Novel multiplex polymerase chain reaction and an oligonucleotide array for specific detection of the dominant foodborne bacterial pathogens in chicken meat

Oligonucleotide array hybridisation and multiplex polymerase chain reaction (m-PCR) can be used to screen and detect multiple foodborne pathogens. In our study, m-PCR and oligonucleotide array assays for the specific detection of the dominant foodborne bacterial pathogens, including Escherichia coli, Listeria monocytogenes, Salmonella spp., and Shigella spp., in chicken meat were developed. The combination of m-PCR and an oligonucleotide array targeting the 16S rRNA, uspA, prfA, fimY, and ipaH genes displayed a high discriminatory power among the aforementioned genera and species with low or no incidence of false negative results. Our combined methods could detect all 4 target bacteria at amounts as low as 1 ng of each from mixed genomic DNA extracted from pure cultures, which is equivalent to 10 4 -10 6 CFU/ml. After enrichment steps for the target bacteria, E. coli, L. monocytogenes, and Salmonella sp. could be detected simultaneously from fresh chicken samples. Combining the two methods could enhance accuracy and sensitivity for foodborne pathogen detection and identification. The problems of cross-reactivities from non-target bacteria isolated from an enrichment culture and the difficulties in result interpretation by m-PCR could be solved using our oligonucleotide array hybridisation method.

The most common tools of standard methods used for pathogen detection are cultural based method, immunelogical based method, and molecular based methods (United States Food and Drug Administration, 1998;Lazcka et al., 2007).Classical cultural methods including step of pre-enrichment and isolation of presumptive colonies of bacteria on solid media, and final confirmation by biochemical and/or serological identification have been applied to detect foodborne pathogens (United States Food and Drug Administration, 1998;Boera and Beumer, 1999;Lazcka et al. 2007).Conventional methods for detecting enteropathogens are very laborious and time consuming.To overcome these limitations, multiplex polymerase chain reaction (m-PCR), real-time PCR, and oligonucleotide arrays have been applied to detect multiple pathogens simultaneously (Yoo et al., 2004;Huang et al., 2007;Mao et al., 2008;You et al., 2008;Severgnini et al., 2011).
m-PCR is a reaction that amplifies more than one target gene simultaneously by mixing multiple primer pairs.m-PCR-based methods have been widely used and adapted for the rapid detection of single and multiple bacterial species, for example, E. coli, Salmonella spp., Shigella spp., and L. monocytogenes (Yeh et al., 2002;Li and Mustapha 2004;Thiem et al., 2004;Jofré et al., 2005;Li et al., 2005;Germini et al., 2009).Although m-PCR can amplify multiple targets in a single tube, its detection capability is still restricted to only a few targets per assay due to the complexity of the amplification (Wang et al., 2007;Settanni and Corsetti, 2007).For these reasons, typically only 2 (Jofré et al., 2005) or 3 (Li and Mustapha 2004;Li et al., 2005) bacterial species are simultaneously detected using m-PCR.These different m-PCR amplicons could be differentiated by real-time PCR with a high efficiency (Huang et al., 2007).However, real-time PCR requires special and expensive equipment, specific fluorescent probes, fluorescent detectors to detect several m-PCR products and expensive reagents (Nugen and Baeumner, 2008;Bai et al., 2010;Suo et al., 2010;Hu et al., 2012).Therefore, simple methods are required to improve the sensitivity and accuracy of m-PCR.An essential feature of the DNA array technique is the hybridisation of the labelled target DNA fragments with the array's immobilised probes.It can then be applied for multiple pathogens and microbial community detection in food samples (Gauthier and Blais, 2003;Cremonesi et al., 2009).Nucleic acid hybridisation occurs between the target DNA from the target organisms and DNA probes of approximately 15-30 nucleotides on the array (Boera and Beumer, 1999).The signal generated by the bound and labelled target on the array allows for identifications based on the known locations of the probes (Rasooly and Herold, 2008).
Among many pathogenic bacteria, consensus sequences can be amplified using a single pair of universal primers (Hong et al., 2004;Chiang et al., 2006;Hu et al., 2012).However, the limitation of using consensus sequences is a cross-reactivity with some other closely related bacteria, such as the cross-reactivity between Salmonella spp.and E. coli when the 23S rRNA gene is used as the target (Hong et al., 2004) or between E. coli and Shigella spp.(Chiang et al., 2006;Hu et al., 2012) when the 16S rRNA or groEL genes are used as the targets.Therefore, combinations of m-PCR amplification of species-and genus-specific genes with a DNA microarray were used in this study.Previously, these combined methods have been applied for multiple pathogen detection in meat product samples (Suo et al., 2010) and clinical samples (Kim et al., 2010) using fluorescent signal detection.Several laboratories have also addressed the development of simple and specific methods with minimal instrumentation requirements (Hong et al., 2011).
In this study, a low-density pathogen detection method using a m-PCR-oligonucleotide array to simultaneously detect 3 foodborne pathogens, including Shigella, Salmonella, L. monocytogenes, and 1 microbial food safety indicator, E. coli, which are frequently found in fresh chicken meat were developed and evaluated.Digoxigenin (DIG) was used to label the DNA.No special equipment was required for the material array construction or for signal detection.The m-PCR products for the 16S rRNA, uspA, prfA, fimY, and ipaH genes were distinguished from each other by DIG post-PCR labelling and hybridised to the oligonucleotide array.The applicability of this assay to fresh chicken samples was also addressed.

