INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Infection by the Borrelia burgdorferi sensu lato complex is spread via the bite of infected ticks and manifests initially as a flu-like illness that, if untreated, can develop into a chronic state consisting of arthritic and neurological complications (25, 41-43). In the United States, the prevailing isolate found in areas where the disease is endemic is B. burgdorferi sensu stricto, although other sensu lato isolates, such as B. andersonii (7, 26), have recently been identified in Missouri and in several Southern states, along with other Borrelia spp. identified in Europe, Eurasia, and Japan (9, 30, 46). The worldwide distribution of the Lyme disease spirochete and its associated morbidity has provided the impetus to develop a vaccine to combat this disease. A great amount of work has been devoted toward testing the efficacy of various B. burgdorferi lipoproteins as protective immunogens; these proteins include OspA, OspB, OspC, OspD, OspE, and OspF and, more recently, DbpA and DbpB (2, 11, 12, 15, 18, 19, 29, 47). OspA was initially tested as a vaccine candidate based on its surface exposure and abundance on in vitro-cultivated B. burgdorferi sensu stricto (3, 12). In the laboratory, tick-infected animals do not generate a significant antibody titer against OspA, suggesting that ospA is not expressed in vivo, thereby implying that OspA vaccination may not protect against natural B. burgdorferi infection (11, 27, 34). Consistent with this finding, humans infected with B. burgdorferi do not exhibit a high titer antibody response to OspA early in infection (5). More recently, OspA has been purported as an arthropod-specific vaccine (11); that is, antibody against OspA binds and kills B. burgdorferi within the midgut of the tick during the blood meal of the immunized host. OspA has been evaluated as a protective immunogen in human trials, and the results indicate that the vaccine has marked efficacy (36, 44). However, the vaccine requires multiple boosts since anti-OspA titers may decrease over time. Because ospA is not expressed in the infected host, B. burgdorferi infection does not function as an immunological boost; as such, a preexisting high titer to OspA is essential for clearance of the spirochetes from the midgut of the tick. More recently, cross-reactivity between OspA and human leukocyte function-associated antigen 1 (hLFA-1) suggests that immunization with OspA may contribute to an autoimmune disorder (17).

The limitations of the OspA vaccine highlight the need to evaluate additional protective antigens, particularly those expressed in mammals, to protect against B. burgdorferi infection. Along these lines, we have identified nine genetic loci of B. burgdorferi that are preferentially recognized by rabbits that have developed infection-derived immunity to low-passage, virulent B. burgdorferi B31 (38). It is conceivable that subsets of these antigens are targets for borreliacidal antibodies and, as such, may function as protective immunogens against B. burgdorferi. Consistent with this idea, one of the clones identified encoded dbpA, which has recently been shown to function as a protective immunogen (18, 19). Yet another of these clones encodes a 451-amino-acid protein antigen that contains 21 consecutive and invariant 9-amino-acid repeats near its amino terminus that we have designated VraA for virulent strain-associated repetitive antigen A (designated BBI16 by The Institute for Genomic Research [TIGR]; see the TIGR website at http://www.tigr.org/tdb/CMR/gbb/htmls/SplashPage.html). In this study we report that VraA is a surface exposed antigen that is slightly induced in B. burgdorferi B31 at 37°C relative to 32°C. We also demonstrate that antibodies directed against either full-length or the amino-terminal half of VraA recognize antigens of various molecular masses in several B. burgdorferi sensu lato isolates. Furthermore, we demonstrate that VraA functions as a partially protective immunogen in the mouse model of Lyme borreliosis.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacteria and plasmids. B. burgdorferi sensu stricto strain B31 was used in the majority of studies presented in this report unless otherwise indicated. All B. burgdorferi sensu stricto and sensu lato isolates used in this study are listed in Table 1. Most strains used were passaged no more than three times in vitro and were subsequently isolated from infected rabbits or mice to determine their infective phenotype unless otherwise indicated. B. burgdorferi sensu lato strains were cultured in BSK II media supplemented with 6% normal rabbit serum (Pel-Freez Biologicals, Rogers, Ark.) at 32°C in a 1% CO₂ atmosphere as previously outlined (38). Clonal isolates of B. burgdorferi were obtained by plating diluted cultures into agarose overlays as previously described (24). Two clones, designated MSK5 and MSK7, were obtained after intradermal inoculation of mice with 10³ B. burgdorferi strain B31 passage 3 and isolation of infected skin at 2 weeks postinfection, followed by plating in BSK II agarose. The entire plasmid profile of both MSK5 and MSK7 was determined by designing primer pairs that were specific for each individual plasmid followed by resolution of the amplimers by agarose gel electrophoresis (24).

