Emergence of Salmonella enterica serovar 4,[5],12:i:- as the primary serovar identified from swine clinical samples and development of a multiplex real-time PCR for improved Salmonella serovar-level identification

Abstract

Rapid identification of the infecting Salmonella serovar from porcine diagnostic samples is vital to allow implementation of appropriate on-farm treatment and management decisions. Although identification at the serogroup level can be rapidly achieved at most veterinary diagnostic laboratories, final Salmonella serovar identification often takes several weeks because of the limited number of reference laboratories performing the complex task of serotyping. Salmonella serogroup B, currently the dominant serogroup identified from swine clinical samples in the United States, contains serovars that vary from highly pathogenic to minimally pathogenic in swine. We determined the frequency of detection of individual group B serovars at the Iowa State Veterinary Diagnostic Laboratory from 2008 to 2017, and validated a multiplex real-time PCR (rtPCR) to distinguish pathogenic serogroup B serovars from those of lesser pathogenicity. Our results indicate that, since 2014, Salmonella enterica ssp. enterica serovar 4,[5],12:i:- has been the dominant serovar identified from swine clinical samples at the ISU-VDL, with S. Typhimurium now the second most common serovar identified. We developed a rtPCR to allow rapid differentiation of samples containing S. 4,[5],12:i:- and S. Typhimurium from samples containing serovars believed to be of less pathogenicity, such as S. Agona and S. Derby. When combined with enrichment culture, this rtPCR has the ability to significantly improve the time to final serovar identification of the 2 most commonly identified pathogenic Salmonella serovars in swine, and allows rapid implementation of serovar-specific intervention strategies.

Introduction

Non-typhoidal Salmonella enterica infections in swine can result in clinical disease (i.e., salmonellosis) or asymptomatic infections, with clinical disease most commonly observed in weaned, growing, and finishing pigs.²³ Although clinical disease typically occurs as fever with diarrhea, the clinical manifestation of Salmonella enterica infection is highly dependent on the infecting serovar and age of the pigs.^19,23 Subclinical infections present unique diagnostic interpretation challenges given that pigs are infected with Salmonella without any clinical signs of disease but will often shed the organism in their feces for an extended period of time.^8,45 Variation between pigs and inconsistencies in shedding patterns make it difficult to confidently confirm that an animal is truly negative for a current Salmonella infection,^9,31 which poses a risk of continued environmental contamination leading to exposure of cohorts. Hence, it is not surprising that a high prevalence of non-typhoidal Salmonella serovars has been documented among finishing pigs at slaughter.^3,29,42

Salmonella shedding around the time of slaughter, including transport and lairage, increases the risk of contamination of pork products with Salmonella. This is of concern given that Salmonella enterica is the most commonly reported cause of foodborne illness in humans.¹¹ There are significant differences in the contribution of each serovar to human disease risk; those that lack host restriction, such as S. Typhimurium, are of primary importance. S. Typhimurium has been widely recognized as a cause of foodborne illness in humans for many years, especially with regard to contaminated pork products.^13,30,44 More recently, the monophasic variant of S. Typhimurium, S. 4,[5],12:i:-, has been implicated as an increasingly common cause of human illnesses.^6,24,33,36

It is widely accepted that S. 4,[5],12:i:- is a variant of S. Typhimurium, based on molecular subtyping through pulsed-field gel electrophoresis, multilocus sequence typing, phage typing, and plasmid characterization.^2,27,28,41 Although S. Typhimurium expresses the “i” phase 1 flagellar antigen and the “1 and 2” phase 2 flagellar antigens, the 4,[5],12:i:- serovar lacks phase 2 flagellar expression.⁴¹ Although S. 4,[5],12:i:- is considered a S. Typhimurium variant, worldwide, it has been demonstrated that there are many clones originating from multiple independent events. Such events resulted in deletions or mutations in various genomic regions including: hin (which encodes a recombinase that regulates inversion of the fljAB promoter), fljB (which encodes the phase 2 flagellar proteins), or the fljAB promoter region.^37,41 Minimal data exists detailing the emergence of S. 4,[5],12:i:- in the United States. Of the 51 isolates recovered from swine in 2014–2016 that were evaluated via whole genome sequencing, 48 were part of a single emerging clade.¹⁷ However, that previous study was limited to samples submitted from the midwestern United States only, and it is unclear how many production units were represented in the dataset and thus whether the study was representative of the current population of S. 4,[5],12:i:- circulating in the United States.

