Borrelia burgdorferi vlsE Antigenic Variation Is Not Mediated by RecA

Abstract

RecA is a key protein linking genetic recombination to DNA replication and repair in bacteria. Previous functional characterization of Borrelia burgdorferi RecA indicated that the protein is mainly involved in genetic recombination rather than DNA repair. Genetic recombination may play a role in B. burgdorferi persistence by generation of antigenic variation. We report here the isolation of a recA null mutant in an infectious B. burgdorferi strain. Comparison of the in vitro growth characteristics of the mutant with those of the wild-type strain under various conditions showed no significant differences. While the RecA mutant was moderately more sensitive to UV irradiation and mitomycin C than the wild-type strain, the lack of RecA abolished allelic exchange in the mutant. Absence of RecA did not affect the ability of the mutant to infect mice. However, the RecA mutant was attenuated for joint infection in competitive-infection assays with the wild-type strain. vlsE sequence variation in mice was observed in both wild-type and RecA mutant spirochetes, indicating that the mechanism of antigenic variation is not homologous genetic recombination.

Borrelia burgdorferi, the etiologic agent of Lyme disease, is transmitted by the bite of Ixodes ticks infected with this spirochetal pathogen (4). Early stages of infection produce a variety of symptoms in humans, including a localized inflammation at the tick bite site (erythema migrans), which may be followed by bacterial dissemination, manifested as cutaneous, cardiac, neurologic, or joint involvement (53). Antibiotic treatment during early disease is curative, but in untreated patients the disease may become chronic, with spirochetes persisting for months to years despite an active immune response (53).

Persistence of spirochetes in infected tissues has also been shown in murine models of Lyme disease (3). Immune evasion is mediated, in part, by antigenic variation of the VlsE surface lipoprotein. Evidence suggests that in mice B. burgdorferi undergoes continuous antigenic variation at the vlsE locus located on plasmid lp28-1 (59). The mechanism of antigenic variation appears to be a random, segmental recombination between variable regions of the middle cassette region of vlsE and corresponding variable regions of adjacent nonexpressed vls cassettes. These gene rearrangements resemble gene conversions, producing thousands of B. burgdorferi clones, each presenting a random mosaic variant of a VlsE protein on its surface (34, 61). Spirochetes incapable of this antigenic variation, due either to loss of lp28-1 or to functional inactivation of the vlsE locus, are initially infectious to immunocompetent mice, but the infection cannot be effectively maintained in the animals (2, 13, 26, 28). Mutants lacking lp28-1 grow normally in severe combined immunodeficient mice (26) and in dialysis membrane chambers implanted into rat peritoneal cavities (41). Furthermore, vlsE antigenic variation does not occur in infected ticks (15, 37). These findings indicate that vlsE antigenic variation occurs only in response to humoral immunity (2). Little is known about the mechanism of recombination at the vlsE locus. To date, neither a vlsE-specific recombinase nor any other vlsE-associated factor has been identified. A recent study by Bankhead and Chaconas concluded that other genes carried by lp28-1 are not involved in vlsE antigenic variation (2). The genome of B. burgdorferi contains a reduced number of loci involved in recombination and DNA repair compared to that of Escherichia coli (12). Such genomic reduction trends have been observed in other host tissue-associated bacteria and correlate with living in stable environments (45).

RecA is a central player in bacterial recombination and DNA repair (24). In E. coli under normal growth conditions, RecA mediates the repair of double-stranded DNA breaks that occur at stalled replication forks during DNA replication (6, 23). In enterobacteria and Bacillus spp., an ATP-dependent coprotease activity of RecA is activated under stress conditions induced by extensive DNA damage. This activity is part of a specialized form of DNA repair designated the SOS response (1). In E. coli, this response involves the transcriptional repressor LexA, whose autocatalytic cleavage is enhanced by activated RecA following DNA damage (27). This results in the coordinate expression of approximately 40 different genes, including recA and lexA. In contrast, the B. burgdorferi genome does not contain a lexA ortholog, suggesting that the SOS response is likely absent in this organism (12).

B. burgdorferi recA encodes a predicted protein of 365 amino acids that shares 56% identity and 77% similarity with E. coli K-12 RecA. All functional domains of RecA that have been assigned by both structural and mutational studies of E. coli are conserved in B. burgdorferi (22, 35). Interestingly, despite extensive predicted amino acid conservation between E. coli RecA and B. burgdorferi RecA, antibodies directed against E. coli RecA do not cross-react with B. burgdorferi RecA (52). recA was not induced in B. burgdorferi following UV irradiation, in contrast to the situation in E. coli (30). In the absence of the SOS response, the role of RecA in DNA repair in B. burgdorferi is unclear since this organism is exquisitely sensitive to UV exposure (30).

The role of RecA in B. burgdorferi DNA recombination has not been investigated, and its possible role in vlsE variation is not known. In order to study RecA function in B. burgdorferi and to assess its role in homologous recombination in immune evasion, a recA null mutant was constructed and the effect of this mutation on DNA repair and vlsE antigenic variation was investigated.

MATERIALS AND METHODS

Bacterial strains and culture.

