Rapid detection of Yersinia pestis with multiplex real-time PCR assays using fluorescent hybridisation probes


*Corresponding author. Present address: Institut für Mikrobiologie der Bundeswehr, Neuherbergstraße 11, 80937 München, Germany. Tel.: +49 (89) 31 68-38 08; Fax: +49 (89) 31 68–32 92, E-mail address: herbert.tomaso@web.de


The objective of the present study was to establish a system of real-time polymerase chain reactions (PCRs) for the specific detection of Yersinia pestis using the LightCycler™ (LC) instrument. Twenty-five strains of Y. pestis, 94 strains of other Yersinia species and 33 clinically relevant bacteria were investigated. Assays for the 16S rRNA gene target and the plasminogen activator gene (resides on the 9.5-kb plasmid) and for the Y. pestis murine toxin gene and the fraction 1 antigen gene (both on the 100-kb plasmid) were combined for the use in two multiplex assays including an internal amplification control detecting bacteriophage λ-DNA. Applying these multiplex assays, Y. pestis was selectively identified; other bacteria yielded no amplification products. The lower limit of detection was approximately 0.1 genome equivalent. Rat or flea DNA had no inhibitory effects on the detection of Y. pestis. The results obtained using the multiplex real-time assays showed 100% accuracy when compared with combinations of conventional PCR assays. We developed and evaluated a highly specific real-time PCR strategy for the detection of Y. pestis, obtaining results within 3 h including DNA preparation.


Yersinia pestis is the causative agent of bubonic plague usually transmitted by bites of the rat flea Xenopsylla cheopis[6,19]. After dissemination to the lungs, the disease can be spread through aerosols leading to outbreaks of highly infectious pneumonic plague with a case fatality rate of 50–100% if untreated [5]. Y. pestis is enzootic in many countries and has the potential of being a biological warfare agent [14,23]. The genus Yersinia consists of 11 species, but only Y. pestis, Yersinia pseudotuberculosis and Yersinia enterocolitica are considered to be pathogenic for humans [30]. Yersinia spp. provoke false or doubtful identification in most commercial biochemical identification systems [26]. Conventional polymerase chain reaction (PCR) methods for detecting Y. pestis have been developed [7,16,24,27,35]. Real-time PCR is a more rapid alternative and is also less prone to cross-contamination. Hybridisation probes add greater specificity. The detection of ciprofloxacin resistance in Y. pestis using LC-PCR has been developed, recently [25].

Most Y. pestis strains contain three virulence plasmids of approximately 9.5, 64, and 100–110 kb length [4,9]. The Y. pestis-specific 9.5-kb pPla (also referred to as pPCP1 or pPst) codes for a plasminogen activator/coagulase (Pla). This protease may be important for invasiveness and dissemination of Y. pestis transmitted by flea bites because pPla strains are avirulent in mice infected via the subcutaneous route [34]. The 64-kb pCD1 plasmid (pYV or pCad) contains a number of genes coding for several virulence factors including a type III secretion system [20]. The proteins are referred to as Yops (Yersinia outer proteins) and are required for full virulence in the mouse [18]. Yops are directly injected into eukaryotic cells and modulate host defence mechanisms such as phagocytosis and stress-activated signalling pathways [2,13]. This plasmid is shared by all pathogenic members of the species Y. pestis, Y. enterocolitica and Y. pseudotuberculosis[9]. The caf1 locus on the Y. pestis-specific 100–110-kb plasmid pMT1 (pFra) codes for the highly immunogenic fraction 1 capsule antigen (F1), a glycoprotein that is expressed at temperatures above 33°C. F1-deficient highly virulent Y. pestis strains have been reported [38]. The Y. pestis murine toxin gene codes for a 61-kDa protein [8]. Purified Ymt is highly toxic for mice when injected intraperitoneally [16]. Recent data define Ymt as a phospholipase important for the survival of Y. pestis in the flea vector and thus crucial for arthropod-borne transmission [19].

We present the development and evaluation of a system of real-time PCRs using hybridisation probes that encompass the detection of the plasmid-encoded genes pla, caf1 and ymt and the chromosomal 16S rRNA gene. For the rapid detection of Y. pestis two multiplex assays were established for pla and the 16S rRNA gene, and caf1 and ymt, respectively, both including an internal amplification control using a bacteriophage λ-DNA PCR assay.