PCR. PCR was conducted essentially as described elsewhere (24), and the oligonucleotide primers used are listed in Table 2. Briefly, the template was prepared by pelleting infectious B. burgdorferi B31 (fewer than four in vitro passages) at 5,800 × g for 10 min and washing it in an equal volume of phosphate-buffered saline (PBS), followed by centrifugation and resuspension in sterile water to yield a final concentration of 5 × 10⁶ B. burgdorferi per µl. The sample was then boiled for 5 min, the insoluble material was pelleted by centrifugation (16,000 × g) for 2 min, and the supernatant was placed in a new tube. A 1-µl volume (or a 5 × 10⁶ cell equivalence) was then added to the PCR. In some instances individual colonies of infectious B. burgdorferi were instead added to the PCR. Each appropriate primer set, in a 1 µl volume and at a final concentration of 0.2 µM per primer, was incubated with 1 µl of the appropriate B. burgdorferi template and brought to a final volume of 20 µl by adding 2 µl of sterile water and 16 µl of Gibco-BRL SuperMix (Gaithersburg, Md.) which contains buffer, all deoxynucleotides, and Taq polymerase. Samples were then denatured at 94°C for 1 min, followed by 35 cycles of the following: 94°C for 1 min, 50°C for 1 min, and 72°C for 2 min. Incubation at 72°C for 6 min served as the final extension step in the reaction. Products were resolved on 0.8 to 1% agarose gels buffered in Tris-acetate-EDTA containing 0.5 µg of ethidium bromide per ml.

Construction of the GST-vraA fusion proteins. Three separate glutathione S-transferase (GST)-VraA fusion proteins were constructed by using the oligonucleotide primers listed in Table 3. A plasmid encoding a fusion of GST to full-length, mature VraA that lacked the VraA leader peptide and the initial cysteine residue was constructed by PCR amplification using the primers Sal/GST-FL and Not/GST-FL in conjunction with B. burgdorferi template DNA as indicated above (Sal and Not refer to SalI and NotI restriction enzyme sites engineered at the 5' ends of the oligonucleotides; see Table 3). The Sal/GST-FL and Not/GST-FL primers amplified a 1,302-bp vraA fragment, encoding the mature VraA protein (i.e., minus the leader peptide), and this fragment was subsequently cloned into the pCR2.1-TOPO vector (Invitrogen) and designated pCRL3-FL. The ligated product was transformed into TOP10F' cells (Invitrogen), and transformants were selected for on LB agar containing kanamycin and X-Gal (5-bromo-4-chloro-3-indolyl--D-galactopyranoside) according to the manufacturer's instructions. White colonies (putative positives) were grown in LB broth, and DNA was purified from these cells by alkaline lysis to screen for those containing the desired insert.

Purification of GST-VraA fusion proteins. A representative clone of the three GST-VraA fusions, i.e., full-length VraA (residues 18 to 451 of VraA fused to GST), the amino-terminal half (residues 18 to 260 of VraA fused to GST), and the carboxy-terminal half (residues 253 to 451 of VraA fused to GST) were transformed into BL21(DE3)pLysE (Novagen), and 2 liters of cells (with appropriate antibiotic selection) was grown to early log phase (optical density at 600 nm of 0.3 to 0.4). The same methodology was applied to each recombinant GST-VraA fusion protein separately as outlined below. The appropriate GST-VraA fusion protein was induced by the addition of 1 mM IPTG and subsequent incubation for 2 h at 37°C. The resulting cells were pelleted by centrifugation at 6,000 × g, resuspended in 40 ml of PBS (pH 7.4), and subjected to one cycle of freezing and thawing to convert the cells to spheroplasts. The cells were then subjected to two passages through the French pressure cell at 16,000 lb/in². The clarified supernatant was centrifuged at 3,000 × g for 10 min to pellet the unlysed E. coli, and the resulting supernatant was recentrifuged at 40,000 × g for 20 min to pellet the total membrane component of E. coli. After we determined that all three forms of the GST-VraA fusion proteins were soluble upon induction, the supernatants (i.e., soluble protein) were then separately added to glutathione-Sepharose 4B beads and rocked gently overnight at 4°C to promote binding of the GST-VraA fusion proteins to immobilized glutathione. After the overnight incubation, the beads were pelleted at 500 × g, the supernatant was removed, and the beads were washed with 25 ml of PBS (pH 7.4). The wash step was repeated and, following resuspension with PBS, the resulting suspension was poured into a column and the column was washed with 25 ml of PBS. Following the wash step, 5 mM reduced glutathione in PBS was added, and 1-ml column fractions were collected. Fractions containing the various GST-VraA proteins were identified by subjecting 2.5 µl of each fraction to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), followed by staining with Coomassie blue and destaining. Fractions containing the different recombinant GST-VraA samples were further purified from low-molecular-weight contaminants by preparative SDS-PAGE. Briefly, 1 mg of the given GST-VraA fusion protein was resolved on a 16-cm-by-20-cm SDS-polyacrylamide gel (Bio-Rad, Richmond, Calif.) with a 13-cm wide preparative gel comb, and the gel was stained with Coomassie blue in double-distilled, sterile water. The band corresponding to the appropriate GST-VraA molecule was excised from the gel and electroeluted using the S&S Elutrap Electro-separation System from Schleicher & Schuell, Inc. (Keene, N.H.) as outlined by the manufacturer.