Bacterial strains
The reference and isolated bacterial strains used to validate the m-PCR and oligonucleotide array probe detection are listed in Table 1.All isolated strains were identified as described by the United States Food and Drug Administration -Bacteriological Analytical Manual (United States Food and Drug Administration, 1998).All target bacteria except for Clostridium perfringens were grown on trypticase soy agar (TSA), composed of tryptone (15 g/l), proteose peptone (5 g/l), sodium chloride (15 g/l), and agar (15 g/l), at 37°C for 24-48 h.The cultivation of C. perfringens was performed on tryptose sulphite cycloserine agar (TSC; Biomark, Pune, India) under anaerobic conditions at 37°C for 24 h.

Primer and probe design
To obtain the consensus sequence of each pathogen, the sequences were downloaded from the National Center for Biotechnology Information (NCBI) database and aligned using MegAlign DNAStar Lasergene 7 (DNASTAR Inc., Madison, Wisconsin, USA).Specific genes and 16S rDNA primers (Table 2) were designed using PrimerSelect DNAStar Lasergene 7 (DNASTAR Inc., Madison, Wisconsin, USA) based on the conserved regions of each specific gene and the conserved regions of all the target bacteria, which contained variable regions in the amplicons.All the primers in Table 2 were tested for their specificity with the reference and isolated bacterial strains (Table 1).For the oligonucleotide array, probes specific for each pathogen (Table 3) were designed based on the variable regions of the 16S rDNA and the conserved regions of each target gene using the PICKY oligonucleotide design program (Chou et al., 2004).

Target gene amplification by m-PCR
Genomic DNA (gDNA) from cultures grown on TSA or TSC (Biomark) for 16-24 h was extracted using a phenol-chloroformbased method (Liu et al., 2011).The concentrations and purity of the gDNA and m-PCR products were detected by measuring the absorbance at 260 and 280 nm using a NanoDrop Spectrophotometer ND-1000 (NanoDrop Technologies, Wilmington, DE, USA).Genomic DNA was used as a template for the target gene amplification by m-PCR.The reactions were performed in a total volume of 25 μl and contained 1× GoTaq Flexi buffer (Promega, Madison, WI, USA), 1 mM MgCl2, 0.2 mM dNTPs (Promega), 0.5 U GoTaq Flexi DNA polymerase (Promega), 100 ng DNA templates and primers.In all the m-PCR reactions, the amplified 16S rRNA gene was used as control.The concentrations of each primer pair and the annealing temperature were optimised.The PCR reactions were maintained at 95°C for 3 min and then 35 cycles of 95°C for 30 s, 50-59°C for 45 s, and 72°C for 60 s were performed followed by a final extension step at 72°C for 5 min.The m-PCR products were analysed by electrophoresis on a 4% (w/v) agarose gel and purified using a QIAquick PCR Purification kit (Qiagen, GmbH, Hilden Germany).