B. burgdorferi B31-A3, a clonal derivative of B31MI (11), was an infectious isolate at passage 4. A high-passage-number B31MI isolate was also used. Both strains were grown in BSK-S, a modified Barbour-Stoenner-Kelly (BSK) liquid medium (58) or on solid BSK-S-1% agarose plates with 0.7% agarose overlays at 34°C in a 5% CO₂ incubator. E. coli DH5α (recA1 gyrA thi1 relA1), BL21 (ompT gal lon; λ prophage with T7 RNA polymerase), and DB3.1 (gyrA endA ΔrecA ara gal proA) were grown at 37°C in Luria-Bertani (LB) broth or on solid LB medium containing 1.5% agar. When necessary, the LB medium was supplemented with ampicillin or kanamycin (Sigma, St. Louis, MO) at 100 μg/ml or 40 μg/ml, respectively. Spirochetes were enumerated by dark-field microscopy as previously described (49).

Construction of a B. burgdorferi recA mutant.

A 3.09-kbp DNA fragment of B. burgdorferi strain B31MI containing the 1,095-bp recA coding region and 1,090 bp and 905 bp of 5′ and 3′ flanking segments, respectively, was PCR amplified with primers p132F and p129R (Table 1) and was introduced into pGEM-T (Promega, Madison,WI) by T/A cloning according to the manufacturer's protocol. Recombinant vectors were transformed into competent E. coli DH5α, and white, ampicillin-resistant colonies were purified and tested for the presence of the plasmid by restriction endonuclease digestion and PCR using the above primers. The appropriate plasmid containing the recA insert, designated pGEM-T-recA_Bb, was used to generate an insertionally inactivated copy of the B. burgdorferi recA (recA_Bb) gene. This derivative was constructed by first cleaving pGEM-T-recA_Bb at a unique BlnI site in recA (at position 212 in the recA coding region) and filling in the 5′ overhangs with T4 DNA polymerase. A 1,040-bp kanamycin resistance cassette (kan) from Streptococcus faecalis was PCR amplified using primers pKNF and pKNR as the forward and reverse primers as previously reported (38). The cassette ends were blunted with T4 DNA polymerase, mixed with BlnI-digested pGEM-T-recA_Bb, and incubated with T4 DNA ligase. The resulting plasmid, designated pGEM-T-recA_Bb::Kn^r, was isolated from a kanamycin-resistant colony of E. coli DH5α transformed with the ligation reaction. The ampicillin resistance cassette was inactivated by ScaI digestion and insertion of a 200-bp nonspecific DNA fragment from E. coli. The orientation of the inserts relative to the pGEM-T backbone and the sequences of the inserted fragments were determined by DNA sequencing (Davis Sequencing, LLC, Davis, CA).

TABLE 1.

Oligonucleotide primers employed in this study

Primer	Sequence (5′-3′)	Purpose	Source or reference
p132F	GCGGGATCCAAAAGGAATAGTGGATC	PCR amplification of region containing recA	This work
p129R	GCGGGATCCCCAAGCTTAGCAATAGTG
pKNF	ATAGCTTGTAAATTCTATCATAATTGTGGT	PCR amplification of kanamycin gene	39
pKNR	TAGGTACTAAAACAATTCATCCAGTAA
pnTM17F	GTGGATCTATTGTATTGTATTAGATGAGGCTCTCG	PCR amplification of internal fragment of recA	31
pnTM17R	GCCAAAGTTCTGCAACATTAACACCTAAAG
pKNPF	GTGGGAGAAAATGAAAACC	Preparation of kan probe	This work
pKNPR	GCCATCGGCCTCACTCATGAG
F4120	AGTAGTACGACGGGGAAACCAGA	PCR amplification of vlsE	61
F4066	GAGTCTGCAGTTCGCAAAGT
pBAf	CACCATGTCAAAGTTAAAGGAAAAAAGAG	PCR amplification of full length recA	This work
pBAr	CTCAGATTCATCTTCTTTAAATTC

Plasmid pGEM-T-recA_Bb::Kn^r was electroporated into competent B. burgdorferi B31-A3 prepared as described by Samuels (48). Each reaction mixture contained 5 × 10⁸ cells and 20 μg of plasmid DNA. Cells in transformation mixtures were grown in liquid BSK-S medium containing 200 μg/ml of kanamycin at 34°C for 4 weeks in 96-well plates, and antibiotic-resistant clones were identified based on the color change of the growth medium. Individual clones were purified by limiting dilution in BSK-S medium containing kanamycin as described by Hubner et al. (14). The plasmid content of transformed B. burgdorferi B31-A3 clones was determined by PCR as previously reported (16); the mutant contained all the plasmids present in parental B31-A3, including plasmids lp25 and lp28-1.

Southern blot analysis.