Antibody production. Three separate rabbits were prebled and subsequently immunized with each of the three affinity-purified GST-VraA fusion proteins obtained. In all cases, 100 µg of the GST-VraA protein, in complete Freund adjuvant, was injected subcutaneously (four sites, 25 µg per site). After 6 weeks, the rabbits were boosted with 100 µg of the same GST-VraA protein in incomplete Freund's adjuvant. Two weeks later the rabbits were bled to obtain serum specific for either full-length VraA fused to GST, the amino-terminal half of VraA fused to GST, or the carboxy-terminal half of VraA fused to GST. Each serum was tested qualitatively by Western blotting to determine the specificity of the antiserum for both homologous and heterologous forms of recombinant GST-VraA, as well as native VraA (see below).

Protease accessibility. Protease accessibility studies were conducted as previously described (4) with the following modifications. B. burgdorferi B31 passage 7 cells (approximately 2 × 10⁸) were centrifuged at 4,300 × g for 15 min at 4°C, the supernatant was removed, and the pellet was resuspended in an equal volume of PBS (pH 7.4)-5 mM MgCl₂-50 mM sucrose. The wash was repeated again, and the samples were split into equal volumes. Either 50 µl of sterile water or proteinase K (to a final concentration of 200 µg/ml) was added to each sample, and both were incubated at 20°C for 40 min. After 40 min, phenylmethylsulfonyl fluoride (PMSF) was added to a final concentration of 1 mM, and the samples were examined by dark-field microscopy to assess their motility (i.e., as an indirect test of viability). The samples were then centrifuged and washed again as described above except that the solution also contained 1 mM PMSF. The final pellets were resuspended in Laemmli sample buffer, and the proteins were separated by SDS-PAGE and analyzed by immunoblotting as described below.

Growth phase induction of vraA. B. burgdorferi B31 MSK5 (passage 2) was inoculated at an initial cell density of 5 × 10⁵ per ml in BSK II media and was grown in separate cultures at 32 and 37°C. Aliquots at different time points were removed at the following cell densities in a volume equal to a final count of 10⁸ B. burgdorferi cells (the numbers listed are approximate per-milliliter values): 5 × 10⁶, 1 × 10⁷, 2.5 × 10⁷, 5 × 10⁷, 1 × 10⁸, 2 × 10⁸, or 3 × 10⁸ organisms. Each sample was pelleted at 5,800 × g for 20 min, washed in PBS (pH 7.4), and resuspended in Laemmli sample buffer. Equivalent amounts of B. burgdorferi whole-cell lysates per time point were resolved by SDS-PAGE, immunoblotted, and probed with anti-VraA serum as described below.

pH induction of vraA. Experiments to determine whether vraA was pH inducible were conducted as described by Carroll et al. (8).

SDS-PAGE and Western blotting. SDS-PAGE and Western blotting with the different anti-GST-VraA sera was done essentially as described previously (38, 40). The dilution of anti-VraA serum used was 1:5,000 regardless of the GST-VraA immunogen used to immunize rabbits. Rabbit anti-p66 serum was generously provided by Sven Bergström, Umeå University, and was diluted 1:5,000. Mouse monoclonal antibody directed against B. burgdorferi endoflagella was kindly provided by Alan Barbour, University of California at Irvine, and was used at a 1:50 dilution. Protein A, conjugated to horseradish peroxidase, and diluted 1:1,000, was used to detect all immobilized immune complexes. Blots were developed using Amersham's enhanced chemiluminescence (ECL) system (Amersham Pharmacia Biotech, Piscataway, N.J.).

Mouse immunization with VraA. Two groups of 24 8-week-old female C3H/HeN mice (Charles River Laboratories, Inc., Wilmington, Mass.) were each immunized with 25 µg of either GST or full-length VraA fused to GST (GST-VraA FL). In addition, 24 naive mice served as controls for B. burgdorferi infection. After 10 weeks, the mice were immunized with the same amount of protein (with the exception of the untreated controls). Representative mice (four were randomly chosen per group) were bled on day 15 postboost, and individual sera were used in a Western blot assay to determine qualitatively whether the animals had a humoral response to recombinant VraA. Since all sera from immunized mice showed the appropriate reactivity to either GST or GST-VraA at day 17 postimmunization, 12 mice per group were challenged with either 10² or 10⁴ infectious B. burgdorferi B31 isolate MSK5 which contains all of the known borrelial plasmids (as determined by PCR amplification [24]). At 2 weeks after challenge, the mice were sacrificed and their abdomen skins, spleens, and bladders were aseptically removed and cultured at 32°C in 10 ml of BSK II medium supplemented with 6% normal rabbit serum. After 4 days, 0.5 ml of the culture was passaged blindly into 10 ml of BSK II medium, and the samples incubated at 32°C. Protection was assessed by the absence of cultured B. burgdorferi from each tissue sample (both the initial culture and the blindly passaged sample) out to a 6-week period.

