Genotypic characterization and antimicrobial resistance profile of Salmonella isolated from chicken, pork and the environment at abattoirs and supermarkets in Chongqing, China

Background Salmonella is one of the most important foodborne pathogens, causing outbreaks of human salmonellosis worldwide. Owing to large scales of consumption markets, pork and poultry that contaminated by Salmonella could pose a tremendous threat to public health. The aim of this study was to investigate the contamination of Salmonella from chicken, pork and the environment in slaughtering and retail processes in Chongqing, China. Results A total of 115 Salmonella isolates were recovered from 1112 samples collected from pork, chicken and the environment. Compared with the isolation rate of samples from chicken (9.50%) and the environment (6.23%), samples from pork had a significant higher isolation rate (44.00%). The isolation rates in slaughterhouses (10.76%) and in supermarkets (10.07%) showed no statistical difference. Thirty different serotypes were identified among all the isolates. S. Derby (n = 26), S. London (n = 16) and S. Rissen (n = 12) were the dominant serotypes. Antimicrobial susceptibility testing revealed that 73.04% isolates were resistant to tetracycline, followed by 66.96% to ampicillin and 59.13% to doxycycline. More than half (50.43%) of the isolates were multidrug resistant (MDR), and most of the MDR isolates were from supermarkets. Multilocus sequence typing results showed 24 out of 115 isolates were ST40, which was the most prevalent. Furthermore, isolates from supermarkets had 20 different sequence types while isolates from slaughterhouses only had 8 different sequence types. Conclusion Our study highlighted that Salmonella was more frequently isolated in pork production chain than that in chicken. Compared with isolates from slaughterhouses, isolates from supermarkets had more MDR profiles and represented a wider range of serotypes and sequence types, indicating that the retail process had more diverse sources of Salmonella contamination than that of slaughtering process.


Background
Salmonella, a foodborne pathogen, causes diarrhoeal diseases even death in both humans and animals [1]; it can survive in a dry environment for several weeks or even in water for several months [2]. According to previous surveys, the aetiological agent of salmonellosis largely attributed to contaminated food, which mostly were poultry and pork [3,4]. In China, pork is the mainstream of meat consumption. Meanwhile, the consumption of poultry is rising year by year. Contamination by Salmonella in slaughtering and retail processes of chicken and pork could be a potential pathway to threat public health.
Antibiotics are widely used to improve human and animal health, and also are commonly incorporated into animal feed to improve growth rate and feed efficiency in many countries [5,6] A previous report showed that China approximately consumed more than 162,000 tons of antibiotics annually, and husbandry industry account for 52.00% of the total consumption. Within the husbandry industry consumed antibiotics, 52.20% antibiotics were in pork production and 19.60% were in chicken production. As for the categories of antibiotic, sulfonamides, tetracyclines, fluoroquinolones, macrolides, β-lactams, and other antibiotics shared 5.00, 7.00, 17.00, 26.00, 21.00, and 24.00% of the total usage [7]. Long-term exposed to antibiotics has led selection pressure to environmental bacterium, which causes antimicrobial resistance and even multidrug resistance (MDR). Antimicrobial resistance in Salmonella is a global issue. Large amount of Salmonellae resistant to extendedspectrum-β-lactams (ESBLs) and fluoroquinolones, which are important in treating salmonellosis [8]. Drug-resistant Salmonella, especially multidrug-resistant Salmonella, has been a menace to food safety and human health.
Typing methods used to investigate the characterization of Salmonella can help to enrich our knowledge of its regularity of dissemination. Serotyping presents a well-established methodology for typing of Salmonella [9]. To date, approximately 2600 serotypes have been discovered. The traditional method for serotyping, the Kauffmann-White-Le Minor Scheme requires a series of antisera, consuming time and money. Hence, a variety of typing methods were established to study the molecular epidemiologic characterization of Salmonella with its transmission dynamics, including pulsed-field gel electrophoresis (PFGE), restriction fragment length polymorphism (RFLP), amplified fragment length polymorphism (AFLP), whole-genome sequencing (WGS), and multilocus sequence typing (MLST) [10][11][12][13]. Compared with other methods, MLST is a highly repeatable typing method that based on sequence analysis of selected housekeeping genes. Recently, approximately 224,516 Salmonella strains has been uploaded by users in the MLST database [14], which becomes a convenient tool for researchers.
Contamination and antimicrobial resistance of Salmonella isolated from food-producing animals is severe worldwide [15], particularly in China [16][17][18]. Furthermore, several studies reported that Salmonella isolates could be recovered from farms, slaughterhouses and retail markets [19,20] and a previous study indicated that Salmonella isolates could transmitted from slaughterhouses to retail markets in pig production chain [17]. However, few studies focused on the comparison of Salmonella contamination inpork and chicken as well as their slaughtering and retail chains. Therefore, the intention of this study was to compare the antimicrobial resistance, and genetic relationship of Salmonella isolates recovered from the environment, chicken and pork at abattoirs and supermarkets located in Chongqing, China.

