Effectiveness of BOX-PCR in Differentiating Genetic Relatedness among Salmonella enterica Serotype 4,[5],12:i:- Isolates from Hospitalized Patients and Minced Pork Samples in Northern Thailand

Salmonella enterica Serotype 4,[5],12:i:-, a monophasic variant of S. Typhimurium, with high virulence and multidrug resistance is distributed globally causing pathogenicity to both humans and domesticated animals. BOX-A1R-based repetitive extragenic palindromic-PCR (BOX)-PCR proved to be superior to three other repetitive element-based PCR typing methods, namely, enterobacterial repetitive intergenic consensus (ERIC)-, poly-trinucleotide (GTG)5-, and repetitive extragenic palindromic (REP)-PCR (carried out under a single optimized amplification condition), in differentiating genetic relatedness among S. 4,[5],12:i:- isolates from feces of hospitalized patients (n=12) and isolates from minced pork samples of S. 4,[5],12:i:- (n=6), S. Typhimurium (n=6), and Salmonella Serogroup B (n=4) collected from different regions of northern Thailand. Construction of phylogenetic trees from amplicon size patterns allowed allocation of Salmonella isolates into clusters of similar genetic relatedness, with BOX-PCR generating more unique clusters for each serotype than the other three typing methods. BOX-, (GTG)5-, and REP-PCR indicated significant genetic relatedness between S. 4,[5],12:i:- isolates 1 and 9 from hospitalized patients and S. 4,[5],12:i:- isolate en 29 from minced pork, suggesting a possible route of transmission. Thus, BOX-PCR provides a suitable molecular typing method for discriminating genetic relatedness among Salmonella spp. of the same and different serotypes and should be suitable for application in typing and tracking route of transmission in Salmonella outbreaks.


Introduction
Nontyphoidal Salmonella (NTS) is a cause of gastroenteritis, particularly in young children, the infection arising from consumption of contaminated food or unhygienic practices [1]. Salmonella enterica Serotype 4, [5],12:i:-is an emerging serotype with distribution worldwide and a significant infection rate of humans and domestic animals [2][3][4][5] including outbreaks in human populations of many countries [6].
Serologically related to S. Typhimurium, S. enterica 4, [5],12:i:-expresses O 4, 5, 12 antigens but not fljB (encoding phase 2 flagellum) due to defective phase switching [7]. e bacteria manifest multidrug resistance phenotype in many regions of the world including ailand [8,9]. e organism has been isolated from various animal species, e.g., chicken, cattle, swine, and turtles, and also from food items, such as raw poultry, pork, and pork sausage [10]. Furthermore, there exists evidence of genetic relatedness between Salmonella isolated from imported ( ai) pork products and (Danish) patients, suggesting an important route of Salmonella transmission across continents [11].
Molecular typing of Salmonella spp. is the usual assay performed to examine genetic relatedness, able to discriminate closely related Salmonella isolates, and reveal source-to-person strain transmission with sufficient precision to identify the specific source responsible for foodborne outbreaks [12]. A number of PCR-based typing techniques have been applied, such as direct sequencing of PCR amplicons, restriction fragment length polymorphism (RFLP)-PCR, amplified fragment length polymorphisms (AFLP)-PCR, random amplified polymorphic DNA (RAPD)-PCR, arbitrary primed (AP)-PCR, and pulsed-field gel-electrophoresis (PFGE)-PCR [13][14][15][16], the latter being the most popular technique and is commonly classified as the standard method due to its high discrimination and reproducibility, but the method requires specialized equipment, specific technical expertise, and lengthy (days) turn-around time. Other techniques have been developed to take advantage of known genetic elements, often noncoding intergenic repetitive sequences located in close proximity to one another, scattered across the genome, and using several PCR primers to amplify several families of repeated sequences. Examples of such methods include BOX-A1R-based (BOX)-, enterobacterial repetitive intergenic consensus (ERIC)-, poly-trinucleotide (GTG) 5 -, and repetitive extragenic palindromic (REP)-PCRs [16,17]. e variability of genomic DNA sequences is identified by differences in sizes of the amplified fragments, and analysis of the different DNA fragment profiles can be performed using computer-assisted algorithms to cluster different patterns and construct phylogeny trees [18]. ose PCR primers can be utilized in different PCR protocols to evaluate their discrimination ability, sensitivity, and robustness [19].
e study sought to simplify identification of genetic relatedness with high discrimination between S. enterica 4, [5],12:i:-isolates from two different sources by comparing four different repetitive element-based PCR methods, namely, BOX-, ERIC-, (GTG) 5 -, and REP-PCR. Clustering power and discriminatory index of each technique were evaluated using the S. 4, [5],12:i:-isolates, together with S. Typhimurium and S. Serogroup B isolates. In addition, phylogenetic trees were constructed to determine relationship of clusters with other data sets, such as antibiogram profile and carriage of antibiotic-resistant genes.  [20], and S. 4, [5],12:i:-(n � 6), S. Typhimurium (n � 6), S. Serogroup B (n � 3; S. Agona, S. Saintpaul, and S. Schwarzengrund), and one unknown Salmonella serotype from minced pork samples collected from retail markets in five different provinces of northern ailand [21] (Figure 1), kept at 4°C until used.

