INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Lyme disease caused by the spirochete Borrelia burgdorferi sensu lato is the most common tick-borne disease in humans in the Northern Hemisphere. B. burgdorferi causes a multisystem inflammatory ailment, although the precise mechanisms of tissue damage are not well understood (10). Lyme disease is characterized by some or all of the following manifestations: an initial erythematous rash, the major presenting feature of the illness, neurological complications, arthritis, or carditis. It is clear that the organism is present at the site of inflammation in many organs and that many of the features of the illness are relieved by antibiotic therapy. However, in addition to pathogenetic processes initiated directly by the spirochetes, there is accumulating evidence that immune reactions triggered by B. burgdorferi may significantly contribute to disease development (9).

While in most cases the diagnosis of Lyme disease is made on the basis of clinical symptoms and serology, the spirochetes can also be cultivated from different clinical specimens and the nucleic acid has been detected by PCR in different tissues and body fluids (5, 6). However, none of the borrelia detection methods currently used in diagnostic settings allow the quantification of bacteria. This would clearly be advantageous for (i) the early efficacy control of the therapeutic regimen, (ii) the correlative analysis of the seriousness of symptoms and bacterial burden, and (iii) the study of the influence of immune modulation on bacterial replication or elimination in animal models of Lyme disease.

Several reports describe the detection of B. burgdorferi by qualitative PCR, in most cases by using the genes of either outer surface proteins (Osps) or flagellin as target sequences (2, 5, 6). To the best of our knowledge, only one group has reported a PCR-based method that allows the quantitative analysis of spirochete burden in tissue samples (12). Based on the amplification of sequences coding for OspA, the authors established a competitive PCR applying incorporation of ³²P-labelled nucleotides followed by polyacrylamide gel electrophoresis and subsequent quantification by phosphoimage analysis. Although potentially useful in experimental settings, this technique displays several features that hamper a broader application in diagnostic laboratories, such as the use of radioactive isotopes.

The problem of quantifying specific DNA target sequences might be overcome by the recently available 5' nuclease PCR assay (7). In this assay, a specific probe which hybridizes internally to the amplified fragment is added to the PCR mixture. During PCR, the 5'-to-3' exonuclease activity of the Taq DNA polymerase hydrolyzes the hybridized probe. By using a fluorogenic probe, which consists of a 5' reporter dye and a 3' quencher dye, the hydrolysis of the probe abolishes the suppression of the reporter dye. This can be monitored by measuring the fluorescence emission of the sample during the PCR. The amount of fluorescence detected is proportional to the amount of accumulated PCR product. In contrast to endpoint analysis where only the plateau phase of the PCR can be detected, real-time PCR allows monitoring of the exponential phase. The quantitative information in a PCR comes only from those few cycles where the amount of DNA grows logarithmically from barely above the background to the plateau. Often only 4 to 5 cycles out of 40 will fall in this log-linear portion of the curve. Since the complete PCR is monitored during real-time PCR, the log-linear region can be easily identified in each single reaction. In addition, no postamplification steps are necessary, eliminating the risk of cross contamination and reducing labor-intensive detection methods. While several reports concerning the use of this method for qualitative detection have appeared (3, 11), only one application of this principle for quantitative clinical diagnostics of a microbial pathogen has been reported. This method has been compared with conventional-method and nested PCRs for Mycobacterium tuberculosis (4). We report the development of a real-time PCR assay for the quantitative detection of B. burgdorferi that represents an accurate and easy diagnostic tool for this pathogen. Furthermore, we demonstrate its applicability in a mouse model of Lyme disease in which we compared (i) two different inbred strains of mice known to display different disease susceptibilities and (ii) antibiotic-treated and nontreated mice.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Mice. Female mice of the inbred strains BALB/c and C3H/HeJ were obtained from Charles River Breeding Laboratories, Sulzfeld, Germany. All mice were 6 to 8 weeks old at the time of infection.

B. burgdorferi and control bacteria. The N40 isolate (1) of B. burgdorferi sensu strictu (kindly provided by D. Postic, Institut Pasteur, Paris, France) and the PKo strain of Borrelia afzelii (kindly provided by B. Wilske, Ludwigs-Maximillian Universität, Munich, Germany) were grown in modified Barbour-Stoenner-Kelly II (BSK II) medium and used at low passage (five or fewer passages). Spirochetes were washed and enumerated with a blood cell counting chamber under dark-field microscopy. For control purposes, Borrelia recurrentis and Treponema bryantii (both obtained from the Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH, Braunschweig, Germany) were used.

