Pentaplex real‐time PCR for differential detection of Yersinia pestis and Y. pseudotuberculosis and application for testing fleas collected during plague epizootics

Abstract Upon acquiring two unique plasmids (pMT1 and pPCP1) and genome rearrangement during the evolution from Yersinia pseudotuberculosis, the plague causative agent Y. pestis is closely related to Y. pseudotuberculosis genetically but became highly virulent. We developed a pentaplex real‐time PCR assay that not only detects both Yersinia species but also differentiates Y. pestis strains regarding their plasmid profiles. The five targets used were Y. pestis‐specific ypo2088, caf1, and pst located on the chromosome, plasmids pMT1 and pPCP1, respectively; Y. pseudotuberculosis‐specific chromosomal gene opgG; and 18S ribosomal RNA gene as an internal control for flea DNA. All targets showed 100% specificity and high sensitivity with limits of detection ranging from 1 fg to 100 fg, with Y. pestis‐specific pst as the most sensitive target. Using the assay, Y. pestis strains were differentiated 100% by their known plasmid profiles. Testing Y. pestis and Y. pseudotuberculosis‐spiked flea DNA showed there is no interference from flea DNA on the amplification of targeted genes. Finally, we applied the assay for testing 102 fleas collected from prairie dog burrows where prairie dog die‐off was reported months before flea collection. All flea DNA was amplified by 18S rRNA; no Y. pseudotuberculosis was detected; one flea was positive for all Y. pestis‐specific targets, confirming local Y. pestis transmission. Our results indicated the assay is sensitive and specific for the detection and differentiation of Y. pestis and Y. pseudotuberculosis. The assay can be used in field investigations for the rapid identification of the plague causative agent.