2Materials and methods


Strains and DNA were obtained from the American Type Culture Collection (ATCC, MD, USA), the German Collection of Microorganisms and Cultures (DSMZ, Braunschweig, Germany), E. Carniel (Institut Pasteur, Paris, France), B. Niederwöhrmeier (WIS, Munster, Germany), J. Prior (DSTL Chemical and Biological Sciences, Porton Down, UK), and D. Tsereteli (NCDC, Tbilisi, Georgia). DNA of 25 Y. pestis (Table 1), 49 Y. pseudotuberculosis, 15 Y. enterocolitica strains, and strains of the species Yersinia aldovae (three), Yersinia bercovieri (four), Yersinia frederiksenii (four), Yersinia intermedia (four), Yersinia kristensenii (six), Yersinia mollaretii (four), Yersinia rohdei (four), and Yersinia ruckeri (one) were investigated. A panel of type strains representing pathogens causing plague-like clinical syndromes was also tested (Brucella abortus (ATCC 23448), Francisella tularensis (ATCC 6223), Burkholderia mallei (ATCC 23344), Burkholderia pseudomallei (ATCC 23343)), and other clinically relevant bacteria (Acinetobacter baumannii (DSMZ 7324), Aeromonas hydrophila subsp. hydrophila (ATCC 7966), Burkholderia cepacia (ATCC 25416), Burkholderia gladioli (ATCC 10248), Citrobacter freundii (DSMZ 30039), Citrobacter koseri (DSMZ 4595), Enterobacter aerogenes (DSMZ 12058), Enterobacter cloacae (ATCC 13047), Enterococcus faecalis (DSMZ 2570), Escherichia coli (ATCC 25922), Francisella philomeragia (ATCC 25015, ATCC 25017), Fusobacterium naviforme (DSMZ 20699), Klebsiella pneumoniae subspecies pneumoniae (DSMZ 30104), Listeria monocytogenes (DSMZ 12464), Moraxella catarrhalis (DSMZ 9143), Morganella morganii (DSMZ 6675), Mycobacterium tuberculosis (ATCC 27294), Pasteurella multocida (DSMZ 5281), Plesiomonas shigelloides (DSMZ 8224), Proteus mirabilis (DSMZ 4479), Proteus vulgaris (DSMZ 30118), Pseudomonas aeruginosa (DSMZ 11810), Serratia proteomaculans (DSMZ 1636), Salmonella typhimurium (ATCC 13311), Shigella flexneri (DSMZ 4782), Staphylococcus aureus (DSMZ 346), Staphylococcus epidermidis (DSMZ 1798), Stenotrophomonas maltophilia (DSMZ 50170) and Vibrio parahaemolyticus (DSMZ 10027)). Strains were cultured on standard agars at 37°C. Mycobacterial DNA was kindly provided by U. Reischl, Institute for Medical Microbiology and Hygiene, Regensburg, Germany.

Table 1. Y. pestis strains used for the evaluation of real-time PCR assays using hybridisation probes and the results for the chosen targets
  1. (p): plasmid-deficient strains, IP*: Institute Pasteur, Paris.

Y. pestis strainsSourceChromosome 16S rRNA genepPlapMT1pCD1
PKH 4, 519, IP*Kurdistan+++++
Turquie 10–1, 521, IP*Turkey++++−(p)
Senegal Thierno, 523, IP*Senegal+++++
Senegal Fay, 524, IP*Senegal+++++
Kenya 147, 537, IP*Kenya+++++
Congo Belge Lita, 543, IP*Zaire+++++
PKR 25, 564, IP*Kurdistan+++++
Exu 21, 567, IP*Brazil+++++
Congo Belge Elisabeth, 678, IP*Zaire+++++
Hambourg 9, 695, IP*France+++++
GB, Porton DownUK+++++
EV 76, Porton DownUK+++++
Colorado 92USA+++++
Java 9Indonesia++−(p)−(p)+
Tjiwidej (TW)Indonesia++++−(p)
EV-A, Porton DownUK+++++
EV 76Madagascar+++++
Y. pestoides G-8786Georgia+−(p)++−(p)


Artificially bred specimens of the oriental rat flea, X. cheopis, were provided by E. Schein and B. Habedank, Institute for Parasitology and Tropical Veterinary Medicine, Berlin, Germany. Eight to 10 specimens were pooled to give a total weight of 0.6 mg.


50 mg of muscle, liver, and lung tissue were isolated from five Rattus norvegicus rats.