Determination of Antibiotic Resistance Profile.
Susceptibility to antibiotics of twelve S. 4, [5],12:i:-originally isolated from hospitalized patients was performed using a disk diffusion method following the Clinical and Laboratory Standards Institute (CLSI) [26] with ampicillin (AMP) l0 μg, cefotaxime (CTX) 30 μg, chloramphenicol (C), streptomycin (S) 10 μg, sulphamethox/trimethoprim (SXT) 1.25 μg/ 23.75 μg, tetracycline (TE) 10 μg, and colistin (COL) 10 μg (Oxoid, Hampshire, UK). Escherichia coli ATCC 25922 was used as a negative control strain. e ESBL test was performed using the combination disk method according to CLSI criteria with both ceftazidime (30 μg) and cefotaxime (30 μg) alone and combined with clavulanic acid (10 μg) (Oxoid, Hampshire, UK). In-house known ESBL-producing Escherichia coli and ESBL-negative Escherichia coli strains ATCC 25922 were used as controls. 5 -, and REP-PCR Assays. DNA was extracted from Salmonella isolates as previously described [27]. In brief, the overnight culture (1 ml) was centrifuged, the pellet was washed twice with 400 μl of TE buffer (10 mM Tris HCl, pH 8.0, 1 mM EDTA), and then the pellet was resuspended in 400 μl of TE buffer. e resuspended solution was incubated at 80°C for 20 minutes. At room temperature, 50 μL lysozyme (10 mg/mL) was added to the solution which was then incubated at 37°C for one hour with occasionally shaking followed by the addition of 75 μL of 10% SDS/proteinase K solution with vigorous vertexing and incubation at 65°C for 10 minutes. en, 100 μL of 5 M NaCl and 100 μL of prewarmed (65°C) CTAB/NaCl solution were added and additionally incubated at 65°C for 10 minutes. 750 μl of chloroform/isoamyl alcohol (24 : 1) was added, and the solution was centrifuged for 5 minutes at 13,000 rpm at 4°C. e upper aqueous solution was collected, and then ethanol precipitation was performed. Finally, the pellet was resuspended with 50 μl doubledistilled water and the DNA solution was kept at −20°C until being further used.

BOX-, ERIC-, (GTG)
To perform PCR reactions, each PCR mixture contained 0.1 μL of DNA, different concentrations of each primer set (

Amplicon Profile Analysis and Phylogenetic Tree
Construction. Analysis of amplicon patterns generated by PCRs described above and construction of phylogenetic tree were carried out using curve-based algorithm (Pearson correlation) (Applied Maths, Sint-Martens-Latem, Belgium) to create a similarity scale and an unweighted pair group using arithmetic averages algorithm (UPGMA) for cluster analysis.
2.6. 3D Coordinate Space Window Construction. 3D visualization of similarity to dataset of BOX-PCR clustering based on multidimensional scaling (MDS) was performed using a Metric algorithm (Applied Maths), and the coordinate space window was calculated based on the similarity matrix. Coordinate space window displayed each S. 4, [5],12:i:-isolates as dots in a cubic coordinate system and also as 3D spheres to enable visualization of 3D clustering in a realistic perspective.

