Increase in Resistance to Extended-Spectrum Cephalosporins in Salmonella Isolated from Retail Chicken Products in Japan

Extended-spectrum β-lactamase (ESBL)-producing Salmonella are one of the most important public health problems in developed countries. ESBL-producing Salmonella strains have been isolated from humans in Asian countries neighboring Japan, along with strains harboring the plasmid-mediated extended-spectrum cephalosporin (ESC)-resistance gene, ampC (pAmpC). However, only a few studies have investigated the prevalence of ESC-resistant Salmonella in chicken products in Japan, which are the main vehicle of Salmonella transmission. The aim of this study was to investigate the prevalence of ESBL-producing, pAmpC-harboring, or carbapenem-resistant Salmonella in chicken products in Japan. In total, 355 out of 779 (45.6%) chicken product samples collected from 1996–2010 contained Salmonella, resulting in 378 distinct isolates. Of these isolates, 373 were tested for resistance to ESCs, cephamycins, or carbapenems. Isolates that showed resistance to one or more of these antimicrobials were then examined by PCR and DNA sequence analysis for the presence of the bla CMY, bla CTX-M, bla TEM, and bla SHV resistance genes. Thirty-five resistant isolates were detected, including 26 isolates that contained pAmpC (bla CMY-2), and nine ESBL-producing isolates harboring bla CTX-M (n = 4, consisting of two bla CTX-M-2 and two bla CTX-M-15 genes), bla TEM (n = 4, consisting of one bla TEM-20 and three bla TEM-52 genes), and bla SHV (n = 1, bla SHV-12). All pAmpC-harboring and ESBL-producing Salmonella isolates were obtained from samples collected after 2005, and the percentage of resistant isolates increased significantly from 0% in 2004 to 27.9% in 2010 (P for trend = 0.006). This increase was caused in part by an increase in the number of Salmonella enterica subsp. enterica serovar Infantis strains harboring an approximately 280-kb plasmid containing bla CMY-2 in proximity to ISEcp1. The dissemination of ESC-resistant Salmonella containing plasmid-mediated bla CMY-2 in chicken products indicates the need for the development of continuous monitoring strategies in the interests of public health.


Introduction
Salmonella infection remains a significant public health concern worldwide. In many countries, the incidence of human infections caused by extended-spectrum cephalosporin (ESC)-resistant Salmonella has increased dramatically [1,2]. ESCs are administrated to treat salmonellosis caused by fluoroquinolone-resistant Salmonella, or in children and infants [3,4]. However, ESC-resistance genes such as bla CTX-M , bla TEM , and bla SHV , which contribute to extended-spectrum β-lactamase (ESBL) production, have been detected in Salmonella isolated from humans in Asian countries neighboring Japan [5,6]. Some of these isolates also contain plasmid-mediated ampC (pAmpC), another ESC-resistance gene [7]. In Japan, ESC-resistant Salmonella containing pAmpC or ESBL genes, such as bla CTX-M- 14 and bla CTX-M-15 , have been isolated from humans [8,9].
ESC-resistant Salmonella have also been isolated from food-producing animals and their products in many countries [10,11]. Reports from Japan over the last decade show that these isolates are especially prevalent in chickens [12]. Moreover, ESBL genes such as bla CMY , bla CTX-M , bla TEM , and bla SHV have been detected in Salmonella from food-producing-animal products in other Asian countries, and throughout the rest of the world [11,13]. In Japan, however, only a few studies have investigated the prevalence of ESC-resistant Salmonella in chicken products [14,15].
Because of the public health importance of these resistant salmonellae, it is imperative to determine their prevalence in chicken products in Japan. Therefore, the aim of this study was to investigate the prevalence of β-lactam antibiotic (cephalosporins or carbapenems) resistance in Salmonella from chicken products in Japan using molecular detection of ESBL genes and pAmpC.

