Introduction

Lyme borreliosis caused by the spirochete Borrelia burgdorferi sensu lato (s.l.) is the main tick-borne disease in the Northern hemisphere. The B. burgdorferi s.l. complex meanwhile comprises at least 11 different species [3, 17]. Three of these species, B. burgdorferi sensu stricto (s.s.), B. afzelii and B. garinii, are known to be pathogenic for humans [3, 6, 17, 49], while for B. valaisiana a human pathogenic potential is suspected only [35]. B. burgdorferi s.s., the only species causing Lyme disease in North America, is homogeneous regarding the outer surface protein (Osp) A, while in Europe at least seven different OspA types have been observed among the human pathogenic B. burgdorferi s.l. species [49]. B. burgdorferi s.s. and B. afzelii are both homogeneous for OspA type, corresponding to OspA type 1 and 2 respectively. By contrast, B. garinii is more heterogeneous comprising OspA types 3–7 [46, 49].

The heterogeneity of European strains has important implications for research on pathogenesis (organotropism), diagnostic test systems, or vaccine development. There is strong evidence for an association between B. afzelii and acrodermatitis chronica atrophicans [1, 2, 40], while an association of B. garinii with the development of neuroborreliosis [9, 40, 49, 50, 51] and of B. burgdorferi s.s. with Lyme arthritis is still subject of controversial discussion [10, 23, 41].

A vaccine based on recombinant OspA from a B. burgdorferi s.s. strain, used until recently in the USA [38], does not protect against most of the causative strains present in Europe. Since OspA antibodies derived from one species or OspA type are not cross-protective against others, all prevalent and pathogenic Borrelia species and OspA types have to be considered for European OspA vaccine development [11, 21, 22, 37]. Valid data about the prevalence of different Borrelia species and OspA types in ticks and patient material is, therefore, a basic requirement for further research on this topic. However, only a few studies on the distribution of the OspA types in ticks or patient material are available [9, 49, 50].

Thus, rapid and sensitive methods are needed for detection and differentiation of the clinically relevant B. burgdorferi s.l. species and OspA types. Various methods for typing of B. burgdorferi s.l. such as RFLP [19, 28], PFGE [4, 5], reverse-line blot [23], and LightCycler [31] have been introduced in the past. However, none of these methods allows a reliable differentiation of the B. garinii-associated OspA types. Here we describe and evaluate a method for sensitive detection and reliable differentiation of all relevant European B. burgdorferi s.l. species and OspA types based on a heminested ospA-PCR with subsequent restriction enzyme analysis.

Material and methods

B. burgdorferi s.l. strains and cultivation

Thirteen strains comprising all European species and OspA types (Table 1) were cultured at 33°C in MKP medium [29]. Cells were harvested at a density of 107 cells/ml by centrifugation (12,000 rpm, 20 min) washed three times in 200 μl phosphate-buffered saline (PBS, pH 7.4) and finally resuspended in 200 μl PBS.

Table 1. Predicted RFLP-pattern for the different Borrelia burgdorferi species and OspA types (RFLP restriction fragment length polymorphism, OspA outer surface protein A, s.s. sensu stricto)
Full size table

Collection regions and ticks

A total of 529 Ixodes ricinus ticks were collected by flagging at three locations in Southern Germany in spring 2000, 2001 and 2002 (Table 2). The collection regions had a size of about 500 m2 each. Regions EG1 and EG2 have a distance of about 1 km and are both located in the English garden, a recreational park with woods, lawns and hedges, located in the city of Munich. The collection area Tölz is located in a countryside forest about 40 km south of Munich. After collection, the ticks were separately stored in 1.5-ml reaction tubes at –20°C until use. Before DNA extraction each tick was crushed, using a separate pipette tip.