Five micrograms of total DNA purified from wild-type B31-A3 and the recA mutant was digested with AvaII overnight at 37°C. The DNA fragments were resolved by electrophoresis on a 1% agarose gel in 1× Tris-borate-EDTA buffer. Following a 30-min denaturation in 0.4 N NaOH-1.5 M NaCl and a 30-min neutralization with 1.0 M Tris-HCl-1.5 M NaCl (pH 8.0), the DNA fragments were transferred to positively charged nylon membranes (Roche Molecular Biochemicals Indianapolis, IN) in 10× SSPE (1× SSPE is 0.18 M NaCl, 10 mM NaH₂PO₄, and 1 mM EDTA [pH 7.7]) transfer buffer. DNA was cross-linked to the membrane by a 20-s UV irradiation and hybridized overnight at 55°C with either a recA or kan cassette probe in 10 ml of 5× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate)-0.1% laurylsarcosine, 0.02% sodium dodecyl sulfate (SDS), and 1% dry milk. A recA probe was prepared by PCR amplification of B. burgdorferi DNA with primers pnTM17F and pnTM17R as previously described (29), and a kan probe was prepared as previously described (38), except that primers pKPF and pKPR were employed (Table 1). Both probes were labeled with digoxigenin by random primer labeling according to the manufacturer's protocol (Boehringer Mannheim Inc., Indianapolis, IN). After hybridization, membranes were washed in 2× SSC-0.1% SDS followed by 0.5× SSC-0.1% SDS at 55°C, blocked for 1 h in 0.1 M Tris-HCl-0.15 M NaCl-2% milk (pH 7.5), and treated with antidigoxigenin antibody conjugated with alkaline phosphatase (diluted 1:10,000) for 30 min at room temperature. The blots were washed twice in 0.1 M Tris-HCl-0.15 M NaCl (pH 7.5) and once in 0.1 M Tris-HCl-0.1 M NaCl-0.05 M MgCl₂ (pH 9.5), and the chemiluminescent substrate CDP-Star (diluted 1:100) (Roche Molecular Biochemicals, Indianapolis, IN) was added. Reactive bands were visualized by exposure to X-ray film for 2 to 5 min.

Preparation of B. burgdorferi anti-RecA antibodies.

The full-length B. burgdorferi recA was PCR amplified from strain B31MI using pBAf and pBAr as the forward and reverse primers, respectively (Table 1), and the PCR product was directionally cloned into a pENTR/SD/D-TOPO vector (Gateway technology; Invitrogen) and transformed into E. coli DH5α. Ampicillin-resistant colonies were tested by PCR for the presence of recA_Bb, and plasmid DNA was isolated. Purified pENTR-TOPO-recA_Bb DNA was mixed with pET-DEST 42 (l attR1-attR2 V5-His₆ epitope) (Gateway technology; Invitrogen, Carlsbad, CA), and the resultant pET-DEST 42-recA_Bb recombinants were transformed into E. coli BL21. Expression of recombinant B. burgdorferi RecA in E. coli was induced by growth of cells in 0.2% arabinose and 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) for 4 h at 37°C. Recombinant RecA_Bb was purified on Ni-nitrilotriacetic acid matrix columns (Qiagen Sciences, MD) according to the manufacturer's instructions. Antiserum against purified recombinant RecA_Bb was produced in rabbits by a commercial laboratory (Pocono Rabbit Farm & Laboratory Inc., Canadensis, PA). Nonspecific reactivity of the rabbit antiserum was removed by absorption to cell lysates of E. coli DB3.1, a recA deletion mutant.

Immunoblot analysis.

Wild-type and recA mutant B. burgdorferi cultures grown in BSK-S at 34°C to late log phase (2 × 10⁸ cells) were centrifuged at 8,000 × g and washed three times with phosphate-buffered saline (PBS) at 4°C. Cell pellets were suspended in 0.1 ml of 0.05 M Tris-HCl-0.3% SDS-0.01 M dithiothreitol (pH 8.0), and cells were disrupted by sonication. Protein lysates were separated on 10 to 20% SDS-polyacrylamide gels and electrotransferred to polyvinylidene difluoride membranes (Millipore Corp., Bedford, MA). Membranes were probed with polyclonal anti-RecA_Bb (diluted 1:1,000) or anti-FlaB (diluted 1:10,000) rabbit serum. Alkaline phosphatase-conjugated goat anti-rabbit antibodies (diluted 1:2,000) (Kirkegaard & Perry Laboratories, Inc., Gaithersburg, MD) were employed as secondary antibodies for 1 h at room temperature. Following three washes with 1× Tris-buffered saline-0.05% Tween 20 (pH 7.6), the immunoblots were developed using 5-bromo-4-chloro-3-indolylphosphate-nitroblue tetrazolium solution (Kirkegaard & Perry Laboratories, Inc., Gaithersburg, MD) for 3 to 5 min.

Determination of UV and mitomycin C sensitivity.

B. burgdorferi B31-A3 and the recA mutant were grown in 10 ml BSK-S medium at 34°C to a density of 1 × 10⁷ organisms/ml. Cells were harvested by centrifugation and resuspended into 1 ml of PBS at a concentration of 1 × 10⁸ cells/ml, and 0.1 ml of this solution was rapidly spread into 32- by 10-mm culture dishes and exposed to UV (254 nm) irradiation at doses of 400 to 1,600 μJ/cm² with a Spectrolinker XL-1000 UV cross-linker (Spectronics, Westbury, NY). After UV exposure, cells were kept in the dark, recovered by aspiration, and serially diluted in BSK-S medium such that each plate contained approximately 100 organisms per 0.1 ml. Each aliquot was added to 2 ml of BSK-S medium containing 0.8% molten agarose (at 50°C) and was overlaid onto 60- by 15-mm plates containing solid BSK-S medium. The agar plates were incubated at 34°C for 3 weeks in a 5% CO₂ incubator, and colonies were enumerated. Duplicate measurements were performed for three biological replicates.

For measurement of mitomycin C sensitivity, 1-ml portions of exponentially growing B. burgdorferi cultures containing 1 × 10⁷ spirochetes were incubated for 12 h at 34°C in BSK-S in the presence of 2.5 to 15 ng/ml of mitomycin C (Sigma, St. Louis, MO). Prior to plating, the cultures were serially diluted in BSK-S; 0.1 ml of each dilution was overlaid on BSK-S plates and incubated at 34°C for 3 weeks in a 5% CO₂ incubator, and colonies were counted. B. burgdorferi not exposed to mitomycin C served as the control.

