The gltA sequences of 22 Rickettsia species were aligned by using the multiple-sequence alignment program ClustalW supported by the Infobiogen website (http://www.infobiogen.fr). Within the alignments, primers and probe were selected that were specific for R. prowazekii.

R. prowazekii strain Breinl (ATCC VR-142) was grown in Vero cell monolayers cultured in minimal essential medium supplemented with 4% fetal calf serum and 2 mmol/L L-glutamine as previously described (16). Infected cells were harvested by using sterile glass beads and sonicated. Cell fragments were removed by centrifugation, and the supernatant was centrifuged for 10 min at 7,500 × g. The resulting pellet was resuspended in 20 mL phosphate-buffered saline, pH 7.5. R. prowazekii inoculum was quantified by using either the plaque assay method (17) or comparatively by 10-fold serial dilutions of a known plasmid standard of R. prowazekii containing 2.0 × 10⁷ copies per sample in an independent real-time PCR as previously described (18).

In this assay 4 R. prowazekii strains, 21 strains of Rickettsia spp., and 14 strains of bacteria from genera other than Rickettsia were evaluated (Table). We also included 31 lice from an outbreak of epidemic typhus in Rwanda in 2004 and 10 R. prowazekii laboratory-infected lice (19), 10 R. typhi laboratory-infected lice (20), and 10 Borrelia recurrentis laboratory-infected lice (21). Finally, we also included 30 lice received in June 2005 from Rwanda, which were tested only with the quantitative PCR (qPCR) assay (Table). Negative controls included 10 pathogen-free lice, distilled sterile water, and PCR mixture. All experiments were repeated 4 times. For mice samples, DNA samples extracted from blood of uninfected mice were used as negative controls.

We have also tested R. prowazekii–infected mice by using a currently available experimental model similar to the previous model described for R. typhi (22). We used 7-week-old female BALB/C mice (Charles River Laboratories, Arbresle, France) that were maintained in cages with sterile food and water. All experiments were performed in a BSL-3 laboratory. Twelve mice were injected with 1.8 × 10⁵ PFU/mL R. prowazekii strain Breinl (ATCC VR-142), and 6 mice were injected with uninfected cells. The solution containing bacteria was injected into the retroorbital venous plexus over a period of 30 s. We collected 200 μL of blood from each mouse at day 3 postinfection (PI) and at day 6 PI and stored it in EDTA at –20°C for PCR.

Total genomic DNA from bacterial strains was extracted with the Qiagen QIAamp Blood Kit (Qiagen, Hilden, Germany), and lice DNA and blood and biopsy samples from infected mice were extracted by using the Qiagen QIAamp Tissue Protocol (Qiagen). PCR was performed by using a LightCycler instrument (Roche Biochemicals, Mannheim, Germany).The PCR mixture included a final volume of 20 μL with 10 μL of the Probe Master kit (Qiagen), 0.5 μL (10 pmol/μL) of each primer, 2 μL (2 μmol/μL) probe, 5 μL distilled water, and 2 μL extracted DNA. The amplification conditions were as follows: an initial denaturation step at 95°C for 15 min, followed by 40 cycles of denaturation at 95°C, annealing and elongation at 60°C for 120 s, with fluorescence acquisition in single mode. Each sample was also tested with a standard PCR that was performed on a PCR instrument (Eppendorf, Mastercycler, Hamburg, Germany) using primers of the gltA gene (23).

The R. prowazekii inoculum used in this study was of 1.8 × 10⁵ PFU/mL using the plaque assay quantification method (17) and contained 1.16 × 10⁶ copies per sample when quantified with our plasmid standard (18). The selected primers and probe of the gltA gene specific only for R. prowazekii were as follows: RproF (5´-TCGGTAAAGATGTAATCGATATAAG-3´), RproR (5´-CATATCCTCGATACCATAATATGC-3´) and Rp.probe (FAM-ACTTTACTTATGATCCGGGTTTTATG-TAMRA), leading to a PCR product size of 154 bp. When qPCR assay was used, only R. prowazekii strains were positive, whereas the standard PCR assay detected all rickettsial species (Table). The standard PCR assay was positive for all 20 laboratory-infected lice (R. typhi or R. prowazekii), while qPCR assay was positive only for the 10 R. prowazekii laboratory-infected lice. Finally, blood samples obtained from our experimental model of R. prowazekii–infected mice at days 3 and 6 PI were also positive by using the protocol described above. The mean number of cycle thresholds (Ct value) for mice sampled at day 3 PI was 32.47 ± 2.11; at day 6 PI, the Ct value was of 35.52 ± 2.01 (p = 0.001). All uninfected lice, B. recurrentis–infected lice, and mice samples were negative with both assays.

