Molecular Epidemiology of Salmonella enterica in Poultry in South Africa Using the Farm-to-Fork Approach

The presence of the zoonotic pathogen Salmonella in the food supply chain poses a serious public health threat. This study describes the prevalence, susceptibility profiles, virulence patterns, and clonality of Salmonella from a poultry flock monitored over six weeks, using the farm-to-fork approach. Salmonella was isolated using selective media and confirmed to the genus and species level by real-time polymerase chain reaction (RT-PCR) of the invA and iroB genes, respectively. Antimicrobial susceptibility profiles were determined using Vitek-2 and the Kirby–Bauer disk diffusion method against a panel of 21 antibiotics recommended by the World Health Organisation Advisory Group on Integrated Surveillance of Antimicrobial Resistance (WHO-AGISAR). Selected virulence genes were identified by conventional PCR, and clonality was determined using enterobacterial repetitive intergenic consensus PCR (ERIC-PCR). Salmonella was present in 32.1% of the samples: on the farm (30.9%), at the abattoir (0.6%), and during house decontamination (0.6%). A total of 210 isolates contained the invA and iroB genes. Litter, faeces, and carcass rinsate isolates were classified as resistant to cefuroxime (45.2%), cefoxitin (1.9%), chloramphenicol (1.9%), nitrofurantoin (0.4%), pefloxacin (11.4%), and azithromycin (11%). Multidrug resistance (MDR) was observed among 3.8% of the isolates. All wastewater and 72.4% of carcass rinsate isolates were fully susceptible. All isolates harboured the misL, orfL, pipD, stn, spiC, hilA, and sopB virulence genes, while pefA, spvA, spvB, and spvC were absent. In addition, fliC was only present among the wastewater isolates. Various ERIC-PCR patterns were observed throughout the continuum with different subtypes, indicating the unrelated spread of Salmonella. This study concluded that poultry and the poultry environment serve as reservoirs for resistant and pathogenic Salmonella. However, there was no evidence of transmission along the farm-to-fork continuum.


Introduction
Salmonella infections remain one of the most common foodborne diseases, which constitutes a global public health concern [1]. As early as the 1950s, Salmonella was highlighted as an important zoonotic pathogen with economic implications by the World Health Organisation (WHO) and the Food and Agriculture Organisation (FAO) of the United Nations [2]. e genus Salmonella is divided into two species, i.e., S. enterica and S. bongori. S. enterica is also subdivided into six subspecies and comprises more than 2600 serovars [3]. e pathogenesis of Salmonella is mediated by various genes that promote host cell invasion, intracellular survival, and colonisation [4,5]. Salmonella has been reported in intensive animal husbandry such as poultry production [6]. In poultry farms, flocks can become infected via vertical or horizontal transmission [7,8]. Vertical transmission occurs when the

Ethical Considerations.
is study was part of an overarching research project on Antibiotic Resistance and One Health; ethical clearance was obtained from the Animal Research Ethics Committee (reference AREC 073/016PD) and the Biomedical Research Ethics Committee (reference BCA444/16) of the University of KwaZulu-Natal (UKZN). e study was also registered with the South African National Department of Agriculture, Forestry, and Fisheries (reference 12/11/1/5 (879)).

Study
Area. An intensive poultry farm located in the uMgungundlovu district in Northern KwaZulu-Natal was identified and agreed to participate in this study. e sampling strategy adopted was according to the WHO-AGISAR guidelines [16].