Discriminatory Index Determination.
In order to calculate the average probability that the molecular typing methods will assign a different type from two unrelated strains randomly sampled from the Salmonella isolates, a discriminatory index (D) was calculated at different levels of similarity index according to the formula [28]: where D � index of discriminatory power, N � number of unrelated strains tested, S � number of different types, and x j � number of strains belonging to j th type. D value in a range of 0 (identical type) to 1.0 indicates that the typing method of interest is capable of distinguishing each member of a population from all other members of that population.  Schwarzengrund, S. Agona, and S. Saintpaul and one unknown was either from the feces of hospitalized patients or minced pork collected from 5 different provinces of the northern ailand ( Figure 1). Most isolates of S. 4, [5],12:i:showed multidrug resistance with five Salmonella isolates from hospitalized patients characterized as CTX-M group 1 producing Salmonella spp.; in addition, one S. Typhimurium isolate from minced pork in Nan province was characterized as CTX-M group 9 producing Salmonella spp. (Table 2). ree other Salmonella Serogroup B, S. Schwarzengrund, S. Agona, and S. Saintpaul, and one unknown, were included in the selection in attempt to generate out group cluster.

Molecular Typing of S. 4,[5],12:i:-Isolates from Hospitalized Patients and from Minced Pork Samples Collected in
Northern ailand. Four different molecular typing methods, namely, BOX-, ERIC-, (GTG) 5 -, and REP-PCR, performed under the same optimized annealing temperature (54.0°C for 2 minutes), were applied to eighteen S. 4, [5],12:i:-isolates from hospitalized patients and from minced pork samples collected in northern ailand, generating 9-28 amplicons of different sizes (100-1,500 bp) ( Figure 2), with BOX-PCR demonstrating the highest mean number of amplicons, followed by REP-PCR, GTG 5 -PCR, and ERIC-PCR (Table 3). In order to compare the capability of each molecular typing method to differentiate among all Salmonella isolates, D was calculated from each constructed phylogenetic tree at three levels of similarity (50, 75, and 90%) using a curve-based algorithm (Pearson correlation) to create a similarity scale. A phylogenetic tree was constructed from each of the four PCR amplicon profiles (Figure 2), which showed BOX-PCR and GTG 5 -PCR with D > 0.9 at 75% and 90% similarity, while ERIC-PCR and REP-PCR have D > 0.9 only at 90% similarity (Table 4). Both the high average number of amplicons bands and high value of D suggest BOX-PCR and GTG 5 -PCR as e UPGMA algorithm was applied to each molecular typing method in grouping into clusters of Salmonella spp. of the same serotype from same or different sources. At 50% similarity, BOX-PCR and GTG 5 -PCR were capable of differentiating S. Typhimurium and S. 4, [5],12:i:-isolates from minced pork into 2-4 clusters, while ERIC-PCR and REP-PCR placed Salmonella isolates of same serotype into one cluster each with D value � 0 (Table 4). Interestingly at 50% similarity, GTG 5 -PCR was capable of generating up to three clusters of six S. Typhimurium isolates with D value � 0.733 compared to one cluster for the other three PCR methods. At 80% similarity, all four molecular typing methods were able to differentiate the same serotype into different clusters except for ERIC-PCR that generated one cluster for six S.       Table 1. Phylogenetic trees were constructed using curve-based algorithm (Pearson correlation). e number at the branch node indicates percent amplicon profile similarity. Dark blue shade represents high cluster similarity. Light blue shade represents low cluster similarity. ID, Salmonella strains: en, from the minced pork sample; numeral, from feces of hospitalized patients.  (Figure 4).

Relatedness of Phylogenetic Tree Constructed from BOX-PCR Amplicon Profiles with Antibiogram Profile and ESBL
Production of Salmonella Isolates. e phylogeny tree constructed from BOX-PCR amplicon profiles of S. 4, [5],12:i:isolates from feces of hospitalized patients (n � 12) and minced pork samples (n � 6), S. Typhimurium isolates from minced pork samples (n � 6), and other Salmonella Serogroup B isolates from minced pork samples (n � 4) showed 50% similarity with three clusters of S. 4, [5],12:i:isolates, one of S. Typhimurium isolates, and 3 of S. Serotype Table 4: Differentiation into clusters by the four molecular typing methods of Salmonella isolates of the same serotype collected from the same source and two different sources.  2, 1, 1, 1, 1), 0.7576 a From phylogenetic tree ( Figure 2). b [28].