Homologous recombination.

The ability of the recA mutant to mediate allelic exchange by homologous recombination was compared to that of the wild type. Two suicide plasmids, designated pGEM-T-BB0243::Spc^r and pGEM-T-BBE02::Spc^r, were constructed to generate disruption mutants. pGEM-T-BB0243::Spc^r contained a 2.5-kbp DNA fragment that included the BB0243 locus from B31MI, which was disrupted by insertion of a 1.2-kbp spectinomycin resistance cassette driven by the flgB promoter (flgB-aadA). pGEM-T-BBE02::Spc^r contained a 3.8-kbp DNA fragment that included the BBE02 locus from B31MI, which was disrupted by insertion of the same spectinomycin cassette. Both plasmids were electroporated into competent wild-type and recA mutant B31-A3 as described by Samuels (48). Cells in transformation mixtures were grown in liquid BSK-S medium containing 100 μg/ml of streptomycin (aadA provides spectinomycin resistance in E. coli and streptomycin resistance in B. burgdorferi) at 34°C for 3 weeks in 96-well plates, and antibiotic-resistant clones were identified based on the color change of the growth medium. Individual clones were purified by limiting dilution in BSK-S medium containing 100 μg/ml of streptomycin as described by Hubner et al. (14). Allelic exchange in streptomycin-resistant clones was confirmed by PCR.

Animal infection studies.

Groups of three C3H/HeJ mice each were inoculated intradermally with 0.1 ml of BSK-S containing 1 × 10², 1 × 10³, or 1 × 10⁴ wild-type or recA mutant B31-A3 cells. Three C3H/HeJ mice injected intradermally with BSK-S medium alone served as negative controls. For competitive-infection assays, four C3H/HeJ mice were inoculated with 5 × 10³ mutant spirochetes in the left flank and 5 × 10³ wild-type spirochetes in the right flank. Ear punches were obtained from each animal at days 7 and 14 postinfection, washed in 70% ethanol and 1× PBS, and placed in 3 ml of BSK-S medium without kanamycin for culture. Blood was drawn from each animal at day 28 for serology. Seroconversion in the mice was confirmed with immunoglobulin G Marblot strips (MarDx Diagnostics Inc., Carlsbad, CA). All mice were sacrificed on day 28, and samples of joint and bladder were obtained for culture and PCR. Tissues were cultured for 3 to 4 weeks in BSK-S medium in a 5% CO₂ incubator at 34°C. Spirochetes were detected by dark-field microscopy (Nikon; Eclipse 400). The 50% infectious dose (ID₅₀) was determined using the NCBI ID₅₀ calculator algorithm, version 5 (http://www.ncbi.nlm.nih.gov/). All animal experiments were approved by the New York Medical College Institutional Animal Care and Use Committee.

Effect of recA inactivation on vlsE-mediated antigenic variation.

DNA was extracted from wild-type and mutant B. burgdorferi cultivated from ear biopsy samples of mice obtained at day 14 postinfection using the IsoQuick DNA extraction kit (ORCA Research Inc., Bothell, WA) and from in vitro-cultivated wild-type B31-A3 spirochetes. The vlsE cassette regions in these isolates were PCR amplified using primers F4120 and R4066 (59). Amplified vlsE cassettes were cloned into pGEM-T vector by T/A cloning and transformed into E. coli DH5α, and the vlsE sequences in individual clones were determined by a commercial service (Davis Sequencing, LLC, Davis, CA).

RESULTS

Construction of a B. burgdorferi recA mutant.

To investigate the function(s) of B. burgdorferi RecA, an isogenic RecA null mutant was constructed in strain B31-A3. This was accomplished by allelic exchange of the wild-type recA with an insertionally inactivated copy (Fig. 1A). The recA coding region was disrupted at a unique BlnI restriction site located at residue P72 (corresponding to the highly conserved P55 of E. coli recA) by insertion of a kanamycin resistance cassette. This resulted in a plasmid designated pGEM-T-recA_Bb::Kn^r. Electroporation of competent B. burgdorferi B31-A3 with the pGEM-T-recA_Bb::Kn^r plasmid resulted in five kanamycin-resistant clones. One of these clones contained a disrupted recA allele yielding the expected 1,262-bp product (Fig. 1B). DNA sequencing revealed that the kan cassette was inserted in the same orientation as the recA reading frame. Insertion of the cassette resulted in complete disruption of the downstream recA reading frame, thus eliminating the production of a kanamycin-RecA fusion protein.

FIG. 1.