Oligonucleotide array preparation and detection
Nylon membrane (Roche, Mannheim, Germany) was used as the array matrix.Single stranded probes were heated at 95°C for 5 min, and 200 pmol was spotted at a specific position on a dry nylon membrane (Figure 2A).The membranes spotted with the probes were exposed to UV for 3 min to allow for cross-linking.Two hundred nanograms of purified m-PCR product was denatured at 99°C for 10 min and quickly chilled on ice.The denatured DNA was labelled with 2 µl of DIG High Prime (Roche) followed the manufacture protocol for 1 h at 37°C.Membranes with spotted probes were pre-hybridised in a pre-warmed DIG Easy hybridisation solution (Roche) at 35°C with gentle shaking for 30 min.Ten microlitres of the labelled PCR product reaction was heated to 99°C for 5 min, then immediately cooled on ice and added to 2 ml of the newly pre-warmed hybridisation solution.The hybridisations were performed with gentle rotation at 35°C for 4 h.After hybridisation, the membranes were washed twice for 5 min each in 2× SSC (Roche) and 0.1% sodium dodecyl sulfate (SDS) (25°C), twice for 10 min each in 0.5× SSC (Roche) and 0.1% SDS (45°C) and briefly washed in washing solution (Roche) at room temperature.Then, the membranes were incubated for 30 min in blocking solution (Roche) and 30 min in antibody solution (Roche).After 2 washes in a washing solution (Roche) for 15 min each, the membranes were equilibrated in detection buffer (Roche) for 2 min and in a freshly prepared NBT/BCIP (Roche) colour substrate solution in the dark for 4 h.The results were visualised and photographed.

Application of the oligonucleotide arrays
Four fresh chicken meat samples including 2 breasts (Cb1 and Cb2), 1 wing (Cw3), and 1 thigh (Ct4), were divided into 2 portions and used as natural samples (non-bacteria spiked sample) and target bacteria spiked samples.For spiked samples, a 10-fold dilution series of each bacterial culture including Salmonella serotype enteritidis (S. enteritidis) JCM 1652, L. monocytogenes DSM 12464, and Shigella boydii DMST 28180 were prepared using 0.85% sodium chloride solution.One hundred microliters of each cell dilution solution was spread onto TSA plates for viable cell count.At the same time, 25 g of each divided portion of each meat sample was placed in a stomacher bag and spiked with 100 µl cell dilution solution (ranging from 1-200 CFU) of each target bacteria.Sample Cb1_1 and Cb2_1 were chicken breast sample 1 (Cb1) and sample 2 (Cb2) (25 g each) spiked with L. monocytogenes 1 CFU, S. boydii 1 CFU and S. enteritidis 20 CFU, respectively.Samples One hundred microlitres of BPW culture was transferred to 10 ml Rappaport-Vassiliadis broth (RV; Himedia, Mumbai, India) and 10 ml tetrathionate (TT) broth (Himedia), followed by incubation at 42°C for 24 h for Salmonella detection.For L. monocytogenes detection, 100 μl of HF culture was transferred to 10 ml Fraser broth (OXIOD) and incubated at 37°C for 24 h.After 24 or 48 h incubation, an aliquot of each enrichment culture from each sample was subjected to the conventional analyses and oligonucleotide array assay.
The accuracy of m-PCR-oligonucleotide array assay was evaluated and compared with the conventional analysis.The cultures of Shigella broths were streaked on MacConkey agar (Himedia) for the conventional analysis of S. boydii detection.For E. coli detection, the BPW cultures were streaked on Eosin-Methylene Blue agar (EMB; Himedia).For Salmonella detection, the cultures of RV and TT broth were streaked on xylose lysine deoxycholate (XLD) agar (OXIOD) and bismuth sulphite (BS) agar (OXIOD).L. monocytogenes was detected by streaking the Fraser culture on PALCAM agar (OXIOD).The inoculations of the target bacteria on selective agar were incubated at 37°C for 24 h for E. coli, Shigella, and Salmonella detection and 48 h for L. monocytogenes detection.The suspected colonies of each target bacterium on the selective agar were re-streaked.Single colonies were picked and mixed in 20 µl water, heated at 100°C for 10 min and 1 µl of supernatant was used directly as templates in the m-PCR reactions for bacterial colony confirmation.
For the oligonucleotide array assay, 1 ml of BPW, RV, TT, Shigella, and Fraser culture were separately collected.Cell pellets were harvested by centrifugation and washed once in 0.85% sodium chloride solution, and gDNA was extracted using a phenolchloroform-based method (Liu et al., 2011).The gDNA pellet was dissolved in 50 µl TE, pH 8.An equal volume of the gDNA solution obtained from each enrichment culture was mixed, and 1 µl of the gDNA mixture was used as the template for the m-PCR amplification.For L. monocytogenes detection, 1 µl of the gDNA extracted from the Fraser culture was used separately as a template.Ten microlitres of the m-PCR products from the mixed enrichment culture and Fraser culture were individually labelled and applied to separate oligonucleotide arrays.The hybridisation patterns of both arrays were combined for the 4 target bacteria detected for each sample.

