INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Spirochetes of the genus Borrelia are host-associated microorganisms that cycle between vertebrate hosts by means of arthropod vectors (10). Borrelia burgdorferi sensu lato, including B. burgdorferi (16), Borrelia garinii (5), Borrelia afzelii (18), and possibly several other closely related genospecies, are the etiologic agents of Lyme disease (50). A phylogenetically distinct group that includes Borrelia hermsii and Borrelia turicatae (39) are causes of relapsing fever (10).

The genome of Borrelia species consists of a 1-Mb linear chromosome (13, 25) complemented by a large number of linear plasmids (lp's) and circular plasmids (cp's) 10 to 60 kb in size (7, 46, 56, 61). lp's and cp's have been shown to share common sequences (8, 47, 48, 56, 61, 63, 64) and, in the case of a B. hermsii plasmid, can exist in both linear and circular forms (26). As lp's and the chromosome appear to have exchanged genetic information (21) and some of the cp's carry genes usually found on chromosomes of other prokaryotes (36), Borrelia plasmids have been likened to minichromosomes (6).

Both lp's and cp's are close to if not equimolar with the chromosome (32, 34). Peculiar in this respect is a family of cp's about 32 kb in size (cp32s) which are almost identical in genetic information. In strain B31, eight of a total of nine distinct plasmids (numbered cp32-1 to cp32-9) have been identified within a clonal bacterial culture (22, 56). Similar cp32s or variants thereof have been found in other B. burgdorferi sensu lato isolates (24, 28, 37, 47, 54). Recently, B. hermsii was shown to contain multiple related cp32s (51). This presence of several very similar yet not identical plasmid sequence entities within one cell leads to a host of paralogous gene families.

Three separate cp32-encoded loci have been sequenced and analyzed to different extents. The most in-depth studies to date have dealt with members of the Erp lipoprotein family, which are related to OspE and OspF, differentially expressed and immunogenic in mammals (52, 56). Two open reading frames (ORFs) originally identified as part of a genetic locus termed 2.9 present on seven cp32s of B. burgdorferi 297, 2.9orfA and 2.9orfB (37), have recently been shown to code for proteins with hemolytic activity in B. burgdorferi B31 (BlyA and BlyB [30]).

A third locus consisting of five ORFs (ORF-1, -2, -C, -3, and -E) flanked by inverted repeats was identified in part by Dunn and colleagues (24) and in our earlier studies on repeated DNA of Lyme disease spirochetes (48, 63, 64). The functions of the ORFs remain unknown. ORF-2 has been postulated to represent a RepC homolog (24), while ORF-C resembled bacterial proteins involved in plasmid partitioning (8, 53, 64). The last ORF in this locus, originally named ORF-E, featured multiple internal 33- and 21-bp-long direct in-frame repeats, suggesting a repeated protein motif domain. Two complete ORF-E copies were identified on one of the multiple homologous cp32s and the related lp56 of B. burgdorferi B31. The predicted polypeptides had 64% overall amino acid identity. Intriguingly, they varied in the number of direct repeat units and thus in size (24.1 and 20.6 kDa, respectively) (64). We hypothesized that these two ORF products represented members of a protein family. To date, a total of 29 ORF-E homologs have been described under various names in strain B31 (20, 28) and other B. burgdorferi sensu lato isolates (1, 37, 54, 57) as well as other Borrelia species (19, 51).

To define the common and divergent characteristics of this protein family, we first undertook a comprehensive comparison of all known paralogs (i.e., intraspecies homologs) and orthologs (i.e., interspecies homologs). We then investigated their possible biological functions, using polyclonal antibodies raised against each of two B. burgdorferi B31 paralogs. Finally, we examined whether these proteins are targeted by the antibody response during Lyme disease. Our results demonstrate that they indeed represent a class of cross-reactive Borrelia proteins that are expressed in vitro and in vivo. Based on their common features, we suggest a common, unambiguous name, Bdr (for Borrelia direct repeat) for all members of this family.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

DNA sequence computer analysis. DNA and protein sequences were obtained from the database sources listed in Table 1. Predictions of protein molecular weight, isoelectric point (pI), and secondary structure features as well as protein sequence identity comparisons (Higgins default values) were performed with the MacDNASIS Pro program suite (version 3.6; Hitachi Software Engineering). Unrooted phylogenetic trees were constructed with the Clustal X program suite (58): multiple pairwise sequence alignments (default values) were subjected to neighbor-joining and bootstrap analysis (number of bootstrap trials = 1,000). The resulting tree data were plotted with the DRAWTREE algorithm in the PHYLIP program suite (version 3.5; J. Felsenstein, University of Washington, Seattle). Transmembrane domains were predicted with the TMPred program (33).