Insertional inactivation of B. burgdorferi recA. (A) Schematic representation of the chromosomal region containing recA (BB0131). recA was disrupted by insertion of a kanamycin cassette (kan) at a unique BlnI site corresponding to P72 of recA. The transcriptional orientation of all genes is from right to left. Locations of primers are indicated by arrows. Vertical lines indicate AvaII cut sites. Bracketed regions indicate expected sizes of the regions used for disruption mutant and AvaII digestion products (see panel C). (B) PCR confirmation of recA inactivation. Total DNA from the wild type and recA mutant was amplified with primers pnTM17F and pnTM17R. Lane 1, negative control (no DNA); lane 2, wild type; lane 3, recA mutant; lane M, molecular size markers. (C) Southern blot analysis of the recA mutant. Genomic DNA from the wild type and recA mutant was digested with AvaII, and the fragments were resolved by agarose gel electrophoresis and blotted to a nylon membrane, which was probed with either a recA (lanes 1 and 2) or kan (lanes 3 and 4) probe. Lanes 1 and 3, wild type; lanes 2 and 4, recA mutant. (D) Immunoblot of the wild type and recA mutant. Total cell lysates from both strains were processed by SDS-polyacrylamide gel electrophoresis. Proteins were transferred to nylon membranes, which were sectioned and probed with either rabbit polyclonal anti-RecA (top) or rabbit anti-FlaB (bottom) antibodies. The vertical line indicates the transition point on the manually assembled membrane from the wild-type to recA mutant lysate. Lanes 1 and 3, no lysate; lane 2, 6 μg of wild-type lysate; lane 4, 6 μg of recA mutant lysate.

Allelic exchange at the recA locus was confirmed by Southern blot analysis of the mutant. Hybridization of AvaII-digested wild-type DNA with a recA probe resulted in a single DNA fragment of 2,851 bp (Fig. 1C, lane 1). This fragment was absent when the kan-specific probe was used for hybridization (Fig. 1C, lane 3). The 1,040-bp kan cassette was inserted at nucleotide 2409 of the 2,851-bp AvaII fragment (Fig. 1A) and contains two additional AvaII sites 70 bp apart. Hybridization of the mutant DNA with the recA probe resulted in the appearance of two bands (Fig. 1C, lane 2). The larger band of 3,024 bp (the sum of the 2,409-bp fragment and a 615-bp kan cassette AvaII fragment) and a smaller band of 807-bp (the remaining 442 bp of the recA AvaII fragment plus 365 bp of the second AvaII fragment of the kan cassette) contain both recA and kan cassette sequences and confirm the insertion of the kan cassette and its orientation with respect to recA. The absence of the RecA protein in the mutant was demonstrated by Western immunoblotting (Fig. 1D, lane 4) employing B. burgdorferi RecA-specific rabbit antiserum.

In vitro sensitivity of B. burgdorferi to DNA-damaging conditions.

The B. burgdorferi recA mutant was tested for the genetic stability of the kanamycin resistance gene inserted at this locus by serial culture for 10 passages in BSK-S medium lacking kanamycin. The recA disruption remained genetically stable in the absence of selection (data not shown). In vitro growth characteristics of the wild-type and recA mutant at 34°C in BSK-S medium without antibiotic selection were compared. The wild-type B31-A3 and mutant growth curves were virtually superimposable, with doubling times of approximately 6 h (data not shown).

The DNA repair capabilities of both wild-type and recA mutant B. burgdorferi were evaluated by exposure to increasing levels of UV or mitomycin C followed by enumeration of survivors on solid BSK-S medium. The plating efficiencies of wild-type and unexposed mutant cells on solid BSK-S medium were not significantly different (data not shown). Both wild-type and mutant cells rapidly lost viability with increasing doses of UV irradiation (Fig. 2A). Loss of viability indicates that the accumulated DNA damage cannot be effectively repaired even in the presence of wild-type recA. At a UV dose of 1,600 μJ/cm² complete loss of viability was observed for both the wild type and mutant. The recA mutant was moderately more UV sensitive than the wild type at UV doses of 400 to 1,200 μJ/cm², but these differences did not reach statistical significance.

FIG. 2.

Sensitivity of B. burgdorferi recA mutant to DNA-damaging agents. (A) UV irradiation. (B) Mitomycin C. ⧫, wild type; ▪, recA mutant. An asterisk indicates a significant difference between the wild type and mutant.

The reduced role of RecA in DNA repair in B. burgdorferi is further supported by the moderate response of the mutant to mitomycin C exposure. Repair of cross-links in DNA caused by mitomycin C requires homologous recombination (33). Exposure of wild-type and mutant spirochetes to mitomycin C resulted in reduced viability for both strains (Fig. 2B). While the toxicity was greater in the recA mutant than in the wild type, this difference reached statistical significance only at the lowest dose of 2.5 ng/ml (P = 0.048). At higher mitomycin C concentrations, both strains showed similar patterns of resistance, resulting in approximately 25% survival at the highest concentration of the drug tested. These results suggest that RecA-mediated homologous recombination may be important at relatively low levels of DNA damage but that, at elevated levels of DNA interstrand cross-linking, RecA-independent repair may be functioning.

B. burgdorferi RecA mediates allelic exchange.

The ability of the recA mutant to promote homologous recombination was compared to that of the wild type. Two genetic loci were targeted for insertional inactivation by allelic exchange. Plasmids pGEM-T-BB0243::Spc^r and pGEM-T-BBE02::Spc^r were electroporated into wild-type and recA mutant B. burgdorferi, and streptomycin-resistant clones were selected by limiting dilution in BSK-S medium containing streptomycin and confirmed by PCR. Four of 96 transformation reactions had a disruption of BB0243, and 8 of 96 transformation reactions had a disruption of BBE02. In contrast, no allelic exchange was detected in 384 transformation reactions for either gene target in the recA mutant. The results demonstrate that allelic exchange in B. burgdorferi requires a functional RecA.

Effects of B. burgdorferi recA inactivation in vivo.