Optimisation of m-PCR
The specificities of the fimY, invA, ipaH, prfA, uspA, and virA genes (Table 2) were tested using the gDNA templates extracted from the pure cultures of E. coli, L. monocytogenes, Salmonella spp., Shigella spp., and the non-target bacteria (Table 1).The fimY, ipaH, prfA, and uspA were suitable target genes for detection of Salmonella spp., Shigella spp., L. monocytogenes, and E. coli because of the specificity and ability of amplification in the m-PCR reaction.The optimum annealing temperature was 52°C and the optimum concentrations of the primers in the m-PCR reaction were 0.02 µM ipaH, 0.036 µM fimY, 0.06 µM uspA, 0.12 µM prfA, and 0.4 µM 16S rRNA.The 16S rRNA gene amplified from all the target bacteria was used as a control for the presence of amplifiable bacterial DNA in the m-PCR amplification.Using m-PCR amplification, only the 16S rRNA gene product was detected from the non-target bacteria (data not shown).The expected PCR products of 884, 489, 422, and 398 bp were detected from the specific amplification of the reference and isolated strains of E. coli, Salmonella spp., Shigella spp., and L. monocytogenes, respectively (Figure 1, lanes 2-5).The amplification of the uspA gene fragment, which encodes for a highly conserved universal stress protein present in all E. coli (Chen and Griffiths, 1998), was also detected from the Shigella spp.This gene could be amplified not only from E. coli but also from all 4 Shigella species due to the high identity of the genes between E. coli and Shigella (Chen, 2007).However, Shigella can be differentiated from E. coli by the presence of the ipaH gene product.These results demonstrated that the specific detection of E. coli, Salmonella spp., L. monocytogenes, and Shigella spp.could be performed using the m-PCR developed during this investigation.However, our results indicated that the separation of all 5 amplicons on an agarose gel by electrophoresis was less sensitive and not sufficient (Figure 1, lanes 7-9).Therefore, oligonucleotide array was used to solve the problem of m-PCR result interpretation.

Probe validation and specificity testing
The target genes used for probe design were the 16S rRNA genes and genus-or species-specific genes included fimY, ipaH, prfA, and uspA genes.Our preliminary results indicated that the detection of E. coli and Salmonella using the probes targeted to the 16S rRNA genes resulted in some cross-reactivity with the non-Salmonella and non-E.coli bacteria from the enrichment culture (data not shown).Moreover, E. coli and the Shigella spp.could not be differentiated using the 16S rDNA probes (Figure 2B).A reliable genus-or species- specific gene was required for differentiation between Shigella and E. coli.To detect multiple target bacteria using a combination of m-PCR and an oligonucleotide array, oligonucleotide array probes specific for each gene and that would bind within the amplicon were designed (Table 3).DNA amplified from the bacterial strains listed in Table 1 was employed to evaluate the performance of the assay.After hybridisation, the signals on the array were unambiguously distinguished (Figure 2B).Crossreactivities of the m-PCR products from Shigella with the E. coli probes (UA and EC probes) were found for all 4 species of Shigella (Figure 2B).But Shigella can be differentiated from E. coli through a positive signal from the IH probes (Figure 2B).A mixture of gDNA from each target bacteria was also used as a mixed template for the detection of multiple target bacteria.The hybridisation patterns were determined to be accurate (Figure 2C).These results indicated that the developed oligonucleotide array could enhance the accuracy and simplicity of the resultant interpretation of the m-PCR detection.Using these techniques, the detection of the PCR products did not solely rely on the length of the PCR products but also required the fragments to contain sequences that were complementary to the oligonucleotide probes on the microarray (Kim et al., 2010).
In previous reports using DIG or biotin for the oligonucleotide array assay, only conserved genes, including the 16S rRNA (Chiang et al., 2006), 23S rRNA (Hong et al., 2004) and groEL genes (Hu et al., 2012), were used as targets.The detection of multiple pathogens was performed in pure culture, food samples, and foodborne infectious samples (Hong et al., 2004;Chiang et al., 2006;Hu et al., 2012).However, the problem of a low discriminatory ability among target and non-target bacteria was reported.Considering this problem, in our work primers and probes identifying the 4 target bacteria were also designed against genes specifically found in their respective pathogens to prevent false-positive and falsenegative results.

