Antimicrobial Resistance in Salmonella Isolated from Food Workers and Chicken Products in Japan

Salmonella is an enteric bacterial pathogen that causes foodborne illness in humans. Third-generation cephalosporin (TGC) resistance in Salmonella remains a global concern. Food workers may represent a reservoir of Salmonella, thus potentially contaminating food products. Therefore, we aimed to investigate the prevalence of Salmonella in food workers and characterize the isolates by serotyping and antimicrobial susceptibility testing. Salmonella was isolated from 583 (0.079%) of 740,635 stool samples collected from food workers between January and December 2018, and then serotyped into 76 Salmonella enterica serovars and 22 untypeable Salmonella strains. High rates of antimicrobial resistance were observed for streptomycin (51.1%), tetracycline (33.1%), and kanamycin (18.4%). Although isolates were susceptible to ciprofloxacin, 12 (2.1%) strains (one S. Infantis, one S. Manhattan, two S. Bareilly, two S. Blockley, two S. Heidelberg, two S. Minnesota, one S. Goldcoast, and one untypeable Salmonella strain) were resistant to the TGC cefotaxime, all of which harbored β-lactamase genes (blaCMY-2, blaCTX-M-15, blaCTX-M-55, and blaTEM-52B). Moreover, 1.3% (4/309) of Salmonella strains (three S. Infantis and one S. Manhattan strains) isolated from chicken products were resistant to cefotaxime and harbored blaCMY-2 or blaTEM-52B. Thus, food workers may acquire TGC-resistant Salmonella after the ingestion of contaminated chicken products and further contaminate food products.


Introduction
Non-typhoidal Salmonella is an enteric bacterial pathogen that causes foodborne illnesses worldwide. Salmonellosis is caused by non-typhoidal Salmonella enterica serovars such as Salmonella enterica subsp. enterica serovar Enteritidis, Salmonella Typhimurium, and Salmonella Infantis. Invasive non-typhoidal Salmonella is estimated to cause 3.4 million cases of infections annually, worldwide [1]. Although non-typhoidal Salmonella infections typically cause acute self-limiting gastroenteritis, they may cause invasive bacteremia, with severe symptoms in certain cases. Antimicrobial therapy is typically not provided in many cases of salmonellosis. However, antimicrobial therapy may be lifesaving in patients with bacteremia or an extra-intestinal focal infection. Ampicillin, third-generation cephalosporins (TGCs), fluoroquinolones, and trimethoprim-sulfamethoxazole are commonly used to treat salmonellosis [2]. Therefore, antimicrobial-resistant Salmonella poses an important issue in the chemotherapy treatment of humans.
A primary route of Salmonella transmission involves ingestion of contaminated chicken meat and eggs [3]. Mori et al. [4] isolated Salmonella from 286 (55.9%) of 512 chicken meat samples collected in Japan between June 2015 and January 2016. In their study, the most frequently isolated serovar was S. Infantis, followed by S. Schwarzengrund, together accounting for 78.2% (243/311) of the isolates; 74.9% (182/243) of these two serovars are resistant to two or more antimicrobials, and seven S. Infantis strains were resistant to cefotaxime, a TGC. TGC-resistant Salmonella has additionally been isolated from fecal samples from healthy adults engaged in food handling work and diarrheic patients in Japan between 2012 and 2015 [5,6]. TGCs are classified as "critically important" by the World Health Organization [7]. In Japan, ceftiofur, a TGC, was approved in 1996 and was used for chemotherapy in cattle and pigs with bacterial infections. Moreover, the off-label use of ceftiofur in combination with in-ovo vaccination or vaccination of newly hatched chicks was performed in certain hatcheries until its use was abandoned in March 2012 [8]. The recent isolation of TGC-resistant Salmonella from chicken meat revealed TGC-resistant Salmonella may have survived in hatcheries and chicken farms after the withdrawal of ceftiofur in hatcheries. Food workers including food handlers who work in food and beverage companies that are infected with antimicrobial-resistant Salmonella after consuming or handling contaminated food may serve as reservoirs and pose a risk for food contamination. Therefore, in this study, we aimed to clarify the prevalence of Salmonella in food workers and characterize the isolates through serotyping and antimicrobial susceptibility testing. To estimate the origin of the isolates, we characterized Salmonella strains isolated from local chicken products by serotyping and antimicrobial susceptibility testing, and then compared the characteristics between human-and chicken-derived strains. Based on the findings of this study, the prevalence of Salmonella in stool samples of food workers was 0.079% between January and December 2018. Additionally, food workers may represent a reservoir of TGC-resistant Salmonella.
Antimicrobial susceptibility testing of chicken-derived strains showed that 174 (87.0%) of the 200 strains were resistant to at least one antimicrobial, and 141 (70.5%) strains were resistant to two or more antimicrobials. High antimicrobial resistance rates were observed for streptomycin (73.0%), tetracycline (67.5%), and kanamycin (53.5%). Four (1.3%) strains (three S. Infantis and one S. Manhattan strain) were resistant to cefotaxime. The three products contaminated with these cefotaxime-resistant S. Infantis strains were transported from the same chicken-processing plant on different days. The product contaminated with cefotaxime-resistant S. Manhattan was transferred from a different chicken processing plant. Among S. Schwarzengrund strains, the prevalence of strains resistant to streptomycin, kanamycin, and tetracycline was greatest (32.2%), sharing similarity with human-derived strains. Among S. Manhattan strains, the prevalence of strains resistant to streptomycin and tetracycline was greatest (50.0%), bearing similarity with human-derived strains. Among S. Infantis strains, the prevalence of strains resistant to streptomycin and tetracycline was greatest (36.7%), which differed from that in human-derived strains. All strains were susceptible to colistin and ciprofloxacin.