Sample Collection.
A flock of one-day-old Cobb breed hatchlings, placed in a clean chicken house, was the target population. A total of 162 samples were collected using block sampling to ensure that the entire flock was represented within the poultry house over a six-week period. e sample collection included the following: weekly poultry faeces and litter, truck, and crate swabs when chickens were transported to the abattoir; carcass rinsate during slaughter at the abattoir; ceca from postslaughter and ready-to-cook packaged samples of the whole carcass, neck, and thigh. e wastewater sample was collected from the main drainpipe during house cleaning after the flock was removed. Briefly, ten litter and ten faecal samples (5 g of each sample) were collected per week (5 weeks � 50 litter, 50 faeces). On week 5, ten samples of each of the following were collected: truck swabs, crate swabs, ceca (25 g of each ceca sample), neck (25 g of each neck sample), thighs (25 g of each thigh sample), and whole carcasses. A single 40 ml sample was collected from carcass rinsate during slaughter and wastewater during the house decontamination. All samples were collected aseptically into sterile containers, labelled clearly, placed into cooler boxes, then transported immediately to the UKZN Antimicrobial Research Unit (ARU) laboratory. Samples were processed within four to six hours of collection. Salmonella prevalence was calculated as a percentage of Salmonella culture-positive samples among the total number of samples collected.

Sample Processing and Isolation of Salmonella.
e samples were inoculated in 40 ml of nonselective tryptic soy broth (TSB) (Oxoid Ltd., Basingstoke Hampshire, UK) and incubated for two hours at 37°C with shaking at 100 rpm. ereafter, 0.1 ml was inoculated into 9.9 ml of Rappaport-Vassiliadis broth (Oxoid Ltd., Basingstoke, UK), a selective enrichment media for Salmonella and incubated at 42°C for 24 hours [17]. A loopful of the sample was streaked onto Brilliance Salmonella Agar (Oxoid Ltd., Basingstoke Hampshire, England) and incubated at 37°C for 24 hours [18]. Single presumptive purple/pink colonies typical of Salmonella were picked and subsequently streaked onto Hektoen Enteric Agar (HEA) (Oxoid Ltd., Basingstoke Hampshire, England) and Salmonella Shigella agar (SSA) (Oxoid Ltd., Basingstoke Hampshire, England) plates and incubated at 37°C for 24 hours. Culture plates were examined for the presence of typical colonies based on morphological characteristics, i.e., clear colonies with a black centre on HEA and SSA. Single colonies were subcultured onto nutrient agar and subsequently stored in TSB supplemented with 10% glycerol (Merck, USA) at −60°C until further analysis. Salmonella enterica subsp. enterica serovar Choleraesuis ATCC 10708 (ATCC, Manassas, Virginia, USA) was included as a control strain.  [19]. Colonies from an overnight culture grown on nutrient agar were briefly suspended into 200 μl of distilled water and lysed at 100°C for 15 minutes.

Isolation and Identification of
is was followed by centrifugation at 13000 rpm for 5 minutes; thereafter, 120 μl of supernatant was transferred to a clean microcentrifuge tube and stored at 4°C until further analysis.
Isolates were genotypically confirmed as Salmonella using RT-PCR of the genus-specific invA gene (284 bp) and the S. enterica species-specific iroB gene (606 bp), using the QuantStudio 5 RT-PCR machine ( ermo Fisher Scientific, Waltham, MA, USA). e following primer sequences were used for invA F: 5′ GTGAAATTATCGCCACGTTCGGG-CAA 3′and invA R: 5′ TCATCGCACCGTCAAAGGAACC 3′; iroB F: 5′ TGCGTATTCTGTTTGTCGGTCC 3′and iroB R: 5′ TACGTTCCCACCATTCTTCCC 3' [20,21]. A 10 ul reaction mixture containing 5 μl of Luna ® Universal qPCR Master Mix (New England Biolabs, Ipswich, MA, USA), 0.5 μl of each of the forward and reverse primers (20 μM) (Inqaba Biotechnical Industries (Pty) Ltd., Pretoria, South Africa), 1 μl of nuclease-free water (Lonza Rockland, ME, USA), and 3 μl of template DNA was prepared. e following cycling conditions were used: an initial activation at 95°C for 10 minutes, followed by 50 cycles of denaturation at 95°C for 10 seconds (s), annealing at 64°C (15 s), extension at 72°C (25 s), and a final extension at 72°C for 5 min. Melting was done by ramping from 72 to 90°C, with a 0.1°C rise at each step, a premelt hold for 90 s in the first step, followed by a hold for 2 s in the next steps [22]. Real-time melt curves were obtained and the presence of the respective genes was analysed. e melting temperatures for the genes ranged from 85 to 86°C for invA and 88 to 89°C for iroB. Salmonella enterica subsp. enterica serovar Choleraesuis ATCC 10708 (ATCC, Manassas, Virginia, USA) was used as a positive control reference strain and a tube containing nuclease-free water ( ermo Fisher Scientific, Vilnius, Lithuania) instead of template DNA was included as a negative control.