| INTRODUC TI ON
Yersinia pestis, a Tier 1 select agent, is the bacterial causative pathogen of a plague that infects rodents and can be transmitted to other mammals, including humans through flea biting Stenseth et al., 2008). Plague causes serious illness or death if not treated promptly, especially if it develops into the pneumonic plague.
The plague has had a big impact on human history through three devastating pandemics  causing millions of deaths, as well as by locally endemic occurrence which results in hundreds to thousands of annual cases worldwide. Currently, the plague is endemic in Asia, Africa, and the Americas (Bertherat, 2016).
Along with Y. pestis, Y. pseudotuberculosis, and Y. enterocolitica are the other two related species within the genus Yersinia that can cause diseases in both animals and humans . Nevertheless, symptoms of disease caused by enteropathogens Y. pseudotuberculosis and Y. enterocolitica are vastly different from the plague. Belonging to the same genus, these organisms, especially Y. pestis and Y. pseudotuberculosis, are genetically closely related (Moore & Brubaker, 1975). Previous studies have indicated that Y. pestis has diverged recently from Y. pseudotuberculosis through events of genome reduction and gene gain; thus, the high genomic similarity were observed between the two species (Achtman et al., 1999;Brubaker, 2004;Chain et al., 2004;Hu et al., 1998;Moore & Brubaker, 1975;Parkhill et al., 2001). In the genetic structure, all three species possess a single circular chromosome and a common virulence plasmid termed pCD1, or pIB, or pYV in Y. pestis, Y. pseudotuberculosis, and Y. enterocolitica, respectively (Cornelis & Wolf-Watz, 1997;Iriarte & Cornelis, 1996;. The most prominent genetic difference between Y. pestis and the enteropathogenic Yersinia species is the presence of pPCP1 and pMT1, two newly acquired plasmids in most but not all strains of Y. pestis (Ben-Gurion & Shafferman, 1981;Ferber & Brubaker, 1981;Filippov, Oleinikov, Motin, Protsenko, & Smirnov, 1995;Filippov, Solodovnikov, & Protsenko, 1990;Hu et al., 1998;Portnoy & Falkow, 1981). These plasmids contain different genes that play important roles regarding pathogenesis. The plasmid pCD1 contains the highly conserved type three secretion system (T3SS) injectosome and effector proteins Yops (Rajanna et al., 2010); pPCP1 contains pla gene and pst gene that encode plasminogen activator (PLA) protease and the pesticin protein (Bearden, Fetherston, & Perry, 1997); pMT1 contains caf1 and ymt, encoding fraction 1 capsule antigen and murine toxin, respectively, as well as other putative virulence determinants (Hu et al., 1998).
Most Y. pestis strains isolated from humans, animals, or fleas contain all three typical plasmids. However, Y. pestis isolates lacking one or more plasmids may cause mild or even fatal disease, as previously been reported in South Africa and the United States (Beesley & Surgalla, 1970;Isaacson et al., 1973;Williams, Harrison, & Cavanaugh, 1975;Williams et al., 1978;Winter, Cherry, & Moody, 1960). Such variants have been reported infrequently. The natural occurrence of strains with atypical plasmid profiles might be greater than present data suggest.
The extreme pathogenicity of Y. pestis makes accurate and rapid detection of this bacterium a priority. Traditional methods of detecting Y. pestis include standard microbiological techniques (Norkina et al., 1994) and immunofluorescent staining (Feodorova & Devdariani, 2000). However, these methods take a long time and are relatively insensitive. In contrast, PCR-based assays have the advantage of rapid detection with high sensitivity. Furthermore, PCR is often preferable with the select agent and biosafety concerns. PCR assays for Y. pestis detection are not uncommon. Several PCR assays have been developed (Campbell, Lowe, Walz, & Ezzell, 1993;Matero et al., 2009;Neubauer et al., 2000;Radnedge, Gamez-Chin, McCready, Worsham, & Andersen, 2001;Riehm et al., 2011;Stewart, Satterfield, Cohen, O'Neill, & Robison, 2008;Tomaso et al., 2003;Woron et al., 2006). The pPCP1-situated pla gene, known to be present in 150-200 copies per bacterium (Parkhill et al., 2001), has been widely used due to the very high sensitivity (Riehm et al., 2011;Stewart et al., 2008;Woron et al., 2006). Ideally, if we are dealing with typical Y. pestis strains that harbor all three plasmids, a plabased assay could be reasonable. However, a good portion (~25%) of Y. pestis strains existing in nature are deficient in one or more of the three plasmids (Filippov et al., 1990;Stewart et al., 2008) due to various host growth temperatures, as well as other unknown factors (Iqbal, Chambers, Goode, Valdes, & Brubaker, 2000). As such, a plasmid-based assay may yield false-negative results. On the other hand, a pla assay may cause false-positive results because pla may also be present in other Enterobacteriaceae such as Escherichia coli and Citrobacter koseri (Armougom et al., 2016;Hänsch et al., 2015).
Further, a homolog of pla has been reported in rodents of multiple species (Giles et al., 2016). In addition to plasmid-based genes, chromosomal genes, such as entF3, were also applied for Y. pestis detection (Woron et al., 2006). The problem with using a chromosomal gene alone is it may involve non-specific amplification due to the genetic relatedness of Y. pestis to the other Yersinia species.
Amplification with a chromosomal gene may not indicate which species is present (Woron et al., 2006).
In the current study, we developed a pentaplex real-time PCR assay that includes three Y. pestis-specific targets, one Y. pseudotuberculosis-specific target and one internal control for flea DNA since the assay is intended to test field-collected fleas. The three Y.
pestis-specific targets in the assay are pPCP1-situated pst, pMT1-situated caf1, and chromosomal gene ypo2088. ypo2088 codes for a putative methyltransferase and is considered Y. pestis-specific (Chain et al., 2004;Matero et al., 2009). The Y. pseudotuberculosis-specific gene used in the assay is chromosomal gene opgG. Yersinia pestis lost this gene during its speciation from Y. pseudotuberculosis; hence, opgG can differentiate Y. pestis strains and Y. pseudotuberculosis strains (Quintard et al., 2015). For the internal control of flea DNA, the very conservative 18S ribosomal gene was used for a wide application for any flea species in general. The objective of this study was to allow accurate and rapid identification of all Y. pestis strains from field-collected fleas; to discriminate Y. pestis from Y. pseudotuberculosis; and to differentiate Y. pestis strains in regard to their plasmid profiles.
2 | E XPERIMENTAL PROCEDURE S 2.1 | Gene selection, primer, and probe design Five genes were selected for developing the pentaplex real-time assay: Yersinia pst, caf1, ypo2088, and opgG, and flea 18S rRNA.
Sequences of primer and probe used in the assay were designed for the current study except for those for caf1 which were adapted from assays published (Liu et al., 2016) (Eisen et al., 2006;Engelthaler, Hinnebusch, Rittner, & Gage, 2000). For the other 16 Y. pestis strains, both pMT1 and pPCP1 were present in 11 strains; both pMT1 and pPCP1 were absent in one strain; pPCP1was absent in four more strains (Table 2). Yersinia pseudotuberculosis was represented by strain B15 and four other strains (Table 2). Two Y. enterocolitica strains were included for specificity checking (Table 2) Leptospira interrogans for specificity testing (Table 3). The reason we included these species is that the transmission route of these bacteria is similar to the Yersinia species, either by fleas or other bloodfeeding vectors (Bai et al., 2017;Brook et al., 2015;Livengood, TA B L E 1 Description of the genes selected for the pentaplex real-time PCR and the validation PCR, and the primers/probes sequences