Serovar
Origin N

Discussion
The results of the present study showed that the prevalence of Salmonella in stool samples of food workers was 0.079% between January and December 2018, and the prevalence was greater from August to October than that from January to March. Outbreaks of Salmonella-associated food poisoning in Japan occur year-round, peaking from May to November [3]. Xu et al. [9] have suggested Salmonella may cause human infections in symptomatic and asymptomatic states, as Salmonella strains isolated from symptomatic and asymptomatic individuals are genetically and phenotypically indistinguishable. Kariuki et al. [10] reported an elevated level of relatedness between Salmonella strains isolated from symptomatic patients and asymptomatic carriers by phylogenetic analysis. Food workers evaluated in the present study may have been infected with Salmonella via ingestion of foods that can cause food poisoning, although they were asymptomatic when submitting their stool samples. Excluding S. Typhimurium monophasic variants, S. Cubana, and S. Newport, the top 14 serovars in human-derived strains, are known to be prevalent in Japanese broiler and laying hen flocks [11,12]. Notably, three serovars, S. Schwarzengrund, S. Infantis, and S. Manhattan, accounted for over 98% of chicken-derived strains in the present study. The most common antimicrobial resistance profiles of S. Schwarzengrund and S. Manhattan in human-derived strains were the same as those in chicken-derived strains. However, the primary antimicrobial resistance profiles of S. Infantis in human-and chicken-derived strains were different. S. Infantis strains susceptible to all antimicrobials tested represented the most common isolates among human-derived strains, whereas they represented the second-most common strains among chicken-derived strains. In our previous study [11], S. Infantis represented the second-most common serovar among Salmonella strains isolated from Japanese laying hen flocks, with 92.9% of S. Infantis strains susceptible to all antimicrobials tested. Therefore, the distribution of antimicrobial resistance profiles in S. Infantis isolated from human stool samples in the present study may be attributed to strains originating from Japanese broilers and laying hen flocks. Shimojima et al. [13] recently reported that Salmonella prevalence was greater in local chicken products (57.9%) than in imported products (8.5%). According to the Ministry of Agriculture, Forestry and Fisheries of Japan [14], local chicken and chicken egg products accounted for approximately 65% of total chicken meat distribution and 96% of total chicken egg distribution, respectively, in Japan in FY 2018. The results of this study suggest many food workers may be infected with Salmonella via ingestion of local chicken and egg products. Moreover, as S. Typhimurium, its monophasic variant, and S. Newport are prevalent in local cattle and pigs [15], beef and pork may represent sources of Salmonella infections among food workers.
Of the 583 human-derived Salmonella strains isolated in the present study, 12 (2.1%) were resistant to cefotaxime. Cefotaxime resistance was observed in seven serovars and one untypeable Salmonella strain. As the untypeable strain was assigned to sequence type 32, to which most S. Infantis strains belong, the strain may be a variant of S. Infantis. This assumption is based on multi-locus sequence typing (MLST) in accordance with public databases for molecular typing and microbial genome diversity (https://pubmlst.org, accessed on 23 November 2021). In the present study, cefotaximeresistant S. Infantis and S. Manhattan strains were obtained from local chicken products, and harbored the same bla genes (bla CMY-2 or bla TEM-52B ) as strains isolated from human stool samples. Shigemura et al. [16] reported that bla CMY-2 -harboring S. Infantis and bla TEM-52 -harboring S. Manhattan strains were isolated from local chicken products, and one bla CMY-2 -harboring S. Heidelberg strain was isolated from an imported chicken product. According to the Trade Statistics of Japan (https://www.customs.go.jp/toukei/info/ index_e.htm, accessed on 23 November 2021), chicken products imported from Brazil accounted for approximately 75% of imported chicken products in 2018. In Brazil, S. Minnesota, S. Infantis, and S. Heidelberg are prevalent in broiler flocks and chicken products, some of which show TGC resistance [17][18][19][20][21]. Shimojima et al. [13] reported the cefotaxime resistance rate in Salmonella isolated between 2009 and 2017 was lower in local chicken products (1.9%) than in imported chicken products (28.0%); additionally, seven S. Heidelberg and three S. Minnesota strains, isolated from imported chicken products, contain bla genes. Moreover, in their study, bla-harboring Salmonella strains were not isolated from beef and pork samples. Therefore, local and imported chicken products may represent major sources of TGC-resistant Salmonella that infect food workers.
Shigemura et al. [22] recently reported the prevalence of Salmonella in human stool samples obtained from food workers between January and October 2017 was 0.113% (164/145,220), and seven strains were resistant to TGCs. Of the seven TGC-resistant strains, one Salmonella Senftenberg and three Salmonella Haardt strains harbored bla CTX-M-14 and bla CTX-M-15 , respectively. The remaining three strains, namely Salmonella Agona, S. Infantis, and an untypeable Salmonella strain, harbored bla CMY-2 . Although their detection method differed from ours, their results share similarity to those reported in the present study, suggesting food workers may be infected with TGC-resistant Salmonella after the ingestion of contaminated chicken products and subsequently serve as reservoirs. Although the prevalence of TGC-resistant Salmonella in food workers may be low, they routinely serve food to consumers. Moreover, estimating the prevalence and characteristics of Salmonella in the Japanese population including food workers is possible. Therefore, monitoring the prevalence and characteristics of Salmonella in food workers is useful for risk management of Salmonella-associated food poisoning.