Statistical analysis. A 3×2 contingency table and ² test of independence were used to test the null hypothesis, "Protection against B. burgdorferi challenge is independent of the vaccinogen used." The null hypothesis was rejected when P values of <0.05 were obtained, indicating statistical significance.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Sequence analysis of vraA. Previously we had identified nine distinct proteins that were antigenic in infection-immune rabbits (38). One of these antigens showed extensive redundancy at the nucleotide level. The complete nucleotide sequence indicated that this locus contained a hydrophobic core sequence at the amino terminus connected to the putative leader peptidase II cleavage sequence FLAC, suggesting that this gene encoded a lipoprotein antigen. Further assessment of this gene indicated that the mature domain contained a 27-bp invariant, repetitive sequence, as shown in the Pustell DNA matrix plot in Fig. 1A, that encodes the following primary sequence: Glu-Glu-Glu-Leu-Lys-Lys-Lys-Gln-Gln. Elucidation of the entire B. burgdorferi genome sequence by TIGR corroborated our nucleotide sequence and indicated that this genetic locus, named BBI16 by TIGR (see TIGR website and also reference 14), was located on linear plasmid 28-4 (lp28-4). Because of these repetitive domains, we have designated this gene vraA for virulent strain-associated repetitive antigen A. Although the predicted molecular mass of mature VraA (i.e., minus the putative amino-terminal leader peptide) is 52,375 Da, VraA migrates at an apparent molecular mass of approximately 70 kDa, presumably due to the conformation of the repetitive domain. This type of anomalous migration has been observed previously in other repetitive proteins (33, 48). The vraA locus (BBI16) is a member of a B. burgdorferi paralogous family containing 17 genes (family 60; see TIGR website); however, the homology between the paralogues resides predominantly within the predicted carboxy-terminal domain (i.e., no other paralogue contains a repetitive domain like that observed in vraA). One of the paralogues, encoded by BBI28, contains one imperfect copy of the repetitive domain observed in vraA with 88% identity at the nucleotide level, resulting in a single amino acid substitution relative to the repetitive domain in VraA (Fig. 1B). With the noted exception of the single small domain in BBI28, no homology is observed at the nucleotide or amino acid level within the repetitive domains of vraA with any other sequence in the database.

View larger version (28K): [in this window] [in a new window]   FIG. 1.   Sequence analysis of vraA repetitive domain. (A) Pustell DNA matrix analysis of vraA aligned against itself. Lines framing the main diagonal show the location of the repetitive domain from nucleotides 157 through 723 of vraA encoding the 21 consecutive 9-amino-acid repeat EEELKKKQQ. The number of lines above or below the diagonal indicates the number of repeat units in vraA. The absence of any breaks within the upper and lower diagonals demonstrates that the repeats are consecutive and invariant. The short perpendicular line in the lower right corner indicates a short imperfect inverted repeat sequence located at nucleotides 1222 to 1240 of vraA. (B) Comparison between the single region of homology of BBI28 and the vraA repeat unit. Underlined nucleotides and amino acid indicate differences in the vraA and BBI28 sequences, respectively.

Production of antiserum specific for VraA. In order to obtain sufficient amounts of recombinant VraA for immunization purposes, we constructed a translational fusion of the gene encoding GST to vraA using the pGEX-4T-1 vector as described in the Materials and Methods section. After induction of the fusion protein, the soluble component, containing the GST fused to full-length VraA recombinant protein [GST-VraA (FL)], was affinity purified using a glutathione-Sepharose column (Fig. 2). The fusion protein shown in Fig. 2 represents a carboxy-terminal fusion to GST of residues 18 to 451 of VraA. Panel A shows the overproduction of VraA (lane 2) relative to the induction of GST in the vector only control (lane 1). We then assessed the solubility of recombinant VraA and found that it was associated essentially with the soluble component (Fig. 2B, lane S) relative to the membrane fraction (Fig. 2, lane M) when produced in E. coli. Affinity purification resulted in a single pure fraction containing recombinant VraA that we used for subsequent immunization and vaccination studies (see below). We also made similar separate constructs that joined residues 18 to 260 of VraA and residues 253 to 451 of VraA to GST, designated amino-terminal VraA and carboxy-terminal VraA, respectively, and purified these fusion proteins to homogeneity (data not shown). Purified full-length, amino-terminal, and carboxy-terminal VraA (all fused to GST) were used to immunize rabbits separately, and the polyclonal antiserum obtained was tested to confirm that these antibody reagents were specific for VraA. Clonal isolates of B. burgdorferi B31, derived from infected C3H/HeN mouse skin, designated MSK5 and MSK7, were analyzed by SDS-PAGE and Western blot with all three sera obtained. We have previously determined that MSK5 contains all known plasmids, whereas MSK7 is lacking lp28-4, the linear plasmid that encodes vraA (24). As predicted, the MSK7 sample showed no immunoreactivity to the 70-kDa full-length VraA species when probed either with serum against full-length VraA (Fig. 3B, lane 2) or with serum directed against the amino- or carboxy-terminal constructs (data not shown). In contrast, a 70-kDa antigen and a probable 50-kDa degradation product was observed in MSK5 when probed with antiserum generated against rabbits immunized with either full-length (Fig. 3B, lane 1) or amino-terminal (data not shown) VraA. However, no reactivity was observed when serum from rabbits immunized with the carboxy-terminal half of VraA was used as the primary antibody (data not shown). In all instances, sera directed against the three different GST-VraA proteins contained antibodies reactive with each homologous recombinant protein (data not shown). This indicated that the rabbits immunized with GST fused to either full-length, amino-terminal, or carboxy-terminal VraA generated a humoral response specific for the appropriate VraA construct. Since serum directed against the carboxy-terminal half of VraA did not react with native VraA from B. burgdorferi B31 and only weakly reacted with full-length and amino-terminal recombinant VraA fused to GST, it is most likely that the immunoreactivity observed is restricted to the GST moiety (data not shown). As controls, we also determined that neither rabbit preimmune serum nor antiserum to GST recognized any B. burgdorferi proteins, indicating that the immunoreactivity observed was restricted to VraA (data not shown). Furthermore, infection derived rabbit serum recognized full-length and amino-terminal recombinant VraA fused to GST, but not carboxy-terminal VraA fused to GST (data not shown). These results, taken together, suggest that the repetitive domain is the predominant immunogenic component of VraA.

