Strains of Salmonella and non-Salmonella organisms, including S. Enteritidis, S. Gallinarum, S. Pullorum, S. Dublin, S. Meleagridis, S. Uganda, S. Anatis, S. London, S. Rissen, S. Typhimurium, S. Derby, S. Choleraesuis, S. Sinstorf, S. Indiana, S. Newlands, S. Dumfries, S. Muenster, S. Yoruba, S. Kentucky, S. Agona, S. Senftenberg, S. Thompson, S. Blockley, S. Inchpark, S. Farsta, S. Dabou, S. Virchow, Mycobacterium tuberculosis, Brucella abortus, Listeria monocytogenes, Campylobacter jejuni, and Escherichia coli, were commercially available or previously isolated in our routine monitoring (Table 1).

The bacterial culture and DNA isolation were performed as previously described (Xiong et al., 2016). Briefly, all strains used in the study were inoculated in Brain Heart Infusion broth (Becton, Dickinson and Company, Sparks, MD, USA) or Luria-Bertani broth (Oxoid, Basingstoke, Hampshire, England) at 37°C overnight with an agitation speed of 180 rpm. Bacterial DNA was harvested with a TIANamp Bacterial DNA kit (TianGen, Beijing, China). The purity and concentration of the extracted DNA were determined using a NanoDrop ND-1000 (Thermo Scientific, Wilmington, DE, USA), and DNA samples were stored at −20°C until use.

To establish a sequence- and PCR-based Salmonella serotyping method for discrimination of S. Enteritidis, S. Pullorum/Gallinarum, and S. Dublin individually, the basic local alignment search tool (BLAST) algorithm (NCBI, non-redundant nucleotide collection) was applied. The tcpS, lygD, and flhB nucleotide sequences were used against the nucleotide collection (nr/nt) database, respectively. The aligned sequence number was set to the maximal value 20,000 with other parameters set to the default values. The three pairs of primers for the targets were designed using Primer Premier 5 (Premier, Palo Alto, CA, USA).

PCRs were conducted in a 25 μL reaction volume, consisting of 100 ng of genomic DNA template, 1 × polymerase buffer, 1 U of Taq polymerase (Takara Biotechnology Co., Dalian, China), 200 μM of deoxynucleoside triphosphate, and 80 nM of tcpS/lygD/flhBinner primers. PCR were conducted with a T100 Thermal Cycler (Bio-Rad, Hercules, California, USA) as follows: 94°C for 5 min, 30 cycles of 94°C for 45 s, 55°C for 45 s, and 72°C for 1 min, followed by 72°C for 10 min. The amplified PCR products were analyzed on the 1% agarose gel in 1 × TAE buffer.

The specificity of the three pairs of primers in the multiplex PCR system was evaluated by detecting genomic DNA from 29 different Salmonella strains, which included 27 Salmonella serovars and eight non-Salmonella species (Table 1).

The genomic DNA of S. Enteritidis strain C50041, S. Pullorum strain S06004 and S. Dublin strain SL5928 were 10-fold continuously diluted from 58.5 ng/μL to 585 fg/μL, respectively. Each dilution (2 μL) was used in the multiplex PCR assay. This assay was to determine the minimum limit of DNA that can be detected by the multiplex PCR method.

S. Enteritidis strain C50041, S. Pullorum strain S06004, and S. Dublin strain SL5928 were cultured overnight and the bacterial concentration was evaluated by the plate count assay. The bacterial culture was washed with phosphate buffered saline (PBS) twice, 10-fold serially diluted to the concentrations 2 × 10⁷ to 2 × 10² CFU/mL, and boiled for 10 min to collect the genomic DNA. In the multiplex PCR method, each dilution (5 μL) was used as templets to determine the minimum cells of Salmonella that can be detected.

Additional clinical Salmonella strains with unknown serovars were obtained from three farms, one pig farm, one chicken farm, and one cattle farm in Yangzhou, China. The Salmonella isolates were collected from feces, floors and lairage, and identified as previously described methods (Cai et al., 2016; Li et al., 2016; Xiong et al., 2016). In brief, each sample was suspended in 50 mL buffered peptone water (Difco, BD, Sparks, MD, USA) and followed by incubation at 37°C for 16–18 h. This broth culture (0.1 mL) was subcultured in 10 mL of Rappaport–Vassiliadis enrichment broth (Difco, BD) at 42°C for 24 h. After incubation, the RV broth was streaked onto xylose lysine tergitol 4 (Difco, BD), and cultured at 37°C for 24–48 h. The presumptive Salmonella colonies were picked from all plates and followed by biochemically confirmation using an API-20E test kit (bioMérieux, Marcy l'Etoile, France).