| Sensitivity testing and specificity assessment
Sensitivity was estimated using the limit of detection (LOD

| Application of the assay for testing fieldcollected fleas
To evaluate the utility of the pentaplex real-time PCR assay, we

| Optimization of primer concentration
The amplification was grouped tightly by the concentration of DNA for each target (Figure 1). Y. pestis-associated targets (pst, caf1, and ypo2088) primers, 600 nM for opgG primers, and 500 nM for 18S rRNA primers (Table 4).

| Sensitivity and specificity
All targets showed extremely high sensitivity ( Figure 2). Among the three Y. pestis-specific targets, pst was the most sensitive with a LOD of 1 fg per reaction, followed by caf1 (LOD 50 fg) and ypo2088 (LOD 100 fg) ( Table 4). The Y. pseudotuberculosis-specific opgG also showed very high sensitivity, with a LOD of 5 fg per reaction; the internal control target 18S rRNA for flea DNA had a LOD of 1 fg.
Standard curves showed the Ct values were correlated to the DNA concentration for each target (Figure 2).
To validate the above testing results, a PCR was performed using the widely used pla gene. The PCR detected 1 fg Y. pestis DNA as the LOD, which is identical to that of pst, suggesting that pst gene and pla gene are equivalent in terms of the sensitivity.
All primer/probe sets showed 100% specificity to the targets accordingly. DNA of Y. pestis strain CO96-3188 was positive to pst, caf1, and ypo2088 but negative to opgG and 18S rRNA; DNA of Y.
pseudotuberculosis strain B15 was positive to opgG but negative to all other genes; DNA of X. cheopis fleas was positive to 18S rRNA but negative to all other genes; no amplification to any of the genes was observed in Y. enterocolitica ( Table 2) and any of the other 12 species of non-Yersinia microorganisms (Table 3). pestis-specific targets, which was associated with their plasmid profile (BEI resource): All strains were positive to the chromosomal gene ypo2088; the strain with pMT1 and pPCP1 absent was negative to both caf1 and pst; four strains with pPCP1 absence were negative to pst. None of the Y. pestis strains were positive to opgG (Table 2). Testing of Y. pseudotuberculosis DNA-spiked flea DNA showed that opgG and 18S rRNA were amplified with LOD 5 fg for opg. The LOD was identical to that observed in the sensitive testing. All DNA in this group was negative to ypo2088, pst, and caf1.