Infection of mice, measurement of joint swelling, and antibiotic treatment. Six mice per group were infected subcutaneously with 5 × 10⁵ spirochetes in 50 µl of phosphate-buffered saline injected into the right hind footpad. The development of arthritis was monitored by measuring the thickness of the infected and noninfected contralateral tibiotarsal joints by means of a vernier caliper (Kroeplin, Schlüchtern, Germany) and calculated by dividing the thickness of the infected joint by the thickness of the control joint. In the case of antibiotic treatment, mice received 10,000 IU of penicillin G in 100 µl of phosphate-buffered saline subcutaneously twice a day for 14 days starting on day 8 after infection.

Preparation of DNA. The QiaAmp tissue kit obtained from Qiagen (Hilden, Germany) was used in accordance with the manufacturer's instructions. The elution volume was 200 µl of water.

Conventional nested PCR. A nested PCR that amplifies a part of the highly conserved flagellin gene was used as described by Huppertz et al. (5). Briefly, external primers yielding a 730-bp product were used in the first PCR (30 cycles); this was followed by a second PCR leading to the amplification of an internal 290-bp fragment which was visualized by agarose gel electrophoresis and ethidium bromide staining.

Real-time PCR. Primers and probes were selected for the flagellin gene of B. burgdorferi (GenBank accession no. X15660). The upstream primer corresponds to the region from base 588 to base 609 (TCTTTTCTCTGGTGAGGGAGCT). The reverse primer corresponds to the region from base 636 to base 657 (TCCTTCCTGTTGAACACCCTCT). The internal oligonucleotide probe corresponds to the region from base 611 to base 634 [(FAM)AAACTGC(TAMRA)TCAGGCTGCACCGGTTC; labels, which are defined below, are indicated in parentheses]. For quantitation of the murine host, -actin was chosen (accession no. M12481). The upstream primer corresponds to the region from base 398 to base 422 (TCACCCACACTGTGCCCATCTACGA). The reverse primer corresponds to the region from base 722 to base 745 (GGATGCCACAGGATTCCATACCCA). The internal oligonucleotide probe corresponds to the region from base 427 to base 453 [(FAM)TATGCTC(TAMRA)TCCCTCACGCCATCCTGCGT]. PCR reagents were purchased from Perkin-Elmer (Weiterstadt, Germany). Primers and probes were from TIB Molbiol (Berlin, Germany). Probes were 5' end labelled with 6-carboxyfluorescein (FAM) and internally labelled with 6-carboxy-N,N,N',N'-tetramethylrhodamine (TAMRA). The PCR mixture (25-µl total volume) consisted of 300 nM primer, 200 nM probe, 200 nM deoxynucleoside triphosphates, 3.5 mM MgCl₂, 1 µl of DNA, 1 U of AmpliTaq Gold, and 1× PCR buffer (50 mM KCl, 10 µM EDTA, 10 mM Tris-HCl [pH 8.3]). Amplification and detection were performed with an ABI 7700 system with the following profile: 95°C for 10 min and 45 cycles of 95°C for 15 s and 60°C for 1 min. Actin and flagellin PCRs were performed in separate reaction mixtures. The fluorescence was continuously monitored, and signals for each dye were extracted from the emission spectrum. The FAM signal was standardized to the passive reference ROX, which was included in the reaction buffer. Furthermore, the background fluorescence, which was calculated from cycles 3 to 10, was subtracted. Quantification was performed by determining the threshold cycle (C_t). This is defined as the cycle at which the FAM fluorescence exceeds 10 times the standard deviation of the mean baseline emission for cycles 3 to 10. Calculation of cell and spirochete numbers was based on standard curves of threefold dilution series representing the plot of C_t values versus the log of copy numbers included in each PCR run (see, as an example, Fig. 1).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Sensitivity and specificity of the real-time PCR. After testing different target genes, we decided to use flagellin of B. burgdorferi as the target sequence for establishing a real-time PCR assay. DNA was prepared from a threefold dilution series of B. burgdorferi bacteria from 1 × 10⁷ down to 2 × 10² bacteria. To avoid inhibition of the PCR, we used 1 µl as the template; therefore, the theoretical number of bacteria in our sample was 200 times lower. As shown in Fig. 1, plotting the obtained C_t values relative to the number of bacteria resulted in a linear correlation with an R² value of 0.9832. Greater deviations were observed, with very low copy numbers (one and three) most likely due to statistical fluctuations. To evaluate the robustness of this assay, we performed the same experiment on different days with different experimenters, including one sample of unknown copy number. The result is given in the inset in Fig. 1. Even with experiments with lower correlation coefficients included, the standard deviation is only about 16.5%. Excluding experiments 1 and 4 would decrease the standard deviation below 8%, which is well within the range of conventional diagnostic assays. The sensitivity of the real-time PCR was identical when B. burgdorferi and B. afzelii (strain PKo) were compared (data not shown). To evaluate the specificity of our assays, we tested DNA from Borrelia hermsii and T. bryantii. No amplification could be observed in either case (data not shown).