Bacterial strains and culture conditions. B. burgdorferi B31 (ATCC 35210), Sh-2-82 (43), N40 (27), and HB19 (12), B. garinii Ip90 (5), B. afzelii ACA1 (3), B. hermsii HS1 (ATCC 35209), B. turicatae Oz1 (17), and the Osp-less B. burgdorferi B31 derivative B313 (41) were grown in BSKII medium supplemented with 12% rabbit serum at 24 or 34°C. Escherichia coli XL1-Blue (Stratagene), DH5 (Gibco), BL21(DE3)(pLysS) (Novagen), and InvF' (Invitrogen) were grown in LB Lennox medium at 30 or 37°C.

Nucleic acid analysis. B. burgdorferi plasmid-enriched DNA was obtained by diethylpyrocarbonate extraction (9) and dot blotted onto nylon membranes (Nytran; Schleicher & Schuell), using standard procedures (4). DNA probes specific for B. burgdorferi B31 cp32-1 through cp32-9 (22, 56) were enzymatically labeled and used in Southern hybridizations (ECL [enhanced chemiluminescence] direct labeling and detection system; Amersham). Hybridizations were carried out in glass tubes in a Mini hybridization oven (Autoblot; Bellco) under standard stringency conditions (0.5× SSC in primary wash; 20× SSC is 0.3 M sodium citrate plus 3 M NaCl [pH 7]). PCRs with oligonucleotide primer pairs E155 plus E168 (cp32-2 specific) and E703 plus E704 (cp32-7 specific) were performed as described by Stevenson et al. (52, 53). DNA probes and primers were a gift from Brian Stevenson, University of Kentucky, Lexington.

Recombinant Bdr (rBdr) proteins. bdrA and bdrV (Table 1) were PCR amplified from recombinant plasmids pOMB14 and pOMB65 (63, 64), using Pwo polymerase (Boehringer Mannheim) and primer pairs bdrA-fwd (5'-GGAATTCCATATGACTAATTTAGCGTAC-3') plus bdrA-rev (5'-CCGAATTCTCATTAATGATGATGATGATGGTGTTTGATATATTG-3') and bdrV-fwd (5'-GGAATTCCATATGAATAGTTTGAC-3') plus bdrV-rev (5'-CCGAATTCTCATTAATGATGATGATGATGGTCTTTGCCACCTTGT-3'), respectively. Fragments were digested with NdeI and EcoRI, ligated into pET29b (Novagen), and cloned in E. coli XL1-Blue (Stratagene), resulting in recombinant plasmids pWRZ101 (bdrA) and pWRZ102 (bdrV).

Whole-cell protein preparations. Borrelia whole-cell lysates were prepared from early-stationary-phase BSKII cultures. Cells were harvested by centrifugation for 30 min at 2000 × g (Sorvall RT7), washed twice with phosphate-buffered saline (PBS) containing 10 mM MgCl₂ (PBS-Mg), concentrated 50-fold by resuspension in sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) loading buffer, and boiled for 5 min. Whole-cell protein preparations from low-passage infectious B. burgdorferi B31 initially grown at 23°C and then shifted to 35°C (55) were a gift from Brian Stevenson. For assaying surface exposure by proteinase K accessibility, intact or sonicated cells were treated with 200 µg of proteinase K (Boehringer Mannheim) per ml of cell suspension as previously described (11). Fivefold-concentrated E. coli whole-cell proteins were prepared as described above.

Human sera, rabbit sera, and monoclonal antibodies. A panel of reference sera from North American individuals with 42 documented cases of Lyme disease and five control sera was provided by Barbara Johnson (Centers for Disease Control and Prevention [CDC], Fort Collins, Colo.). The cases were clinically classified into the stages of early localized or disseminated and persistent infection according to previously described guidelines (49). Additional patient sera included in this study were from four Swiss individuals with disseminated Lyme disease (provided by Peter Erb, Institute for Medical Microbiology, University of Basel, Basel, Switzerland) and a New Jersey patient with chronic Lyme arthritis (provided by Lisa Dever, Veterans Affairs Medical Center, Orange, N.J.). Sera from seven healthy adults living in an area where Lyme disease is not endemic and a commercially available pool of normal human sera (Omega Scientific) were used as additional controls.

