Cross-Species Surface Display of Functional Spirochetal Lipoproteins by Recombinant Borrelia burgdorferi

Abstract

Surface-exposed lipoproteins of relapsing fever (RF) and Lyme borreliosis Borrelia spirochetes mediate certain interactions of the bacteria with their arthropod and vertebrate hosts. RF spirochetes such as Borrelia hermsii serially evade the host's antibody response by multiphasic antigenic variation of Vsp and Vlp proteins. Furthermore, the expression of Vsp1 and Vsp2 by Borrelia turicatae is associated with neurotropism and higher blood densities, respectively. In contrast to RF Borrelia species, the Lyme borreliosis spirochete Borrelia burgdorferi is amenable to genetic manipulation. To facilitate structure-function analyses of RF surface lipoproteins, we used recombinant plasmids to introduce full-length vsp1 and vsp2 as well as two representative vlp genes into B. burgdorferi cells. Recombinant B. burgdorferi cells constitutively expressed the proteins under the control of the B. burgdorferi flaB promoter. Antibody and protease accessibility assays indicated proper surface exposure and folding. Expression of Vsp1 and Vsp2 conferred glycosaminoglycan binding to recombinant B. burgdorferi cells that was similar to that observed with purified recombinant proteins and B. turicatae expressing native Vsp. These data demonstrate that the lipoprotein modification and export mechanisms in the genus Borrelia are conserved. They also validate the use of recombinant B. burgdorferi in studies of surface lipoprotein structure-function and the biogenesis of spirochete membranes.

Since Braun's description of a lipoprotein (Lpp) in the cell envelope of Escherichia coli (10), the important roles of lipoproteins in bacterial physiology and pathogenesis have been increasingly appreciated. Lpp and other prominent lipoproteins such as Pal are anchored in the inner leaflet of the outer membrane, where they interact with peptidoglycan in the periplasm to stabilize the cell envelope (17). Studies on lipoprotein export and modification have revealed that Sec-dependent translocation is followed by a three-step modification of the polypeptide on the periplasmic face of the inner membrane, resulting in a mature lipoprotein with an acylated amino-terminal cysteine anchor (42). In Escherichia coli and likely other diderm (i.e., double membrane) bacteria (29), lipoprotein sorting and export to the outer membrane are determined by the identity of the penultimate N-terminal amino acid (37) and occur by the Lol pathway (37, 52, 53).

Only a few bacterial genera have actually been shown to deploy lipoproteins to their surface, and the mechanisms for this are largely unknown. The Klebsiella oxytoca pullulanase PulA is transported to the outer surface through a complex type II secretion apparatus (42). Other characterized surface lipoproteins of bacterial pathogens include Neisseria meningitidis TbpB, which acts as part of a two-component transferrin receptor (2), the subtilisin-like protease SphB1 of Bordetella pertussis (20), a potential adhesin of Campylobacter jejuni, JlpA (31), and multiple polypeptides expressed by Mycoplasma spp. (38).

Lipoproteins are particularly abundant in the spirochetal genera Treponema, Brachyspira, Leptospira, and Borrelia (30). Several of them localize to the bacterial surface, i.e., the host-pathogen interface, and have thus received considerable attention as potential virulence determinants and vaccine targets. In Borrelia spp., the agents of arthropod-borne Lyme borreliosis and relapsing fever (RF) (4), surface lipoproteins are important factors in pathogen transmission and persistence. For example, the Lyme borreliosis spirochete Borrelia burgdorferi expresses outer surface protein OspA in the unfed tick; another lipoprotein, OspC, is upregulated during tick feeding and thought to be involved in tick-to-mammal transmission (28, 48). RF spirochetes such as Borrelia hermsii repeatedly evade the host's immune response by serial surface expression of immunodominant and antigenically variant lipoproteins, the variable small (Vsp) and large (Vlp) proteins (4). Two Vsps of Borrelia turicatae are associated with niche selection; cells expressing Vsp2 (previously named VspB) grow to high densities in the blood, while those expressing Vsp1 (VspA) exhibit a neurotropic phenotype (14-16, 40).

