part of the Pan American Health Organization Network for Monitoring/Surveillance of Antibiotic Resistance. However, very few of them are conducting surveillance for Campylobacter species. In this context,  (Fernández & Pérez-Pérez, 2016). Available data shows that antimicrobial resistance in Campylobacter jejuni varies among Latin American countries. The higher rates are described for fluorquinolones in several countries besides Brazil, including Ecuador, Argentina and Peru (Pollett et al., 2012;Zbrum et al., 2015;Fernández & Pérez-Pérez, 2016;Vinueza-Burgos et al., 2017). Resistance to aminoglycosides is usually lower among these countries (Zbrum et al., 2015;Vinueza-Burgos et al., 2017;Toledo et al., 2018). Resistance rates for erythromycin and tetracycline is variable according to the country (Pollett et al., 2012;Zbrum et al., 2015;Vinueza-Burgos et al., 2017).
The individual maximum and minimum multipleantibiotic resistance (MAR) indices for the isolates were 0.7 and 0.2, respectively, with an average index of 0.5, regardless of the interpretative criteria used. According to Proroga et al. (2011), the MAR index is a good risk assessment tool and can be applied to differentiate low-(MAR < 0.2) and high-risk (MAR > 0.2) regions where antibiotics are overused. Our results (overall MAR = 0.5) may indicate high antibiotic usage and high selective pressure in the poultry chain, but the practical significance of this finding may be lost, because antibiotic use is widespread in developing countries, including Brazil (Davis & Brown, 2016).
Based on the CLSI results, 13% (7/54) of the strains were multidrug-resistant (MDR), whereas 16.7% (9/54) of the strains were classified as MDR using the ECOFF values. Emerging resistance to the antimicrobial agents of choice for treating Campylobacter infections is becoming a serious threat to public health (Du et al., 2018). The frequency of MDR strains found in this study is similar to that in previous reports from Brazil (Sierra-Arguello et al., 2015). Given such results, Brazilian authorities should consider establishing an integrated surveillance network for antibiotic resistance in Campylobacter.
Considering that poultry is the major source of human Campylobacter spp. infection and that antimicrobial-resistant strains can be easily transmitted to humans via the food chain, our results show the need for the implementation of prudent antimicrobialuse policies in the Brazilian food production chain.

DISCLOSURE STATEMENT
No potential conflict of interest was reported by the authors.