Table 2. Prevalence rates of B. burgdorferi s.l. at the different locations [s.l. sensu lato, EG1 English Garden Munich area 1, EG2 English Garden Munich area 2, Tölz Town Bad Tölz, (Southern Germany), 2000/2001/2002 year of collection]
Full size table

DNA extraction

DNA from the cultured strains and the ticks was extracted with 'High Pure PCR Template Preparation Kit' (Roche Molecular Biochemicals, Mannheim, Germany) according to the manufacturer's instructions. The final volume of eluted DNA was 200 μl.

Polymerase chain reaction

The presented PCR was developed on the basis of a protocol of Trebesius et al. [39]. Published and additionally designed primers were evaluated using a DNA sequence alignment consisting of 34 published complete ospA sequences (obtained from the National Centre for Biotechnology Information) and 17 complete sequences performed in our laboratory, representing all relevant European Borrelia species and OspA types. For optimization of the PCR protocol, annealing temperatures from 45.1°C to 56.9°C were tested in a gradient cycler (Mastercycler gradient, Eppendorf, Hamburg, Germany); MgCl2 concentration was varied from 1.5 to 2.1 mM.

The optimized PCR mixture (total volume, 50 μl) contained 5 μl isolated DNA, 5 μl of 10× buffer (Roche Molecular Biochemicals, Mannheim, Germany), 200 mM of each nucleotide (Roche Molecular Biochemicals, Mannheim, Germany), 10 pmol of each primer V1a, V1b, R1, R37 (Table 3) (Metabion, Martinsried, Germany) and 0.5 U Taq polymerase (Roche Molecular Biochemicals). PCR was done in an automated DNA thermal cycler (Gene Amp PCR Systems 2700, Applied Biosystems, Weiterstadt, Germany). Initially, DNA was denatured at 95°C for 5 min and then amplified for 30 cycles, each consisting of denaturation at 94°C for 45 s, primer annealing at 48°C for 45 s, and extension at 72°C for 1 min, followed by a final extension for 7 min. The heminested amplification with primers V3a,V3b, R1, and R37 (Table 3) (100 pmol each) was performed with 5 μl of the amplification product from the first PCR using the same conditions as outlined for the first PCR.

Table 3. PCR primer
Full size table

The amplified products were visualized on a 2% agarose gel (Sea Kem LE agarose, Biozym, Hessisch Oldendorf, Germany) stained with 1 mg/ml ethidium bromide (Bio-Rad, Munich, Germany) and documented with a Gel documentation System (Herolab, Wiesloch, Germany).

To determine the detection limit of the PCR assay, tenfold serial dilutions of nine reference strains representing B. burgdorferi s.s., B. afzelii, B. valaisiana, B. lusitaniae, and all five OspA types of B. garinii from 107 to 103, 2×102 and 2×101 borreliae/ml (corresponding to 5×104 to 5, 1, 0.1 borreliae/PCR, respectively) were tested twice. The quantity of borreliae was determined by dark-field microscopy since insufficient growth of borreliae on solid media prevents control by colony-forming units. To determine the sensitivity of the method for the detection of multiple infections, combinations of different Borrelia species (B. burgdorferi s.s. with B. afzelii, B. burgdorferi s.s. with B. garinii, B. burgdorferi s.s. with B. valaisiana, B. afzelii with B. garinii, and B. valasiiana with B. garinii) were tested in different relative concentrations. To assure specificity related spirochetes (B. hermsii, B. duttonii, B. anserina, B. parkeri, B. turicatae, B. recurrentis, B. litoralis, eight Leptospira serovars as well as Treponema phagedenis and T. denticola) were tested at a concentration of 107 spirochetes/ml each.

Restriction enzyme analysis

Aliquots of each amplification product (7 μl) were digested separately with 0.5 U of the restriction enzymes Kpn21 (MBI Fermentans, Lithuania), BglII, SspI, HindIII, and SfuI (Roche Molecular Biochemicals) overnight according to the manufacturer's instructions. The digested products were visualized and documented as described for the PCR amplification products (Fig. 1).