In a review of case data from clinical submissions to the ISU-VDL during an 18-mo period of 2016–2017, a statistically significant positive association between histologic lesions consistent with enteric salmonellosis and isolation of S. 4,[5],12:i:- was noted.¹ A small-scale pathogenicity study utilizing a multi-drug resistant isolate from a human outbreak also reported the ability of S. 4,[5],12:i:- to cause disease in swine similar to that seen in previous studies utilizing S. Typhimurium in the same laboratory.⁴⁰ Such findings are consistent with an in vitro study that compared the ability of a S. 4,[5],12:i:- isolate and a S. Typhimurium isolate to adhere to and invade cultures of porcine intestinal epithelial cells, determining that there was no difference between the 2 serovars in their cytotoxicity, colonizing ability, or effect on pro-inflammatory chemokine release.¹⁴ Thus, without additional available data to the contrary, it seems reasonable to consider S. 4,[5],12:i:- to be of similar pathogenicity to S. Typhimurium in swine. However, although vaccines to aid in the prevention of disease caused by S. Typhimurium in swine are available commercially in the United States, there are no vaccines similarly labeled for control of S. 4,[5],12:i:-. Given that no research has been published to date on the potential for cross-protection of vaccination against S. Typhimurium on the incidence of S. 4,[5],12:i:- in swine operations to our knowledge, and although pathogenicity may be equivalent, successful prevention strategies may not be identical between the serovars. In addition, evaluation of antimicrobial resistance patterns of S. 4,[5],12:i:- isolates from humans has shown these isolates to be more highly resistant to several different classes of antibiotics compared to human isolates of S. Typhimurium, thus effective treatment of clinical disease in swine may differ between the serovars.¹¹

Both S. Typhimurium and S. 4,[5],12:i:- are part of Salmonella serogroup B. From the late-1990s through the mid-2000s, S. Typhimurium and S. Derby were reported as the group B serovars isolated most frequently from swine in the United States.²⁰ Interestingly, Salmonella 4,[5],12:i:- was rarely identified prior to the mid-1990s.⁴¹ This serovar first appeared in Europe, where it is now the third most frequently isolated serovar from human salmonellosis cases.¹⁸ In the United States, it was the fifth most frequently isolated serovar from human enteric salmonellosis in 2014, increasing in prevalence by 194% from 2005 to 2015.¹² From July 2006 through June 2015, the Minnesota Veterinary Diagnostic Laboratory reported the most frequent Salmonella serovars isolated from swine sample submissions (n = 2,537) as S. Typhimurium var. 5- (28.2%), S. Agona (14.7%), and S. Derby (12.1%).²⁶ The Minnesota study²⁶ as well as a national study incorporating both human and veterinary data⁴⁶ noted an increase in isolation of S. 4,[5],12:i:- similar to that initially reported by the Iowa State University Veterinary Diagnostic Laboratory (ISU-VDL; Ames, IA) in late 2016 (Krull A, et al. Increased frequency of isolation of multi-drug resistant Salmonella I 4,[5],12:i:- from swine with histologic lesions consistent with salmonellosis. Proc Am Assoc Vet Lab Diagn Ann Conf; Oct 2016; Greensboro, NC). According to the 2013 NARMS Retail Meat Interim Report, S. 4,[5],12:i:- was identified as one of the serovars most commonly isolated from retail pork in the United States.¹² It is worth noting that S. 4,[5],12:i:- has been detected in cattle^15,26,32 and poultry^12,15 in addition to swine.^5,26,43

Rapid detection and identification of Salmonella, particularly serovars such as S. Typhimurium and S. 4,[5],12:i:- that are believed to be pathogenic to both pigs and humans, is critical for preventing foodborne illness outbreaks. Implementation of suitable treatment and prevention protocols on farms is reliant on rapid detection and identification of pathogens, which subsequently aids in reducing potential pork contamination. To date, protocols for Salmonella isolation, identification, and serogrouping take 3–5 d on average.³⁴ However, serotyping at a reference laboratory can take 4–6 wk. The extended time to final identification at the serovar level limits the ability of veterinarians and producers to begin the most appropriate treatment and prevention protocols. Our objectives were: 1) to determine the frequency of detection of various group B serovars commonly identified from swine at the ISU-VDL from 2008 to 2017, and 2) to develop a multiplex real-time PCR (rtPCR) to rapidly detect and differentiate Salmonella serovars likely to be pathogenic in swine (i.e., S. Typhimurium and S. 4,[5],12:i:-) from those of lesser pathogenicity (e.g., S. Derby) from clinical specimens.