View larger version (20K): [in this window] [in a new window]   FIG. 1.   Standard curve for dilution series of B. burgdorferi bacteria. Quantification was performed by determining the threshold cycle (Ct). This is defined as the cycle at which the fluorescence exceeds 10 times the standard deviation of the mean baseline emission for cycles 3 to 10. One representative curve along with its R2 value is given. The same experiment, which included an unknown sample, was repeated six times. The determined copy number of this sample and R2 values of each experiment are listed in Table 1. Means and standard deviations (SD) were calculated from these six experiments.

Spiking experiments. In a clinical diagnosis setting, the DNA of the pathogen has to be amplified with the background of host DNA. Therefore, we performed experiments in which the same dilution series as that described above was spiked with a dilution series of mouse spleen cells. As shown in Fig. 2A, the number of host cells does not influence the amplification of B. burgdorferi DNA even when mouse cells are present in large excess over bacteria.

View larger version (26K): [in this window] [in a new window]   FIG. 2.   Absence of interference between mouse cells and B. burgdorferi in the real-time PCR. Dilutions of mouse spleen cells were mixed with dilutions of bacteria, and DNA was prepared from this mixture. Determinations of the Cts for flagellin (A) and mouse -actin (B) are shown.

Comparison of the PCR methods. To compare the overall sensitivity of the newly developed real-time PCR with that of the nested PCR described previously (5), both methods were employed to detect spirochete DNA in several tissue samples of B. burgdorferi-infected C3H/HeJ mice and BALB/c mice during the course of infection. As shown in Table 1, there was a high degree of concordance between the two methods (>94%). Discrepant results with the two methods were obtained only with tissue samples in which the spirochete burdens were obviously very low (<1,000 B. burgdorferi genome equivalents per 10⁶ cells), approaching the detection thresholds of the methods. Since the rates of single positive results with either the conventional or the real-time PCR were not significantly different, the quantitative PCR strategy displays a sensitivity similar to that of the nested qualitative PCR.

Analysis of kinetic tissue distribution of spirochetes in different inbred mice. To analyze the kinetics of tissue distribution of spirochetes in mice, disease-resistant BALB/c mice and arthritis-developing C3H/HeJ mice were infected with B. burgdorferi and the spirochete burden in several tissues was monitored with the real-time PCR over the course of infection. As depicted in Fig. 3, the overall patterns of bacterial burden were similar in both inbred strains of mice. An initially and sustained high level of spirochetal DNA was detected in foot tissue where the infection was initiated and in the regional draining lymph node. In contrast, the contralateral footpad and lymph node as well as the spleen and kidney were essentially free of B. burgdorferi until day 8 of infection, and a relatively constant level of spirochete burden was reached as late as 2 weeks after infection. However, quantitative differences were detected when tissues from susceptible C3H/HeJ and resistant BALB/c mice were compared. On day 8 after infection, the numbers of B. burgdorferi bacteria in the infected foot and the regional draining lymph node were 241-fold and 4-fold higher, respectively, in C3H/HeJ mice than in BALB/c mice. One week later, C3H/HeJ mice had 1.5- to 6-fold more spirochetes in the foot tissue and lymph nodes on the infected side and 15- to 23-fold more bacteria in the respective contralateral tissues than BALB/c mice did. On day 55 after infection, the differences in the numbers of spirochetes between tissues from the two inbred strains became smaller but the C3H/HeJ mice still displayed significantly higher bacterial burdens.

View larger version (27K): [in this window] [in a new window]   FIG. 3.   Kinetic distribution of spirochetes in tissues from mice. The right hind footpads of C3H/HeJ (A) and BALB/c (B) mice were infected with 105 spirochetes. At the time points indicated, two mice per group were sacrificed, and DNA was isolated from the spleens, kidneys, lymph nodes (LN), and foot tissues. The numbers of borreliae as well as the amounts of mouse -actin DNA were determined by real-time PCR as described in Materials and Methods. The mean numbers of spirochetes per 106 mouse cells (calculated on the basis of -actin standard curves) are given. , from noninfected side.

Influence of antibiotic therapy on spirochete burden in different tissues. C3H/HeJ mice, which are known to develop severe arthritis after infection with B. burgdorferi, were treated daily with penicillin G from day 8 to 21 after the infection. As can be seen in Fig. 4A, the treatment almost completely abolished the development of joint swelling and there were no signs of inflammation in the treated group of mice, in comparison to the control, infected C3H/HeJ mice. The most prominent reduction in spirochete burden (>99%) in penicillin G-treated mice was seen on day 15 in the lymph nodes on the noninfected side. Reductions of >70% caused by the antibiotic treatment were detected on days 8 and 15 in lymph nodes and foot tissues of the infected and contralateral sides. However, it is important to note that there was no complete elimination of spirochetes after penicillin G treatment and the differences between the treated and the control groups vanished on day 55 after infection, i.e., 4 weeks after the antibiotic treatment had been finished. The fact that borreliae could be cultivated from tissues of mice treated with penicillin G 8 weeks after the infection (data not shown) showed that PCR positivity is a good indicator for the presence of viable spirochetes.