Antimicrobial Susceptibility Testing (AST)
. AST was carried out using the automated Vitek-2 system according to the manufacturer's instructions and the Kirby-Bauer disk diffusion method (for antibiotics absent from the Vitek-2 panel) following the established Clinical and Laboratory Standards Institute (CLSI) guidelines [23]. e WHO-AGISAR panel of antibiotics recommended for testing is based on well-established monitoring and surveillance systems on antimicrobial resistance in foodborne bacteria. e selection of antibiotics provides a harmonised standard that allows for data comparison between laboratories and countries [16]. In brief, a bacterial inoculum matched to a 0.5 McFarland standard was prepared from a pure overnight culture. For Vitek-2, the inoculum and AST-N255 cartridges were loaded into the instrument and programmed to run overnight. For disk diffusion, the suspension was evenly spread on a Mueller-Hinton agar plate using a sterile cotton swab and allowed to air-dry before dispensing the antibiotic disks onto the surface. Plates were incubated at 37°C for 18-24 hours. e minimum inhibitory concentrations (MIC) were determined by Vitek-2 and inhibition zones were measured for disk diffusion. e results were interpreted as susceptible, intermediate, or resistant according to the CLSI breakpoints [23]. E. coli ATCC 25922 (ATCC, Manassas, Virginia, USA) was used as a quality control strain. Isolates were classified as MDR if they displayed resistance to one or more antibiotics belonging to three or more different antibiotic classes [24,25].

Molecular Detection of Virulence Factors
2.5.1. DNA Extraction. Genomic DNA was purified from bacterial cultures grown overnight in 1 ml of TSB. e genomic DNA GeneJET extraction kit ( ermo Fisher Scientific, Vilnius, Lithuania) was used, following the manufacturer's instructions for Gram-negative bacteria. All DNA was stored at −20°C until further analysis by conventional PCR for virulence genes and clonality by ERIC-PCR.

Detection of Virulence Genes.
Twelve virulence genes were analysed by conventional PCR using specific primer pairs as described previously [4,26,27]. e primer sequences, cycling conditions, and product sizes are listed in Table 1. All PCR reactions were performed in a final volume of 15 μl containing the following: 8 μl of DreamTaq Master Mix ( ermo Fisher Scientific, Vilnius, Lithuania), 0.5 μl of each the forward and reverse primers (20 μM), 3.5 μl of nuclease-free water ( ermo Fisher Scientific, Vilnius, Lithuania), and 2.5 μl of the template DNA. Salmonella enterica subsp. enterica serovar Choleraesuis ATCC 10708 was used as quality control. e amplified PCR products were electrophoresed on a 1.5% agarose (Lonza Rockland, ME, USA) gel, containing 5 μl of ethidium bromide (Sigma-Aldrich, St. Louis, MO, USA) in 1X TAE buffer (BioConcept Ltd., Allschwil, Switzerland), at 100 volts (V) for 1 hour. e bands were visualised using the Bio-Rad gel documentation system (Bio-Rad, Hercules, CA, USA) and Image software (Bio-Rad, Hercules, CA, USA). A 100 bp molecular weight ladder (New England Biolabs, Ipswich, MA, USA) was used to determine the size of the PCR fragments. Salmonella enterica subsp. enterica serovar Choleraesuis ATCC 10708 (ATCC, Manassas, Virginia, USA) was used as a positive International Journal of Microbiology 3 control strain and a tube containing nuclease-free water ( ermo Fisher Scientific, Vilnius, Lithuania), instead of template DNA, was included as a negative control.