Isolation and serotyping of Salmonella from samples
A total of 115 Salmonella isolates were recovered from 1112 samples collected from slaughterhouses and supermarkets, the isolation rates form pork, chicken and the environment were 44.00, 9.50 and 6.23%, respectively. Samples from pork had significant higher isolation rates than that from chicken and the environment in both slaughterhouses and supermarkets (Table 1). Within the different sources of environmental samples, Salmonellae were isolated only from floor, knives and tables at slaughterhouses, but isolated from all the sources of environmental samples at supermarkets, especially chopping boards and ice ( Table 2). These results indicated that the environment in supermarkets had more diverse contamination sources than that in slaughterhouses.
Thirty serotypes in 108 isolates were successfully identified, except that seven isolates were failed in serotyping. Salmonella Derby (n = 26), London (n = 16), and Rissen (n = 12) were the most commonly observed serotypes in this study. Isolates from slaughterhouses had 10 serotypes (Fig. 1a) and supermarkets had 25 serotypes (Fig. 1b). Five serotypes (Derby, Typhimurium, London, Rissen, and Jerusalem) were shared both at supermarkets and slaughterhouses. Five serotypes were only detected at slaughterhouses, while 20 serotypes were only detected at supermarkets. Taken together, isolates from supermarkets showed more diverse serotypes than that from slaughterhouses.

Antimicrobial susceptibility testing
Antimicrobial susceptibility testing of 115 Salmonella isolates to 13 antimicrobials was performed. Overall, 85.22% (98/115) isolates were resistant to at least one antibiotic and 50.43% (58/115) were MDR. For samples from both slaughterhouses and supermarkets, isolates showed resistance to tetracycline (73.04%, 84/115) was the highest, followed by 66.96% (77/115) to ampicillin and 59.13% (68/115) to doxycycline (Fig. 2a). The rates of strains that were sensitive or resistant to one or two classes of antibiotics were displayed in Fig. 2b

Multilocus sequence typing
Multilocus sequence typing was used to identify the relatedness of Salmonella in pork, chicken and the environment from slaughterhouses and supermarkets. Twenty-three different sequence types (STs) were identified among 108 isolates, while 7 isolates with failed serotyping results also defected in MLST were excluded for further analysis (Fig. 3). All sequence types consist of 15 from supermarkets, 3 (ST543, ST365 and ST516) from slaughterhouses and 5 (ST19, ST34, ST40, ST155 and ST469) from both sites. Most of the sequence types were detected in less than 10 isolates, except for ST40, ST155 and ST469. The largest population of isolates were ST40 (n = 24), followed by ST155 (n = 19) and ST469 (n = 17). Some Salmonella isolates presented similar sequence types that belonged to the same serovar. For example, all ST40 isolates were S. Derby, all S. Rissen isolates belonged to a single cluster (ST469) and 15 out of 16 S. London isolates were ST155. As for the relationship between MLST and antibiotic resistance, we found that ST17 and ST155 were resistant to a wide range of  antibiotics in this study, especially that 16 out of 18 ST155 isolates were MDR strains (Fig. 3).