Sample Collection and Salmonella Isolation
A total of 740,635 stool samples (approximately 60,000 samples per month) were collected from food workers in all 47 prefectures in Japan between January and December 2018. The food workers included cooks and servers in restaurants and food factory workers. The individuals periodically (mostly monthly) collected their stool samples using transport swabs with a modified Cary-Blair transport medium (Eiken Chimical Co., Tokyo, Japan) and then submitted them to one of two laboratories (the Incorporated Foundation Tokyo Kenbikyo-in and the BML Food Science Solutions) during the study period. We were not provided with the precise number of individuals who submitted stool samples or the number of facilities involved in sampling. The individuals did not report symptoms of enteritis upon submission of stool samples. Each sample was tested in laboratories within 72 h upon arrival. A tip of each swab was streaked onto a modified Salmonella-Shigella agar plate (Eiken Chemical) and incubated at 37 • C for 24 h. Putative Salmonella colonies were biochemically identified as described previously [23]. One strain per sample was suspended in 20% glycerol and stored at −80 • C until ready for serotyping and antimicrobial susceptibility testing.

Local Chicken Products and Salmonella Isolation
A total of 235 local chicken products (liver, breast, thigh, neck skin, and minced meat) were collected from 25 retail stores and 6 chicken processing plants between January 2018 and October 2021. The products were transported under refrigeration to the National Institute of Health Sciences. Each sample was tested within 24 h after arrival at the laboratory. Twenty-five grams of the sample was mixed in 225 mL of buffered peptone water (Oxoid Ltd., Hampshire, UK) and incubated at 37 • C for 18 h for pre-enrichment. After incubation, 0.1 and 1 mL of the culture was added to 10 mL of Rappaport-Vassiliadis broth (Oxoid) and 10 mL of Hajna tetrathionate broth (Eiken Chemical), respectively, then incubated at 42 • C for 20 h. After incubation, each culture was streaked onto two selective isolation agar plates: xylose-lysine-deoxycholate agar (Oxoid) and CHROMagar TM Salmonella (CHRO-Magar, Paris, France) then incubated at 37 • C for 24 h. For the selective isolation from minced meat, Mannitol-Lysine-Crystal-Violet-Brilliant-Green agar (Nissui Pharmaceutical, Tokyo, Japan) was used in place of xylose-lysine-deoxycholate agar. Putative Salmonella colonies were biochemically identified as mentioned above. One strain per sample was suspended in 20% glycerol and stored at −80 • C until ready for serotyping and antimicrobial susceptibility testing.