SDS-PAGE and Western immunoblot analysis. Proteins were separated and visualized on SDS-12% polyacrylamide gels as described above. For Western immunoblots, proteins were electrotransferred (Bio-Rad Trans-Blot) to nitrocellulose membranes (Immobilon-NC; Millipore). Membranes were rinsed in Tris-buffered saline (TBS; 20 mM Tris, 500 mM NaCl [pH 7.5]) and either air dried or processed directly; 5% dry milk in TBST (TBS with 0.05% Tween 20) was used for membrane blocking and subsequent incubations (1 h each), and TBST was used for intermittent washes. Human patient and control sera were used at 1:100 dilutions, rabbit antisera were used at 1:500 (anti-Bdr) and 1:5,000 (anti-OspA) dilutions, and mouse monoclonal antibody hybridoma supernatants (anti-OspC) were used at a 1:10 dilution. Alkaline phosphatase-conjugated protein A/G (ImmunoPure protein A/G-calf intestinal phosphatase conjugate; 1:5,000; Pierce) as the second ligand and a stabilized alkaline phosphatase substrate solution (1-Step NBT/BCIP; Pierce) were used for colorimetric detection.

IFA and GIA. For indirect immunofluorescence assays (IFA) with unfixed cells, B. burgdorferi cells were washed as described above, resuspended in PBS-Mg containing 2% bovine serum albumin (BSA) (PBS-Mg-BSA) to a concentration of 2 × 10⁷ cells/ml, and incubated with dilutions of polyclonal rabbit sera at room temperature (RT) for 1 h with gentle rotation. After a wash with PBS-Mg-BSA, the cells were suspended in a 1:15 dilution of the secondary antibody, fluorescein-labeled goat anti-rabbit immunoglobulin G (heavy plus light chain) [IgG(H+L); Kirkegaard & Perry Laboratories], and incubated for 30 min. Subsequently, the cell suspension was adjusted to a volume of 300 µl with PBS-Mg-BSA and mounted for microscopy.

ELISA. Tissue culture flat-bottom 96-well plates (Costar) were coated overnight with rBdrA (200 ng/well) in carbonate buffer and blocked with 1% dried nonfat milk in PBS pH 7.4 (1% milk-PBS) for 2 h. Subsequently, the plates were incubated for 2 h with dilutions of sera in 1% milk-PBS followed by a 1-h incubation with a 1:1,000 dilution of alkaline phosphatase-conjugated goat anti-human IgG(H+L) antibodies (Kirkegaard & Perry Laboratories). All incubations were carried out at RT with intermittent washes with PBS pH 7.4. Detection was carried out in the dark, using 1 mg of Sigma 104 phosphatase substrate per ml in DEA buffer (9.7% diethanolamine, 0.01% MgCl₂, 0.02% sodium azide [pH 9.8]). Plates were read on a SPECTRAmax 340 microplate reader (Molecular Devices) at 405 nm, using 490 nm as reference, and analyzed with the SOFTmax PRO software package (version 1.2.0 for Macintosh). Enzyme-linked immunosorbent assay (ELISA) titer values 3 standard deviations (SD) above the mean of the negative control values were considered positive.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Comparative analysis of the Bdr protein family. To further define the common and divergent characteristics of the ORF-E-related protein family, we comprehensively compared the known paralogs and orthologs. Including the two initially identified copies on one of the multiple homologous cp32s and the related lp56 of B. burgdorferi B31 (ORF-E [64]), a total of 29 related sequences have been determined, mostly as part of the B. burgdorferi B31 genome sequencing project but also in other B. burgdorferi sensu lato strains and the relapsing fever spirochetes B. hermsii and B. turicatae (Table 1). Strain B31 ORFs BBF03 and BBJ10 (28) are likely pseudogenes, since they do not possess a consensus Shine-Dalgarno sequence upstream of the AUG start codon, and were therefore excluded from further analysis.