2.4DNA preparation

Crude DNA of strains was prepared by transferring one distinct colony from an agar plate to 200 μl lysis buffer (5× buffer D (PCR Optimation Kit, Invitrogen, DeShelp, The Netherlands; 1:5 diluted in aqua dest.); 0.5% Tween 20; 2 mg proteinase K per ml (Roche Diagnostics, Mannheim, Germany)). After incubation at 56°C for 1 h and inactivation for 10 min at 95°C, 0.2 μl of the cleared lysate were added to the PCR mixture. Purified DNA was prepared using the QIAamp Tissue Kit according to the manufacturer's instructions (Qiagen, Hilden, Germany). 1 μg of purified DNA was calculated to represent 6×104 genome equivalents. For the disruption of fleas or rat tissue the Bio101 BioPulverizer System (Dianova, Hamburg, Germany) was used. Briefly, eight to 10 fleas were filled up to 1 ml with sterile water in the sampler tubes. The tubes were shaken in the Bio101 Mini Beadbeater at 5000 rpm for 20 s and placed on ice for a further 20 s. This procedure was repeated five times. Afterwards the sampler tubes were centrifuged at 14,000 rpm for 20 min at 4°C. 100 μl of supernatant and 100 μl of lysis buffer (5× buffer D, PCR Optimation Kit; Invitrogen, DeShelp, The Netherlands; 1:5 diluted in aqua dest.; 0.5% Tween 20) containing 2 mg of proteinase K per ml (Roche Diagnostics, Mannheim, Germany) were mixed by inversion. 1 μl of clear supernatant of rat DNA and 1 μl of 10-fold dilutions of Y. pestis EV 76 DNA were added to the reaction mix to evaluate inhibitory effects of rat DNA. This procedure was repeated for X. cheopis DNA to evaluate inhibitory effects of the flea vector DNA on the detection of Y. pestis. Human DNA from 200 μl aliquots of whole blood was prepared using the QIAamp blood mini kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol and was resuspended in 200 μl buffer AE. All samples for multiplex real-time PCR assays were spiked with bacteriophage λ-DNA (Bethesda Research Laboratories, Gaithersburg, MD, USA). DNA concentrations were determined spectrophotometrically at 260 nm by using a GeneQuant II RNA/DNA Calculator™ (Pharmacia Biotech, Cambridge, UK).

2.5Real-time PCR

A real-time hot-start PCR was performed with the LightCycler FastStart DNA Master Hybridisation Probes Kit™ which contains a modified form of thermostable recombinant Taq DNA polymerase (Roche Diagnostics, Mannheim, Germany) in a LightCycler Instrument™ (Roche Diagnostics, Mannheim, Germany). Primers and probes were designed based on alignment studies using BLAST programme and the sequence database of the National Centre for Biotechnology Information (NCBI). The primer pairs showing best results in pilot tests were used for further evaluation of the assays. The nucleotide sequence accession numbers of all targets, primers, and probes are shown in Table 2. Primers and probes were obtained from and designed in cooperation with TIB MOLBIOL (Berlin, Germany). Magnesium2+ concentration and reaction conditions were optimised for each real-time PCR in pilot experiments. The 20 μl reaction mixture contained 2.5 mM MgCl2, 2 μl 10× LightCycler FastStart Reaction Mix Hybridisation Probes, 0.5 μl of each primer (20 pmol μl−1), 0.5 μl of each hybridisation probe (8 pmol μl−1) and 2 μl template. 32 capillaries can be tested simultaneously in each run. No template controls that contained 2 μl of water instead of DNA and positive controls that contained DNA of Y. pestis were included in each run to detect any amplicon contamination or amplification failure. Optimisation resulted in reaction conditions of 10 min at 95°C, followed by 45 cycles of 10 s at 95°C, 10 s at 45°C and 12 s at 72°C. Fluorescence increase, i.e. creation of a specific product, was measured during the annealing step at 45°C. A melting curve analysis was performed after the last amplification cycle with 95°C for 0 s, 45°C for 30 s, and 95°C for 0 s. Temperature change rates were 20°C s−1 for all but the last step, where the rate was 0.1°C s−1. The low annealing temperature made it possible to design an internal amplification control with a melting temperature of 49°C which is clearly discernible from targets for the detection of Y. pestis. Data analyses were performed with LightCycler software version 5.32. In simplex real-time PCR assays the baseline for calculating the threshold cycle number (CT) was set at a fluorescence value of 0.04 for reasons of standardisation. In multiplex real-time PCR assays the baseline was set manually above the background signal because of possible inter-run variability. Those samples that scored positive by the instrument were confirmed by visual inspection of the graphical plots showing cycle numbers versus fluorescence values and the melting curves generated by the instrument. Human DNA, R. norvegicus and X. cheopis DNA were spiked with 10-fold dilutions of Y. pestis EV 76 DNA in triplicates to determine a possible influence on the real-time PCR assays.