The structural features that underlie the biological functions of these and other borrelial surface lipoproteins have only begun to emerge. Our previous notion that RF Borrelia Vsps and Vlps share primary and secondary structural features with Lyme borreliosis Borrelia OspC and VlsE proteins (13, 54) was confirmed by X-ray crystallography. In contrast to OspA, which has a repetitive antiparallel beta topology (35), OspC (24, 33) and B. burgdorferi VlsE (23) are highly alpha-helical. Our recent determination of the B. turicatae Vsp1 structure showed that Vsps share a dimer four-helix bundle fold with OspC (C. L. Lawson, W. R. Zückert, and A. G. Barbour, unpublished data; 54). The observed variation between known Vsp alleles is predicted to occur mainly in the intervening loops and short beta-sheets distal to the bacterial outer membrane. We therefore concluded that the variation within these loops is likely responsible for the different biological functions of the paralogs belonging to this protein family (54).

The structure-function analysis of Vsp and Vlp proteins has been hampered by the lack of a genetic system for RF spirochetes and the limitations of currently available lipoprotein surface display options. While Escherichia coli has been used as a surrogate host for the surface expression of a variety of target proteins (19), this approach has been unsuitable to date, as borrelial lipoproteins are associated with the cytoplasmic membrane (22). Bunikis et al. showed that cultivable spirochetes can be decorated with exogenous recombinant lipoproteins (12), yet this “resurfacing” led only to a transient, nonheritable change in phenotype and required the purification of lipidated recombinant proteins.

Basic molecular tools for the mutation and extrachromosomal complementation of B. burgdorferi genes are now available (9, 49). Since both Lyme borreliosis and RF borreliae have abundant surface lipoproteins, we asked whether genetically engineered B. burgdorferi could express and present proteins of RF Borrelia spp. and potentially other spirochetes. If so, this would be evidence that their lipoprotein export machineries are compatible. At least the initial steps of lipoprotein translocation and modification appear to be conserved: orthologs of all essential components of the Sec translocase complex as well as the three enzymes required for lipoprotein biosynthesis are present in B. burgdorferi (26), as well as in the syphilis agent Treponema pallidum (27). Furthermore, spirochetal lipoprotein signal sequences and signal II peptidase recognition sites (lipoboxes) have characteristics in common (30). The rules and mechanisms for lipoprotein sorting and transport to a spirochete's outer surface, however, are unknown.

In this study, we introduced recombinant, autonomously replicating plasmids containing full-length vsp or vlp genes into B. burgdorferi. The resulting recombinant B. burgdorferi express and display functional Vsp or Vlp proteins on their surfaces, indicating that lipoprotein pathways in Borrelia spp. are conserved. B. burgdorferi could therefore serve as a model organism to study lipoprotein export and membrane biogenesis in this genus and maybe even in other spirochetes. Furthermore, this surface display system for recombinant lipoproteins provides a useful tool for the characterization of lipoprotein domains important in the interactions of borreliae with their arthropod and vertebrate hosts.

(This work was presented in part at the 9th International Conference on Lyme Borreliosis and Other Tick-Borne Diseases, 18 to 22 August 2002, in New York, N.Y.)

MATERIALS AND METHODS

Cloning of recombinant plasmids containing P_flaB-vsp and -vlp fusions.

Fusions of the B. burgdorferi flaB promoter to full-length B. turicatae vsp1 and vsp2 as well as B. hermsii vlp7 and vlp21 were constructed by PCR as described previously (55) with Pwo proofreading DNA polymerase (Roche Molecular Biochemicals) and the oligonucleotide primers listed in Table 1. The flaB promoter was amplified from pJLB12A (9). Full-length vsp1, vsp2, vlp7, and vlp21 genes were amplified from previously described recombinant plasmids (15, 32, 40, 41). The P_flaB-vsp1 and -vsp2 fusion products were digested with BamHI and HindIII, the P_flaB-vlp7 and -vlp21 fusions were digested with BamHI and XbaI and cloned in pBSV2 (49). E. coli TOP10 (Invitrogen) transformants were grown in Luria-Bertani broth (46) containing 30 μg of kanamycin per ml, and expression of lipoproteins was checked by Western blot analysis of whole-cell lysates with monoclonal antibodies against Vsp1 (1H12), Vsp2 (5F12), Vlp7 (H12915), and Vlp21 (H10022) (8, 15).