View larger version (61K): [in this window] [in a new window]   FIG. 1.   Features and comparison of Bdr paralog and ortholog proteins. (A) Features of Bdr proteins, showing B. burgdorferi B31 BdrA as an example. Amino acid positions: 1 to 42, N-terminal domain; 43 to 179, central direct repeat domain; 180 to 210, C-terminal domain. Residues in italics are putative phosphorylation sites. Boxed rectangles indicate the 7- and 11-aa repeat units. The black bar with white letters indicates the predicted C-terminal transmembrane (Tm) domain. (B) Comparative analysis of known Bdr paralogs and orthologs (Table 1). B. burgdorferi B31 paralogs are listed in bold. Protein sequence alignments were bootstrapped by using Clustal X. An unrooted phylogenetic tree was drawn by using the PHYLIP DRAWTREE algorithm. (C) Comparison of Bdr paralogs and orthologs. Sequences are listed according to the homology groups in panel B. Levels of homology of N and C termini are indicated with letters and numbers as follows: different letters separate groups of sequences with more than 30% amino acid identity within a group; different numbers separate subgroups of sequences with more than 60% amino acid identity among members. Boxed shaded triangles indicate the 7- and 11-aa repeat units in the central direct repeat domain.

Heterologous expression and cross-reactivity of BdrA and BdrV. Based on the common features and homologies in the direct repeat domain, we surmised that the Bdr proteins could constitute a cross-reactive protein family. To test this hypothesis with two closely related proteins, we chose two previously cloned group I paralogs from strain B31, BdrA and BdrV (Table 1; Fig. 1B) (64), for overexpression in E. coli. Whole-cell preparations of the E. coli harboring the expression construct pWRZ101 or pWRZ102 showed that BdrA and BdrV could be expressed in this system (Fig. 2A). The observed sizes corresponded to the expected molecular masses of BdrA (24.1 kDa) and BdrV (20.6 kDa).

View larger version (49K): [in this window] [in a new window]   FIG. 2.   Heterologous expression and cross-reactivity of B31 Bdr proteins. (A) Coomassie blue-stained SDS-12% polyacrylamide gel. Lanes: m, molecular weight marker (Gibco high molecular weight); 1, B. burgdorferi B31 whole-cell lysate; 2, E. coli BL21(DE3)(pLysS)(pET29b) whole-cell lysate; 3, E. coli BL21(DE3)(pLysS)(pWRZ102) whole-cell lysate (expressing rBdrV); 4, E. coli BL21(DE3)(pLysS)(pWRZ101) whole-cell lysate (expressing rBdrA); 5, purified rBdrA. Asterisks indicate rBdrV and rBdrA. (B) Western immunoblot with anti-BdrA, anti-BdrV, and control rabbit sera (1:500). Lanes are as in panel A.

Further characterization of Bdr proteins. Since the expression of many immunogenic Borrelia proteins (e.g., OspC) is differentially regulated by environmental factors such as temperature, we assayed for differences in Bdr expression in cultured B. burgdorferi B31 in cultures grown at a constant temperatures of either 24 or 37°C. As shown in Fig. 3A, Bdr expression was essentially the same at 24 and 37°C. Interestingly, protein bands in the 35- to 60-kDa range, likely corresponding to Bdr paralog dimers, were seen in lysates of cells grown at 24°C but not at 37°C. We repeated the experiment, this time assaying for Bdr expression immediately after a temperature shift from 23 to 35°C, which leads to an upregulation of OspC (44, 55). As expected, we detected OspC at 35°C but not at 23°C. The level of Bdr proteins, however, remained constant under these conditions as well (results not shown).

View larger version (77K): [in this window] [in a new window]   FIG. 3.   Lack of temperature-dependent expression and proteinase K susceptibility of Bdr proteins. (A) Western immunoblot of an SDS-12% polyacrylamide gel with whole-cell lysates of B. burgdorferi B31 grown at 24 or 37°C probed with anti-BdrA antiserum (1:500). Protein sizes indicated to the left are derived from a molecular weight marker (Gibco high molecular weight). (B) Proteinase K susceptibility of Bdr proteins. Coom. Blue, Coomassie blue-stained SDS-12% polyacrylamide gel with whole-cell lysates of B. burgdorferi B31 either untreated () or previously treated (+) with proteinase K (200 µg/ml of cell suspension); anti-BrdA and anti-OspA, Western immunoblots of gels with anti-BdrA (1:500) and anti-OspA (1:20,000) antisera. An asterisk indicates OspA.