Virulence
Genes. Eight of the 12 virulence genes were detected among the isolates. e misL, orfL, pipD, stn, spiC, hilA, and sopB genes were present in all isolates from all sources, whereas pefA, spvA, spvB, and spvC were absent. e chromosomal virulence gene fliC was only present in the week 6 wastewater isolates.

ERIC-PCR.
e ERIC-PCR yielded patterns consisting of 3-13 bands, with sizes ranging from 150 bp to 3 Kb. Diverse patterns were observed along the continuum. Clusters were source-specific based on a similarity index of 70% ( Figure S1).

Discussion
e contamination of poultry and poultry products with Salmonella can occur at various stages of the production process. Although there are food safety standards, Salmonella continues to persist in the poultry processing industry [32].  [39,40]. e proportion of Salmonella isolated from poultry can be influenced and varies by geographical location, environmental contamination, nature of farm management systems, type of production systems, breed of the flock, age of birds, sample size, sampling procedures, and the methods used for isolation [6,11]. e biological nature of Salmonella in an infected host and its shedding pattern can also influence its isolation frequency, which may be seasonal and    determined by environmental factors [25]. After contamination in poultry farms, Salmonella can colonise the intestines of chickens and result in contamination of the carcass during the slaughtering and meat processing stages [32]. In this study, Salmonella was not detected on the farm during week 1 and week 4. Sporadic isolation may be associated with various factors on the farm. Salmonella colonisation in chickens can be influenced by factors such as the age of the chicken, physiological and environmental stresses, survival of Salmonella through the gastric barrier, health and disease status of the chickens, vaccination, use of antimicrobials as growth promoters and prophylaxis, diet, and genetic background of the chickens [8,41,42]. In a previous study conducted in Ethiopia, young chicks were infected with 10 7−9 CFU/ml of S. typhimurium. Salmonella was detected two weeks postinfection, suggesting that the immunity of young chicks can provide early protection. However, over time, at four weeks, a decline was observed, indicating that resistance to Salmonella infection by older chickens may  be associated with the activation of cellular and humoral immunity [43]. is scenario may provide information on the immune status of the poultry flock evaluated in our study.