TABLE 1.

Oligonucleotides used in this study

Name	Target	Sequence (5′ to 3′)a
BamPflaB-fwd	5′ end of PflaB	CGGGATCCTGTCTGTCGCCTCTTG
PflaBNdevsp1/2-fwd	PflaB-vsp fusion	TGGAGGAATGACATATGAAAAGAATTAC
PflaBNdevsp1/2-rev	PflaB-vsp fusion	GTAATTCTTTCCATATGTCATTCCTCCA
PflaBNdevlp7/21-fwd	PflaB-vlp fusion	TGGAGGAATGACATATGAGAAAAAGAAT
PflaBNdevlp7/21-rev	PflaB-vlp fusion	ATTCTTTTTCTCATATGTCATTCCTCCA
Hindvsp1-rev	3′ end of vsp1	CCCAAGCTTCATTAGTTAGATTGAACAG
Hindvsp2-rev	3′ end of vsp2	CCCAAGCTTCATTAGTTAGATTGAGCAG
Xbavlp7-rev	3′ end of vlp7	GCTCTAGATTATCACTTACTTGATTCTG
Xbavlp21-rev	3′ end of vlp21	GCTCTAGATCATTATTGCTGACCTGC

^aUnderlined nucleotides are introduced restriction sites, and bold nucleotides are start and stop codons.

B. burgdorferi transformants.

B. burgdorferi cells were transformed by electroporation with established protocols (49). Briefly, approximately 10¹⁰ electrocompetent high-passage B. burgdorferi B31 (ATCC 35210) or B313 (45) cells were electroporated at 12.5 kV/cm in 2-mm cuvettes with 10 to 20 μg of plasmid DNA. Cells were immediately diluted in 6 ml of BSK-II (3), incubated for 24 h without selection, and then plated in the top agar of selective BSK-II plates containing 200 μg of kanamycin per ml. Colonies were picked approximately 8 days after plating and used for inoculation of 6-ml cultures of selective BSK-II.

Pre- and posttransformation B. burgdorferi B31 and B313 plasmid profiles were determined by PCR with plasmid-specific oligonucleotide primer pairs (34, 43).

Antibody and protease accessibility assays.

Borrelia cells were grown in BSK-II and harvested as described previously (3). Intact Borrelia cells were treated in situ with trypsin (Roche Molecular Biochemicals) or proteinase K as described previously (11, 54). Whole-cell proteins were probed with the anti-Vsp and -Vlp monoclonal antibodies listed above as well as a monoclonal antibody against B. burgdorferi FlaB (H9724) (6).

Antibody agglutination assays were performed with a procedure modified from that of Barbour and Bundoc (5). Briefly, Borrelia cells were incubated for 1 h at room temperature on a rotary shaker in undiluted monoclonal antibody hybridoma supernatants or 1:10 diluted ascites fluids in phosphate-buffered saline containing 5 mM MgCl₂. Cells were then observed under phase contrast at 400x magnification with an Olympus BX60 microscope and PM-30 camera.

Protein gel electrophoresis and immunoblot analysis.

Proteins were separated by sodium dodecyl sulfate-12% polyacrylamide gel electrophoresis (SDS-PAGE) and visualized by Coomassie blue staining. For immunoblots, proteins were electrophoretically transferred to nitrocellulose membranes (Immobilon-NC; Millipore) as described previously (56). Membranes were rinsed in 20 mM Tris-500 mM NaCl, pH 7.5 (TBS), and either air-dried or processed directly; 5% dry milk in TBS with 0.05% Tween 20 was used for membrane blocking and subsequent incubations for 1 h each. TBS with 0.05% Tween 20 alone was used for the intervening washes. Mouse monoclonal antibody hybridoma supernatants were used at a 1:25 dilution, and ascites fluids were used at a 1:100 dilution. Alkaline phosphatase-conjugated protein A/G (ImmunoPure protein A/G calf intestinal phosphatase conjugate; Pierce) at a 1:5,000 dilution was used as the second ligand, and a stabilized alkaline phosphatase substrate solution (1-Step nitroblue tetrazolium/5-bromo-4-chloro-3-indolylphosphate [NBT/BCIP]; Pierce) was used for colorimetric detection.