Groups of C3H/HeJ mice were infected with 10-fold serial dilutions of either the wild-type B31-A3 or recA mutant, and ID₅₀ was determined (Table 2). Both the wild type and mutant showed comparable ID₅₀ values: (2.77 ± 1.4) × 10³ and (3.17 ± 1.4) × 10³, respectively. PCR amplification of DNA isolated from day 28 cultures of ear tissues from mice infected with either wild-type or mutant spirochetes was positive, indicating that both strains were capable of dissemination from the original site of inoculation (Fig. 3). Thus, the absence of functional RecA did not significantly affect the ability of B31-A3 to infect mice.

TABLE 2.

Effect of recA inactivation on B. burgdorferi infection in C3H/HeJ mice

Strain (ID50)	Inoculum (no. of cells)	No. of culture-positive tissue samples/no. tested for:				No. of culture-positive mice/no. tested
		Ear	Joint	Bladder	Total sites
Wild type ([2.77 ± 1.4] × 103)	1 × 104	6/6	4/6	5/6	15/18	6/6
	1 × 103	2/6	2/6	0/6	4/18	2/6
	1 × 102	0/6	0/6	0/6	0/18	0/6
recA mutant ([3.17 ± 1.4] × 103)	1 × 104	6/6	4/6	4/6	14/18	6/6
	1 × 103	2/6	1/6	1/6	4/18	1/6
	1 × 102	0/6	0/6	0/6	0/18	0/6

FIG. 3.

PCR amplification of positive cultures from mouse tissues harvested on day 28 postinfection. PCR for recA was performed on cultured isolates from wild-type- or mutant-infected mice using primers pnTM17F and pnTM17R. Lane 1, wild-type control; lane 2, recA mutant control; lanes 3 to 5, cultures from mice infected with wild-type B31-A3; lanes 6 and 7, cultures from mice infected with the recA mutant; lane M, DNA molecular size markers.

Since the recA mutant is as infectious as the wild type when independently injected, the fitness of the mutant was explored in a coinfection model. Mice were simultaneously infected with a mixture of 5 × 10³ wild-type and 5 × 10³ mutant B. burgdorferi spirochetes. Ear biopsy samples taken at day 14 and joint and urinary bladder tissue taken after sacrifice of the animals at day 28 were cultured in BSK-S medium without kanamycin. Positive cultures were tested for the presence of recA and the kan cassette by PCR. Results of the competition assay are presented in Table 3. All ear biopsy cultures taken from seven infected animals (one mouse died) at day 14 were PCR positive for recA and six of seven were PCR positive for the mutated recA (as indicated by a larger recA-specific PCR product [Fig. 4]) and the kan cassette (as a second marker for the recA mutant allele). This indicates that all but one of the mice were infected with both wild-type and mutant spirochetes. Positive cultures were obtained from joint and urinary bladder tissue from these animals 2 weeks later (day 28). In contrast to the day 14 findings, none of the six recA PCR-positive cultures from the joints were PCR positive for the kan cassette. In addition, only one of five cultures from urinary bladder tissues contained the mutant (Fig. 4). These results suggest that functional RecA increases the fitness of B. burgdorferi in these tissues.

TABLE 3.

Competition between wild type and recA mutant during coinfection in C3H/HeJ mice^a

Animal	Resultb for:
	Earc			Jointd			Bladderd
	Culture	recAe	kane	Culture	recAe	kane	Culture	recAe	kane
M1	−	−	−	−	−	−	−	−	−
M2	+	+	+	−	−	−	+	−	+
M3	+	+	+	+	+	−	−	−	−
M4	+	+	+	+	+	−	−	−	−
M5	+	+	+	+	+	−	+	+	−
M6	+	+	+	+	+	−	+	+	−
M7	+	+	−	+	+	−	+	+	−
M8	+	+	+	+	+	−	+	+	−

^aIndividual mice were coinfected with 5,000 wild-type and 5,000 recA mutant spirochetes each. Mouse M1 was inoculated with BSK-S medium only and served as a negative control.

^bPositive or negative for B. burgdorferi (culture), recA, or the kan cassette.

^cDay 14 postinfection.

^dDay 28 postinfection.

^eAs determined by PCR.

FIG. 4.

PCR analysis of cultures obtained from tissues of mice infected with equal amounts of wild-type and recA mutant B. burgdorferi harvested on day 28 postinfection. (Top) PCR for recA using primers pnTM17F and pnTM17R; (bottom) PCR for the kanamycin cassette using primers pKNF and pKNR. Lane 1, wild-type DNA control; lane 2, recA mutant DNA control; lane 3, mouse M2 bladder; lane 4, mouse M3 joint; lane 5, mouse M4 joint; lane 6, mouse M5 joint; lane 7, mouse M5 bladder; lane 8, mouse M6 joint; lane 9, mouse M6 bladder; lane 10, mouse M7 joint; lane 11, mouse M7 bladder; lane 12, mouse M8 joint; lane M, DNA molecular size markers.

Role of RecA in vlsE antigenic variation.