View larger version (104K): [in this window] [in a new window]   FIG. 4.   Immunofluorescence microscopy of fixed and unfixed B. burgdorferi cells. (A) Methanol-fixed, permeabilized B. burgdorferi B31 cells reacted with anti-OspA (1:1,000), anti-BdrA (1:200), and control (1:200) polyclonal rabbit sera. Secondary antibody was a fluorescein isothiocyanate-labeled goat anti-rabbit antibody. (B) Unfixed, intact B. burgdorferi B31 cells reacted with anti-OspA (1:1,000), anti-BdrA (1:100), and control (1:100) polyclonal rabbit sera.

Cross-reactive Bdr proteins in B. burgdorferi B31. In the immunoblot experiments described above, we also included whole-cell lysates of B. burgdorferi B31. Interestingly, anti-BdrA detected several protein bands with estimated sizes of 30.5, 28, 27, 25.5, 24.5, 24, and 19.5 kDa (Fig. 2B). On a Coomassie blue-stained gel, however, the corresponding bands were barely detectable (Fig. 2A). The range of protein band sizes corresponds with the expected sizes of Bdr paralogs described in this strain (Table 1 and Fig. 1C). To corroborate this observation and to audit the antigen specificity of anti-BdrA, we repeated the experiment including B. burgdorferi B313, a derivative of B31. This strain had been previously shown to lack several lp's, including lp29 (i.e., the lp28s [28]), lp38 (lp36 and lp38), and lp49 (lp54 and lp56), and thus to lack bdrT, bdrU, bdrX, bdrV, and bdrW. Circular plasmids such as cp30 (i.e., the multiple cp32s), however, were still present (41). To determine the B313 complement of bdr paralogs, we used cp32-specific DNA probes (22, 56) and PCR oligonucleotide primers (52, 53) to determine its cp32 plasmid profile.

View larger version (125K): [in this window] [in a new window]   FIG. 5.   cp32 plasmid profile of B. burgdorferi B31 derivative B313. Plasmid-enriched DNA of B. burgdorferi B31 and its derivative B313 was dot blotted with nucleotide probes (1 through 9) specific for the multiple homologous 32-kb-long circular plasmids (cp32-1 through cp32-9). Probes were also blotted as a control for probe specificity. Note that probe 2/7 does not discriminate between cp32-2 and cp32-7. PCR analysis with oligonucleotide primers specific for cp32-2- or cp32-7 revealed that both B31 and B313 carried cp32-2 but not cp32-7 (see text).

View larger version (59K): [in this window] [in a new window]   FIG. 6.   Cross-reacting Bdr paralog and ortholog proteins in other Borrelia isolates. (A) Coomassie blue-stained SDS-12% polyacrylamide gel. Borrelia isolates are indicated above the lanes: m, molecular weight protein marker (Gibco high molecular weight); B. burgdorferi sensu stricto isolates B31, B313, Sh-2-82, N40, and HB19; B. garinii Ip90; B. afzelii ACA1; tick-borne relapsing fever isolates B. turicatae Oz1 (Bt) and B. hermsii HS1 (Bh). (B) Western immunoblot of an identical gel with anti-BdrA antiserum (1:500). Lanes are as in panel A.

Cross-reactive Bdr proteins in Borrelia species. As mentioned above, genes belonging to the B. burgdorferi B31 Bdr family have also been found in other Lyme disease or relapsing fever Borrelia isolates (Table 1). We therefore checked for the presence of cross-reacting proteins in three additional B. burgdorferi strains, Sh-2-82, N40, and HB19, as well as in B. garinii Ip90, B. afzelii ACA1, and two relapsing fever spirochete species, B. turicatae Oz1 and B. hermsii HS1. Immunoblots with anti-BdrA against whole-cell lysates revealed proteins in every isolate tested, ranging in size from approximately 17.5 to 33 kDa (Fig. 6B). This indicates that the cross-reactivity of the Bdr protein family also extends to other Borrelia species in general. Intriguingly, each of the isolates had a distinct immunoreactive protein pattern ranging from, e.g., only one 27-kDa protein band in Ip90 to seven bands ranging from 18 to 30 kDa in HB19.