Densitometric analysis of protein bands was performed on a scanned digital gel image with an Epson Perfection 2450 photo scanner and a Macintosh computer running the public domain NIH image program (http://rsb.info.nih.gov/nih-image).

Glycosaminoglycan-binding assays.

Binding of bacteria to immobilized glycosaminoglycans was assayed according to established protocols (36). Briefly, B. burgdorferi cells were metabolically labeled with 100 μCi of [³⁵S]methionine and [³⁵S]cysteine (Amersham Biosciences) per ml in BSK II, washed with phosphate-buffered saline, and stored at −80°C in serum-less BSK II and 20% glycerol. Heparin (Sigma, H9399), chondroitin sulfate B (Sigma, C3788), chondroitin sulfate A (Sigma, C8529), and chondroitin sulfate C (Sigma, C4384) were bound to microtiter plates (Nunc Maxi-Sorp 96-well break-apart). After blocking the wells with bovine serum albumin, approximately 10⁶ labeled spirochetes were added per well, incubated, and washed. Inhibition of binding by exogenous glycosaminoglycans was assayed by adding heparin and chondroitin sulfate B to the cells at a final concentration of 1 mg/ml prior to incubation with immobilized glycosaminoglycans. Bound cells were quantified by scintillation counting in Biosafe-II (Research Products International) scintillation fluid.

RESULTS

Expression of B. turicatae Vsp and B. hermsii Vlp proteins by recombinant B. burgdorferi.

In the absence of molecular tools to directly manipulate sequences in RF spirochetes, we employed the Lyme disease spirochete B. burgdorferi as a host for heterologous expression of surface lipoproteins. We used derivatives of the E. coli/B. burgdorferi shuttle vector pBSV2 (49) to introduce four representative RF lipoprotein genes, B. turicatae vsp1 and vsp2 as well as B. hermsii vlp7 and vlp21, into B. burgdorferi type strain B31 and one of its derivatives, B313 (45). B313 lacks most of the linear plasmids (45) and some of the circular plasmids (56) found in its parent, B31.

To obtain constitutive expression of these four lipoproteins in B. burgdorferi, their full-length genes (i.e., the sequences including their signal sequence and lipoboxes) were separately fused to the B. burgdorferi flaB promoter (P_flaB) by overlap PCR. Whole-cell lysate proteins were separated by SDS-PAGE, and immunoblots were probed with Vsp- and Vlp-specific monoclonal antibodies. Novel protein bands representing recombinant Vsp1, Vsp2, Vlp7, and Vlp21 were observed (Fig. 1).

FIG. 1.

Expression of RF Borrelia surface lipoproteins Vsp1, Vsp2, Vlp7, and Vlp21 in B. burgdorferi. Total cellular proteins of B31 cells harboring pBSV2 alone (negative control, −) and pBSV2 clones harboring fusions of P_flaB with vsp1, vsp2, vlp7, and vlp21 were separated by SDS-PAGE, and transferred proteins were probed with Vsp- and Vlp-specific monoclonal antibodies. Vsp and Vlp bands are indicated by asterisks (*). The sizes of protein molecular size markers (Invitrogen) are indicated to the left (in kilodaltons).

Based on a densitometric analysis of Coomassie blue-stained polyacrylamide gels of B. burgdorferi B31 harboring pBSV2.vsp1 and B. turicatae serotype 1, the expression levels of recombinant Vsp and Vlp proteins in recombinant B. burgdorferi ranged from 15 to 20% of their respective wild-type levels in B. turicatae (Fig. 2B). Expression levels were consistent for all four lipoproteins expressed. A recent study by Fischer et al. (25) observed similar levels of expression of DbpA and -B lipoproteins when driven by an identical flaB promoter. Given the usual abundance of FlaB in Borrelia whole-cell protein preparations, such a low level of expression from a flaB promoter fusion is surprising and deserves further investigation. Our own preliminary Northern blot analyses of mRNA isolated from B. turicatae serotype 1 and B31 pBSV2.vsp1 suggest that this lower-than-expected level of expression is not due to protein degradation, but rather to lower levels of transcript (J. E. Lloyd and W. R. Zückert, unpublished data).