Materials and methods

Retrospective analysis of ISU-VDL data

The ISU-VDL receives >75,000 case submissions annually, ~75% of which are samples from all types of swine production systems throughout the United States. The ISU-VDL Laboratory Information Management System provided data for the analysis of the frequency of detection of specific serovars of Salmonella, using search criteria that included: 1) the period from January 1, 2008 through December 31, 2017; 2) all porcine cases from which a Salmonella species was isolated from a clinical specimen (primarily feces or tissues); 3) verification of the isolate as Salmonella and identification of the serovar by the National Veterinary Services Laboratories (NVSL); and 4) the results of antimicrobial susceptibility testing performed on the isolate. All submissions considered research cases or cases in which Salmonella isolate serotyping was unable to be performed were eliminated from analysis. When multiple Salmonella isolates were isolated from a single case, the standard laboratory protocol was to send only a single isolate to NVSL for serotyping, therefore, each isolate listed in our analysis represents a unique case submission to the ISU-VDL.

Culture, isolation, and DNA extraction of bacterial samples

All bacterial isolates used for validation were selected from clinical samples submitted to the ISU-VDL. Porcine samples submitted for routine enteric culture were plated onto 4 different agar plates at various atmospheric conditions for isolation of pathogens associated with enteric disease. These included: 1) trypticase soy agar (TSA) with 5% sheep blood (Remel Products, Lenexa, KS) incubated under 5% CO₂; 2) TSA with 5% sheep blood incubated anaerobically; 3) brilliant green agar with novobiocin (BGN; in-house) incubated aerobically; and 4) tergitol-7 agar (T7; Remel Products) incubated aerobically. All plates were incubated at 35°C for a minimum of 48 h. Colonies consistent with Salmonella were selected from the BGN and/or T7 plates, subcultured to TSA to obtain pure cultures, and saved in 10% glycerol stocks for long-term storage at −80°C for further use.

Upon request from a diagnostician with clinical suspicion of salmonellosis, a 24-h enrichment with either tetrathionate broth or buffered peptone water (BPW) was also included in select cases. For these samples, the enrichment culture was incubated at 42°C and then was subcultured onto BGN, which was incubated aerobically for 24 h.

Confirmation of suspect isolates was done via matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) following the manufacturer’s recommendations (Bruker Daltonics, Billerica, MA). A minimum MALDI-TOF MS confidence score of 2.10 was required for confirmatory identification. Confirmation of Salmonella identification was then followed by serogrouping via slide agglutination testing (BD Diagnostics, Sparks, MD; SSI Diagnostica, Herredsvejen, Denmark) at ISU-VDL and serotyping at NVSL.

Bacterial isolates from routine cases were saved at −80°C in brain-heart infusion (BHI) broth with 10% glycerin. Stored isolates selected for use in validation of the rtPCR were cultured onto TSA with 5% sheep blood agar and incubated at 35°C for 24 h to ensure purity prior to DNA extraction. DNA extraction of pure culture of bacterial isolates was achieved first by suspension of multiple colonies of the culture in phosphate-buffered saline (PBS). DNA extraction of the samples was then performed (RNA DNA pathogen extraction kit; Kingfisher rapid throughput DNA extraction system; Thermo Fisher Scientific, Rochester, MN) according to the manufacturer’s recommendations.

Development of rtPCR

The rtPCR assay was designed to detect the following genes or regions: invA (present in all Salmonella),²¹ fliA (present in both S. Typhimurium and S. 4,[5],12:i:-),^16,25 fljB (present in S. Typhimurium, may be absent in S. 4,[5],12:i:-),^7,28,29,41 the intergenic space between hin and iroB (present in S. Typhimurium, may be absent in S. 4,[5],12:i:-),^7,29,41 and an internal control (Table 1). To be considered S. 4,[5],12:i:-, one or both of either the fljB or hin-iroB targets must be absent (negative) to result in the monophasic variant. The primers and probes were designed and validated previously (Supplementary Table 1).^35,38 Controls for each rtPCR assay included an internal amplification control, 2 positive extraction controls, a negative extraction control, and a negative amplification control. The internal amplification control consisted of G-block gene fragments 125–500 bp (XIPC_IVT; Integrated DNA Technologies; 149551620) based on the GenBank sequence DQ883679, as described previously.³⁹ The positive extraction controls included one S. Typhimurium isolate and one S. 4,[5],12:i:- isolate, both at a concentration of 10⁴ CFU/mL suspended in Luria-Bertani broth. The negative extraction control was nuclease-free water (Life Technologies, Waltham, MA). The negative amplification control was the prepared rtPCR master mix.