Antibody response to Bdr in Lyme disease. Many of the plasmid-encoded, immunogenic proteins studied so far have proven to be antigens expressed at one point during the parasitic life cycle of Borrelia, either in the tick vector or in the vertebrate host. We therefore determined whether Bdr was immunogenic during the course of Lyme disease. For that purpose, we obtained 4 European and 43 American Lyme disease patient sera as well as 13 human control sera; 42 of the patient sera were from a well-characterized CDC Lyme disease patient serum panel. The sera were first used in an immunoblot assay against whole-cell lysates of E. coli expressing rBdrA and rBdrV. None of the control sera produced reactive bands the sizes of rBdrA and rBdrV, whereas 51% (24 of 47) (chi square = 9.04, P = 0.003) of the patient sera reacted with both BdrA and BdrV, indicating that Bdr proteins are indeed immunogenic. Representative immunoblots for a positive serum as well as a negative control serum are shown in Fig. 7A.

View larger version (24K): [in this window] [in a new window]   FIG. 7.   Western immunoblots and ELISA with Lyme disease patient sera. (A) Left, Western immunoblot with a representative positive Lyme disease (LD) patient serum (1:100). Lane 1, B. burgdorferi B31 whole-cell lysate; lane 2, E. coli BL21(DE3)(pLysS)(pET29b) whole-cell lysate; lane 3, E. coli BL21(DE3)(pLysS)(pWRZ102) whole-cell lysate (expressing rBdrV); lane 4, E. coli BL21(DE3)(pLysS)(pWRZ101) whole-cell lysate (expressing rBdrA); lane 5, purified rBdrA. Right, Western immunoblot of an identical blot with a representative human control serum. (B) Distribution of rounded ELISA OD values obtained with LD patient and control sera. A triangle on the x axis indicates the cutoff OD (mean of control sera values plus 3 SD = 0.20) for positive sera.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In this study, we further characterized the Borrelia spp. Bdr protein family. The genes encoding these proteins so far have been found exclusively on plasmids. B. burgdorferi B31, e.g., carries a large number of bdr paralogs on several lp's as well as the multiple homologous 32-kb-long cp's (20, 28). Interestingly, most of the cp32s carry two bdr copies in two different plasmid loci separated by approximately 5 kb (see below and reference 20). Related genes have been found on circular and linear plasmids of other B. burgdorferi sensu lato isolates (1, 37, 54, 57) as well as relapsing fever spirochetes B. hermsii (51) and B. turicatae (19) (Table 1). So far, no homologs have been found in spirochetes other than Borrelia, such as Treponema pallidum (29), or in other prokaryotes, which suggests that this family of proteins is specific for the genus Borrelia.

All Bdr proteins have a characteristic central domain consisting of various numbers of tandem 7- and 11-aa repeat units and thereby alter in size from about 18 to 31 kDa. Another common feature is a variable yet structurally conserved C-terminal putative transmembrane domain. Our comparative analyses show that the currently known members of the Borrelia Bdr protein family fall into five major homology groups, mainly due to variations in their N- and C-terminal domains. Interestingly, the differentiation between homology groups is consistent with their respective linear or circular plasmid locations. Whereas group I and group II homologs are both found on the cp32s (and related plasmids), they are part of different loci: group I paralogs are part of the ORF-1-2-C-3-E locus (24, 64); group II paralogs are located approximately 5 kb upstream, as part of a locus related to the 2.9 locus identified in B. burgdorferi 297 (37).

We have shown that antibodies raised against two B31 Bdr paralogs from the same homology group, BdrA and BdrV, cross-react. Furthermore, anti-BdrA antibodies also detect related proteins in other B. burgdorferi sensu lato isolates and relapsing fever spirochetes. We were able to account for all visible reactive protein bands in strain B31 as well as its derivative B313 by linking them to the encoding bdr paralogs, and the size ranges of proteins seen in the other isolates correspond with each other and the one seen in strain B31. This strongly suggests that the polyclonal antiserum used is specific for members of the Bdr protein family. Given the significant overall sequence variability of Bdr proteins, the observed cross-reactivity is likely due to the more conserved direct repeat domain.