View larger version (23K): [in this window] [in a new window]   FIG. 2.   Overproduction and purification of full-length VraA fused to GST. (A) Coomassie blue stain of an SDS-10% polyacrylamide gel showing the overproduction of GST alone (lane 1) and full-length VraA fused to GST [GST-VraA(FL); lane 2, indicated by an arrow] following IPTG (isopropyl--D-thiogalactopyranoside) induction as outlined in Materials and Methods. (B) Solubility of recombinant GST-VraA(FL). After induction, E. coli cells synthesizing GST-VraA(FL) were subjected to breakage with a French pressure cell and the insoluble component cleared by centrifugation as indicated in the Methods section. The resulting membrane material (M) and soluble component (S) were resolved by SDS-PAGE, and the gel was stained with Coomassie blue. Note the appearance of an approximately 100-kDa species in the lane containing the soluble fraction (lane S, indicated by an arrow). (C) Purified GST-VraA(FL). GST-VraA(FL) was affinity purified with glutathione-Sepharose beads, electroeluted from an unfixed SDS-polyacrylamide gel to eliminate lower-molecular-mass contaminants, reseparated by SDS-PAGE, and stained with Coomassie blue. Purified GST-VraA(FL) was then used to immunize both rabbits and mice to generate anti-VraA serum and in protection studies, respectively. Numbers on the left refer to the molecular mass of protein markers (in kilodaltons).

View larger version (47K): [in this window] [in a new window]   FIG. 3.   Specificity of antiserum directed against full-length VraA. Whole-cell lysates of protein derived from B. burgdorferi B31 MSK5 (wild type) and MSK7 (lp28-4) were separated by SDS-PAGE and stained with Coomassie blue (A) or immunoblotted and probed with rabbit serum directed against full-length VraA fused to GST (B). Lane 1, MSK5; lane 2, MSK7. The numbers on the left refer to the molecular masses of the protein markers (in kilodaltons).

Surface localization of VraA. To determine whether VraA was a surface exposed antigen in B. burgdorferi B31, we made changes to the protease accessibility methodology previously published (4) inasmuch as B. burgdorferi began to clump and were nonmotile after incubation in PBS (pH 7.4)-5 mM MgCl₂. Based on our previous experience with outer membrane protein localization and outer membrane purification studies (37, 39, 40), we hypothesized that the loss of motility was due to the loss of the structural integrity of the B. burgdorferi outer membrane. B. burgdorferi cells were therefore incubated in several different buffers, and their viability was assessed as a function of motility. We found that B. burgdorferi resuspended in PBS (pH 7.4)-5 mM MgCl₂ containing 50 mM sucrose was active and motile for periods of time exceeding 1 h with or without added proteinase K. Under these conditions, addition of proteinase K resulted in a near-complete reduction of VraA (Fig. 4A), modified the surface exposed protein P66 (Oms66) (39) (Fig. 4B), but did not alter the levels of endoflagella (Fig. 4C), a known subsurface marker of spirochetal bacteria (4). This result implies that VraA is a surface-exposed protein and that the B. burgdorferi cells were structurally intact, respectively.