The multiplex PCR method was applied to detect the genomic DNA of Salmonella isolates from one pig farm (24 Salmonella isolates), one chicken farm (24 Salmonella isolates), and one cattle farm (11 Salmonella isolates). The obtained results of the developed PCR method were compared with traditional Salmonella serotyping approach.

The traditional serotyping for all isolated Salmonella strains from the pig, chicken and cattle farms were conducted by slide agglutination assay using the specific antisera (Tianrun Bio-Pharmaceutical, Ningbo, China) following the White-Kauffmann-LeMinor instructions (Grimont and Weill, 2007).

In silico analysis showed that tcpS exists only in S. Enteritidis, S. Pullorum/Gallinarum, and S. Dublin, and has 98–100% sequence similarity across the three Salmonella serovars in the database. Although tcpS in one E. coli strain showed 89% DNA sequence identity to the Salmonella tcpS, it does not contain the C-terminal region of tcpS or a match to the tcpS-R primer site (data not shown). lygD gene exists only in S. Enteritidis, and shares 98–100% sequence similarity among this serovar in the database (data not shown). Our previous study showed that flhB gene of S. Pullorum/Gallinarum lacks a unique region flhBinner compared with other serovars, and could be used to identify S. Pullorum/Gallinarum (Xiong et al., 2016). Therefore, three pairs of oligonucleotide primers distinguishing three specific Salmonella serovars were designed based on the three targets tcpS, lygD and flhBinner (Table 2).

The specificity of the multiplex PCR method was determined by detecting 29 Salmonella strains and eight non-Salmonella species. The results showed that S. Enteritidis generated three specific bands for tcpS, lygD and flhBinner, and S. Dublin generated two specific bands for tcpS and flhBinner, while S. Pullorum/Gallinarum generated only one specific band for tcpS. In contrast, only one band of flhBinner was generated in the other 23 Salmonella serovars, and no amplification product was observed in all non-Salmonella pathogens (Figure 1).

To evaluate the sensitivity of the multiplex PCR method, genomic DNA of S. Enteritidis, Pullorum, and Dublin were consecutively diluted and used as templates. The targeted fragments could be amplified at the lowest concentration of 58.5 pg/μL (Figure 2A), suggesting 58.5 pg/μL of genomic DNA was needed to detect S. Enteritidis, Pullorum, or Dublin using this method. Besides, the minimum cells of S. Enteritidis, Pullorum, and Dublin that could be detected using this multiplex PCR method were determined. After detecting various dilutions of Salmonella cells, we validated that the least cells that could be detected was 100 CFU (Figure 2B).

To determine the effectiveness of the developed multiplex PCR method, samples from one pig farm (24 Salmonella isolates), one chicken farm (24 Salmonella isolates), and one cattle farm (11 Salmonella isolates) were examined. The PCR results showed that three isolates from the pig farm generated three specific bands of tcpS, lygD, and flhBinner, suggesting that the three isolates were S. Enteritidis. Five samples from the chicken farm generated three specific bands of tcpS, lygD, and flhBinner, and 11 samples generated only one specific band of tcpS, suggesting that the five isolates and the other 11 isolates were S. Enteritidis and S. Pullorum/Gallinarum, respectively. Among the isolates from the cattle farm, only one sample generated two specific bands of tcpS and flhBinner, suggesting that this isolate were S. Dublin (Table 3).

The serotypes of Salmonella isolates from the three farms were identified using slide agglutination assays using specific O and H antisera. The isolates from the pig farm were obtained from lairage, floors, and feces, and included three strains of S. Enteritidis, nine strains of S. Derby, five strains of S. Typhimurium, two strains of S. London, and five strains of S. Rissen. The isolates from the chicken farm were obtained from fecal samples and floors, and included 11 strains of S. Pullorum, five strain of S. Enteritidis, five strains of S. Indiana, and three strains of S. Thompson. The isolates from the cattle farm were obtained from fecal samples, and included one strain of S. Dublin, eight strains of S. Newlands, and two strains of S. Muenster. The traditional serotyping results showed complete concordance with the developed multiplex PCR methods for all samples (Table 3).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.