Fig. 1.
figure 1

Restriction fragment length polymorphism analysis of Borrelia burgdorferi sensu stricto, B. afzelii, B. valaisiana, B. lusitaniae and the five different OspA types of B. garinii with the restriction enzymes SspI, SfuI, BglII, Kpn21, HindIII (M molecular weight marker VIII, OspA outer surface protein A)

Full size image

Statistical analysis

Fisher exact test (EpiInfo Version 6, [7]) was performed for statistical analysis of the obtained data.

Results

Optimization and evaluation of the PCR

Optimized PCR conditions are outlined in Material and methods. Sensitivity testing performed with the published primers [39] revealed that B. valaisiana, B. lusitaniae and B. garinii OspA types 3 and 7 were insufficiently detected (>10 borreliae/PCR). Replacing the forward primers of the seminested amplification V2a and V2b by the newly designed primers V3a and V3b improved the detection of B. valaisiana and B. lusitaniae. Addition of the reverse primer R37 improved the detection of OspA types 3 and 7. After optimization, the detection limit of the PCR assay was 2×102 borreliae/ml (corresponding to 1 borrelia/PCR) for all Borrelia species and different OspA types except B. afzelii, where the limit was 103 borreliae/ml (corresponding to 5 borreliae/PCR). The same sensitivity was found when combinations of different OspA types were tested. Positive PCR resulted in a single DNA fragment between 798 and 807 bp, depending on the OspA type. Related Borrelia spp., Treponema spp. and Leptospira spp. were not amplified at a concentration of 107 spirochetes/ml (5×104 spirochetes/PCR).

Evaluation of the restriction enzyme analysis

The restriction enzyme analysis of the amplicons of the tested strains resulted in typical RFLP patterns for B. burgdorferi s.s., B afzelii, B. valaisiana, and B. lusitaniae, whereas B. garinii strains were found to be heterogeneous (Table 1; Fig. 1), corresponding to the high diversity of ospA sequences present in B. garinii [46]. Each of the five OspA types associated with B. garinii [49] was represented by one RFLP pattern.

Typing of mixed cultures with different species and OspA types resulted in a mixture of the characteristic RFLP patterns, which could easily be identified. Combinations of different known concentrations and species were tested to quantify the efficiency of the method for detection of double infections. Double infections were detected and differentiated at a ratio of at least 1:10 to 1:100 for all tested combinations of different Borrelia species.

Ticks

Of the 529 I. ricinus ticks, 67 were tested positive in the ospA-PCR, corresponding to a total infection rate of 13%. Infection rates at the different locations ranged from 5% for Tölz 2000 to 26% for EG1 2000. The infection rates increased from larvae (2–14%) to nymphs (4–23%) and adult ticks (5–42%) (Table 2).

All amplicons of the positive ticks were subjected to RFLP analysis. The most common species in all regions was B. garinii (61% of the positive ticks), followed by B. afzelii (25%) and B. burgdorferi s.s. (11%). B. valaisiana was present at one location only while B. lusitaniae was not detected (Table 4).

Table 4. Distribution of species and OspA types of B. burgdorferi s.l. at the different locations
Full size table

Regarding B. garinii, the most frequent OspA types were type 6 (32%) and type 4 (22%). Type 4 showed an unusual high prevalence at the location EG1 2000 (47%), while at the other locations the prevalence of this OspA type was significantly lower (0–13%; P=0.034). Low infection rates were found for OspA type 5 (6%) and 3 (1%) (Table 4). No OspA type 7 was detected.

The rate of double infections was 7% without obvious preference of any combination of Borrelia species or OspA types. Significant differences in the distribution of the Borrelia species and OspA types were found both between the closely situated locations EG1 and EG2 (OspA type 4 EG1 2000 vs EG2 2000, P=0.003; OspA type 5 EG1 2000 vs EG2 2000, P=0.004), and between the location EG and Tölz (B. afzelii EG1 2000 vs Tölz 2000, P=0.0022), which are about 40 km apart. The distribution also changed at the same location at different collection times (B. afzelii EG1 2000 vs EG1 2002, P=0.0019) (Table 4).