Discussion
For the purpose of this study, Salmonella isolates were recovered from scores of sites, including chicken, pork and the environment at abattoirs and supermarkets. Our results indicated that Salmonella was more frequently isolated in pork, and supermarkets exhibited a higher MDR Salmonella isolation rate and more diversity in serotypes and sequence types than slaughterhouses. The overall isolation rate of Salmonella in our study was 10.25%, which was lower than previous studies conducted in Sichuan province [11] and Yangzhou city [17], but close to surveys in three provinces of central China [21] and Germany [22]. It should be noted that although isolation rate at slaughterhouses (10.76%) was similar to that at supermarkets (10.07%), Salmonella appeared more frequently in pork (44.00%) than that in chicken (9.50%).
Other studies also showed that Salmonella contamination rates in pork varies from 30 to 70% at retail markets [23,24] and from 10 to 50% at slaughterhouses [25,26], indicating that poor control measures were performed in slaughtering and retail chains of pork. For example, poor general hygiene and unsuitable storage conditions were commonly detected in pig slaughterhouses. Also, lacking of appropriate storage methods and regular disinfection increased the risk of Salmonella-colonizing activity at retail markets. In general, all of the differences attributed to collection seasons, amounts of samples and types, isolation methods and management.
S. Derby, generally detected in pork, could cause salmonellosis in many countries [27], and S. Rissen was generally considered to be transported through pig products in European countries [28]. In this study, S. Derby and S. Rissen were isolated from pork, chicken and the environment, and most of S. Derby isolates were detected in the environment. S. Derby was the dominant serotype, which was similar to other studies [11,17]. S. Typhimurium and S. Enteritidis are the main serotypes causing acute human infection [29], and in this study 11 isolates were detected to be S. Enteritidis or S. Typhimurium, which had potential threats to public health. In addition, more categories of serotypes and sequence Serotype results of isolates from chicken, pork and the environment in supermarkets types were detected in isolates from supermarkets than that from slaughterhouses, demonstrating the various and abundant sources of contamination in retail process. Some STs in this study related to specific serovars, for instance S. London with ST155, ST469 with S. Rissen, and ST40 with S. Derby. Our results supported the conjecture that multilocus sequence typing could be an alternative method for serotyping in the future [30].
In this study, most of the isolates showed resistance to tetracycline and ampicillin, which was similar to a previous study [31]. The high resistance rate to these two antibiotics is reasonable, since these two antibiotics has been largely used in the husbandry industry in China [7]. It was noteworthy that more than half of the isolates exhibited MDR profiles, and compared with slaughterhouses, MDR Salmonella contamination occurred at supermarkets was much more frequently. The sanitation control strategies in supermarkets need to be carried out in this area to improve the safety of animal products.

Conclusion
In summary, the results of this study indicated that the contamination of Salmonella occurred in pork made it become a potential reservoir for human infection to some extent. In addition, although the isolation rate at supermarkets was close to that at slaughterhouses, isolates from supermarkets presented a high frequency of MDR profiles and a wider range of serotypes and sequence types; these results indicated that Salmonella isolates from supermarkets were more threatening and their sources were much more complicated than that from slaughterhouses. Therefore, strict hygiene method and HACCP management in retail process should be taken into consideration to prevent foodborne infection caused by Salmonella.

Isolation and serotyping
After a pre-enrichment step of each sample in 10 mL sterile buffered peptone water (BPW) and incubated overnight at 37°C. 0.2 mL of each pre-enriched suspensions were added into 2 ml of Rappaport-Vassiliadis enrichment Broth (RVB) and 2 ml of Tetrathionate broth (TTB) respectively, then incubated at 42°C for 24 h. One loopful of each RVB and TTB culture was then streaked onto Xylose Lysine Tergitol 4 (XLT-4) agar plates, which were incubated at 37°C for 24 to 48 h. Among suspected colonies, only one was picked up from a plate and confirmed by specific gene through Polymerase Chain Reaction (PCR) of Salmonella using assays. Each isolate was serotyped by slide agglutination based on the Kauffmann-White-Le Minor Scheme [32].

Multilocus sequence typing
Protocols used for MLST of Salmonella were described online [33]. Seven housekeeping genes were amplified by PCR, including thrA, purE, sucA, hisD, aroC, hemD, and dnaN. PCR products were purified and sequenced by Sanger method, and the alleles and STs were assigned according to the MLST scheme [34]. The unweighted pair group method with arithmetic means analysis (UPGMA) was utilized to infer relationships among the isolates through MEGA7 software [35].

Statistical analysis
All statistical analyses were done using SPSS 20.0 (SPSS Inc., Chicago, IL), and the chi-squared test was applied to assess any statistically significant (P < 0.05) differences in this study.