Factor H (fH) is an important regulator of the alternative complement cascade. Several human pathogens have been shown to bind fH to their surface, a process that facilitates immune evasion or cell to cell interaction. Among the pathogens that bind fH are some Borrelia species associated with Lyme disease and relapsing fever. The fH-binding proteins of the Lyme spirochetes form two classes (I and II). In Borrelia burgdorferi B31MI, class I includes the outer surface protein E (OspE) paralogs, L39, N38, and P38, whereas the class II group includes A68 and additional proteins that have not yet been identified. To identify the OspE determinants involved in fH and OspE-targeting infection-induced Ab (iAb) binding, deletion, random, and site-directed mutagenesis of L39 were performed. Mutations in several different regions of L39 abolished fH and or iAb binding, indicating that separable domains and residues of OspE are required for ligand binding. Some of the mutants that lost the ability to bind fH, iAb, or both had only a single amino acid change. Site-directed mutagenesis of three putative coiled coil motifs of OspE revealed that these higher order structures are required for fH binding but not for iAb binding. The data presented within demonstrate that the binding of fH and iAb to the OspE protein is mediated by higher order structures and protein conformation. These studies advance our understanding of fH binding as a virulence mechanism and facilitate ongoing efforts to use fH-binding proteins in the development of microbial vaccines.
Lyme disease is a tick-borne zoonosis caused by Borrelia burgdorferi, Borrelia garinii and Borrelia afzelii (1, 2, 3). In 2002, ∼24,000 cases were reported to the Centers for Disease Control and Prevention, and indications are that the endemic regions for this infection are spreading. A vaccine is not available for use in humans; hence, reducing tick exposure is the basis for disease prevention. Infection with the Lyme disease spirochetes is persistent, and consistent with this, mechanisms associated with evasion of both the acquired and innate immune defenses have been identified (4, 5, 6, 7, 8). Some species of the Lyme disease spirochetes bind the complement-regulatory proteins, factor H (fH)3 and fH-like protein 1 (fHL-1) to their surface (9, 10, 11, 12, 13, 14, 15). In mammals, fH and fHL-1 regulate the complement cascade through several different mechanisms; they serve as cofactors in the factor I-mediated cleavage of C3b, inhibit the binding of factor B to C3b, and dissociate preformed C3bBb complex (16, 17, 18, 19, 20). Through these activities, the production of C3b is regulated, thus keeping the alternative complement cascade in check. In terms of pathogenesis, the binding of fH and fHL-1 increases the effective concentration of these complement regulators at the pathogen cell surface, resulting in increased efficiency of C3b cleavage and decreased availability of C3b for opsonization and phagocytosis. The ability of surface-bound fH to participate in the cleavage of C3b, and hence the biological significance of fH binding, has been demonstrated for several Borrelia species (8, 13).
Differential binding of fH and fHL-1 has been reported for the Borrelia species associated with Lyme disease (B. burgdorferi, B. garinii, and B. afzelii) and relapsing fever (Borrelia hermsii, Borrelia parkeri, and Borrelia turicatae) (8, 9, 11, 13). Most B. burgdorferi, B. afzelii, B. hermsii, and B. parkeri isolates are able to bind fH (9, 11). Each of these species disseminate efficiently during infection. In contrast, the majority of isolates of species that localize in the CNSs (B. garinii and B. turicatae) do not bind fH. The inability to bind fH may favor residence in CNS, an environment that is devoid of complement. As eluded to above, fH and fHL-1 binding as a virulence mechanism is not unique to the Borrelia and plays an important role in the pathogenesis of several human infections or diseases. The Gram-positive streptococci bind fH through the M and β proteins (21, 22). The M protein also binds fHL-1 (23), a process that facilitates intracellular invasion by group A streptococci (24). The Gram-negative bacteria, Neisseria gonorrhoeae, increases its serum resistance by binding fH with porin protein 1A serving as the fH-binding protein (fHBP) (25, 26). Candida albicans binds fH and fHL-1, but its fHBPs have not been identified (27). Several parasites, including microfilariae of Onchocerca volvulus, the causative agent of river blindness, also exploit fH binding in their pathogenesis (28). Lastly, fH binding may contribute to complement resistance of tumors, facilitating their persistence (29). The diversity of organisms that exploit fH in their pathogenesis highlights the broad significance of this virulence mechanism.
The Lyme spirochetes produce several fHBPs (7, 8, 9, 10, 30) that can be divided into two distinct classes (11). The class I fH binding outer surface protein E (OspE) lipoproteins are a highly polymorphic family of proteins (31, 32, 33, 34). In B. burgdorferi B31MI, the OspE protein family contains three members, designated by The Institute for Genomic Research as N38, L39, and P38 (35). The class II fHBPs are more conserved (our unpublished data) and include the A68 protein (also referred to as CRASP-1 (14)) and other unidentified proteins (11, 14). In addition to their role in complement evasion, the OspE proteins also aid in evasion of the humoral immune response through antigenic variation. Sung et al. (5) demonstrated that ospE genes undergo mutation during infection, resulting in the generation of genes encoding new antigenic variants.