Table 1.

Expected multiplex real-time PCR results for Salmonella enterica serovars Typhimurium and 4,[5],12:i:- compared to all other serovars based on presence or absence of target genes.

Gene target	Interpretation	Expected PCR result
		Salmonella 4,[5],12:i:-	Salmonella Typhimurium	Salmonella non-Typhimurium, non-4,[5],12:i:-
fliA	Present in both Typhimurium and 4,[5],12:i:- only	Positive	Positive	Negative
fljB †	Always present in Typhimurium, may be absent in 4,[5],12:i:- and other serotypes	Positive or negative*	Positive	Positive or negative
hin-iroB	Always present in Typhimurium, may be absent in 4,[5],12:i:- and other serotypes	Positive or negative*	Positive	Positive or negative
invA	Present in all Salmonella	Positive	Positive	Positive

^*S. 4,[5],12:i:- can be negative for either hin-iroB, fljB, or both hin-iroB and fljB.

^†If fljB has > 8 cycle threshold (Ct) difference from the fliA gene, then the isolate is likely S. 4,[5],12:i:-; if there is < 5 Ct difference, then the isolate is likely S. Typhimurium.

Real-time PCR assays were carried out in 25-µL reactions (QuantiTect virus + ROX kit; Qiagen, Waltham, MA) according to the manufacturer’s recommendations, with each target primer at a final concentration of 200 nM, each target probe at 100 µM, and XIPC primers and probe at final concentrations of 160 and 60 nM, respectively. A thermocycler system (Rotor Gene-Q 5-plex HRM; Qiagen) was used with the following cycling conditions: 20 min at 50°C, 5 min at 95°C, then 40 cycles of alternating 15 s at 95°C and 1 min at 60°C. The initial 20-min step was included to allow simultaneous testing with other reverse-transcription rtPCR assays run in the testing laboratory. The threshold for analysis was set at 0.02. The following interpretation criteria were used for all sample types. The negative cutoff for detection of any Salmonella DNA, regardless of serovar, present in the sample was maintained at a cycle threshold (Ct) value of 40, as is standard in the ISU-VDL; the sample was considered positive for the presence of any Salmonella DNA below this threshold. A sample was considered positive and able to be identified at the serovar level if the Ct was <30. If the Ct was 30–40, it was considered positive for Salmonella DNA but inconclusive for serovar identification. Therefore, further testing utilizing additional culture and/or enrichment followed by NVSL serotyping or repetition of rtPCR on pure culture was necessary for serovar identification in some samples.

Inclusivity and exclusivity

To assess the inclusivity of the rtPCR, 61 S. 4,[5],12:i:- isolates and 45 S. Typhimurium isolates were tested. To assess the exclusivity of the rtPCR, 38 isolates, representing 28 non-Typhimurium, non-4,[5],12:i:- serovars of Salmonella and 7 non-Salmonella organisms commonly found in feces, were tested (Supplementary Table 2). Salmonella serovars selected for exclusivity testing included the 10 serovars isolated most commonly from porcine samples at the ISU-VDL as well as serovars that are closely related to the serovar Typhimurium. All isolates tested were obtained from previous ISU-VDL case submissions using the culture techniques described above.

Limit of detection studies

Two limit of detection (LOD) studies were completed, one with and one without enrichment in BPW prior to DNA extraction and rtPCR. Enrichment in BPW has been shown in the ISU-VDL to be a superior method, relative to tetrathionate broth, for enriching of Salmonella sp. in clinical and environmental samples prior to rtPCR (Krull A, et al. Use of enrichment and quantitative PCR to improve detection of Salmonella in referral hospitals. Am Coll Vet Intern Med (ACVIM) Forum Research Report; June 2016; Denver, CO). To determine the non-enriched LOD, the following method was repeated with one S. Typhimurium and one S. 4,[5],12:i:- isolate each using Salmonella-negative feces (confirmed by enrichment culture) as the sample medium. To prepare the dilutions, a 0.5 McFarland standard (~1.5 × 10⁸ CFU/mL) was prepared from a pure culture and then serially diluted 1:10 in PBS. The actual amount of Salmonella in each dilution was determined using the standard plate count method on TSA with 5% sheep blood agar. Salmonella, in the form of 0.5 mL of the serially diluted McFarland standard, was added to 0.2 g of porcine feces to create 10-fold dilutions of Salmonella from 5 × 10⁷ to 5 × 10¹ CFU/mL. Each dilution was created in triplicate. Salmonella was also added to feces to create 5-fold dilutions of Salmonella from 1 × 10⁴ to 1 × 10¹ CFU/mL, again in triplicate. DNA from the fecal samples was extracted using the bead-beating method recommended in the Thermo Fisher total nucleic acid kit insert, using the RNA DNA pathogen extraction kit and 100 µm of acid-washed zirconium bead-filled tubes (OPS Diagnostics, Lebanon, NJ).