Our studies with strain B31 indicate that several if not all Bdr paralogs are expressed by B. burgdorferi. Expression is likely contemporaneous, as is the case for the Erp proteins encoded downstream of the bdr paralogs (52). Under all conditions examined in this work and in a recent study with the B. turicatae ortholog (19), Bdr expression levels are constant and low compared to other Borrelia antigens. Expression is not upregulated in an OspC-like manner (55) by an increase in temperature. Nevertheless, there could be other environmental factors that lead to a transient upregulation of Bdr expression in the human host.

Several lines of evidence indicate that Bdr proteins are not surface exposed: in intact B. burgdorferi cells, they are not susceptible to proteolytic degradation by external proteinase K and inaccessible to polyclonal antibodies. Due to the lack of a periplasmic export sequence and the structurally conserved C-terminal predicted transmembrane domain, the proteins are predicted to be anchored in the cytoplasmic membrane. In a previous study, Bledsoe and colleagues isolated inner and outer membranes of B. burgdorferi B31 by isopycnic centrifugation (14). They found proteins 20 to 60 kDa in size that appeared to localize specifically to the inner membrane. Most of these proteins had apparent pIs more acidic than that of OspA. Intriguingly, two of the protein clusters identified by two-dimensional nonequilibrium pH gel electrophoresis-SDS-PAGE (II and III) formed a vertical ladder, indicating various sizes (approximately 15 to 32 kDa) but similar pIs. These parameters would agree with the predicted values for the Bdr paralogs expressed by this strain.

In the last part of this study, we showed that Bdr proteins are also expressed in vivo and recognized in the course of Lyme disease. Approximately half of the tested patient sera reacted with one or more Bdr proteins in immunoblots and an IgG-based ELISA. All of a statistically significant number of control sera were negative in both assays. Our data may suggest an increase in seroconversion from early localized to late disseminated Lyme disease. However, this is also true for several other B. burgdorferi antigens recognized in North American as well as European Lyme disease (23, 31), and this difference is not statistically significant due to the low number of sera from the latter patient group.

Among the immunodominant B. burgdorferi antigens are several outer surface proteins (OspA to -F), an outer membrane porin (p66), decorin- and fibronectin-binding proteins (DbpA and BBK32, respectively), a periplasmic flagellar protein (FlaB), a cytoplasmic membrane-associated protein (BmpA), heat shock proteins (DnaK and GroEL), and a protoplasmic cylinder-associated protein (p93) (23, 31). In comparison to these abundant proteins, the individual Bdr proteins are present in relatively scarce amounts in culture-grown B. burgdorferi and therefore might go undetected in standard diagnostic immunoblots. In a recent study, Western blots with patient sera were subjected to densitometric analysis revealing approximately 40 visually distinguishable bands, including several sofar uncharacterized and unassigned immunogenic proteins in the 20- to 30-kDa range (31). It is possible that Bdr paralogs account for some of these protein bands. The polyclonal anti-Bdr sera described here may be helpful in further analysis.

The function of the Bdr proteins remains unclear, as our current knowledge of Lyme disease and relapsing fever bdr homologs is limited. Proteins with similarly sized repeat motifs can be found in both prokaryotic and eukaryotic pathogens. Among others are the surface-located IgA proteases of several Streptococcus species (38), a cell surface antigen of Bacteroides forsythus (45), and a repetitive organellar protein of Plasmodium chabaudi (59). It has been proposed that repetitive domains of antigens direct the immune response towards a less protective T-cell-independent B-cell activation (42). However, a similar function for the Bdr proteins is at odds with the observed lack of surface exposure in vitro.

Alternatively, the presence of some of the bdr genes in plasmid loci containing homologs for plasmid segregation and partition proteins (8, 53, 64) may suggest that Bdr paralogs are part of this machinery. As is the case for other genes in this locus, the observed sequence divergence and thus plasmid specificity of each Bdr paralog might ensure that each cp32 plasmid is segregated properly, resolving plasmid incompatibility issues. The predicted subcellular location of Bdr proteins in the cytoplasmic membrane could be a further hint in this direction. The significance of these and other predicted Bdr features such as the putative phosphorylation sites separated by multiples of alpha-helical turns (19, 57, 62) remains unknown. Together, these questions warrant further investigation of the role of Bdr proteins in both Lyme disease and relapsing fever as well as in their respective etiological agents.