View larger version (32K): [in this window] [in a new window]   FIG. 4.   Surface Localization of VraA. Intact B. burgdorferi B31 MSK5 either left untreated (lane 1) or treated with proteinase K (lane 2), resolved by SDS-PAGE, and immunoblotted. Identical blots were then probed with either anti-VraA (full-length VraA fused to GST) (A), anti-p66 (B), or anti-endoflagellum (C) serum. The numbers on the left refer to the molecular mass of protein markers (in kilodaltons).

Temperature- and growth-phase-dependent induction of vraA. We initially evaluated whether vraA was induced at 23, 32, and 37°C and found that VraA was overproduced at 37°C (data not shown). However, subsequent studies to repeat this observation yielded variable results. This led us to surmise that vraA may be subjected to growth phase regulation, pH regulation, or both. To assess this, cultures of infectious B. burgdorferi were inoculated at a density of 5 × 10⁵ per ml, and samples were taken at various time points commensurate with early-, mid-, late-, and stationary-phase growth at both 32 and 37°C using a previously published methodology (21, 31). Identical amounts of protein for each time point were resolved by SDS-PAGE and either stained with Coomassie blue (data not shown) or immunoblotted to ensure that equivalent levels of B. burgdorferi protein were present and to assess levels of VraA synthesized at each time point, respectively. Our results, after four independent experiments, indicate that VraA is only slightly overproduced when cells are grown at 37°C and that vraA expression is not affected by the growth phase of the culture (Fig. 5). Additionally, VraA appears to be proteolytically degraded more efficiently at 37°C as seen by the increase in the approximate 50-kDa degradation product, perhaps due to the induction of a temperature-dependent protease (Fig. 5). Alternatively, it is possible that the 50-kDa degradation product accumulates at 37°C due to decreased proteolysis of this lower-molecular-weight species independent of full-length VraA. Only three samples, at various growth phases, ranging from approximately 1 × 10⁷ (Fig. 5, lanes 1 and 2) to 3 × 10⁸ cells (Fig. 5, lanes 5 and 6), are shown in Fig. 5; however, other intermediate samples at different growth phases were also evaluated and yielded identical immunoblot profiles at 32 and 37°C relative to the samples shown in Fig. 5 (data not shown). These results, taken together, indicate that vraA expression is not regulated by growth phase.

View larger version (51K): [in this window] [in a new window]   FIG. 5.   vraA induction at 37°C relative to 32°C. B. burgdorferi B31 MSK5 (passage 2) was inoculated at an initial cell density of 5 × 105 per ml in BSK II medium and cultivated as outlined in Materials and Methods at both 32 and 37°C. At various phases of growth, samples were removed and protein from whole-cell lysates was subjected to SDS-PAGE and immunoblotting with anti-VraA serum. Lane 1, 1.25 × 107 B. burgdorferi grown at 37°C; lane 2, 1.9 × 107 B. burgdorferi grown at 32°C; lane 3, 1.22 × 108 B. burgdorferi grown at 37°C; lane 4, 1.37 × 108 B. burgdorferi grown at 32°C; lane 5, 3.14 × 108 B. burgdorferi grown at 37°C; lane 6, 3.1 × 108 B. burgdorferi grown at 32°C. The numbers on the left refer to the molecular mass of protein markers (in kilodaltons).

VraA confers partial protection in mice. To test whether VraA could function as a protective immunogen, we immunized 24 C3H/HeN mice with recombinant full-length VraA fused to GST, i.e., GST-VraA(FL) (FL represents full length) as described in Materials and Methods. The controls used included 24 mice immunized with GST alone and 24 naive mice. Prior to the challenge, we collected serum from several randomly chosen mice to qualitatively test whether a humoral response had been generated against the appropriate immunogen [GST or GST-VraA(FL)]. In all cases, mice immunized with GST or GST-VraA(FL) generated antibodies to the appropriate antigen, respectively (data not shown). Twelve mice from each set were needle challenged with either 10² or 10⁴ of a clonal isolate of B. burgdorferi B31 MSK5 passage 3 that contains all known B. burgdorferi plasmids (24). The results indicated that nearly all of the mice challenged with 10⁴ infectious B. burgdorferi were susceptible regardless of the immunogen given (Table 4). In contrast, only 8.3% (1 of 12) of the mice immunized with GST-VraA(FL) were infected when challenged with 10² B. burgdorferi (Table 4), whereas the naive and GST-immunized controls were infected 40% and 63 to 67% of the time (depending on the infected tissue assessed), respectively (Table 4). To assess statistical significance, a 3×2 contingency table was used to test the null hypothesis "Protection against B. burgdorferi challenge is independent of the vaccinogen used." Values corresponding to skin samples obtained using a 10² challenge inoculum were used in this analysis. The calculated ² of 8.67 exceeded that of ²_0.025,2 of 7.378, so the null hypothesis was rejected (0.01 < P < 0.025; see Table 4), indicating that the protection observed in the VraA immunized mice was statistically significant. Similar results were obtained with bladder tissue and, to a lesser extent, spleen tissue (see Table 4).