Serotyping
Salmonella strains were tested for somatic antigens by slide agglutination using O antisera (Denka Co., Tokyo, Japan, and SSI Diagnostica, Copenhagen, Denmark). Salmonella strains were further tested for flagella antigens by tube agglutination using H antisera (Denka). Serovars were determined based on the reaction between O and H group antigens according to the Kauffmann-White scheme [24]. Strains agglutinated with anti-O4 and anti-H:i serum but not anti-H:1 or anti-H:2 serum were confirmed as monophasic variants of S. Typhimurium using a previously reported polymerase chain reaction method [25].

Antimicrobial Susceptibility Testing
The susceptibility of the Salmonella isolates to 11 antimicrobial agents was determined by antimicrobial susceptibility testing. Minimum inhibitory concentrations were determined by a two-fold broth microdilution method in 96-well microtiter plates (Dry-plate 'Eiken', Eiken Chemical), as described in the standards of the Clinical and Laboratory Standards Institute (CLSI) [26]. The following antimicrobial agents were tested: ampicillin (range of antimicrobial dilution: 1 to 128 mg/L), cefazolin (1 to 128 mg/L), cefotaxime (0.5 to 64 mg/L), streptomycin (1 to 128 mg/L), gentamicin (0.5 to 64 mg/L), kanamycin (1 to 128 mg/L), tetracycline (0.5 to 64 mg/L), nalidixic acid (1 to 128 mg/L), ciprofloxacin (0.03 to 4 mg/L), colistin (0.12 to 16 mg/L), and chloramphenicol (1 to 128 mg/L). The Escherichia coli reference strain ATCC 25922 was used as a control. The plates were incubated for 24 ± 2 h at 37 • C under aerobic conditions. The resistance breakpoints were defined based on the CLSI standard [27] and the Report on the Japanese Veterinary Antimicrobial Resistance Monitoring System 2016-2017 [28].

Determination of Antimicrobial Resistance Genes and Sequence Types Based on MLST in Cefotaxime-Resistant Salmonella Strains by WGS Analysis
DNA was extracted from cefotaxime-resistant strains using DNeasy ® UltraClean ® Microbial Kit (Qiagen GmbH, Hilden, Germany). WGS analysis was performed as previously described [29]. Sequencing libraries for each strain were prepared using QIAseq FX Library Kit (Qiagen) to obtain pair-end sequences (300 bp × 2) using the Illumina Miseq platform. Draft genome sequences were obtained by assembling the read sequences using A5 miseq [30]. The WGS data were analyzed using the Database of Pathogen Genomics and Epidemiology, GenEpid-J [31], and by using the ResFinder tool available from the Center for Genomic Epidemiology (http://www.genomicepidemiology.org/, accessed on 21 July 2021). MLST was performed using nucleotide sequences of seven housekeeping genes (aroC, dnaN, hemD, hisD, purE, sucA, and thrA) according to protocols available on the MLST database (https://pubmlst.org/organisms/salmonella-spp/, accessed on 23 November 2021).

Statistical Analysis
All statistical analyses were performed using R version 4.1. Confidence intervals were determined using the exact binomial test. Differences between proportions were tested using Fisher's exact test, where p-values of <0.05 were considered statistically significant.

Conclusions
Salmonella was isolated from 0.079% (583/740,635) of stool samples obtained from food workers in Japan. The strains were serotyped into 76 serovars and 22 untypeable Salmonella isolates. The top 14 serovars together accounted for over 70% of human-derived strains. A majority of the top 14 serovars in human-derived strains are prevalent in Japanese broiler and laying hen flocks. Of the 583 Salmonella strains, 12 (2.1%) were resistant to the TGC cefotaxime and harbored the same bla genes as those observed in local and imported chicken products. The characterization of Salmonella isolated from food workers showed that local and imported chicken products may represent the primary sources of TGCresistant Salmonella responsible for infection among food workers. Although the prevalence of TGC-resistant Salmonella in food workers is low, they may contaminate food served to consumers. Therefore, monitoring the prevalence and characteristics of Salmonella in food workers is necessary for the risk management of Salmonella-associated food poisoning.