The post-enrichment LOD was determined by adding 250 µL of the fecal samples created for the non-enrichment LOD studies to 5 mL of BPW. The inoculated BPW was then incubated at 35°C for 18–24 h prior to DNA extraction and rtPCR. Post-enrichment DNA samples in BPW were extracted using the DNA extraction method described above for pure culture.

Sequencing of isolates with unexpected results

Salmonella isolates that did not react as expected on the rtPCR (i.e., results did not match NVSL serotyping) were sequenced in the region amplified by the fljB primer set to determine the cause of the unexpected result. Sanger sequencing was completed at the ISU DNA Facility (Ames, IA). The sequences were then assembled for further analysis using DNASTAR software (DNASTAR, Madison, WI). BLAST analysis (http://blast.ncbi.nlm.nih.gov/Blast.cgi) was subsequently used to compare the consensus sequences to known DNA sequences of S. Typhimurium and S. 4,[5],12:i:-.

Validation using clinical samples (retrospective and prospective)

Further validation of the rtPCR was done via retrospective and prospective analysis of clinical samples submitted to the ISU-VDL. For retrospective validation, porcine cases that met the following criteria were selected: 1) Salmonella was isolated from the sample submitted for culture, and 2) molecular testing was performed for identification of other disease entities on a sample from the same pig that had a Salmonella-positive culture. The use of samples that had molecular testing performed ensured that there was a DNA extract stored at −20°C from which further testing could be completed. The previously extracted DNA was tested via the Salmonella rtPCR as described above to compare to the results of culture.

For prospective validation, feces, colonic mucosal scrapings, and fecal swabs rinsed in 1 mL of PBS were collected from pigs that had gross lesions suggestive of salmonellosis upon postmortem examination. These samples were subsequently tested for the presence of Salmonella both directly and post-BPW enrichment using the rtPCR. Other tissues from the large intestine were also collected for testing that included standard culture for Salmonella. Cultured isolates from the standard testing were confirmed as Salmonella by the ISU-VDL and serotyped by the NVSL.

Results

Retrospective data analysis

A total of 10,194 isolates of Salmonella were confirmed from swine clinical cases at the ISU-VDL during 2008–2017. Of these, 3,476 of 10,194 (34%) of the isolates did not include the state of origin of the sample on the submission form. The remaining 6,718 of 10,194 (66%) of Salmonella isolates originated from farms in the following states: Iowa (45%); North Carolina (17%); Illinois (6%); Minnesota (5%); Arkansas, Kansas, Missouri, and Virginia (4% each); Nebraska and Pennsylvania (2% each); Arizona, California, Colorado, Hawaii, Idaho, Kentucky, Maryland, Massachusetts, Michigan, Montana, New Hampshire, New York, North Dakota, Ohio, Oklahoma, South Carolina, South Dakota, Texas, Utah, Wisconsin, and Wyoming (0.01–1% each).