Discussion

The goal of this study was to develop and evaluate a PCR-based method for sensitive detection and reliable differentiation of all European B. burgdorferi s.l. species and OspA types. At present, only a few authors have investigated the distribution of OspA types in ticks or patients and, therefore, data on this issue are limited in number and geographical source [9, 49, 50].

The target of the developed PCR was the ospA gene located on a linear plasmid. This gene has previously been successfully used as a PCR target for detection of B. burgdorferi s.l. in patient material and ticks [13, 25, 26, 30, 35, 41]. A special advantage for this target gene might be a possible overrepresentation of B. burgdorferi plasmid sequences compared to genomic DNA in clinical specimen [27]. Furthermore, the heterogeneity observed among ospA sequences of different European strains, especially regarding the species B. garinii [46, 49], can be used for specification and differentiation of the amplification product. Since the expected rate of transfer and recombination of ospA between different Borrelia species is rather low [8, 36], clustering strains based on the sequence of ospA is usually accepted and in most cases corresponds to classification based on conserved chromosomal genes [44].

Evaluation of the PCR revealed a high analytical sensitivity and specificity for all European OspA types. In this study a heminested DNA amplification was performed. A nested PCR procedure is a common tool to increase sensitivity and specificity of the amplification process [14, 52]. The developed PCR has been shown to be highly sensitive, detecting one to five spirochetes per PCR. Sensitivity testing was performed with nine defined strains, representing not only all the relevant European species of B. burgdorferi s.l., but also each of the five OspA types of B. garinii, which seems to be necessary regarding the heterogeneity of the ospA gene among this species [46]. This idea is further underscored by the results of our PCR optimization. However, to our knowledge none of the described ospA-PCR protocols has been evaluated with strains of all B. garinii OspA types [9, 18, 30, 31, 35]. Furthermore, analysis of the DNA sequence alignment revealed that, for the development of the RFLP analysis, a long amplicon is necessary, to include the different restriction enzyme cutting sites. By contrast, the published primers allow only the amplification of a shorter part of the ospA gene. By testing a large panel of related spirochetes, a high specificity of the amplification for the B. burgdorferi s.l. complex was demonstrated.

RFLP analysis reliably discriminates the different OspA types, including multiple infections. PCR-based RFLP analysis has been successfully used previously for further specification and differentiation of B. burgdorferi s.l. It can be applied to cultured and uncultured specimens, is simple to perform, and therefore a practical tool for epidemiological and population genetic studies [44]. Digestion of rrl (23S)–rrf (5S) intergenic spacer amplicons is widely used, and allows a reliable differentiation of B. burgdorferi s.l. genospecies [28, 42, 43]. Digestion of the 16S–23S rDNA spacer [19] even allows a further sub-differentiation of B. burgdorferi s.s., the most prevalent species in the United States. Again, none of the existing protocols allows a further differentiation of B. garinii, the most prevalent European species [16]. However, several studies have stressed the importance of the heterogeneity among European B. garinii strains, and have led to the definition of at least five OspA types among this species [46, 49, 50, 51].

The RFLP analysis of the ospA amplicon described here not only allows a reliable differentiation of the European B. burgdorferi s.l. species, but also a sub-differentiation of B. garinii into five groups, corresponding to the five OspA types associated with B. garinii [49]. In addition to the experimental evaluation, analysis of all referred sequences for the cleavage sites of the used restriction enzymes was performed. This revealed possible RFLP variants for B. burgdorferi s.s., B. lusitaniae, B. valaisiana and B. garinii OspA type 7, which indicates that a further sub-differentiation of these types might be possible (Table 1). However, none of these RFLP variants were found among the investigated reference strains and ticks.

Furthermore, a high efficiency of the described method for detection and differentiation of multiple infections with different Borrelia species as well as subtypes could be demonstrated. Multiple infections have been described both in ticks and in patient material [34, 35, 50]. Detection and differentiation of multiple infections can be difficult, depending on the method. There is some evidence that, in case of culture, OspA serotype 4 present in multiple infections might be overgrown by other types [15, 50]. In a study by Liveris et al. [20] the rate of mixed infections in clinical specimens was significantly higher by direct RFLP genotyping, compared to genotyping following prior culture isolation. Thus, assuming an equal amplification of the different types, PCR-based RFLP analysis can bypass such difficulties when detecting and differentiating multiple infections.