Cross-reactivity of VraA antiserum against B. burgdorferi sensu lato isolates. To determine whether other Borrelia spp. encoded a homologue of VraA, we tested protein lysates for reactivity to antiserum directed against full-length strain B31 VraA in a Western blot assay. All B. burgdorferi sensu stricto isolates encoded a cross-reactive VraA-like antigen; however, only strain ECM-NY-86 encoded an antigen of the same molecular mass as that of strain B31 VraA (Fig. 6; note a similar immunoreactive band in both lanes 1 and 5). In contrast, strains 297, JD-1, and NT-1 (Fig. 6, lanes 3, 6, and 7, respectively) encoded antigens that were weakly reactive (only visible after long exposures; not seen in Fig. 6) and larger in molecular mass relative to strain B31 VraA, whereas strains CA-2-87 and 2872-2 (Fig. 6, lanes 4 and 12, respectively) produced strongly reactive homologues that were smaller in molecular mass. Some B. burgdorferi sensu lato isolates such as B. garinii strain Ip90, did not encode a VraA-like molecule, whereas others (Fig. 6, lane 11), including B. afzelii ACA1 (data not shown) and B. andersonii strain MOD-3, MOD-5, and MOD-6, did synthesize reactive homologues (Fig. 6, lanes 8 to 10, respectively). Interestingly, B. andersonii encoded multiple, highly reactive antigens. These strains were obtained from infected rabbit skin samples and represent a polyclonal population. Therefore, the profile observed for these samples may reflect multiple genetic loci that encode VraA-like molecules that are either present in each individual bacterium or a subset of individual clones within the polyclonal pool tested. The reactivity observed for antiserum against full-length VraA was identical to that seen for reactivity against the amino-terminal half of VraA, which is composed mostly of the repetitive domain. In contrast, little or no reactivity was seen for the antiserum generated against the carboxy-terminal half of VraA, implying that the cross-reactivity between the VraA homologues is mediated via the antigenicity of the repetitive motifs (data not shown).

View larger version (31K): [in this window] [in a new window]   FIG. 6.   Presence of VraA or VraA homologues in various B. burgdorferi sensu lato isolates. Whole-cell lysates from B. burgdorferi sensu stricto strain B31 passage 4 (lane 1), strain B31 passage 47 (lane 2), strain 297 (lane 3), strain CA-2-87 (lane 4), strain ECM-NY-86 (lane 5), strain JD-1 (lane 6), strain NT-1 (lane 7), and strain 2872-2 (lane 12) were resolved by SDS-PAGE along with B. andersonii strains MOD3, MOD5, and MOD6 (lanes 8 to 10) and B. garinii strain IP-90 (lane 11). All isolates with the exception of the strain B31 isolates were passaged two times in vitro. The resulting gels were either stained with Coomassie blue (A) or immunoblotted and probed with anti-VraA serum (B). The numbers on the left refer to the molecular mass of protein markers (in kilodaltons).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In this report we describe the characterization of a surface-exposed repetitive motif containing antigen, designated VraA, that we originally identified as an infection associated antigen in infection immune rabbits from B. burgdorferi sensu stricto strain B31 (13, 38). The most striking feature of the vraA locus was the number of completely conserved 27-bp repeats encoding an invariant 9-amino-acid repetitive motif (see Fig. 1). Unlike other antigens, such as the Bdr proteins of Borrelia spp. (32, 51), the Vlp repetitive antigens of Mycoplasma spp. (10, 33), and various proteins from Plasmodium spp. (16, 22, 23), that contain repetitive domains, the repetition in vraA is absolute at the nucleotide level, suggesting that this conservation is important in maintaining some as-yet-unknown function essential for maximal infectivity. Alternatively, the repeat units may represent a site for recombination resulting in antigenic variation or slipped strand mispairing resulting in either antigenic or phase variants. By analogy to other prokaryotic systems, notably the Vlp proteins of Mycoplasma spp. (33, 48) and the Opa antigens of Neisseria spp. (6, 28), VraA may change its antigenicity by varying its length via slipped-strand mispairing. This change in length has been shown previously to change the antigenicity of Mycoplasma Vlp antigens (33, 48) and cell-wall-associated proteins from gram-positive bacteria (20). This antigenic variation could aid in immune evasion resulting in persistent B. burgdorferi infection. Specifically, the repeat region of vraA contains a stretch of six consecutive adenine nucleotides (within a region containing adenines at 15 of 18 locations). A single mismatch within this mononucleotide repetitive domain would yield an antigen with either a modified repetitive primary sequence that would be truncated soon after the repeats or a variant that was abbreviated almost immediately (i.e., a stop codon introduced within a repeat unit). In either case, the resulting modified forms might exhibit reduced or nullified antigenicity. The location of this change, i.e., which repeat unit is modified, would dictate whether any of the repeats in native VraA would be synthesized and to what degree the truncated variants might cross-react with antibodies directed against full-length VraA. This could, in part, explain the resistance to infectious challenge with 100 low-passage B. burgdorferi observed in 1 of the 12 mice immunized with recombinant VraA (Table 4). That is, if a subpopulation of our inoculum changed such that VraA was either antigenically modified by the increased or decreased length in the number of repeats or by frameshift mutations within the repeat units that resulted in truncated antigenic variants, then B. burgdorferi encoding these proteins may not be killed by antibody directed against "full-length" recombinant VraA. To date, no evidence of such changes in native VraA have been observed in B. burgdorferi B31 inasmuch as the frequency of these changes may be similar to that seen in other systems (approximately 10³ to 10⁴ [6, 28]). However, we have recently determined that variants of vraA are present in in vitro-cultivated B. burgdorferi B31 that have additional repetitive domains (i.e., greater than 21 in number; K. L. Swingle and J. T. Skare, unpublished data), suggesting that VraA may be an antigenically variable protein. Additionally, as seen in Fig. 6, other sensu stricto isolates synthesize putative homologues relative to strain B31 that may represent putative variants of B. burgdorferi sensu stricto VraA. The changes predicted for these variants may occur more frequently in vivo (i.e., within an infected animal) not unlike what has been shown for both the vlsE (49, 50) and ospEF-related (erp) loci (45), whose changes occur exclusively during infection. A more detailed analysis of the strain B31 vraA variants and its requisite homologues should indicate whether this hypothesis is valid.