Sensitivity of the m-PCR-oligonucleotide array detection
The detection sensitivity of the assay was determined using a gDNA mixture extracted from S. enteritidis JCM 1652, E. coli TISTR 887, S. boydii DMST 28180 and L. monocytogenes DSM 12464.A 10-fold dilution series of gDNA mixtures ranging from 10-0.001 ng were used as templates for m-PCR amplifications.Ten microlitres of the m-PCR products was labelled with 2 µl of DIG High Prime (Roche) followed by hybridisation with the specific probes.The detectability of the 4 target bacteria from pure cultures by our assay was 1 ng of each gDNA (Figure 3), which corresponds to approximately 2 × 10 5 copies of the bacterial genome and was equivalent to 10 4 CFU/ml S. boydii, 10 5 CFU/ml S. enteritidis and E. coli, and 10 6 CFU/ml L. monocytogenes.The m-PCR products amplified from the mixture of templates (1 ng of each gDNA) were not sufficiently separated, and all the target gene products could not be observed on an agarose gel (data not shown).Thus, the m-PCR method followed by a hybridisation of the labelled products to the oligonucleo-tide array could improve the detectability.Although our detection limit level was less sensitive than that of the microarray using fluorescence detection, as reported by others (Kim et al., 2010;Suo et al., 2010), our system is still simpler and does not require any expensive or special equipment for microarray construction and fluorescent signal detection.