FIG. 2.

In situ protease and antibody accessibility of Vsp1 on recombinant B. burgdorferi and B. turicatae. (A) Intact B31 cells harboring pBSV2.vsp1, pBSV2.vsp2, pBSV2.vlp7, and pBSV2.vlp21 were incubated with (+) or without (−) 200 μg of proteinase K per ml. Total cellular proteins were separated by SDS-PAGE, and transferred proteins were probed with Vsp-, Vlp-, or FlaB-specific monoclonal antibodies. ‡, Vsp dimers (54). The sizes of protein molecular size markers (Invitrogen) are indicated to the left (in kilodaltons). (B) Intact B31 cells harboring pBSV2.vsp1 and B. turicatae (B.t.) serotype 1 cells expressing Vsp1 were incubated with (+) or without (−) 200 μg of trypsin per ml. Total cellular proteins were separated by SDS-PAGE, and transferred proteins were probed with Vsp1-specific monoclonal antibodies. *, Full-length lipidated Vsp1; †, cell-associated C-terminally truncated form; ‡, Vsp1 dimer (54); ▸, OspA. The sizes of protein molecular size markers (Invitrogen) are indicated to the left (in kilodaltons). (C) Intact B31 cells harboring pBSV2 and pBSV2.vsp1 were incubated with Vsp1-specific monoclonal antibodies.

Surface exposure of Vsp and Vlp proteins.

We next assessed whether the Vsp and Vlp proteins were properly exported and displayed on the surface of their B. burgdorferi host. It has previously been shown that surface exposure on intact spirochetal cells can be adequately assessed by the in situ accessibility of the proteins in question to proteases and antibodies (11, 54). Surface proteins or surface-exposed domains of integral membrane proteins can be cleaved by proteases such as proteinase K and trypsin, and incubation with surface protein-reactive antibodies can lead to the agglutination of bacterial cells. In addition, these experiments can also serve as a general assay for the proper two-dimensional folding and assembly of proteins.

To assess protease accessibility, we followed established protocols (11) with proteinase K. In these studies, final proteinase K concentrations of 200 μg/ml cleaved the surface-exposed loops of the integral membrane protein P66 and outer surface lipoprotein OspB to completion, while only traces of OspA were detectable; subsurface proteins such as periplasmic FlaB remained unaffected (11). We therefore incubated B31 cells expressing Vsp1, Vsp2, Vlp7, and Vlp21 for 1 h with 200 μg of proteinase K/ml. As Fig. 2A shows, Vsp1, Vsp2, Vlp7, and Vlp21 expressed by B31 were accessible to proteinase K, while FlaB was protected. Vsp1 and Vsp2 dimer bands disappeared upon protease treatment. A densitometric analysis of immunoreactive protein bands indicated that 50 to 60% of the expressed Vsps and Vlps were cleaved.

Next, we treated B31 expressing Vsp1 and B. turicatae serotype 1 for 1 h with trypsin at a final concentration of 200 μg/ml. While surface-exposed OspA is resistant to trypsin at this concentration (11), treatment of native Vsp1 in situ and recombinant Vsp1 in vitro leads to a protease-resistant core, which is truncated at both the carboxy and amino termini (54). The first cleavage leads to a truncated, cell-bound lipoprotein, while the latter removes Vsp1 from the bacterial surface. As Fig. 2B shows, the protein fragmentation patterns of Vsp1 were identical for B31/pBSV2.vsp1 and B. turicatae serotype 1. As observed with proteinase K, approximately 60% of total Vsp1 was cleaved and the Vsp1 dimer band was no longer detectable after protease treatment.