Discussion
Many types of short-interspersed repetitive DNA sequences have been identified in prokaryotic genomes [24], and BOX elements are characterized as being conserved among diverse bacterial species and serve as potential targets for identifying genetic relatedness in both Gram-negative and Gram-positive bacteria, such as Aeromonas spp. [29], Escherichia coli [30,31], and Streptococcus pneumoniae [32]. e constructed phylogeny tree from BOX-PCR typing effectively differentiated genetic relatedness of S. 4, [5],12:i:isolates as well as grouping them into different clusters according to their origin, feces of hospitalized patient, or minced pork sample. Previous studies in Germany  Figure 1. c In the same cluster as minced pork sample (Figure 2). d Highest value observed from the maximal similarity that each strain ID from minced pork shared with strain ID from patients in Figure 2.   [33]. BOX-, GTG 5 -, and REP-PCR similarly identified two isolates from hospitalized patients (ID 1 and 9) with high genetic relatedness to isolates from minced pork, suggesting the possibility that (some) Salmonella isolates causing human infection could have come from contaminated food (minced pork) as traditional food of northern ai food often contains raw meat, such as raw spicy minced pork. Many studies have shown contaminated raw meat and poultry are causes of Salmonella transmission if there is a lapse in food safety practices, leading to increased risks in salmonellosis outbreaks [34]. Repetitive element-based (RE)-PCR assays were shown to be capable of typing 80 serotypes and five isolates previously not typeable as well as generating amplicon profile heterogeneity within some serotypes [35]. RE-PCR was shown to be a better serotyping method over traditional serotyping of Salmonella isolates during outbreak investigations [36]. Furthermore, the greater discriminative ability of RE-PCR over the standard PFGE protocol indicates the former to be the preferred method to detect Salmonella transmission links [37]. In addition, composite of a number of RE-PCR methods offer even more discriminatory power in estimation of genetic relatedness stemming from different independent genetic information obtained from the different RE-PCR primers [37]. RE-PCR performs better than MLST in subtyping Salmonella Enteritidis isolates of food and human origin [38].
Virulent ESBL-producing S. 4, [5],12:i:-isolates from feces of hospitalize patients highly shared genetic relatedness and formed a unique cluster, with their antibiograms indicating acquisition of blaCTX group 1 as reported in many countries [39,40]. To the best of our best knowledge, ESBL-producing S. 4, [5],12:i:-isolates resistant to meristin and harboring mcr-3 gene is the first observed in northern ailand, which poses the risk of traveler's diarrhea to those returning after travelling in this region of the country [41]. In addition, to the best of our knowledge, this is the first study in which four different RE-PCR typing methods were compared in evaluating genetic relatedness among S. 4, [5],12:i:-isolates from different sources and geography.

Conclusion
e simple BOX-PCR typing method is effective in differentiating genetic relatedness of S. 4, [5],12:i:-isolates from feces of hospitalized patients in Phayao province, northern ailand, and those from minced pork samples obtained at different locations in the same region of the country and should be adopted in tracking transmission during Salmonella outbreaks.
Data Availability e original gel pictures used to support the findings of this study are included within the supplementary information file.  Figure 5: Association of Salmonella isolates with antibiogram profiles and ESBL production. Phylogenetic tree was constructed as described in the legend of Figure 2 using S. 4, [5],12:i:-isolates from feces of hospitalized patients (n � 12) and from minced pork samples (n � 6), S. Typhimurium isolates from minced pork (n � 6), and other Salmonella Serogroup B isolates from minced pork (n � 4). Antibiograms and ESBL-production properties of Salmonella isolates from minced pork were adapted from [21]. Dark blue shade represents high cluster similarity. Light blue shade represents low cluster similarity. AMP, ampicillin; C, chloramphenicol; COL, colistin; CTX, cefotaxime; S, streptomycin; SXT, sulfamethox/trimethoprim; TE, tetracyclin; ESBL, extended-spectrum beta lactamase; MDR, multidrug resistant; p, positive.

Conflicts of Interest
e authors confirm that there are no known conflicts of interest associated with this publication.