Gene conversion at the vlsE locus occurs exclusively in mammalian hosts (60) and is a process driven by the host immune response (2), and VlsE expression is required for the maintenance of infection in mice (25, 40). Isolation of a recA null mutant allowed us to investigate the role of recA in vlsE antigenic variation. Mice were infected with wild-type and mutant B. burgdorferi, and ear biopsy samples collected at day 14 were cultured. DNA isolated from positive cultures was subjected to PCR amplification with vlsE-specific primers. vlsE was also amplified from in vitro cultures of wild-type cells. All vlsE PCR products were cloned into pUC18 and transformed into E. coli, and individual cloned vlsE genes were sequenced. An alignment of derived VlsE amino acid sequences for variable regions 1 to 5 is presented in Fig. 5. As expected no VlsE sequence variation was evident in B31-A3 grown in culture. In contrast, individual vlsE clones obtained from mice infected with either wild-type or mutant B. burgdorferi showed sequence divergence from the vlsE input cassette of B31-A3. These sequence variations included single nucleotide changes, insertions, and/or deletions and occurred exclusively at locations encoding five of the six reported variable regions of the gene (61). No additional sequence differences in the nonvariable regions of the gene were observed in any of the vlsE clones analyzed.

FIG. 5.

Alignment of VlsE amino acid sequences at variable region 1 (VR-1) through VR-5 generated in mice infected with wild-type or recA mutant B. burgdorferi. VlsE sequences were derived from vlsE amplified from mouse ear cultures taken on day 14 postinfection, cloned into pUC-18, and sequenced. The VlsE sequence obtained from vlsE amplified from a clonal isolate of B31-A3 grown only in culture served as the input VlsE amino acid sequence for comparison (top two rows). A dot indicates an identical amino acid at that position, a dash indicates deletion of the amino acid at that position, and an asterisk indicates a change from the input VlsE sequence in at least one isolate.

DISCUSSION

The present study assessed the role of RecA in DNA repair, homologous recombination, and vlsE antigenic variation in B. burgdorferi. The results demonstrate that RecA is minimally involved in DNA repair but is required for homologous recombination that results in allelic exchange. The recA mutant was able to generate VlsE recombinants in mice, indicating that the mechanism of vlsE gene conversion is RecA independent. This is in contrast to antigenic variation at the pilE locus of Neisseria gonorrhoeae, which involves recombination between the complete pilE gene and several partial, silent pilS cassettes and requires the activities of the host RecA (21). A similar requirement for host rad51, the eukaryotic homologue of recA, has been reported for antigenic variation of variant surface glycoprotein genes in Trypanosoma brucei (8). There are several examples of RecA-independent DNA rearrangements involving bacterial phase variations. For example, in Pseudomonas fluorescens motility variants arise in flagellin during root colonization by a RecA-independent mechanism (7), and lipopolysaccharide phase variation observed in Legionella pneumophila is induced by a RecA-independent deletion of a 30-kb fragment from its chromosome (31).

Comparison between VlsE recombinants generated by wild-type B31-A3 and recA mutant cells showed no qualitative differences. In both the wild type and mutant insertions and deletions were observed exclusively in the VlsE variable regions but not in the constant regions. The recombination specificity and mosaic nature of such genetic rearrangements in vlsE are suggestive of a site-specific recombinase/transposase and not of homologous recombination mediated by RecA. The presence of two 17-bp direct repeats in the DNA flanking both the vlsE variable cassette and the silent cassettes and the reported requirement for cis arrangement of the vlsE expression cassette and vls silent cassettes on plasmid lp28-1 support this view (28). RecA-mediated recombination would not be expected to have such molecular constraints but would rely solely on DNA homology wherever it may be located in the genome.

Recombination has been suggested as a mechanism of variation for several B. burgdorferi genes, including vlsE (61), ospA (46), ospB (5), ospC (18), ospD (32), and members of the ospE and ospF gene families (54, 56). A study by Qiu et al. of 18 polymorphic loci in several clinical isolates of B. burgdorferi concluded that genetic recombination was the driving force of diversification at these loci (43). Dykhuizen and Baranton estimated the recombination frequencies of several B. burgdorferi isolates and hypothesized that these spirochetes may employ a limited genetic transfer system that allows the recombination of relatively small (<1-kb) DNA fragments (10). Whether these proposed recombination events are homologous and thus RecA dependent remains an unanswered question. The results of the current study indicate that vlsE recombination is not RecA mediated.

The presence of a functional homologous DNA recombination pathway in B. burgdorferi is indicated by the successful inactivation of numerous B. burgdorferi genes by allelic exchange (47). Here RecA function was shown to be indispensable in this process, since no allelic exchange was observed in the recA mutant. Successful complementation of E. coli recA null mutants with B. burgdorferi recA indicates that this gene encodes a functional protein (30, 42).

There is a lack of information regarding DNA recombination in spirochetes that may be attributed to both their fastidious nature and inherent resistance to genetic manipulation. The only previous report of inactivation of a gene (recA) involved in recombination in a spirochete has been documented for Leptospira biflexa (19). Previous attempts to isolate a recA null mutant B. burgdorferi were unsuccessful despite extensive effort by several laboratories, and the indispensability of recA in this spirochete has been suggested (42). The successful inactivation of recA in B. burgdorferi B31-A3 reported here indicates that this is not the case.

Inactivation of recA in B31-A3 occurred after two transformation attempts. This is in contrast to many previous attempts at recA inactivation in strains B31MI, 297, and N40 and clinical isolates BL206 and B356, all of which had failed. Isolation of a recA null mutant B31-A3 could be ascribed to either serendipity or its ability to somehow tolerate the loss of RecA function. The acquisition of a rare compensatory mutation enabling the spirochete to overcome recA inactivation as a necessary precondition for isolation of a recA null mutant cannot be ruled out. Such a compensatory mutation has been suggested for tolerance of a recA null mutation in Streptomyces lividans (57).