Table 2.  Primers and probes used for real-time PCR
  1. All primers were designed within the genes. Nucleotide positions are according to sequences published by the National Centre for Biotechnology Information. TM, melting point obtained in a melting curve analysis using a LightCycler™ instrument.

 Target (nucleotide sequence accession number)Primer or probeSequence (5′→3′)Size (bp) of PCR productTM (°C)
Chromosomal DNA16S rRNA gene (AE013964)YP 16 F+gAATTTggCAgAgATgCTAAAg16165
pPlaPla (M27820)YPpla SgTAATAggTTATAACCAgCgCTT23264
pMT1caf1 (AF053947)YPcaf STACggTTACggTTACAgCAT24062
  YPcaf lcrLC-Red 705-AggCggTAAACTTgCAgCAggTAA p  
pMT1Ymt (X92727)YPtox UAggACCTAATATggAgCATgAC16862
pCD1YopT (AL117189)YopT UgATCAggAgCCATgCACAA33063
Internal amplification controlBacteriophage Lambda (J02459)Lambda FATgCCACgTAAgCgAAACA27849
  Lambda LCLC-Red 705-CggATATTTTTgATCTgACCgAAgCg p  

To detect inhibitory effects, a real-time PCR assay was established specifically detecting λ-DNA (Table 2). This assay was designed to yield an amplification product with a melting point clearly discernible from the various Y. pestis targeting assays (ΔT>10°C). The melting points measured for pla, yopT, the 16S rRNA gene target, ymt, caf1 and the λ-DNA target were at approximately 64°C, 62°C, 65°C, 62°C, 62°C and 49°C, respectively (Table 2, Fig. 1). The amount of λ-DNA was titrated to be detected repeatedly at a cycle number >30. Readout of LC-Red 640 values was performed in channel F2/F1 and readout of LC-Red 705 values was performed in channel F3/F2. In various experiments positive CT values (crossing cycle numbers) were confirmed using standard gel electrophoresis and ethidium bromide staining (Fig. 2).

Figure 1.

A–D: Melting curves generated by the LightCycler™ software for Y. pestis-specific multiplex real-time PCR assays using hybridisation probes. A and B show melting curves obtained in a multiplex real-time PCR assay for the 16S rRNA gene target on channel F2/F1 (A), and for pla including a λ-DNA PCR assay as an internal amplification control on channel F3/F1 (B). The melting points were approximately 65, 64 and 49°C, respectively. For the pPla-deficient Y. pestis strain G-32 the internal amplification control was negative (B) because the 16S rRNA gene target was simultaneously detected in channel F2/F1 (A). C and D show melting curves obtained in a multiplex real-time PCR assay using hybridisation probes for ymt on channel F2/F1 (C) and for caf1 including the internal amplification control on channel F3/F2 (D). The melting points were approximately 62, 62 and 49°C, respectively. Y. pestis Java 9 yielded no amplification products for ymt and caf1 because it is a pMT1-deficient strain. As displayed by overlaying melting curves, the internal amplification control was detected in this plasmid-deficient Y. pestis strain and the no template control containing λ-DNA, only. NTC, no template control; Lambda, bacteriophage λ-DNA.

Figure 2.

Standard curve for a real-time PCR targeting the 16S rRNA gene of Y. pestis on the LightCycler™ instrument using 10-fold dilution of Y. pestis DNA (ranging from 0.1 to 10,000 genome equivalents per reaction). Amplification products were verified by gel electrophoresis. M, 50–2000 bp molecular marker; lanes 1 to 6 correspond to sample numbers. Yp, Y. pestis EV 76; CFU, colony forming units; NTC, no template control; r, correlation coefficient; H2O, PCR water.