From 2008 to 2017, the number of S. 4,[5],12:i:- isolates identified from clinical cases in swine by the ISU-VDL rapidly increased, from a total of only 26 isolates between January 1, 2008 through December 31, 2010, to 331 isolates identified in 2017 alone. During the same timeframe, isolation of S. Typhimurium decreased from 364 isolates in 2008 to 144 isolates in 2017. In 2008, from a total of 1,060 isolates, the 5 Salmonella serovars most commonly isolated from swine through the ISU-VDL were: Typhimurium (34%, serogroup B); Derby (13%, serogroup B); Choleraesuis (9%, serogroup C1); Agona (7%, serogroup B); and Heidelberg (6%, serogroup B). During 2008, S. 4,[5],12:i:- was the fifth most commonly identified serogroup B isolate and the thirteenth most commonly isolated serovar overall, representing <2% of all isolates of Salmonella from swine. In contrast, by 2017, the 5 most commonly isolated serovars, from among 1,031 total isolates, were: 4,[5],12:i:- (32%); Typhimurium (14%); Derby (9%); Choleraesuis (7%); and Infantis (5%, serogroup C1). Thus, by 2017, 32.1% of all Salmonella isolated from swine at the ISU-VDL were identified as S. 4,[5],12:i:-. Additionally, S. 4,[5],12:i:- made up 50.3% of all group B isolates (Fig. 1). This observation coincided with a proportional decrease in identification of both S. Typhimurium (decreased from 34.3% in 2008 to only 14.0% of isolates in 2017) and all other Salmonella serogroup B serovars as well. Interestingly, the percentage of serogroup B isolates from porcine samples has remained relatively constant between 2008 and 2017, comprising 65% of 1,060 total porcine Salmonella isolates in 2008 and 64% of 1,031 total isolates in 2017. The tipping point in observed dominance between S. Typhimurium and S. 4,[5],12:i:- in swine occurred rapidly between 2013 and 2014, and the trend for increasing actual and relative frequency of isolation of S. 4,[5],12:i:- from swine cases has remained constant since that time.

Figure 1.

The percent contribution of Salmonella Typhimurium and 4,[5],12:i:- to the serogroup B isolates of Salmonella identified from swine clinical cases at the Iowa State Veterinary Diagnostic Laboratory from 2008 to 2017.

Inclusivity and exclusivity of the rtPCR

Results of the inclusivity study indicated that, of the 45 serovar-confirmed S. Typhimurium isolates tested, 44 were correctly identified as S. Typhimurium by the rtPCR assay based on a positive signal for all 4 genes tested (Supplementary Table 3). Of the 61 S. 4,[5],12:i:- isolates tested, 60 were correctly identified as matching the NVSL-confirmed serovar (4,[5],12:i:-) based on a positive signal for the invA and fliA genes, and negative on either one or both of the hin-iroB or fljB targets given that one or both of which must be absent for the isolate to be considered monophasic. Isolates tested for exclusivity reacted as expected (Supplementary Table 2), with all other Salmonella isolates testing positive for the invA gene, negative for the fliA gene, and variable results for the other 2 targets. All non-Salmonella fecal organisms tested for exclusivity also reacted as expected, being negative for all gene targets.

Interestingly, the S. 4,[5],12:i:- isolates were consistently a minimum of 8 Ct values higher for the fljB target than the Ct value of the invA and fliA gene targets (Table 2). This trend was not noticed in the S. Typhimurium isolates. To determine the source of the differences in the Ct values of fljB, 5 S. Typhimurium and 5 S. 4,[5],12:i:- isolates were sequenced in the region amplified by the fljB primers that includes the 5’-coding region of the fljB gene. The fljB gene is 1,521 nucleotides, translating to 506 amino acids. Four base pairs were found to be consistently different between S. Typhimurium and S. 4,[5],12:i:- isolates (nucleotide 38: C to T; nucleotide 73: A to T; nucleotide 103: T to C; and nucleotide 163: C to T). The base pair differences did not result in any changes in the translated amino acid sequence. Additionally, the base pair differences were not located in the primer or probe binding sites, so it is unclear why the change results in differences in Ct values in the rtPCR assay. Several other Salmonella serovars tested on exclusivity testing also exhibited the same difference in Ct values between targets (Supplementary Table 2).

Table 2.

Representative example of real-time PCR results, expressed as cycle threshold (Ct) values, for pure culture of 4 different Salmonella serovars including S. Typhimurium and S. 4,[5],12:i:-.

Salmonella serovar	Representative example of Ct values for each gene target
	invA	fliA	hin-iroB	fljB
4,[5],12:i:-	23.0	24.6	Negative	35.2
Agona	18.5	Negative	Negative	Negative
London	18.3	Negative	18.4	20.9
Typhimurium	21.3	21.3	23.2	21.9

For the 2 isolates tested for inclusivity that did not react as expected in the rtPCR, further sequencing was also performed (Supplementary Table 3). Isolate “A” was positive for all 4 targets, but the fljB gene Ct value was > 8 cycles greater than the invA and fliA genes. The Ct value difference would make the isolate appear to be a S. 4,[5],12:i:- isolate, which matches its identification by NVSL serotyping. However, given that the isolate was positive for all 4 genes by rtPCR, identification as a S. Typhimurium isolate would also be expected. Isolate “B” was negative for hin-iroB but positive for the remaining 3 targets. Given that the isolate was negative for hin-iroB, it would be expected to be a S. 4,[5],12:i:- isolate. However, the lack of a Ct difference between the fljB and invA/fliA genes would make the isolate more likely a S. Typhimurium isolate, which matches its identification by NVSL serotyping. When the sequence of the target region amplified by the fljB primers was determined for both isolates, the identification based on the 4 base pair differences matched the NVSL serovar identification.

