We developed multiplex PCRs targeted to the O, H1, and H2 alleles associated with four S. enterica serovars Enteritidis, Hadar, Heidelberg, and Typhimurium. Specific PCR primers to identify specific Salmonella serogroups, H1 and H2 alleles were designed based on the divergence of the glycosyl synthase genes, the unique linkage between two genes for a specific O-antigen of Salmonella, or allele-specific sequences within the hypervariable region of H1 and H2 antigen genes. In the primer design, a unique amplicon size was selected in order to facilitate development of a multiplex PCR (Table 1, Fig. 1 &2). The ability of the multiplex PCR to correctly identify serogroups (Fig. 1) was evaluated for 239 Salmonella isolates representing forty-three different serotypes which belonged to one of the six major serogroups, A, B, C1, C2, D1 and E1. With the exception of serogroups A and D1, which produce the same size amplicons (Kappa = 0.98), the multiplex PCR accurately distinguished salmonellae belonging to serogroups B, C1, C2, and E1 (Kappa = 1.00) (Table 2). The inability to distinguish serogroups A and D1 is due to the high degree of nucleotide sequence homology between the prt (paratose synthase) genes 20. The fliC multiplex PCRs successfully detected the H1, i, r, or z₁₀, alleles (Fig. 2A) and no amplicons were produced for serovars with other H1, flagellins (Kappa = 1.00) (Table 2). However, the fliC g,m primer set produced the same size amplicon only for salmonellae that possessed both the g and m, or g alone, or either epitope, g or m, in combination with other serotype-specific epitopes, or non-motile salmonellae that possess the fliC g,m allele 21 and therefore it did not have the specificity of the other H1 primer sets (Kappa = 0.58 vs. 1.00) (Table 2). To complement our PCR-based H allelotyping, a fljB multiplex PCR was designed to detect the H2 antigen alleles by targeting conserved regions within fljB alleles encoding the antigen complexes I: 1,2; 1,5; 1,6; 1,7 or II: e,n,x; e,n,z₁₅ and producing unique size amplicons (Table 1, Fig. 2B). The expected size amplicons were produced for only those S. enterica serovars belonging to H2 antigen complexes I: 1,2; 1,5; 1,6: 1,7 and. II: e,n,x; e,n,z₁₅ (Fig. 2B). The H2 multiplex PCR however could not distinguish H2 1,2 allele (Kappa = 0.75) or e,n,x (Kappa = 0.54) among the different H2 alleles within each antigen complex; for example indistinguishable amplicons were produced for Salmonella isolates bearing 1,2 vs 1,5; 1,6; or 1,7 (Table 2).

Validation of the allelotyping method is important for its integration with conventional Salmonella culture and typing methods used in diagnostic and food microbiology 22232425. We therefore assessed the allelotyping multiplex PCR against the standard conventional Salmonella serotyping method in identifying Salmonella O, H1 and H2 antigens for 43 different serovars of salmonellae isolated mainly from chicken carcasses and poultry environments (Tables 2 and 3).