2.6Data and statistical analysis

In order to determine the range of linearity, the lower limit of detection and the intra-assay variation of the assays, three replicates of six 10-fold dilutions of Y. pestis EV 76 DNA in TE buffer (Tris 10 mM, ethylenediamine tetraacetic acid (EDTA) 1 mM), pH 8, containing approximately 0.1–10,000 genome equivalents were assessed simultaneously in a single run. Four runs were performed on different days by using the previous dilutions to investigate inter-assay variation between real-time runs and the coefficients of variance (CV) were calculated. A one-way analysis of variance was performed to evaluate the repeatability of DNA quantification between runs. Standard graphs of the CT values obtained from serially diluted Y. pestis DNA were compiled and the correlation coefficients (r) and the slopes were calculated (Fig. 2). Statistical analyses and plotting were carried out with SigmaStat and SigmaPlot Software 8.0 (SPSS Inc., Chicago, IL, USA).


A total of five real-time PCR assays for the detection of Y. pestis and one for an internal amplification control were evaluated (Table 2). 25 strains of Y. pestis, 94 strains of other Yersinia species and 33 clinically relevant bacteria were subjected to PCR. Detectable products were observed for caf1, ymt and pla in all Y. pestis strains except for plasmid-deficient strains (Table 1). No products were amplified from DNA of other bacteria. The 16S rRNA gene target was amplified in all Y. pestis strains and in various other species of the genus Yersinia (Y. aldovae, Y. bercovieri, Y. enterocolitica, Y. kristensenii, Y. mollaretii, Y. rohdei) but in no other bacteria. YopT located on the plasmid pCD1 was detected in 17 Y. pestis and in 22 Y. pseudotuberculosis strains but did not yield amplification products in any other bacteria. The real-time PCR assay for the detection of λ-DNA as an internal amplification control did not influence Y. pestis targeting PCR assays. Color compensation with the LightCycler™ software corrected mainly the overlap of fluorescence emission spectra of a channel 2 signal into channel 3.

3.1Simplex real-time PCR assays

In order to determine the linear dynamic range of the assays, serial 10-fold dilutions of quantified Y. pestis (EV 76) DNA starting from 10−1 to 105 genome equivalents per reaction were analysed. The amount of transcripts detected were in the range of 10−1 to 105 genome equivalents per reaction for pla, yopT and the 16S rRNA gene target, 10 to 105 for ymt and 102 to 105 for caf1. For all assays the coefficient of determination (r2) was >0.95, indicating a precise log–linear correlation between the input DNA and the amount of transcripts detected (Fig. 2). As indicated by overlaying amplification curves of the Y. pestis DNA concentrations investigated, intra-assay CVs were between 0.1 and 10.1%, inter-assay variation was between 0.6 and 10.6% of the calculated crossing points in the range of linearity. Significant differences between runs were observed by one-way analysis of variance (α=0.05, power=0.8) indicating that quantified Y. pestis DNA has to be included in each run, whenever exact quantification of genome equivalents is necessary.

R. norvegicus DNA was spiked with serial dilutions of Y. pestis DNA. In triplicate assays the lower limit of detection for the assays targeting the 16S rRNA gene, pla, ymt, and caf1 was assessed to be approximately 0.5, 5, 5, and 50 genome equivalents, respectively. When X. cheopis DNA was spiked with serially diluted Y. pestis DNA, the detection limit was approximately 0.5, 0.5, 5, and 50 genome equivalents, respectively.

3.2Multiplex real-time PCR assays

The linear dynamic ranges of the simplex real-time PCR assays for pla and the 16S rRNA gene target, and for caf1 and ymt, were essentially identical (Fig. 3). Each combination was used to evaluate a multiplex PCR assay and included a λ-DNA PCR assay as an internal amplification control. We added 150 pg of λ-DNA to the assays without influencing the lower limit of detection for Y. pestis. The lower limit of detection in both multiplex assays including the internal amplification control was identical to the limit of detection in the respective simplex real-time PCR assays. The coefficient of correlation (r) was calculated by the LC instrument to be >0.95 in both multiplex assays for the CT values of Y. pestis-specific amplification products measured on channel F2/F1.

Figure 3.

The graphs A–E show curves determined by linear regression analysis in the linear dynamic range of real-time PCR assays using hybridisation probes for the Y. pestis-specific gene targets. Cycle threshold numbers (CT values) were calculated from triplicate assays tested in four separate runs and were plotted against logarithmic dilutions of Y. pestis DNA (ranging from 0.1 to 10,000 genome equivalents per reaction). 16S rRNA, 16S rRNA gene; pla, plasminogen activator gene; yopT, Yersinia outer protein T gene; ymt, Yersinia murine toxin gene; caf1, capsule fraction 1 antigen gene.