In an alternative approach, we tested for surface display of Vsp1, Vsp2, Vlp7, and Vlp21 by incubating recombinant B31 with monoclonal antibodies against the four lipoproteins. B31 harboring pBSV2 without an insert was included as a control. As the example of Vsp1 in Fig. 2C shows, the presence of antibodies against Vsp1 led to the agglutination of B31/pBSV2.vsp1 cells, while B31/pBSV2 cells remained unaffected. Interestingly, the anti-Vlp7 antibody did not agglutinate B. burgdorferi cells expressing protease-accessible, i.e., surface-exposed, Vlp7. This might be due to the so far unknown nature of the antibodies' epitope and deserves further investigation. However, agglutination was obtained with cells expressing Vsp2 and Vlp21 in the presence of the respective antibodies (not shown). Together, these assays demonstrate that Vsp1, Vsp2, Vlp7, and Vlp21 are surface exposed. Furthermore, identical tryptic fragment patterns for native and recombinant Vsp1 as well as the antibody agglutination data indicate that the proteins are folded properly on the surface of their recombinant B. burgdorferi host.

Functional studies.

Cross-species surface display of lipoproteins by recombinant bacteria was achieved, but the utility of this system for structure-function studies hinged on whether the displayed proteins were functional. Magoun and colleagues observed that Vsp2, in contrast to Vsp1, had high affinities for certain glycosaminoglycans such as heparin and chondroitin sulfate B and that this was also phenotypically mirrored in the B. turicatae serotypes (36). Accordingly, we tested whether recombinant B. burgdorferi expressing Vsp1 and Vsp2 would show the same differences in glycosaminoglycan affinities.

Two full sets of preliminary glycosaminoglycan binding experiments were performed with B. turicatae serotypes 1 and 2 as well as recombinant B. burgdorferi B31 displaying Vsp1 and Vsp2. Radiolabeled Borrelia cells were incubated with immobilized heparin and chondroitin sulfate A, B, and C. After unbound cells were washed off, attached cells were quantified by scintillation counting. Confirming the measurements by Magoun et al. (36), an average of 19.9% of serotype 2 cells bound to heparin, compared to 3.2% of serotype 1 cells. Serotype 2 cells also bound better to chondroitin sulfate B than serotype 1 cells (means of 5.0 and 1.4%, respectively). The same trend was observed with recombinant B31 cells: the heparin and chondroitin sulfate B binding efficiency of Vsp2-expressing cells (5.1 and 7.8%, respectively) was reduced more than twofold with cells expressing Vsp1 (2.3 and 3.3%, respectively). However, we also observed high background binding to heparin (6.7%) and chondroitin sulfate B (3.7%) by control B31 cells harboring pBSV2 alone. This may be attributed to the recently described glycosaminoglycan-binding properties of B. burgdorferi Bgp and DbpA and -B proteins (25, 39). We concluded that the presence of these proteins on the surface of strain B31 could at least partially obscure the binding activities of the coexpressed Vsps. For more detailed studies, we therefore decided to use the B31 derivative B313 (45).

PCR with plasmid-specific primers (34, 43) indicated that B313 carries only circular plasmids cp26, cp32-1, cp32-2/7, and cp32-3 as well as linear plasmid lp17, confirming previously determined partial plasmid profiles (45, 56). Compared to its parent strain B31, B313 therefore lacks lp54, encoding OspA and -B, as well as DbpA and -B. For reasons yet to be fully understood, B313 also fails to express the cp26-encoded OspC at detectable levels (45). The absence of these major glycosaminoglycan-binding lipoproteins on the B. burgdorferi surface makes B313 an almost ideal host for studying the interactions of heterologously expressed surface lipoproteins, particularly the glycosaminoglycan binding mediated by Vsp1 and -2.

As determined by PCR, B313 transformants harboring pBSV2, pBSV2.vsp1, or pBSV2.vsp2 maintained all plasmids found in B313. Vsp expression levels and surface localization assay results were indistinguishable from the ones shown for B31 (Fig. 1 and 2). As previously observed with B. turicatae serotypes 1 and 2, with recombinant Vsp1 and Vsp2 (36), and with recombinant B31 cells in our preliminary studies, B313 cells expressing Vsp2 bound significantly better to heparin and chondroitin sulfate B than did cells expressing Vsp1 (Fig. 3). The addition of exogenous heparin and chondroitin sulfate B at 1 mg/ml, a concentration previously shown to block Vsp2-glycosaminoglycan interactions (36), interfered with binding of Vsp2-expressing B313 cells (not shown). This indicates that the surface lipoproteins expressed and displayed in this system are fully functional with respect to their glycosaminoglycan-binding properties.