The allelotyping PCR scheme for identifying S. enterica serovars Enteritidis, Hadar, Heidelberg and Typhimurium is envisioned to work as follows. An initial multiplex PCR is performed to determine which O antigen allele that an isolate possesses and a serogroup designation is given or unknown, based on PCR results. If the isolate possesses O alleles for serogroups B, C2, or A/D1, then a 2^nd allelotyping PCR is done to determine the presence of H2 alleles: i; g,m; r; or z₁₀. Based on the results of this 2^nd allelotyping PCR, an H1 allele type can be given an isolate as either being i; g,m; r; z₁₀ or unknown, if no amplicons with the expected size for the H1 allelotyping PCR are produced. If both O and H1 allelotyping PCR detects O and H1 alleles associated with S. enterica serovars Enteritidis, Hadar, Heidelberg, and Typhimurium, then a 3^rd final H2 allelotyping PCR is performed to further differentiate the isolate to serovar level. Therefore, identifying one allele for each O, H1, and H2 allelotyping PCR, as listed in Table 3; it is possible to discern the serovar for isolates typed using this PCR-based scheme. For example, identification of serogroup B, H1 i, and H2 I antigen complex by multiplex PCR presumptively identifies the isolate as S. enterica serovar Typhimurium (Sensitivity = 1.00; Specificity = 1.00) (Table 3). The expansion of O-antigen PCR to detect serogroups C1 and E1, affords a laboratory the opportunity to detect other S. enterica serovars, as the antigenic formula for O, H1 and H2 antigens defines the serovar. Therefore, we were able to identify additional S. enterica serovars with our multiplex PCRs including Haifa [B; z₁₀; 1,2], Infantis [C1; r; 1,5], and Mbandaka [C1; z₁₀, e,n,z₁₅]. We can also identify monophasic S. enterica serovars (ex. Montevideo: [C1; g,m,s; -]) by including a generic Salmonella fljB (H2) PCR test 14. Isolates negative for O, H1, and H2 alleles by our multiplex PCR screen would need to be characterized by traditional serotyping, RFLP PCR 14, or PCR screens that identify the other H1 and H2 alleles 151622. The limitations with our multiplex PCR are that it cannot distinguish among serogroup/serovar variants that arise due to phage conversion and the resulting chemical/antigenic alteration of the somatic O antigen 8 (ex. Hadar vs. Istanbul), or subtle point mutations in H2 antigen gene, fljB responsible for loss of flagellar expression observed in some S. enterica serovar Typhimurium strains 26. Our multiplex PCRs were designed to be used as a rapid screen for S. enterica serovars: Enteritidis, Hadar, Heidelberg, and Typhimurium, targeting key genes/alleles that differentiate these serovars from the rest. As a diagnostic test, our allelotyping PCR was also designed to minimize the cost of this test to a few individual PCR tests, with a minimum number of primers needed for this typing scheme. Unfortunately, our H2 multiplex PCR cannot discern differences within the H2 antigen complexes (Table 2) to make a definitive serovar designation for S. enterica serovars with the same O and H1 antigens as our target serovars (S. enterica serovars: Typhimurium [H2: 1,2] vs. Lagos [H2: 1,5]; Heidelberg [H2: 1,2] vs. Bradford [H2: 1,5], Winneba [H2: 1,6] or Remo [H2: 1,7]; or Hadar [H2: e,n,x] vs. Glostrup [H2: e,n,z15]). Also, the allelotyping primers for H1 g,m allele identifies those H1 alleles bearing g or m in any possible combination (Table 2), therefore H1 multiplex would not be able to discern serogroup D1, S. enterica serovars Enteritidis [H1: g,m] from Blegdam [g,m,q]. While the possibility of encountering these alternate serovars may be remote based on epidemiological data 89, it is still a possibility, and where a reporting laboratory may require confirmatory testing there are additional PCR based tests that can discern these allelic differences to make a final, definitive serovar designation possible 1516. Alternatively, the H2 amplicons can be sequenced to definitively identify the H2 allele. Although several multiplex PCRs have been developed to assist laboratories in identification of S. enterica serovars 15161722, our results are the first to focus on, validate and describe a PCR-based scheme for assisting diagnostic labs in differentiating S. enterica serovars: Enteritidis, Hadar, Heidelberg, and Typhimurium.

The S. enterica isolates used in this study were from multiple animal species, including human, poultry, livestock and wildlife 27282930, and serotyped by the National Veterinary Service Laboratory (NVSL; Ames, IA) using classical methods (Table 2). The isolates were used to test the specificity of PCRs specific for O, H1 and H2 alleles described in Table 1. Additional Salmonella isolates of unknown serovars were obtained from two poultry farms in northeast Georgia 2531 as well as salmonellae isolated from routine submissions to the Poultry Diagnostic and Research Center (PDRC) in Athens, GA.

We sampled the commercial chicken broiler house environment and chicken carcasses for Salmonella as previously described 31. The processing, enrichment, isolation and final diagnostic confirmation of Salmonella from samples is described in detail elsewhere 31. Serotyping was done using standard serological typing procedures for Salmonella O, H1 and H2 antigens 32.

From comparative analysis of the wba operon for Salmonella serogroups A/D1, B, C1, C2, and E1 203334353637 we identified serogroup-specific gene(s) (National Center for Biotechnology Information (NCBI) Accession #: M29682, X56793, X61917, M84642, X60665) for PCR primer design. Similarly, we identified from an alignment within the central variable region 113839 of fliC (H1) and fljB (H2) alleles (NCBI accession #: D13689, M84974, AF15949, AF332601, U06199, U06206, U06225, U06197, M84973, Z15086, D78639, Z15071, Z15072, Z15069, U06205, U06204, AF420426, AF420425, AF045151, U17175, U17171, U17172, AF425736, AF425737), using the DNA analysis software AlignPlus^® version 3.0 (Scientific and Educational Software), candidate sequences to differentiate Salmonella with the H1 flagellin antigens i, g,m, r, z₁₀, and the H2 flagellin antigen complexes 1,2, 1,5, 1,6, 1,7 and e,n,x, e,n,z₁₅ alleles. We analyzed these gene sequences, using the GeneRunner^® (Hastings software; Hastings, NY) DNA analysis software, and identified suitable primer sets that were compatible in a single multiplex PCR reaction and designed to produce an amplicon with size unique for the sequence(s) targeted by a specific primer set (Table 1).