View larger version (31K): [in this window] [in a new window]   FIG. 4.   Influence of penicillin G treatment on the clinical course and bacterial burden in tissues from B. burgdorferi-infected C3H/HeJ mice. Twenty mice in the control group () and 10 in the penicillin G-treated group () were infected as described in the legend to Fig. 3. (A) Swelling of the tibiotarsal joints was measured three times a week as described in Materials and Methods. (B to F) At the time points indicated, two mice per group were sacrificed, DNA was isolated from the organs indicated, and the burdens of B. burgdorferi were determined as described in the legend to Fig. 3.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In this report, we describe a new real-time PCR for the quantitative detection of B. burgdorferi in tissue samples. The sensitivity and the specificity of this method are similar to those of a previously published nested PCR (5). However, the real-time PCR compares favorably with the nested PCR, with key advantages of speed, increased throughput, and decreased risk of false-positive results because of elimination of second-round amplification. Furthermore, our real-time PCR allows for the reliable quantification of spirochete DNA. In combination with a real-time PCR for a host gene, e.g., actin, the quantitative data from the Borrelia PCR could be standardized in order to compare the bacterial burdens of different samples. Currently, we are testing different fluorescent dyes to develop a real-time duplex PCR for the simultaneous detection of host and bacterial DNA in one tube.

The data obtained by our new method provide evidence of a positive correlation between clinical symptoms, measured as arthritis and spirochete burden in the mouse model. First, the numbers of spirochetes were consistently higher in disease-susceptible C3H/HeJ mice than in BALB/c mice, which develop no, or very moderate, arthritis. Second, the spirochete burden peaked in the second week after infection, only shortly before the arthritis score reached its maximum. Furthermore, during the late course of infection, the decline in bacterial burden paralleled the spontaneous lessening of tibiotarsal swelling. The 2-week penicillin G treatment starting 1 week after infection not only abolished disease symptoms but also significantly reduced the B. burgdorferi levels in all tissues analyzed. These effects were most pronounced (>95%) in tissues in which the invasion by the bacteria proceeds during the course of treatment, e.g., the contralateral foot and lymph nodes. This is most likely due to the fact that in these locations the replication of B. burgdorferi is initiated, starting from a very low level, at a time point when penicillin G has already reached bactericidal concentrations in tissue.

The only organ where a significant increase in spirochete burden was detected in penicillin-treated mice was the kidney. This might reflect the increased clearance of bacterial debris during the antibiotic treatment via urine, since we could not cultivate B. burgdorferi from urine (data not shown). A similar situation has also been described for treated Lyme disease patients, where the rate of PCR sensitivity in pretreatment urine samples was only 50% and increased to 90% when urine obtained 3 to 6 days after the onset of therapy was used (6). Furthermore, Maiwald et al. used a method for absorption of soluble DNA to glass beads without previous enzymatic digestion or boiling for preparation of human urine samples which allowed the highest rate of B. burgdorferi detection by PCR (8). These findings support our suggestion that soluble DNA, not whole borrelia cells, is excreted in the urine.

Most importantly, our data show that there is persistence of B. burgdorferi both in mice spontaneously resolving their arthritis and in mice cured by a 2-week course of antibiotic treatment. This appears not to be due to the presence of dead bacterial DNA since (i) there was an increase in bacterial burden after the antibiotic treatment had been finished (Fig. 4) and (ii) we could also cultivate B. burgdorferi from the tissues of these mice 5 weeks after treatment. Thus, despite an efficient control of bacterial replication by the humoral immune response (9, 10), viable spirochetes persist, at least in mice, and can be quantified by real-time PCR up to 7 months after infection (data not shown). In a more limited study, similar conclusions have been drawn by Yang et al., who analyzed the fate of B. burgdorferi in tissues from BALB/c and C3H/HeJ mice by a competitive outer surface protein PCR applying phosphoimage analysis of radioactively labelled PCR products (12). The differences in bacterial burden between C3H/HeJ and BALB/c mice were in the same range (5- to 10-fold) as those detected in our study, although the authors only quantified the bacterial burdens in the heart and ankles of mice at 2 weeks postinfection.

In summary, we developed a new diagnostic method on the basis of the real-time PCR that allowed for the fast, cost-effective, and easy high-throughput quantification of B. burgdorferi in tissue samples. Since we found a good correlation between clinical symptoms and spirochete burden in the mouse model of Lyme disease, we propose this method as a valuable tool in the diagnostics of human Lyme disease. Ongoing research in our laboratory on human samples will further define the value of this approach.