The interaction between fH and fHBPs of the Borrelia has been the focus of several recent studies. Human fH is a 150-kDa glycoprotein that is composed of 20 short consensus repeats. Binding to OspE (10) is mediated specifically by short consensus repeats 19 through 20 (36). The OspE determinants required for fH binding are not as clearly defined. Recent evidence suggests that the fH-binding domain of OspE is conformational or discontinuous (30) and that dispersed lysine residues are involved in a charge based interaction (37). Several studies have demonstrated the requirement for an intact OspE C-terminal domain for fH binding (15, 30, 37). In addition to fH binding, Metts et al. (30) investigated the binding of infection-induced Ab (iAb) to OspE and demonstrated that iAb requires both the N- and C-terminal domains of OspE for optimal binding. We use the term iAb to refer specifically to Ab elicited against the native, membrane-bound form of OspE that is presented by the bacteria during infection. The goals of this report were to further assess the interaction of OspE with fH and iAb and to identify specific domains or residues that are involved in ligand binding. The data demonstrate that binding of fH is dependent on the formation of a series of coiled coil (CC) structural motifs and that perturbation of other aspects of OspE structure results in the loss of fH binding. Although the binding of iAb was also found to be dependent on protein confirmation, the CC domains of OspE are not required for this interaction. The findings of this study advance our understanding of molecular aspects of fH binding as a virulence mechanism and provide important information that can be exploited in the design of vaccines to protect against Lyme disease and other infections caused by organisms that produce fHBPs.
Materials and Methods
OspE nomenclature and random mutagenesis of the OspE paralog, L39
Several nomenclature designations have been applied to the OspE proteins including OspE, p21, Erps, and CRASPs (complement regulator-acquiring surface proteins) (15, 33, 38, 39). In this report, the OspE paralogs are referred to as L39, P38, and N38, a nomenclature assigned by The Institute for Genomic Research (35). Because L39 and P38 are identical, henceforth they are collectively referred to as L39. Recombinant L39 variants generated by mutagenesis are distinguished by a hyphen followed by a number that indicates the Escherichia coli clone from which the gene was recovered (i.e., L39-9 is an L39 variant recovered from E. coli clone 9). To introduce random mutations into L39, the PCR-based GeneMorph II Random Mutagenesis kit was used using the reagents and protocol supplied by the manufacturer (Stratagene, La Jolla, CA). A pET32-Ek/LIC (enterokinase/ligase-independent cloning; Novagen, Madison, WI) plasmid, constructed in an earlier analysis (30), that carries the L39 gene served as template for PCR amplification using the low fidelity Mutazyme DNA polymerase. L39 was amplified with the L39-16+ and L39-173− primers (Table I) as previously described (30). These primers possess extensions that allow for annealing into the LIC series of expression vectors (Novagen). Three rounds of PCR were performed, and after each round the amplicons were purified using the Qiaquick Purification Kit (Qiagen, Valencia, CA). To generate the single-stranded overhangs required for annealing into the pET32-Ek/LIC vector, the gel-purified amplicons were treated with T4 DNA polymerase as directed by the manufacturer (Novagen). The resulting amplicons were annealed into the pET32-Ek/LIC vector and transformed into E. coli NovaBlue DE3 cells (Novagen) under standard conditions. The bacteria were plated on Luria-Bertani (LB) plates containing ampicillin (50 μg/ml), and colonies were picked and cultivated overnight at 37°C in LB broth with ampicillin (50 μg/ml). Induction of expression of the recombinant proteins was accomplished using isopropylthiogalactoside (IPTG). However, in some cases, IPTG induction proved unnecessary as sufficient basal expression of each rOspE protein was observed after overnight cultivation. Cell lysates of the E. coli clones were fractionated by SDS-PAGE and transferred onto polyvinylidene difluoride membranes (Pierce, Rockford, IL) by electroblotting as described below. The pET32-Ek/LIC vector used in this study allows for the production of recombinant proteins that carry both polyhistidine and S-tag fusions. The N-terminal tag adds ∼17 kDa to the molecular mass of the recombinant protein.