Antimicrobial
Susceptibility. Salmonella strains found in poultry have also developed resistance to antibiotics, which are crucial for treating human infections [7]. In SA, antibiotics among livestock are used mainly in intensively farmed poultry and pigs. e AST results revealed that 51% (107/210) of the isolates were resistant to one or more antibiotics, and 3.8% (8/210) were classified as MDR. Resistance was detected for six of the 21 antibiotics tested, which belonged to classes of antibiotics (cephalosporins, macrolides, and quinolones) that are used in the SA poultry production. In comparison to our study, higher rates of MDR Salmonella among poultry isolates have previously been reported in Africa (12.1%-100%), Asia (34.72-43.1%), the USA (9.5-18%), and the EU (38.2%) [12,33,37,[44][45][46][47][48].
In this study, resistance to the second-generation cephalosporins, cefuroxime (45.2%) and cefoxitin (1.9%), among isolates that were susceptible to other β-lactam antibiotics such as ampicillin was observed. A similar resistance phenotype was reported in Saudi Arabia, where clinical and environmental samples showed resistance to cefuroxime (90.9%) and cefoxitin (87.9%), while most isolates were susceptible to β-lactams [49]. Previous studies in Africa, Australia, and the USA have reported a higher frequency of resistance to cefuroxime or cefoxitin among Salmonella-positive poultry isolates [14,39,46,50].
In 2015, pefloxacin was recommended as a reliable surrogate marker to identify susceptibility to fluoroquinolones, as tests with nalidixic acid and ciprofloxacin disks do not reliably detect resistance at low levels in Salmonella spp. [51][52][53][54][55]. is was observed in this study where there was 100% susceptibility to ciprofloxacin and 4.8% (10/ 210) intermediate to nalidixic acid compared with 11.4% (24/210) resistance to pefloxacin ( Table 5), indicating that antibiotic panels may impact the resistance profiles observed. A similar scenario was reported in India by Prabhurajan et al. [56], where a higher rate of resistance to pefloxacin than that of nalidixic acid was observed among clinical Salmonella isolates. It was concluded that different mechanisms and mutations associated with fluoroquinolone resistance can influence atypical resistance phenotypes [56]. A systematic review of fluoroquinolone resistance in Salmonella in Africa by Taddesse et al. [57] also reported that Salmonella strains susceptible to nalidixic acid and resistant to ciprofloxacin have been identified among NTS serovars. e review also indicated that although reports on the distribution of Salmonella strains with unusual AST phenotypes are scarce, these strains occur in Africa [57].
Previous studies have reported varying levels of resistance or susceptibility to nalidixic acid, ciprofloxacin, and pefloxacin among isolates, attributed to the varying resistance mechanisms of fluoroquinolones. Different AST methodologies or reagents have also been identified as possible causes of this variation. Antibiotic discs from different manufacturers have given varying results and the narrow range to classify isolates as resistant or susceptible may introduce subjectivity when reading plates. e CLSI M100 guidelines indicate that no single antibiotic test can accurately detect all types of resistance associated with fluoroquinolones [7,54,[58][59][60]. Automated systems used in laboratories may also have challenges in implementing new lower breakpoints for ciprofloxacin, and therefore, low-level resistance in isolates may be undetected [61].
In comparison to the results of the current study, higher rates of fluoroquinolone resistance (ciprofloxacin and nalidixic) have been reported in poultry in the EU (64.7% and 61.5%), Brazil (86.5% and 89%), and China (25.7% and 46.7%) [37,45,60]. However, Salmonella isolated from broilers in Canada (3% and <1%) and in the USA (0% and <1%) have shown a low resistance frequency to fluoroquinolones, which can be attributed to the restricted use of fluoroquinolones in poultry [39,60,62,63].
In this study, low resistance frequencies were also observed for chloramphenicol (1.9%), nitrofurantoin (0.4%), and azithromycin (11%). Similarly, low rates of chloramphenicol resistance were reported among poultry isolates in Brazil (0.6%) and the USA (0.3%) [39,60]. e low chloramphenicol and nitrofurantoin resistance in this study can be attributed to the prohibited use of these antibiotics in food animals in SA [64]. In South Korea, a slightly higher but comparable azithromycin resistance (17.9%) was reported compared to the results in the present study. In contrast, in the EU, USA, and Brazil, azithromycin resistance in foodproducing animals and food products was either not detected or detected at a very low frequency (<1%) [12,48,60,65]. Azithromycin is used for the treatment of MDR Salmonellosis in humans. Resistance to azithromycin observed in the present study may be associated with the authorised use of other macrolides such as tylosin on poultry farms in SA compared to the EU where macrolides are banned [12,64,66].
However, based on the resistant results obtained in this study associated with second-generation cephalosporins, quinolones, and fluoroquinolones, further molecular analysis of the genes that confer resistance and the mechanisms involved would be needed to explain the unusual AST phenotypes observed. is study also identified the presence of MDR Salmonella strains specifically at the farm level ( Table 2). Antibiotic resistance continues to increase in LMICs, due to the injudicious use of antimicrobials as growth promoters and feed additives to strengthen the intestinal microflora and for the control, prophylaxis, and treatment of infectious diseases in animal husbandry [41]. In this study, Salmonella isolates were resistant to antibiotic classes that are listed as critically and highly important antimicrobials for human medicine by the WHO and as antimicrobials of veterinary importance by the World Organisation for Animal Health (OIE) [67,68]. In Europe and the USA, many of these antibiotic classes are restricted or prohibited for use in food animals [63,66]. In some African countries, antibiotics are used in food animals without strict regulations [14,46]. However, in SA, regulations and policies exist, although enforcement remains challenging [64,69]. 8 International Journal of Microbiology e resistant Salmonella strains in food animals and food products can be transmitted to humans through the food chain, and this poses a health risk to consumers by causing infections that are more severe and difficult to treat [60,70].