Application of the oligonucleotide array
The application of oligonucleotide array was tested with a total of 4 unspiked and 6 spiked fresh chicken samples (Table 4).In raw meat, pathogens are often present at low concentration (1-2 cells/25 g food) in a relatively high background of microbiota (Suo et al., 2010).Therefore, enrichment steps are very important to increase the target bacterial cells in samples.Detection of L. monocytogenes in BPW is poor due to the significant growth of Salmonella (Jofré et al., 2005).Therefore, pre-enrichment and enrichment steps specific for each target bacteria were performed in our study.Performing an enrichment step on a suspect food sample adds time to the overall detection regime and precludes the ability to enumerate the original density of the target pathogen.However, enrichment is necessary and, of course, extremely common for target bacteria detection.
In food sample applications, the total gDNA extracted from enrichment cultures contains both the target and non-target bacteria of a high microbiota background.The presence of these non-target DNAs may interfere with the amplification and/or hybridisation of the target DNAs and, hence, affect the detection sensitivity (Kim et al., 2010).Therefore, optimisation of each primer for the amplification of several target bacteria from food samples was necessary.The optimum primer concentrations for amplification of the multiple target bacteria in the fresh chicken samples using m-PCR were 0.032 µM ipaH and uspA, 0.036 µM fimY, 0.28 µM prfA, and 0.14 µM 16S rRNA.We also found that the efficiency of our assay for L. monocytogenes detection in samples with very low conta-mination levels decreased when all the gDNAs extracted from each enrichment culture were mixed and used as templates (data not shown).Therefore, only the gDNA extracted from the Fraser culture was used as a template for m-PCR amplification prior to the application of the oligonucleotide array.The results for the detection of multiple target bacteria using our protocol are summarised in Table 4.
Our protocol could simultaneously detect 3 target bacteria from the fresh chicken samples.All unspiked and spiked samples were found to be indigenously contaminated with Salmonella and E. coli, which could be detected using our methods and the conventional culture assay.An indigenous contamination of L. monocytogenes was found in only 1 of the unspiked samples (sample Cb2; Table 4).After the enrichment step using our combined methods, the sensitivity of L. monocytogenes detection in the fresh chicken samples was at least 10 CFU of initial contamination in 25 g samples.At this contamination level, positive hybridisation signals from the PA probes were detected while the PCR product for the prfA gene was not visible on agarose gels (data not shown).This result indicated that our oligonucleotide array could increase detectability compared to the PCR method.However, Shigella could not be detected from all the spiked samples using either our assay or conventional culturing.These problems might be due to the lower sensitivity of m-PCR amplification or the choice of the target genes (Ojha et al., 2013).In our preliminary investigation, the selected gene, ipaH, was specific for all 12 strains included reference and isolated strains of Shigella species (data not shown).This result indicated that the ipaH gene was suitable for specific detection of Shigella.Therefore, equal volumes of gDNA extracted from each enrichment culture were mixed and used as a template to individually amplify with each specific primer.An ipaH gene amplicon of the expected size and positive hybridisation signals from the IH probes were observed in the 5 spiked samples (Cb1_2, Cb2_1, Cb2_2, Cw3_1, Ct4_1), which contained an initial cell concentration of at least 1 CFU of S. boydii in a 25 g sample (data not shown).This result indicated that problems was due to the amplification of the target gDNA templates from the fresh chicken samples using m-PCR was less sensitive than using conventional PCR with a single primer pair.In m-PCR, a mixture of several primer sets might lead to a poor amplification efficiency (Chiang et al., 2006).Thus, to increase the specificity and sensitivity of the m-PCRoligonucleotide array for multiple pathogen detection, a determination of how many genes (that is, pathogens) can be used for the m-PCR in a single reaction without sacrificing the sensitivity of the hybridisation to the array is required (Kim et al., 2010).To avoid this problem in future studies, all target genes could be amplified from mixed gDNA templates using a separate pair of primers by conventional PCR.Each target amplicon could be labelled, mixed together and distinguished from each other on a single array.When comparing the conventional culture method to the array, 3 target bacteria could be detected from only 2 of the 6 spiked samples while the oligonucleotide array could detect 3 target bacteria simultaneously from 5 of the 6 spiked samples (Table 4).Thus, the detection of multiple foodborne pathogens using our assay was easier and had a higher accuracy compared to the conventional culture and PCR methods.However, sensitivity of our technique was not sufficient to detect 1 cell of L. monocytogenes and Shigella in 25 g sample.In sample contaminated with very low initial cell concentration, all the factors, including stressed environment in food, antibiotic selection, homogenisation, among others, could make the lag phase of cell growth longer.Therefore, detecting pathogens in food without enrichment or with inappropriate enrichment time and media might result in an underestimation or even a false-negative assessment of the pathogen contaminations in food (Suo et al., 2010).In our further works, optimisation of the enrichment steps of all target bacteria follow by PCR amplification and hybridisation will be tested to improve the sensitivity of simultaneous multiple pathogen detection in food.
In conclusion, oligonucleotide arrays and m-PCR can be successfully applied to detect multiple foodborne pathogens.To avoid cross amplification by m-PCR in food samples with a high bacterial background, a combination of m-PCR and oligonucleotide array hybridisation can be performed to specifically detect multiple target bacteria after enrichment steps.Although multiple pathogen detection using this protocol requires an additional 10-15 h for labelling, hybridisation and signal detection, compared with a conventional PCR method, the analysis time is still shorter and the protocol is simpler compared to traditional cultivation approaches.Our protocol is simple and has minimal instrumentation requirements, and, thus, a general molecular laboratory, especially in a developing country, is sufficient for performing this protocol.

Figure 2 .
Figure 2. Specific hybridisation patterns of the target bacteria.(A) Position of specific probes on the nylon membrane.Positive controls are 0.1 ng of DIG-labelled control DNA (pBR328 DNA, linearised with BamHI) (P) and 200 pmol 16S rDNA forward primer (16S).The abbreviated letters in the grids are the probe names shown in Table 3. (B) Specific hybridisation of individual m-PCR amplification products from each target bacteria with specific probes on the array.(C) Detection of multiple target bacteria using the m-PCRoligonucleotide array hybridisation-based method.

Figure 3 .
Figure 3. Sensitivity of the oligonucleotide array for the detection of multiple target bacteria.Genomic DNA extracted from each target bacteria were mixed at the same final concentration.A series of 10-fold dilutions of gDNA mixtures, ranging from 10-0.001 ng, from the 4 target bacteria were used as templates for m-PCR amplification followed by the oligonucleotide array hybridisation.

Table 1 .
Bacterial strains used for the validation of m-PCR and the oligonucleotide array.

Specie Number of strains Strain name and sources
Salmonella sp.; TT, non-Salmonella bacteria enriched using TT broth and isolated on XLD agar.c Strains isolated from food in Khon Kaen, Thailand: Sh, Shigella sp.

Table 2 .
Primers used for the target gene amplifications by m-PCR.

Table 3 .
Sequences of the probes spotted on the oligonucleotide array.