Genomic DNAs of four biovars (Microtus, Orientalis, Antiqua and Mediaevalis) of Y. pestis were stored in our lab. Closely related or other genomic DNAs used in this study include 9 species of Yersinia (Y. enterocolitica, Y. intermedia, Y. aldovae, Y. bercovieri, Y. frederiksenii, Y. kristensenii, Y. mollaretii, Y. rohdei, Y. ruckeri), 16 different serotypes of Y. pseudotuberculosis, Brucella, Francisella tularensis, Bacillus anthracis, and E. coli DH5α; and DNAs from human blood and mouse. All bacterial strains used in this study were listed in Table 1.

The primers and probes for real-time PCR were designed based on the 3a sequence in the chromosome [26] and caf1 in plasmid pMT1 using Primer Express 2.0 (PE Corporation,USA). Other primers were also designed for cloning the target 3a and calf1 sequences into pGEM-T Easy Vector (Promega, USA). The primers and probes used in this study were listed in Table 2.

The PCR system (20µl) contained 2µl of 10×buffer (500 mM KCl, 100 mM Tris-HCl, 25 mM MgCl₂, 1mg/ml glutin), 1µl of each forward (F) and reverse (R) primer (5µM), 1µl of TaqMan probe (5µM), 1.6 µl of dNTPs (2. 5 mM), 0.2µl of Taq DNA polymerase (5U/µl), 5µl of DNA template, 5µl of enzyme stabilizer mixture (40% trehalose and 20% dextran), and 3.2µl of ddH₂O. Real-time PCRs were all performed on the Roche LightCycler 1.0 with the optimized cycling parameters of pre-denaturation at 94°C for 5 min, 40 cycles of denaturation at 95°C for 5 s, annealing and extension at 60°C for 30 s, and finally cooled at 40°C for 10 s. Signal acquisition mode is “single” at each cycle end of amplification.

DNA fragments flanking amplicons of 3a and caf1 were amplified using primers CF and CR (Table 2), respectively, from the 91001 genomic DNA for cloning into the pGEM-T Easy Vector according to the standard protocols reported elsewhere [18],[27]. The ligation products were transformed into DH5α and positive clones were identified by PCR and sequencing. Plasmids containing target fragments were purified separately and linearized by Sal I digestion. The concentration of linearized plasmid solution was then determined by UV spectrophotometer for calculating the copy numbers of the target DNAs. The quantified plasmid solution was serially diluted by 10-fold to prepare the standard templates with known copy numbers of target DNAs. Therefore, the real-time PCR was performed using these templates for obtaining the standard curves of C_t-Log concentration for 3a and caf1, respectively [18],[27]. C_t, the cycle threshold, refers to the cycle at which the fluorescence from a sample crosses the threshold.

Y. pestis live attenuated strain EV76 was cultivated in Luria-Bertani broth overnight at 26°C and then serially diluted by 10-fold. The number of viable cells was determined by counting colony forming unit (CFU) on the HBI (Heart Brain Infusion) agar plate. Five µl of each sample were directly applied to PCR amplification for evaluating the sensitivity of the PCR systems. All experiments were performed in duplicate. Genomic DNAs listed in Table 1, human and mouse DNAs were employed to evaluate the specificity of the PCR systems.

All reagents needed for real-time PCR were mixed up at appropriate proportion and dispensed into 200 µl microcentrifuge tubes with 15 µl each for a single reaction. The dispensed reaction mixture was then vacuum-dried. For stability evaluation, the dry reagents were kept away from lights at 37°C up to 79 days and the reagents were taken out at different time points to perform PCR by the following method. The reagents were reconstituted by adding 10 µl of ddH₂O, and then, 5 µl of plasmid templates containing 10¹, 10³, 10⁵ and 10⁷ copies/µl were separately added. The real-time PCRs were performed according to the protocol mentioned above. Only the real-time PCR system for 3a primers and probe was evaluated in this experiment. The stability of the real-time PCR reagent is assessed in terms of accuracy and reproducibility.

A standard curve of the PCR system was made using the plasmid templates as mentioned above, and the “log₁₀ concentration” of the freshly prepared reagents was chosen as reference values. Results of “log₁₀ concentration” of the reagents kept at 37°C for different periods were compared with the reference ones for accuracy evaluation.

Four plasmid templates (10¹, 10³, 10⁵, and 10⁷copies/µl) were used to perform PCR in triplicate in parallel. The coefficient of variation (CV) of C_t values was used to evaluate the reproducibility of the 3a PCR systems.

A culture of Y. pestis EV76 was serially diluted by 10-fold, and the number of viable cells was determined. For each of serial dilutions, 100 µl was added to 0.2 g soil samples, and the spiked samples were inoculated at 4°C over night for complete adsorption of the bacteria to soil. The soil samples were collected from the yard of our institute, and they were air dried at room temperature. The dried soil samples were directly spiked with dilution of Y. pestis cultures without any pretreatment for mimicking the real situation that is usually confronted in practice.