The O-antigen multiplex PCR was designed to detect serogroup A/D1, B, C1, C2, or E1 specific genes or alleles (Table 1). The O-antigen multiplex PCR was performed using the Amplitron II Thermolyne thermocycler (Barnstead; Dubuque, IA), using HotStart PCR tubes (Molecular Bio-Products, Inc., San Diego, CA). Each reaction had a final concentration of 1.5 mM MgCl₂, 50 mM Tris, pH 8.3, 0.25 mg/ml bovine serum albumin, 0.5 μM primer, 0.2 mM deoxynucleotides (Boehringer Mannheim; Indianapolis, IN), 0.5 units of Taq DNA polymerase (Boehringer Mannheim), and 1 μl of whole cell template. The PCR was performed with pre-amplification heating as described by D'Aquilla et al. 40. The program parameters for PCR include an initial five minutes incubation at 85°C, to mix the two PCR reaction mixes, followed by 30 cycles of denaturation (94°C for 1 min), annealing (55°C for 1 min), and extension (72°C for 1 min). Amplicons were separated on 1.5% agarose gel with Tris-acetate-EDTA buffer 41 and ethidium bromide (0.2 μg/ml) at 100 V. The 100-bp ladder (GIBCO/BRL, Gaithersburg, MD) was used as a molecular weight (MW) standard for determining the MW of the PCR products. Various S. enterica serovars belonging to serogroups A/D1, B, C1, C2, E1 were used in the PCR to test the specificity of the primer sets.

The H1-1 multiplex PCR was used to identify isolates with antigens i or g, m; while the H1-2 multiplex PCR was designed to detect isolates with antigens r or z₁₀. Finally, the H2 multiplex PCR was created to differentiate isolates with either H2 antigen complexes 1,2; 1,5; 1,6; 1,7; or e,n,x; e,n,z₁₅. In order to identify the H1 and H2 alleles, capillary PCR reaction was performed to amplify the alleles of fliC and fljB by three multiplex PCRs with the Rapidycler™ hot-air thermocycler (Idaho Technologies; Idaho Falls, ID) 42 in 10-μl capacity capillary tubes. We sought to reduce the expense of reagents and reaction time by utilizing a capillary thermocycler that accommodates very low reaction volumes. The 10-μl PCR mix for the fliC i and g,m multiplex consisted of 2.0 mM MgCl₂, 50 mM Tris (pH 8.3), 0.25 mg/ml bovine serum albumin, 0.5 μM of each primer, 0.2 mM deoxynucleotides, 5% DMSO, 1.0 units of Taq DNA polymerase, and 1 μl whole cell template. For fliC r and z₁₀ multiplex, 3.0 mM MgCl₂ and 1.0 μM of each primer were used for each reaction. For the amplification of the H2 alleles, the fljB multiplex consisted of 3.75 mM MgCl₂, 62.5 mM Tris, pH 8.3, 0.31 mg/ml bovine serum albumin, 0.5 μM of each primer, 0.2 mM deoxynucleotides, 5% DMSO, 1.0 units of Taq DNA polymerase, and 1 μl whole cell template in a 10 μl volume. The program parameters for the hot-air thermocycler were an initial heating step of 94°C for 1 min; 94°C for 1 sec, 55°C for 1 sec, and 72°C for 20 sec with a slope of 2.0 for 40 cycles; and a final extension at 72°C for 4 min. Amplicons were detected as described above. The specificity of the PCR detection was tested against various Salmonella serovars possessing the relevant fliC and fljB alleles (Table 2). Escherichia coli LE392 served as a negative control. Whole cell template for all multiplex PCRs was prepared according to the procedures of Hilton et al. 43.

Kappa statistics were calculated to evaluate the agreement between the classical serotyping systems and multiplex PCR for each of the antigen groups examined. Sensitivity and specificity of the allelotyping PCR scheme relative to conventional serotyping was calculated for S. enterica serovars Enteritidis, Hadar, Heidelberg, and Typhimurium.