An important consideration related to protective immunity in B. burgdorferi is the form in which the challenge inoculum is delivered. Recently, it has become apparent that host-adapted B. burgdorferi repress or derepress several genes, notably, ospA or ospC, respectively (34), in the disparate environments that B. burgdorferi occupies (i.e., ticks versus mammals), as well as under experimental ex vivo conditions (27). B. burgdorferi is present at very low levels during mammalian infection; therefore the ability to evaluate gene expression in vivo has required the implementation of indirect methodologies (38). To circumvent this dilemma, Akins et al. have demonstrated recently that B. burgdorferi synthesizes a subset of proteins and/or antigens in dialysis membrane chambers that are implanted within the peritoneal cavity of rats (1). Although temperature certainly upregulates some B. burgdorferi genes, clearly other host signals were required for expression of some of the gene products observed in this model system (1). Our results indicate that a temperature shift in vitro results in the modest induction of VraA (Fig. 5). Previous studies indicated that the growth phase is also an important consideration for assessing gene regulation in B. burgdorferi. Specifically, OspC and BmpD are regulated in the transition from logarithmic-phase growth to stationary-phase growth (21, 31). However, in contrast, our data indicate that VraA is not subject to growth phase regulation (Fig. 5 and data not shown). It is conceivable that vraA may be expressed at high levels in vivo (i.e., within the arthropod vector or mammalian hosts) where nutrients and other important cofactors may be limiting. Whether vraA is subject to such regulation remains to be determined.

Shang et al. recently demonstrated that rabbits immunized with outer membrane vesicles containing surface proteins were protected only when challenged with in vitro cultivated B. burgdorferi; rabbits challenged with skin biopsies containing host-adapted B. burgdorferi were infected in a manner indistinguishable from naive controls (35). One explanation could be alternative gene expression; that is, the expression of new antigens within the infected mammal that are not expressed during in vitro cultivation. Another viable possibility is that B. burgdorferi changes antigens expressed during in vitro cultivation (i.e., antigenic variation) or uses other mechanisms, including phase variation, to alter their antigenic profile. Once inside the host, this effect could manifest in a polyclonal manner such that each B. burgdorferi cell may express a different form of a given antigen, such as VlsE (49, 50) or OspEF (45) or different antigens in various combinations. This hypothesis may, in part, explain the presence of numerous paralogous gene families. Such genetic paralogues may therefore be involved in immune evasion independent of function. The vraA locus belongs to one such paralogous family (family 60; see TIGR website). However, in contrast to the other family members, vraA is the only family member that contains the aforementioned repeat units with the notable exception of the protein encoded by BBI28, which contains a similar but nonidentical single repeat unit relative to VraA (see Fig. 1).

The absolute nature of the repetitive domains at both the nucleotide and the amino acid level suggests an important role for this motif. One possible explanation could be related to an increased avidity of VraA for a target ligand. In this regard, we have no indication that VraA functions as an adhesin in either arthropod or mammalian structures. Alternatively, these repeats may serve as an immunological "smokescreen," as has been proposed for several Plasmodium antigens (22). If so, then the epitope(s) defined by these repeat units would theoretically mediate a robust yet nonprotective response. Our immunization results, indicating partial protection to needle challenge, argue against this; however, the possibility that the repeats of strain B31 VraA change in length in a manner analogous to that observed in other B. burgdorferi sensu lato isolates (see Fig. 6), resulting in antigenic variation, suggest that a strong immune response to a single form of VraA may not protect in all cases. Heterologous challenge against animals immunized with B. burgdorferi B31 VraA could decipher this; these studies and others to determine whether VraA is subject to phase and/or antigenic variation are in progress.