Limit of detection and clinical validation of rtPCR

LOD was determined to be ~500 CFU/g (or mL) of feces tested directly without enrichment. When the same samples were enriched in BPW for 18–24 h prior to running the rtPCR, the LOD was 5 CFU/mL of feces. The calculated R² value was 0.967 for S. Typhimurium and 0.964 for S. 4,[5],12:i:-, indicating a strong inverse correlation between Ct values and CFU/mL (or g) of sample (Supplementary Fig. 1).

Further clinical validation of the rtPCR was completed using both retrospective (no enrichment) and prospective (both direct and enrichment) samples from clinical cases at the ISU-VDL (Table 3). All prospective samples (n = 24) were identified as the same by culture and rtPCR (both with and without enrichment). Specifically, 4 samples were identified as S. 4,[5],12:i:- in both culture and rtPCR. Four samples were identified as Salmonella enterica ser. Heidelberg in culture and as Salmonella non-Typhimurium, non-4,[5],12:i:- in rtPCR. Sixteen of the samples were negative for Salmonella in culture and rtPCR. Although there were differences in the Ct values obtained from the non-enriched version and enriched version of each sample, the same conclusion was reached on all of the samples (data not shown).

Table 3.

Results of prospective and retrospective validation of the multiplex real-time PCR completed directly on clinical samples and compared to NVSL serotyping results.

n	Identification
	Culture result	rtPCR result
Prospective validation
16	No Salmonella	No Salmonella
4	S. 4,[5],12:i:-	S. 4,[5],12:i:-
4	S. Heidelberg	Salmonella, non-Typhimurium, non-4,[5],12:i:-
Retrospective validation
16	S. 4,[5],12:i:-	S. 4,[5],12:i:-
13	Salmonella, non-Typhimurium, non-4,[5],12:i:-	Salmonella, non-Typhimurium, non-4,[5],12:i:-
5	S. Typhimurium	S. Typhimurium
3	S. 4,[5],12:i:-*	Negative
2	S. 4,[5],12:i:-*	Inconclusive†
2	S. 4,[5],12:i:-	S. Typhimurium
1	Salmonella, non-Typhimurium, non-4,[5],12:i:-*	Negative

NVSL = National Veterinary Services Laboratories.

^*Few colonies to low levels of Salmonella sp. growth in culture.

^†Results would have been interpreted as “Inconclusive” given the high cycle threshold (Ct) values > 30.

For the retrospective validation completed on previously extracted DNA, of the 42 samples tested, 34 samples gave identical matches between the culture and rtPCR results. Four of the samples contained very low levels of Salmonella organisms in culture with only a few Salmonella-suspect colonies noted on the original agar plates; all of these samples were negative for Salmonella by rtPCR. Of the 4 remaining samples, 2 had high Ct values on rtPCR (30–40) that would have led to a positive interpretation for the presence of Salmonella DNA in the sample but inconclusive for identification at the serovar level; these samples were positive for S. 4,[5],12:i:- by culture with only single-to-low colony growth present. The remaining 2 samples were classified as S. Typhimurium by rtPCR but as S. 4,[5],12:i:- by culture and serotyping.

Discussion

Our investigation of the most common serovars identified at the ISU-VDL clearly demonstrates that S. 4,[5],12:i:- has become the dominant Salmonella serovar isolated from clinical samples in swine at the ISU-VDL. This finding may have important implications for the development of herd infection prevention strategies given that there are no vaccines labeled for control of infection with Salmonella 4,[5],12:i:- in swine in the United States, to date.