Isolates and sampling
A total of 779 chicken product samples were collected, consisting of chicken meat (n = 683), giblets (n = 72), and processed chicken (tataki, thin slices of raw chicken with a minimally seared surface, n = 24). Of these samples, 622 samples originated from Japanese farms, and animals were slaughtered in Japan, 145 samples were from unrecorded origins, and 12 samples came from animals reared and slaughtered outside of Japan (11 samples were imported from Brazil, and one sample was from an unrecorded origin). All samples apart from two (described below) were collected in Fukuoka Prefecture, Japan, from 1996-2010. A total of 554 of the 779 samples were randomly collected from 282 retail outlets, including supermarkets and butcher shops, by Fukuoka prefectural food hygiene inspectors. A further 225 samples were randomly collected from 42 retail outlets by the authors from 2008-2010. Two of the 225 samples were collected in 2008 from two retail outlets in Nagasaki Prefecture, which neighbors Fukuoka Prefecture, while the remaining 223 samples were collected from 40 retail outlets in Fukuoka Prefecture, Japan.
The bacterial isolation method used for samples collected from 1996-1998 has been reported previously [16]. Isolations carried out from 1999-2010 were conducted in accordance with the method for isolating bacteria from meat samples described in our previous report [17], except for 85 samples collected in 2008. Salmonella isolation was carried out from these 85 samples using a previously described method [16] with a minor modification, as follows: Rappaport-Vassiliadis enrichment broth (Oxoid Ltd., Hampshire, UK) was used for the isolations, along with mannitol, lysine, crystal violet, brilliant green agar (Oxoid Ltd.) and Rimler-Shotts-Maeda agar (Kanto Chemical Co., Tokyo, Japan). The overall incidence of Salmonella isolation from the samples was 45.6% (355/779) (Table 1), with isolation rates of 45.5% (311/ 683), 54.2% (39/72), and 20.8% (5/24) from the chicken meat, giblets, and processed chicken samples, respectively. All salmonellae were isolated at the Fukuoka Institute of Health and Environmental Sciences, except for 52 isolates that were obtained from Nakamura Gakuen University Junior College in Fukuoka Prefecture. All the isolates were serotyped using somatic antisera and flagella antisera (Denka Seiken Co., Tokyo, Japan), as previously described [16].
A total of 378 distinct Salmonella isolates were collected from 355 of the 779 chicken product samples. Twenty-one and one samples yielded two and three distinct isolates, respectively (  Table 1). The remaining five isolates were not tested because of damage incurred during the storage process. The tested isolates were obtained from chicken meat (n = 310), giblets (n = 35), and processed chicken (n = 5) that had been collected from 195 different outlets. The susceptibility-tested Salmonella isolates constituted 16 serovars, along with 28 untypeable strains. The dominant serovars were Salmonella enterica subsp. enterica serovar (S.) Infantis (n = 180), S. Schwarzengrund (n = 70), and S. Manhattan (n = 52) ( Table 1). The incidences of Salmonella from the samples collected from 1996 to 1997, and those from 85 samples collected in 2008, have been reported previously [16,18].

DNA sequence analysis of resistance genes
Nucleotide sequences of the resistance genes were determined using the DNA sequencing primers listed in Table 2 [26,27]. PCR products were purified using Microcon filters (Millipore      Corporation, Bedford, MA) and were sequenced using a BigDye Terminator Cycle Sequencing Reaction v. 3.1 kit (Applied Biosystems, Carlsbad, CA) on a 3500 Genetic Analyzer (Applied Biosystems). The complementary sense and antisense sequences were aligned using the SeqMa-nII program within the Lasergene software package (DNASTAR, Madison, WI). The DNA sequences and deduced amino acid sequences were examined using the BLAST program at the DNA Data Bank of Japan (DDBJ) [28], and the "β-Lactamase Classification and Amino Acid Sequences for TEM, SHV, and OXA Extended-Spectrum and Inhibitor Resistant Enzymes" website [29].

Detection of ISEcp1 and other genes related to bla CMY-2
Genetic variation in the regions upstream and downstream of bla CMY-2 was identified by restriction fragment length polymorphism (RFLP) and sequence analysis. Twenty-five bla CMY-2harboring isolates (23 S. Infantis, one S. Manhattan, and one O-untypeable:r:1,5) were tested. A PCR fragment containing part of bla CMY-2 and its upstream region was generated by PCR using a forward primer located in ISEcp1 (ISEcp1-F: 5 0 -CTATCCGTACAAGGGAGTGT-3 0 ) [31] and a reverse primer located in bla CMY (cmy-R) ( Table 2). A PCR fragment containing part of the bla CMY-2 gene and its downstream region was generated with a forward primer located in bla CMY (cmy-F) ( Table 2) and a reverse primer located in sugE (SugE-R: 5 0 -ATTG-CAGGTTTGCTCGAAGT-3 0 ). PCR was conducted using a HotStarTaq Plus Master Mix Kit (Qiagen, Hilden, Germany), and products were digested with MseI (New England BioLabs, Ipswich, MA) at 37°C for 1 h. Digested fragments were separated by electrophoresis on 1% (w/v) agarose gels at 100 V for 30 min. Additionally, the PCR products of three S. Infantis isolates harboring bla CMY-2 (isolates 1993, 2127, and 2150) and one S. Manhattan isolate harboring bla CMY-2 (isolate 2179) were sequenced as described above. These three S. Infantis isolates were chosen for sequence analysis of the ISEcp1 region because they showed the dominant PFGE profiles (PFP) and LPP characteristics.

Localization of bla CMY-2 by Southern blot analysis
To determine whether bla CMY-2 was located in the chromosome or on a plasmid, S1 nuclease (New England BioLabs) and BlnI (Takara Bio, Otsu, Japan) digestions were prepared from the total DNA of the four isolates selected for ISEcp1 analysis (harboring bla CMY-2 ) (S. Infantis isolates 1993, 2127, and 2150, and S. Manhattan isolate 2179), as well as two susceptible isolates Total DNA was treated with 2 U/ml of S1 nuclease (incubated at 37°C for 45 min) or BlnI (incubated at 37°C for 16 h), followed by PFGE separation [30]. Separated fingerprints were transferred to positively-charged Nylon membranes (Roche Applied Science, Penzberg, Germany) and hybridized with PCR-generated bla CMY-2 digoxigenin-labeled probes (Roche Diagnostics, Basel, Switzerland) using hybridization solution (Roche Diagnostics) according to the manufacturer's instructions and a previous report [32].