The spiked soil samples were vortexed thoroughly and centrifuged at 2, 000 rpm for 4 min and the supernatants were collected. Then, 1 ml of dH₂O was added to the pellet, and the mixture was vortexed completely and centrifuged again at 2, 000 rpm for 4 min. The supernatant was transferred to the tube containing the supernatants collected in the first step. This washing step was repeated once for collecting as many bacteria in the soil as possible. The supernatants collected above were mixed together for centrifuging at 8,000 rpm for 5 min to collect the target bacteria. The pellets were washed by dH₂O once, and 100 µl of dH₂O was added to each sample to resuspend the pellet for boiling in a water bath for 5 min. Five microlitres of this solution were used as templates for real-time PCR. We evaluated the sensitivity and specificity of the PCR system for detecting Y. pestis in the spiked soil samples and a blind experiment was performed for evaluating the feasibility of applying this technique in detection of Y. pestis directly from soil samples.

The spiked soil samples with different concentrations of Y. pestis were prepared and processed following the procedures described above, and the washed supernatants were used for sensitivity evaluation. As for the specificity test, genomic DNA (100 µg each) of the four biovars of Y. pestis, and other bacterial genomic DNAs listed in Table 1 (100 µg each) were all spiked into soil samples separately, and the real-time PCR was performed to test the specificity of the primers and probes. All experiments were performed in triplicates.

Soil samples spiked with different concentrations of EV76 were prepared by our colleague 1, and disordered and renumbered by our colleague 2, and then, the real-time PCR were performed by the first two authors of this article. After reporting the results by the first two authors, the three sides involved in this blind experiment sat together for evaluating agreement of the results to the original preparations. The blind test was performed in triplicates.

As shown in Fig. 1A and Fig. 1B, standard curves of both 3a and caf1 sequences, based on 10-fold serial dilutions of 10–10⁸ copies/µl, have a linear relationship between the log₁₀ concentration and the C_ts, with slopes of −3.3687 and −3.5154, respectively. The efficiency of each reaction is 98% and 93%, and Y-intercepts of 36.298 and 40.317, respectively, indicating a good quantitative determination of the PCR system. The r² (coefficient of determination) are 0.9996 and 0. 9997, for 3a- and caf1-based PCRs, respectively, implying the accuracy of serial dilution and the precision of sampling, and the ideal linearity of our PCR system as well.

The sensitivity of real-time PCR system based on the 3a and the caf1 sequences is 1.5 and 15 CFU per system, respectively, using dilutions of EV76 as template (Table 3). For both 3a and caf1 primers and probes, all four biovars of Y. pestis gave positive signals, and the other genomic DNAs in Table 1, human and mouse DNAs gave negative results, indicating the good specificity of these PCR systems.

The vacuum-dried reagents were put in a 37°C incubator and taken out at different time points. We used the ratio of “log₁₀ concentration” of them against the original “log₁₀ concentration” of the standards as a criterion to evaluate the stability of the reagents. A value close to 1 indicates good stability of the reagents, and we set the ratio of 0.900∼1.100 as the standard of acceptable stability. As shown in Table 4, only the results at the days 18 and 79 for lower concentration (10 copies/µl) were beyond this threshold. However, for the days 23, 32 and 49 at this concentration, the results were good enough to indicate the stability of the reagents. Therefore, we could reasonably attribute the biased results at day 18 for the lower concentration to the experimental errors. For concentration of template higher than 10 copies/µl, the vacuum-dried real-time PCR reagents could remain effective for at least 79 days at 37°C.

For evaluating the reproducibility of the PCR reagents, four concentrations of standard templates (10, 10³, 10⁵, 10⁷copies/µl) were chosen for performing real-time PCR in triplicates. As illustrated in Table 5, the coefficients of variations (CV) of C_t value for all concentrations were less than 3% in the 79-day's storage at 37°C, indicating an excellent reproducibility.

The results above suggest that the vacuum-dried real-time PCR reagent could be transported at room temperature by normal post or express mailing system.

For the spiked soil with different concentrations of Y. pestis EV76 for evaluating sensitivity of our PCR system, the results demonstrated that 5×10⁴ CFU per gram of soil could be detected by both 3a- and caf1-based PCR systems when C_t>35 was considered as a negative result. Genomic DNAs of four biovars of Y. pestis and its closely related and non-related bacterial genomic DNAs were all spiked separately into soil samples for evaluating the specificity of the PCR systems, and only the samples spiked with the four biovars of Y. pestis DNA showed positive results while all the samples spiked with other DNAs gave negative results. In the blind test, all the spiked samples with different amount of Y. pestis EV76 were detected correctly. As expected, the CFU values determined by the real-time PCR were 10 to 100 times lower than the ones spiked into the soil samples. These results implied that the real-time PCR systems could detect Y. pestis sensitively and specifically from soil samples.