FIG. 3.

Glycosaminoglycan binding of Vsp1- and Vsp2-expressing B. burgdorferi. Radiolabeled B313 cells expressing Vsp1 and Vsp2 were added to mock-coated microtiter wells (no glycosaminoglycan [GAG]) or wells coated with heparin or chondroitin sulfate (chon) A, B, or C. B313 cells harboring pBSV2 alone were used as controls. Stably bound bacteria were quantitated by scintillation counting. Each value represents the mean ± 95% confidence intervals of data from four independent experiments. Significant differences in values between pairs of strains are indicated by brackets and P values were derived from two-tailed Student's t tests. F factors and P values obtained from a one-way analysis of variance of heparin and chondroitin sulfate B binding data groups are boxed.

Although B. burgdorferi B313 cells lack DbpA and -B, they likely still express chromosomally encoded Bgp (39). Consequently, the binding of B313 harboring pBSV2 alone to heparin and chondroitin sulfates (Fig. 3) was not surprising. The comparatively reduced binding by B313 cells expressing Vsp1 to chondroitin sulfate B suggests that the surface-displayed recombinant Vsp1 masks the endogenous B. burgdorferi ligand.

DISCUSSION

Recent work by Schwan and colleagues successfully used B. burgdorferi as a host to study the activity of the RF Borrelia-specific GlpQ and GlpT proteins in phospholipid metabolism (47). Here, we used a similar genetic approach to display heterologous spirochetal lipoproteins on the surface of recombinant B. burgdorferi cells. Four representative variable major surface lipoproteins from RF borreliae were properly expressed, processed, and exported to the outer leaflet of the B. burgdorferi outer membrane. Furthermore, the glycosaminoglycan binding phenotypes of Vsp1- and Vsp2-expressing B. burgdorferi cells were similar to those previously observed with recombinant Vsp1 and Vsp2 proteins as well as their respective B. turicatae serotypes, indicating the surface localization of biologically functional spirochetal lipoproteins.

The effects of Vsp1 and Vsp2 expression by B. burgdorferi on the cells' phenotype are consistent with previous hypotheses on the outer surface architecture of borreliae. Bunikis and Barbour have shown that B. burgdorferi OspA is able to protect an otherwise surface-exposed loop of the integral membrane protein P66 from antibodies and proteases (11). This suggested that the outer surface of Borrelia organisms consists of multiple layers of membrane-proximal integral membrane proteins and membrane-distal lipoproteins. Similarly, the expression of “low-affinity” Vsp1 or “high-affinity” Vsp2 appears to mask the endogenous glycosaminoglycan binding of B. burgdorferi DbpA and DbpB (25) and Bgp (39). It is highly unlikely that the altered binding efficiencies are due to unintentional genetic differences between the strains tested; (i) the transformants were obtained in parallel from the same stock of competent cells, (ii) the plasmid profiles of all B313 strains tested were identical, and (iii) the same phenomenon was observed independently with Vsp-expressing B31 and B313 cells. Since the overall structures of the Vsp proteins are conserved (C. L. Lawson, W. R. Zückert, and A. G. Barbour, unpublished; 54), differential masking of the endogenous adhesins by Vsp1 or Vsp2 is improbable as well.

The proper export of RF surface lipoproteins by B. burgdorferi indicates that the pathways of membrane biogenesis in Borrelia spp. are conserved. Thus, B. burgdorferi could serve as a model for studying lipoprotein export in other Borrelia species and potentially other spirochetal genera. We hypothesize that the lipoprotein modification and export machinery in spirochetes has the same evolutionary roots as the ones observed in other diderm bacteria such as E. coli. Protein orthologs present in B. burgdorferi (26) and T. pallidum (27) seem to provide for Sec-dependent translocation and acylation on the periplasmic side of the inner membrane (30).