Previous studies using an E. coli surrogate system and a cloned B. burgdorferi recA demonstrated that the B. burgdorferi gene was able to induce the SOS response in E. coli and to promote lysis of λ lysogens by cleaving both LexA and λ phage CI repressor proteins (30). However, neither the SOS response nor λ phage induction is normally part of B. burgdorferi biology. While the SOS response in B. burgdorferi is apparently absent and no lexA ortholog is annotated in the B. burgdorferi genome, the coprotease activity of RecA is functional. The putative target of this coprotease activity in B. burgdorferi is unknown. recA was not induced in B. burgdorferi following UV irradiation, in contrast to what is observed in E. coli (30). The apparent absence of the SOS response has also been noted in other host-dependent spirochetes such as Treponema pallidum and Treponema denticola (50, 55). In contrast, the free-living spirochete Leptospira biflexa encodes a lexA orthologue and components of a damage-inducible SOS response (19, 44).

Wild-type B. burgdorferi is extremely UV sensitive. This sensitivity suggests that RecA plays a minor role in DNA repair of UV-induced photoproducts (30). B. burgdorferi could potentially recruit the nucleotide excision repair pathway as a backup system to repair DNA. All genes of this pathway (uvrA, uvrB, uvrC, and uvrD) are present in B. burgdorferi (12). However, in the absence of LexA, the role of RecA in the expression of these DNA repair genes is unclear. Alternately, increased UV sensitivity of wild-type B. burgdorferi may be the result of inefficient restart of DNA replication forks blocked by UV-induced lesions. This process, designated UV-induced replisome reactivation, requires the functions of both RecA and RecF proteins, the latter of which is apparently absent in B. burgdorferi (20).

Although the recA mutant of B. burgdorferi was more UV sensitive than the wild type, this reduction in viability was not statistically significant. Interestingly, under similar experimental conditions an L. biflexa recA null mutant showed almost total loss of viability, indicating a critical dependence on RecA function for its survival. This mutant also showed significant in vitro growth defects, compared to the wild type (19). In contrast, no such growth defects were apparent for the B. burgdorferi recA mutant. The differential responses to UV irradiation shown by B. burgdorferi and L. biflexa recA mutants may be due to their very different lifestyles. B. burgdorferi cycles exclusively between tick and mammalian hosts. Neither of these environments would provide an opportunity for exposure to UV irradiation. In contrast, L. biflexa is a free-living aquatic spirochete that would be exposed to UV via sunlight. Thus, reliance of B. burgdorferi on a RecA-mediated SOS-like response to DNA damage may be evolutionarily minimized, whereas dependence on RecA and LexA is critical for L. biflexa. recA is absent in numerous obligate intracellular organisms (e.g., Buchnera aphidicola and Blochmannia pennsylvanicus), suggesting that in some stable environments RecA functions may be dispensable (36).

Repair of interstrand DNA cross-links and base adducts produced by mitomycin C requires recombination and base excision in E. coli (17, 33). Unrepaired, these lesions can lead to single-strand gaps and double-stranded breaks in DNA. Excision repair can repair most of the damage, but homologous recombination is needed for the repair of double-stranded breaks in DNA (51). This may also be the case for B. burgdorferi DNA repair following mitomycin C exposure. Comparison of survival curves for the wild type and recA mutant at increasing levels of mitomycin C suggests that a RecA-mediated homologous recombination repair may be taking place at levels of DNA damage caused by the lowest dose (2.5 ng/ml) of mitomycin C. However, the similar levels of survival shown by both mutant and wild-type B. burgdorferi strains at elevated mitomycin C concentrations suggest that under those conditions either excision repair is activated or an alternate RecA-independent recombination may become functional. A RecA-independent recombination pathway that becomes evident after inactivation of cytoplasmic exonucleases has been recently proposed for E. coli (9, 39). The gene(s) involved in this pathway in E. coli is currently unknown. It is conceivable that a RecA-independent recombination activity may become evident in B. burgdorferi under stress conditions, either when homologous recombination is overwhelmed by extensive DNA damage or when RecA becomes functionally inactive due to mutation. However, this activity does not appear to be involved in allelic exchange since no recombinants were observed in the recA null mutant.

When mice were infected with equal numbers of wild-type and mutant spirochetes, the recA mutants were absent from joint and bladder tissues of the infected mice. This indicates that mutant spirochetes are somehow less fit than their wild-type counterparts. Thus, RecA and homologous recombination must provide functions which B. burgdorferi requires in order to effectively persist in mouse tissues. This is not due to vlsE antigenic variation since such variation occurs equally well in the mutant. On the other hand, the presence of functional RecA cannot overcome the complete loss of vlsE since vlsE is required for persistence in mice (26). Thus, it appears likely that both vlsE antigenic variation and homologous recombination are required for successful persistent infection of the mouse host. The mechanism for the apparent reduction in fitness of recA mutants is currently unclear, but the use of immunodeficient (SCID) mice in competition infection experiments could provide further evidence for the involvement of acquired immunity in eliminating the RecA mutants.

The studies presented here demonstrate that a B. burgdorferi recA mutant is viable and only moderately more sensitive to DNA-damaging treatments (UV irradiation and mitomycin C) than wild-type spirochetes. Furthermore, RecA is not required for generation of VlsE antigenic variants. Despite this, the mutant appears to be less fit than the wild type in terms of persistence in infected mouse tissue. The precise role of RecA in B. burgdorferi biology awaits further elucidation.