Combinations of more than two Y. pestis-specific real-time PCR assays yielded no amplification products for caf1 and ymt and no increase of fluorescence above the background was observed. This may be explained by competitive PCR reactions.

3.3Comparison with conventional PCR assays as a reference method

According to criteria described in EN ISO 16140 which defines the general methods for the validation of alternative methods in the field of microbiological analysis of food, animal feeding stuff and environmental and veterinary samples, we compared previously established combinations of conventional PCR assays with the combinations of real-time PCR assays performed in this study [22,27]. The bacterial strains and the DNA preparation methods used for comparison were identical in both methods. The Y. pestis strains were chosen from a wide range of areas to avoid any local bias (Table 1).

3.3.1Inclusivity, exclusivity and accuracy

The system of real-time PCR assays described here was able to detect all Y. pestis strains without interference from a panel of relevant non-target strains. This was in line with the results obtained using conventional PCR resulting in 100% accuracy.


In the present study, we developed and evaluated a combination of real-time PCRs using hybridisation probes that can be performed in two multiplex real-time PCR assays including an internal amplification control. Many samples like blood, stool or soil may contain high amounts of inhibitors of DNA polymerases used in PCR. Therefore, an internal inhibition control should be included in diagnostical assays. The established λ-DNA real-time PCR assay did not interfere with assays targeting Y. pestis. This is the first description of a multiplex real-time PCR assay for the detection of Y. pestis. The use of hybridisation probes provides the advantage of speed as it requires no restriction digestion or sequence analysis for species confirmation. In contrast to conventional PCR technique, LC-PCR is less prone to carry-over contamination because the amplification of the target is performed within a closed capillary.

The value of plasmid-encoded sequences as a diagnostic tool for the identification of Y. pestis was discussed controversially [27,35–38]. Plasmids can be lost during cultivation under laboratory conditions. It is also known that plasmid-deficient Y. pestis strains exist in nature. In the Caucasus and the Daghestan mountains pPla-deficient Y. pestis strains have been isolated from voles in natural plague foci [11]. In rare cases, Y. pestis strains do not express the F1 antigen, which is encoded by caf1 on the plasmid pMT1 [32]. Java 9 is a pMT1-deficient Y. pestis strain isolated from a rat in Java. However, no detailed history of the strain was available, so a loss of the plasmid during cultivation cannot be ruled out [27]. The third plasmid pCD1 is shared by all pathogenic members of the species Y. pestis, Y. enterocolitica, and Y. pseudotuberculosis and can be lost spontaneously [30,37]. It has been reported that Y. pestis strains deficient of one or more plasmids can cause mild, chronic or even fatal infections [36,37]. Consequently, Y. pestis-specific real-time PCR assays have to be combined for the confirmation of a presumptive diagnosis. Higgins et al. developed a 5′ nuclease PCR assay targeting the pla gene of Y. pestis to detect bacilli in experimentally infected X. cheopis fleas and green monkeys (Cercopithecus aethiops). The specificity of this assay was problematic because several bacteria other than Yersinia (e.g. B. anthracis) were also detected depending on cycle number [15]. We designed primers and probes directed against multiple copy targets to improve the detection limit of our assays. In a recent study the copy numbers for the 16S rRNA gene, pPla (pla), pCD1 (yopT) and pMT1 (caf1, ymt) were estimated to be 6, 186, 4, and 2, respectively [29]. Therefore, we developed and evaluated real-time PCR assays for the chromosomally encoded 16S rRNA gene target and for targets on each plasmid harboured by typical Y. pestis strains. The 16S rRNA gene is a conserved chromosomal gene, displaying the relatedness of genera, species, and even strains. Y. pestis and Y. pseudotuberculosis are closely related because they are pathovars of a single species [3]. The assay for the 16S rRNA gene amplified products of other species of the genus Yersinia, too. Iqbal et al. described a real-time ‘TaqMan’ PCR assay (PE Applied Biosystems) targeting the species-specific pesticin gene located on pPla, which encodes for a bacteriocin. This assay had a detection limit of as few as three copies of the pesticin gene and detected Y. pestis, only [21]. In the present study we decided to use pla as it has proven to be a reliable target in conventional PCR [7,10,16,24,27,28]. Pla was detected in all strains of Y. pestis except for Y. pestis G-32 and Y. pestoides G-8786, which are pPla deficient. The low limit of detection of this real-time PCR (∼0.1 genome equivalents) was well in accordance with those found in conventional PCR assays [27]. Caf1 codes for the F1 capsular antigen, which is still the diagnostic antigen used in immunological tests [31]. Therefore, the presence of the gene in Y. pestis strains should be monitored in endemic areas although it is not necessary for pathogenicity in animal models. On the early stage of infection, the bacterial counts may be too low for the detection of Y. pestis by PCR [28]. Ymt is crucial for the flea cycle; in natural outbreaks it will be present in the agent [19]. Both genes are located on the 100-kb plasmid pMT1 of Y. pestis. The established real-time PCR assays yielded amplification products for Y. pestis only and had detection limits of 10 and 100 genome equivalents, respectively. Only the F1-deficient mutant Java 9 tested negative for ymt and caf1. Caf1 and ymt encoded on pMT1 proved to be stable and specific genetic markers for the detection of Y. pestis.