The mechanisms of spirochetal lipoprotein transport to the outer leaflet of the outer membrane are still unidentified. In E. coli, an inner membrane ABC transporter-like complex, LolCDE, releases outer membrane-targeted lipoproteins from the inner membrane in an ATP-dependent manner (52). The released lipoproteins then form a water-soluble complex with the periplasmic chaperone LolA (53). After crossing the periplasm, this complex interacts with the outer membrane receptor LolB, which mediates anchoring to the inner leaflet of the outer membrane (53). Lipoproteins with aspartate at the penultimate amino-terminal position avoid being recognized by LolCDE and are retained in the inner membrane (37). This pathway appears to be conserved in several gram-negative microorganisms, and Masuda et al. (37) have also referred to potential B. burgdorferi Lol orthologs.

While these orthologs could function similarly in spirochetal lipoprotein export, some differences are evident. First, the primary sequences of spirochetal lipoproteins indicate that sorting signals are different from the ones dictating inner or outer membrane localization in E. coli (30). Indeed, borrelial surface lipoproteins expressed in E. coli are mislocalized to the cytoplasmic membrane (22). A second intriguing observation is the apparent absence of a lolB ortholog in the B. burgdorferi genome. This suggests that gram-negative and spirochetal lipoprotein export pathways may diverge at the outer membrane, which could be reflected in the different localization of E. coli and Borrelia major outer membrane lipoproteins to the inner and outer leaflet, respectively.

If an E. coli Lol-like pathway is in fact responsible for spirochetal lipoprotein export, lipoproteins would be detectable not only on the surface, but also in transit, i.e., in the inner membrane, the periplasm, and possibly the inner leaflet of the outer membrane. This may explain the partial protection of full-length Vsps and Vlps from proteases observed in this and a previous study (54). Together with the proposed “self-decoration” of membranes with previously released lipoproteins (12), it might also account for the so far paradoxical subsurface localization of Borrelia Osp proteins (21). Yet, pending further studies on spirochetal lipoprotein export, alternatives to the Lol pathway, such as a Klebsiella oxytoca pullulanase-like type II secretion system (42) or a completely novel mechanism, cannot be excluded.

Before the genetic manipulation of Borrelia spp. became possible, several other approaches resulting in surface exposure of spirochetal proteins were used successfully. As part of the development of an improved vaccine vehicle with increased immunogenicity of target antigens, B. burgdorferi OspA was expressed by recombinant Mycobacterium bovis strain BCG (50). Phage display of B. burgdorferi peptides has been used to identify candidate integrin-binding proteins (18). Using a nongenetic approach, Bunikis et al. described the decoration of spirochetes with exogenous lipoproteins leading to a nonheritable change in phenotype in a variety of cultivatable Borrelia species (12).

In complementing the approaches listed above, the benefits of using recombinant Borrelia organisms as surface display hosts in structure-function analyses of spirochetal surface lipoproteins are obvious. First, the system takes advantage of the endogenous spirochetal protein export machinery, which renders chimeric protein fusions unnecessary and removes restrictions on peptide size. This might be particularly important for studies of discontinuous conformational antibody epitopes and multimeric proteins such as the Vsps, where functional domains may span several subunits. Second, it allows the exertion of targeted selective pressure on stably expressed surface lipoproteins that would otherwise undergo antigenic switching. This might facilitate the mapping of antibody epitopes by selecting for escape mutants (44) and ultimately in vivo pathogenicity studies once the obstacles to routinely transform infectious Borrelia isolates are overcome. Third, it avoids the need to purify recombinant lipoproteins, which might preclude large-scale screening procedures for functional mutants. Last but not least, and similar to the approach of viral pseudotyping, e.g., inserting Ebola virus surface glycoproteins into murine leukemia viruses (51), transgenic Borrelia paves the way for studies of putative surface virulence factors of uncultivatable spirochetes such as T. pallidum, Borrelia lonestari, the suspected agent of Southern tick-associated rash illness (7), and a Spanish RF agent (1).