The fljB gene did not react as expected based on results from the original published validation.³⁵ During our validation of the rtPCR, this gene was consistently 8–10 Ct cycles greater than that of the fliA and invA genes for all S. 4,[5],12:i:- isolates. Through sequencing of 5 S. Typhimurium and 5 S. 4,[5],12:i:- isolates, we noted that 4 base pairs were consistently different between the 2 serovars. However, none of the differences were in the primer or probe binding regions, and therefore, they do not help to explain this anomaly. It remains unknown why the difference was not observed in the original study.³⁵ However, it is possible that, because the isolates originated from Germany for the original validation, the isolates possessed different mutations or deletions that led to the monophasic phenotype compared to the isolates from the United States that we used. The isolates used in the prior study also originated from human, animal, food, and environmental sources rather than strictly from porcine samples, as was the case in our study. It is also possible that in the original validation, samples were only weakly positive for Salmonella, resulting in Ct values for the fljB gene greater than the negative cutoff. However, based on the consistency of the results achieved in our study, the Ct value difference can be used as an aid in differentiating S. Typhimurium isolates from S. 4,[5],12:i:- isolates. Although we do not believe that this is a significant limitation to the assay based on the current knowledge of mutations that have led to the monophasic phenotype, it is possible that additional mutations exist and have yet to be characterized or will emerge that might interfere with identification utilizing our method. In addition, given that all clones of S. 4,[5],12:i:- are at this time believed to originate from S. Typhimurium, at minimum all should still be identified as potentially pathogenic based on positivity of the invA and fliA genes via the rtPCR.

Given the variability between Ct values of the gene targets, one of the limitations of our rtPCR assay is that samples could not be identified reliably at the serovar level if the Ct values were > 30. Slight differences in Ct cycles between invA, fliA, and hin-iroB, as well as fljB in non–S. 4,[5],12:i:- isolates, raises concerns that, as the standard Ct cutoff value (40) for negative rtPCR tests is approached, it is possible that one of the targets will appear negative whereas others test positive. However, we mitigated this issue by using a BPW enrichment step prior to rtPCR to increase the amount of Salmonella DNA present in the sample, thereby avoiding high Ct values and ensuring more accurate interpretation of results. Based on this limitation, although our rtPCR can be used directly on clinical samples such as feces, intestinal contents, and intestinal scrapings, the more ideal approach is to utilize the rtPCR either following BPW enrichment or as a confirmatory step following standard culture. In our study, use of the BPW enrichment step prior to rtPCR appears to provide advantages for both increasing sensitivity by improving the LOD (from 500 CFU/mL to 5 CFU/mL) and improving specificity by ensuring that enough organism is present to generate Ct values < 30. The lack of enrichment prior to DNA extraction in our retrospective case study highlights the sensitivity challenges with rtPCR when low numbers of Salmonella are present in a sample.

One additional potential limitation of our rtPCR applies to simultaneous infections by more than one serovar of Salmonella, which has been described in pigs and may in fact be commonplace.²² The identification of S. 4,[5],12:i:- by rtPCR is dependent on the fljB and/or hin-iroB targets being negative. Therefore, if a pig had a coinfection with both S. 4,[5],12:i:- and a serovar that possessed those genes, the rtPCR results would be indistinguishable from a pig infected with only S. Typhimurium. We believe that the possibility of mixed infections may explain the difference between rtPCR identification and serotyping observed in 2 of the samples in the retrospective validation that was performed. The standard Salmonella culture protocol of the ISU-VDL includes pursuit of complete identification at the serovar level for only one isolate per sample. Thus, in the case of a mixed infection, it is conceivable that the one serotyped isolate was not representative of the entire Salmonella population present in the original sample. Although we do not at this time know the significance of mixed infections with S. Typhimurium and S. 4,[5],12:i:- in swine, discordant results between NVSL serotyping and rtPCR results in clinical cases may warrant further investigation to determine if multiple serovars may be present in a clinical sample to better direct clinical decision making. Additional research studies have also demonstrated that some isolates of Salmonella may appear phenotypically as monophasic isolates, but still maintain a biphasic molecular status.^4,10 Therefore, it is also possible that the few discordant results identified in our study may be the result of this phenomenon; further research into the reason for this anomaly is also warranted.

Although our rtPCR cannot replace antimicrobial susceptibility testing in selection of the proper antimicrobial agent, it can detect the presence of Salmonella in a sample and identify it to the serovar level for S. Typhimurium and S. 4,[5],12:i:-. This identification can be completed earlier than susceptibility data can be made available. Given that S. 4,[5],12:i:- tends to be a highly resistant organism compared to S. Typhimurium,¹¹ serovar identification provides additional information regarding the common antimicrobial resistance profile to aid in the earlier selection of an antimicrobial likely to be effective, which can minimize the overall effects of an outbreak in a herd. Recognized differences in common antimicrobial susceptibility patterns between the serovars may also provide cause for treatment failures in mixed infections.