YopT, as part of the secretion type III system, is considered to be associated with pathogenicity in Yersinia species and is encoded on the 64-kb Yersinia virulence plasmid [20]. Arnold et al. compared the sequences of yopT in Y. enterocolitica, Y. pseudotuberculosis and Y. pestis revealing a number of mismatches [1]. Therefore, we used this finding to discriminate strains of Y. pestis and Y. pseudotuberculosis from enteropathogenic Y. enterocolitica strains. The yopT real-time PCR assay yielded amplification products from 17 Y. pestis and from 22 Y. pseudotuberculosis strains. Since this target was present in <70% of Y. pestis strains, this assay was not included in multiplex assays.

Hinnebusch et al. developed a quantitative PCR based on competitive PCR reactions to quantify plague bacilli in infected fleas and to explore flea Y. pestis dynamics [10,17,33]. This technique is sophisticated and dependent on the experience of the investigator. Real-time PCR assays including hybridisation probes might be a simple and fast alternative for this technique. Therefore, we investigated whether our assays might be influenced by rat or flea DNA. The detection limit was slightly elevated in all Y. pestis-specific real-time PCR assays. However, it can be assumed that this loss is without relevance since the number of bacteria was 106–1010 bacilli (g tissue)−1 in experimentally infected animals (i.e. the mouse and the guinea pig) [6]. DNA preparations of fleas in various feeding conditions may not inhibit conventional PCR assays [28]. The real-time PCR system described here might provide a rapid and reliable tool for routine plague surveillance which may in turn help evaluate risks for the human population [12]. It is also easily possible to quantify the number of Y. pestis bacteria in a sample using real-time PCR assays.

It is of paramount importance to use diagnostic assays for Y. pestis, which are capable of showing the presence of various targets – on plasmids and on the bacterial chromosome. This necessity is based upon naturally occurring or artificially induced loss of plasmids and/or genetic manipulation of one or more target genes or their regulatory operons. Leal and Almeida developed a conventional multiplex PCR identifying genes on the three virulence plasmids (caf1, lcrV, pla) and a chromosomal pathogenicity island (irp2) [24]. However, the time needed for the detection of Y. pestis DNA with quantitative real-time PCR can be reduced to 3 h including DNA preparation in contrast to approximately 6 h with conventional PCR. The results obtained by the multiplex real-time assays showed 100% accuracy when compared with combinations of conventional PCR for the identification of Y. pestis. The developed real-time PCR assays provide excellent detection limits and a combination of two multiplex assays (pla/16S rRNA gene and caf1/ymt) may be best suited for a selective detection of Y. pestis as well as for initial characterisation of found organism. Although multiplex real-time PCR proved to be a very convenient method, it should be considered that it may lead to a clinically relevant loss of sensitivity. Therefore, it has to be stressed that the assays described here can also be performed as simplex real-time PCR assays within the same run.

In the future the multiplex assays will be applied on clinical samples from regions where plague is endemic in order to assess their value in a clinical setting. However, considering forensic requirements in a bioterroristic scenario and because of the existence of atypical Y. pestis strains in nature, culture and biochemical identification are still mandatory.


We gratefully acknowledge K. Wilke and C. Lodri for their excellent technical assistance. C. Bartling is to be thanked for giving a deeper insight into statistics.