The sensitivity of qPCR and the standard PCR was determined by using 10-fold serial dilutions of our known R. prowazekii inoculum (1.16 × 10⁶ DNA copies per sample). The sensitivity of the qPCR was increased 10-fold over that of the standard PCR. Compared to our plasmid standard, the cutoff detection of the qPCR was 1–5 copies per sample, whereas the cutoff detection was >10 copies for the standard PCR.

Among the 31 lice from Rwanda sampled in 2004, 17 were positive by real-time PCR, whereas only 10 of these 17 lice were positive by standard PCR. The latter 10 samples had a mean number of 1,300 DNA copies (Ct value 26.82–35.22). The 7 samples positive only by real-time PCR had a mean number of 8.5 DNA copies (Ct value 33.72–38.73). The real-time PCR therefore appears to be more sensitive. However, this difference was not significant (p = 0.07) perhaps because of the small number of tested lice. Finally, 5 of the 30 lice received from Rwanda in June 2005 were positive when the qPCR was used (Table).

We developed a real-time quantitative PCR for specific detection of R. prowazekii. The selected primers and probe were 100% complementary to R. prowazekii only and to no other rickettsial strains. We confirmed the specificity of these primers and probe on rickettsial isolates and other common bacteria and repeated the experiments 4 times without discrepancies. Real-time quantitative PCR for rickettsiae was first developed to test antimicrobial drug susceptibility (24) and then was used to detect R. rickettsii and closely related spotted fever group rickettsiae (17) or R. prowazekii strains (25).

Our assay has a greater sensitivity than the standard assay, with a cut-off detection of only 1 to 5 DNA copies per sample, as measured comparatively to plasmid DNA quantification. The sensitivity found with our standard PCR has been previously estimated at 1–10 DNA copies of the gene (23). The first use of standard PCR for detecting R. prowazekii using primers derived from the 17-kDa antigen sequence had a cut-off detection of as few as 30 rickettsiae (26). Cutoff detection of rickettsiae with real-time quantitative PCR ranges from 5 copies (17) to 10 copies (25). Using our LightCycler assay, we detected an extra 7 samples in lice from Rwanda as compared to standard PCR. Only 1 report exists of real-time detection of R. prowazekii using molecular beacon probes targeting the ompB gene (25). In this report, only 2 R. prowazekii strains were tested (25). Moreover, we showed that R. prowazekii can be amplified from blood of experimentally infected mice. This experimental model of R. prowazekii infection and the ability to quantify the bacteria with the real-time PCR could be used to better study the pathogenesis of the organism. We found in this mouse model that the number of bacteria in blood was lower at day 6 PI than that at day 3 PI, which suggests that mice can eradicate infection at this dose.

The assay we describe can be performed wherever a real-time quantitative PCR machine is available. The reagents and the machine are standardized; this method gives rapid results (sequencing is not necessary) and decreases the likelihood of error. This assay was applied successfully in lice received from Rwanda in June 2005. Indeed, using our assay we were able to alert the World Health Organization of the presence of R. prowazekii–positive lice within 1 working day.

Because body lice and their associated diseases are generally encountered in areas where medical and biologic assistance is limited, local assessment of their roles as sources of infection is difficult. Lice are easy to collect and to transport to reference laboratories, where suitable molecular biologic approaches can be used (23). Although sucking lice die within 24 h of their final blood meal, the infecting bacterial DNA will remain intact for extraction for several weeks if the samples are kept dry (15). Upon arrival in the laboratory, the lice can be processed very quickly, and a diagnosis can be established rapidly (DNA extraction and LightCycler PCR take ≈5 h). Several weeks are necessary to obtain bacterial culture and serologic results, and those procedures do not always highlight the presence of bacteria. The usefulness of bacterial DNA detection in lice by PCR has been demonstrated by recent investigations. In central Africa, large outbreaks of lice infections occurred during civil wars in Burundi, Rwanda, and Zaire and preceded the outbreak of epidemic typhus by 2 years (4). Finally, our data obtained in experimentally infected mice suggest that real-time PCR could also be useful for detecting R. prowazekii directly from blood specimens. Because our assay is highly standardized and easily adaptable anywhere and anytime, it could improve epidemic typhus surveillance in public health programs, especially for countries with underdiagnosed or unrecognized human cases (4).