| Detection of Yersinia pestis in fieldcollected fleas
By

| DISCUSS ION
It is of supreme importance using analytical assays for Y. pestis which can show the presence of various targets located on the chromosome and plasmids. A pentaplex real-time PCR assay developed in this study included three Y. pestis-specific genes (pst, caf1, and ypo2088), one Y. pseudotuberculosis gene (opgG), and 18S rRNA as an internal control of flea DNA if the assay is to be applied to test fleas.
Multiplex PCR assays for Y. pestis are not uncommon (Stewart et al., 2008;Tomaso et al., 2003;Woron et al., 2006) but, to our knowledge, this is the first to integrate five biologically meaningful targets.
Many assays have used the pla gene that is located on the pPCP1 plasmid for the detection of Y. pestis due to its high sensitivity with the presence of high copy numbers (150-200 copies) (Parkhill et al., 2001). Finding a homolog of pla in other bacteria or rodents (Armougom et al., 2016;Giles et al., 2016;Hänsch et al., 2015) has raised a concern of false identification, particularly when considering the frequency of flea-feeding on rodent hosts. Instead of using pla, we used pst which is also located on pPCP1. Our Cycles RFU results demonstrated pst is equally as sensitive as the pla gene with the same LOD (1 fg per reaction). Such results suggest that pst is a plausible diagnostic marker to replace pla. Like pla, the pMT1-located caf1 is another commonly used target for Y. pestis detection in multiple studies (Norkina et al., 1994;Stewart et al., 2008;Tomaso et al., 2003;Woron et al., 2006). Similar to other reports, our results also showed that caf1 is highly sensitive, but somewhat less sensitive compared to pst. It is known that caf1 is present in only about two copies per bacterium (Parkhill et al., 2001), explaining the differences in sensitivity between the two genes.

F I G U R E 3
We did not use any pCD1-located genes in this study. This is because pCD1 is possessed by all currently recognized pathogenic Yersinia species. Detection based on pCD1-located genes does not necessarily indicate the species of the presenting organism as previous studies have shown detection of Y. pestis, Y. enterocolitica, and Y.
pseudotuberculosis using pCD1-located lcrV or other targets (Stewart et al., 2008). Further, using a pCD1-located gene may not be as efficient compared to pPCP1-and pMT1-located genes because the pCD1 plasmid is absent in many more Y. pestis strains compared to pPCP1 and pMT1 plasmids in nature due to the intrinsic variability of pCD1 plasmids in Y. pestis (Stewart et al., 2008).
Unlike plasmids, chromosomal genes are not likely to be lost during laboratory cultivation or in nature; thus, chromosomal genes are more reliable targets than plasmid genes. Nevertheless, specific identification can be difficult if substantial genomic similarities are shared among multiple species. In the past, researchers have used the inv gene, the entF3 gene, the wzz gene, and the 16S rRNA gene in the identification of Y. pestis (Matero et al., 2009;Neubauer et al., 2000;Tomaso et al., 2003;Woron et al., 2006). However, non-specific amplification has been observed with these genes. All of them not only amplify Y. pestis, but also Y. pseudotuberculosis. In our assay, we used ypo2088 and opgG, two chromosomal genes that are de- We applied the assay to test 102 fleas that were collected from prairie dog burrows where prairie dog die-offs were reported 5 months before our field investigation. All fleas were successfully amplified with the 18S rRNA. One out of the 102 fleas was positive for Y. pestis by all three Y. pestis-specific genes (pst, caf1, and ypo2088).
The result confirmed the local transmission of Y. pestis in the prairie dog colony. Since the flea collection did not occur until 5 months afterward the prairie dog die-off, the low number of positive samples may have suggested plague activity had diminished gradually after the epizootic peak. Flea numbers were noticeably lower compared to an earlier investigation conducted in the area (around 3 months after the die-off). This indicated that most fleas may have died of starvation. The extremely low number of positive fleas was reasonably expected (Tripp, Gage, Montenieri, & Antolin, 2009). Not surprisingly, no Y. pseudotuberculosis was detected in any of these fleas.
In conclusion, the pentaplex real-time PCR assay developed in this work is highly sensitive and 100% specific in the detection and differentiation of Y. pestis and Y. pseudotuberculosis. Further, the assay allows one for the elucidation of the presence/absence of Y.
pestis pPCP1 plasmid and pMT1 plasmid in a particular strain, which can be applied to testing fleas and other field-collected materials when the plague is suspected.

ACK N OWLED G EM ENTS
The following reagents were obtained through the NIH Biodefense