Results
Antimicrobial susceptibility testing, PCR and sequence analysis of bla genes, and extended-spectrum β-lactamase phenotyping Of the 373 tested isolates, 35 isolates from 31 chicken meat and four giblet samples showed resistance to one or more antimicrobials. These 35 chicken product samples included 27 samples An analysis of the most common resistance genes harbored by each of the serotypes showed that S. Infantis with bla CMY-2 (n = 24) was the most common serotype/resistance gene combination. No other combination occurred in more than two isolates (Table 1, 3). Of the 26 isolates harboring bla CMY-2 , 24 showed resistance to cefoxitin and two showed intermediate sensitivity to cefoxitin (15-17 mm), based on the CLSI criteria [21]. All 26 isolates harboring bla CMY-2 were susceptible to other cephamycins (cefmetazole, cefotetan, and moxalactam). Furthermore, six isolates harboring only bla CMY-2 , which does not contribute to ESBL production, also showed ESBL phenotypes when tested with cefpodoxime-clavulanic acid (n = 5) or ceftazidime-clavulanic acid disks (n = 1).

Confirmation of ISEcp1 and other genes related to bla CMY-2
MseI-RFLP profiles of the regions upstream and downstream of bla CMY-2 were identical in all 25 isolates harboring bla CMY-2 . S. Infantis isolates 1993, 2127, and 2150, showing the dominant PFP C and LPP 1 characteristics, and S. Manhattan isolate 2179, the only S. Manhattan isolate harboring bla CMY-2 , were also used for sequencing of the bla CMY-2 upstream and downstream regions and for Southern blot analysis. A 2,389-bp fragment containing the bla CMY-2 gene from the selected isolates was sequenced. Results showed that these fragments from S. Infantis and S. Manhattan were identical. The fragments contained partial ISEcp1, bla CMY-2, blc, and partial sugE genes. The ISEcp1 sequences from the four selected isolates were 100% identical to the sequence reported for pNF4565 (GenBank accession AY581207; 2,389/2,389 bp, 100%), a plasmid carrying bla CMY-2 from S. Typhimurium (isolated from a human in 1999) [33].

Localization of bla CMY-2 by Southern blot analysis
All the S. Infantis isolates harboring bla CMY-2 carried plasmids of approximately 280 kb, as did the bla CMY-2 -negative S. Infantis isolate (isolate 1737, LPP 1) (Fig. 3). Southern blot analysis using a bla CMY-2 gene probe only produced a hybridization signal for the plasmids from the three S. Infantis isolates in which bla CMY-2 was detected by PCR (Fig. 3). As a result, these plasmids of approximately the same size were classified on the basis of the presence/absence of bla CMY-2 . Southern hybridization using a bla CMY-2 gene probe with BlnI-digested genomic DNA from these isolates showed hybridization signals at approximately 54 kb in the three isolates harboring bla CMY-2 .
One S. Manhattan isolate that contained bla CMY-2 showed two hybridization signals (approximately 310 kb and 60 kb) (Fig. 4). The 310-kb hybridization signal appeared to be hybridization background; however, the point was confirmed as a small peak of density using densitometric analysis (FPQuest Software, Bio-Rad Laboratories) (Fig. 4). Therefore, the S. Manhattan isolate was confirmed to harbor two bla CMY-2 genes in putative plasmids of 310-kb and 60-kb in size. No hybridization signal was observed for the bla CMY-2 -negative S. Manhattan isolate. Southern blot analysis of the S. Manhattan isolate was carried out on two separate occasions, with the same result obtained each time.
Typhimurium, >100 kb) [33], pCVM29188 (S. Kentucky, 101,461 bp) [48], and pUMNK88 161 (E. coli, 161,081 bp) [49]. These studies reported that the mobile element ISEcp1 in the plasmids led to acquisition of various resistance genes in addition to bla CMY-2 . As S. Infantis plasmids from both the resistant and susceptible isolates (Fig. 2) were indistinguishable in size, the transfer of bla CMY-2 between these plasmids should be studied.
The present study suggests some flaws in common screening approaches for ESBL detection. We identified an isolate carrying bla TEM-20 that was only resistant to cefpodoxime and cefepime. Such a strain could escape detection by traditional ESBL tests, which use only cefotaxime and ceftazidime [1,10,50]. Furthermore, this isolate showed intermediate sensitivity (20 mm) to ceftiofur (30 μg) (data not shown), which supports previous findings by Rodriguez et al. [11]. Therefore, routine ESBL screening may be improved by the use of cefpodoxime, cefepime, and/or ceftiofur.
In conclusion, we found that S. Infantis isolates harboring plasmid-borne bla CMY-2 were prevalent in chicken products in Japan after 2005. None of the studied isolates showed resistance to carbapenem antibiotics; however, the dissemination of ESC-resistant Salmonella containing plasmid-mediated bla CMY-2 demands the development of continuous monitoring strategies in the interests of public health.