OspA types showed a broad distribution in ticks. To test the reliability of the developed protocol and to examine the prevalence and the heterogeneity of different Borrelia species and OspA types, 529 I. ricinus ticks were collected from three different regions in Southern Germany and tested by PCR and RFLP analysis. Our results suggest considerable local differences in prevalence and distribution of the different types of B. burgdorferi s.l. Local prevalences varied from 5% to 26%, similar to previously reported rates for Southern Germany [12, 13, 48]. An unusually high infection rate (42% in adult ticks) was found at location EG1 in the English Garden, a recreational park located in the center of Munich, which has been described for this area in 1985 (34%) and 1997 (38%) [12, 13, 48], indicating a constantly high prevalence. Notably these infection rates in ticks showed no correlation to anti-Borrelia antibody rates in the general population of the respective regions determined in a previous study [32]. The most likely explanations are a relatively high occupational exposure (farmers, forest workers) in the rural area of Bad Tölz compared to the urban area of Munich, differences in the recreational activities or not representative selection of the tick collection areas. Regarding the great differences in prevalence rates of closely located areas and the high mobility of the population a larger number of areas must be investigated to obtain a more representative picture.

Since different species and OspA types of B. burgdorferi s.l. have different pathogenic potential [40, 49, 51], information on their distribution in ticks is a basic requirement for local risk assessment [34]. Corresponding to previous European studies, four species of B. burgdorferi s.l., i.e., B. garinii, B. afzelii, B. burgdorferi s.s. and B. valaisiana, were detected in this study [33]. Consistent with other findings in Europe, the most frequently detected species was B. garinii [16]. Further differentiation showed that, except OspA type 7, all different OspA types associated with B. garinii were present in the ticks. However, an OspA type 7 strain (T25) has previously been isolated from a tick collected in the English garden [49]. It is remarkable that the most frequent B. garinii OspA type was OspA type 6. This OspA type seems to be less frequent in patient material, a finding which might reflect a low pathogenic potential of this type [49]. Surprisingly, OspA type 4 was found at an unusually high rate at location EG1, in contrast to very low rates at the other locations. Previously, OspA type 4 has rarely been cultivated from ticks [15], although it has been frequently cultivated from CSF of patients with neuroborreliosis [24, 49, 50]. The focal prevalence of OspA type 4 detected in this study, might be an explanation for the previous difficulties in cultivating this type from ticks. Since OspA type 4 was found in ticks by xenodiagnosis, methodical difficulties in cultivating this type from ticks might be another reason [15].

Interestingly the RFLP pattern obtained from one tick material did not match with any of the tested reference strains, but matched the predicted RFLP pattern according to the published sequences of the Borrelia strain A14S [43] and the B. valaisiana strain M53 [45]. Sequence analysis of the ospA gene showed that it was equivalent to the Borrelia strain A14S. This strain, described by Wang et al. [43], most likely represents a new species [44] and seems to be pathogenic for humans since it was cultivated from a skin biopsy of a patient with erythema migrans from the Netherlands [43]. The detection of this type, which already has been documented in ticks [31] and even in a patient with chronic skin disease [47] from Germany, indicates the need for methods which can assess the distribution of the genospecies Borrelia A14S throughout Europe.

Due to a great heterogeneity of the different species and OspA types found between closely located areas and even in the same area at different collection times, no local predictions about the prevalence of the different species or subtypes can be made. Overall, a broad variety of species and OspA-types was found, indicating that in terms of the development of a European vaccine or serological tests, no species or OspA type can be excluded. Additional studies with larger numbers of ticks and collection areas have to be made to get a more representative picture of distribution and local heterogeneity of the respective OspA types.