Virulence.
e virulence genes present in Salmonella can determine the pathogenicity among the strains and serotypes circulating in poultry and the environment [26,71]. In SA, information on the prevalence of Salmonella virulence genes found in poultry is limited, and only a few studies have investigated virulence in chickens [4,21]. Gene clusters known as Salmonella pathogenicity islands (SPIs) are regions located on chromosomes that are distributed in the Salmonella genome, which are associated with virulence and enable efficient bacterial colonisation in the host. Genes such as hilA are found in SPI-1 that encode proteins involved in cell invasion [27]. Other genes located in the following SPIs, the SPI-2 gene spiC, the SPI-3 gene misL, and the SPI-4 gene orfL, are responsible for replication within macrophages [21,28]. Genes located in SPI-5, such as pipD and sopB, promote enteritis and macrophage invasion accordingly [21,27,28,72].
In this study, all Salmonella isolates possessed these virulence genes, indicating that healthy chickens and their environment serve as carriers of pathogenic Salmonella strains, which can cause systemic infection and enteric infection in the host [28]. Previous studies conducted in SA, UK, Egypt, and China have also reported the presence of these virulence genes in poultry [21,26,28,72].
Other genes present on the Salmonella chromosomes found in this study were stn and fliC, which play a role in enterotoxin production and encode phase-1 flagellin, respectively [26]. All Salmonella isolates contained the stn gene, which has also been reported in all poultry isolates from studies conducted in China, Benin, and India [72][73][74]. However, the fliC gene in our study was only detected in the wastewater isolates obtained during the house decontamination. e fliC gene encodes for the flagellin protein and has been used as a target gene to determine antigenic specificity and genetic diversity in Salmonella. Studies done in Egypt, Malaysia, and Saudi Arabia described that fliC was serovar-specific for S. typhimurium, S. Kentucky, S. Aberdeen, S. Bergen, and S. Kedougou [26,27,75]. Serovars were not determined in this study.

Clonality.
In Columbia, Herrera-Sanchez et al. [76] reported similar ERIC banding patterns consisting of 2-13 bands that ranged from 200 to 4000 bp in size whilst contrasting ERIC results of band sizes 15-1500 bp and 190-1430 bp were observed by Fendri et al. [77] in France and Oliveria et al. [78] in Brazil, respectively. Among the 210 isolates, there were various patterns and clusters. Isolates obtained from litter and faeces over the collection period clustered, indicating clonal spread on the farm. However, the carcass rinsate and wastewater isolates formed separate clusters. Salmonella transmission along the farm-to-fork continuum was not evident as unrelated strains from different sources belonged to different clusters based on a 70% similarity index. Of note, although ERIC-PCR has a shorter turnaround time, it is less discriminatory. erefore, further studies involving more resolute typing approaches, such as serotyping and whole-genome sequencing (WGS), are recommended to elucidate the population structure of the strains.

Conclusion
Our study showed sporadic Salmonella contamination along the farm-to-fork pathway and demonstrated that healthy chickens could serve as a reservoir of resistant and pathogenic Salmonella strains, which harbour virulence factors and exhibit resistance to various antibiotics. e presence of MDR Salmonella strains among litter and faeces in this study indicates that the imprudent use of antibiotics on poultry farms could contribute to the increasing development and spread of antibiotic resistance in food animals and animal food products. erefore, the initial steps to reduce and control Salmonella contamination of poultry meat should be taken at the farm level. SA has developed and implemented an Antimicrobial Resistance National Strategy Framework for 2018-2024 with the strategic objective of promoting the appropriate use of antibiotics in humans and animals using regulations. However, interventions to address antimicrobial use and resistance in food animals need to be expedited.

Data Availability
e data used to support the findings of this study are included within the article.

Disclosure
Any opinions, findings, conclusions, or recommendations expressed in this material are those of the authors and do not necessarily reflect the organizations or agencies' views that provided support for the project. e funders had no role in the study design or the decision to submit the work for publication.

Conflicts of Interest
Professor Sabiha Y. Essack is the chairperson of the Global Respiratory Infection Partnership and a member of the Global Hygiene Council, both sponsored by unconditional educational grants from Reckitt, UK. All other authors declare that they have no conflicts of interest.