Construction of L39 internal deletion variants
Internal deletion mutants of OspE were generated using a PCR-based approach as outlined in Fig. 1. The segments of the gene that flank the region to be deleted were independently amplified. The primer of each primer pair that flanked the region to be deleted was constructed with a 12-base overhang. The overhang introduced into the B1, B2, and B3 primers complemented the overhang introduced into the C1 and C2 primers. After PCR of each gene segment, the two amplicons were mixed together and used as template for amplification using the L39-16+ and L39-173− primers. As a result of the presence of the 12-bp overhangs on the B and C series primers, all resulting deletion constructs encode a recombinant protein with a 4-aa insertion of the sequence ALAG at the deletion site, and all retain the proper reading frame. A control amplicon was also generated that harbored no deletion but carried the 12-bp insertion. The resulting recombinant protein harbors a 4-aa insertion. These amplicons were then annealed into the pET32Ek/LIC vector and propagated in E. coli NovaBlue cells. The resulting plasmids were purified and transformed into E. coli BL21 DE3 cells, and recombinant protein production was induced by IPTG induction as described above.
Site-directed mutagenesis of L39
Site-directed mutagenesis was performed for two different purposes. In one set of experiments, the mutations introduced into L39-8 and L39-9 by random mutageneis that resulted in elimination of iAb and fH binding, were replaced with amino acids of similar or different physical properties. For these analyses, pET32-Ek/LIC plasmids carrying L39-8 or the L39-9 genes were subjected to site-directed mutagenesis. To introduce the mutations, the 5′ portion of the gene was PCR amplified using a primer set that contained a reverse primer that carried the desired sequence change (mutations) near its 3′ end. The mutagenic reverse primer was designed to overlap with the 5′ end of the amplicon derived from the 3′-end gene. The forward primer that was used to amplify the 3′ end of the gene also contained the mutated overlap sequence. All amplification reactions were performed using standard PCR conditions and Pfu polymerase. The resulting amplicons were purified from an agarose gel and combined to serve as template in another PCR. The conditions for this PCR were slightly modified. Eight cycles were run in the absence of oligonucleotide primers. During these first eight cycles, the complementary sequences on the two template amplicons allow for annealing to each other. Hence, each amplicon essentially serves as a large primer. After eight cycles, the L39-16+ and L39-173− primers were added to amplify the full length mutated gene. The resulting amplicons were purified, the single-strand tails were generated by treatment with T4 DNA polymerase as directed by the vector manufacturer (Novagen), and the amplicons were annealed with the linearized pET32-LIC/Ek vector. To propagate the plasmids, the annealed products were introduced into E. coli NovaBlue DE3 cells by transformation and plated on LB plates containing 50 μg/ml ampicillin. To screen for recombinants, E. coli colonies were picked from the plate and boiled in 50 μl of H2O. The presence and size of the inserts in the recombinant plasmids were determined by PCR amplification. Protein expression was induced by the addition of IPTG (0.1 M). After cultivation for 2 h, the cells were harvested by centrifugation. The new mutant variants were then tested as described elsewhere in this report for expression of r-S-tag-OspE fusion proteins and then tested for their ability to bind fH and iAb.
Site-directed mutagenesis was also performed to alter specific residues within each of the three CC domains of L39 using either L39 wild type (wt) or L39-101 as the template. The residues that were mutated within each CC motif are indicated in Fig. 2. The impact of the mutations on CC formation was determined using the COILS program with weighting and unweighting (40). The a and d residues within the heptad repeat are particularly important in CC formation. These residues are nonpolar and properly position the charged e and g residues. The mutagenesis of the internal CC domains (CC1 and CC2) was conducted as described above using a two-step PCR-based approach and primers listed in Table I. Mutagenesis of the C-terminal CC3 domain was conducted using a one-step PCR approach with the L39-16+ forward primer and different reverse mutagenic primers. All primers were constructed with tails to allow for LIC with the pET32-Ek/LIC vector. Template for the PCR was a recombinant plasmid carrying L39 (30).
SDS-PAGE, immunoblot, and fH affinity ligand-binding immunoblot (ALBI) assays
Cell lysates were fractionated in 15% Criterion Precast Gels (Bio-Rad, Hercules, CA) by SDS-PAGE and immunoblotted onto polyvinylidene difluoride membranes as previously described (41). Expression of rOspE proteins was confirmed using S-protein-HRP conjugate (Novagen), which detects the N-terminal S-Tag fusion that is carried by all recombinant proteins used in this study. The S-protein-HRP conjugate was used at a dilution of 1/40,000. For OspE immunoblot analyses, mouse anti-OspE antiserum, generated against L39 as part of an earlier study (30), was used at a dilution of 1/1000. Goat anti-mouse IgG served as the secondary and was used at a dilution of 1/40,000. To analyze the ability of Ab elicited to OspE during infection (iAb) to bind to the different OspE mutants, infection sera generated in an earlier report (30) was used at a dilution of 1/1000. Goat anti-mouse IgG served as the secondary as above. The infection sera were generated by infecting C3h-HeJ mice with a clonal population of B. burgdorferi B31MI. Ab binding was detected by chemiluminescence using the Pierce Super Signal Substrate (Pierce). To test the ability of the OspE proteins to bind human fH, ALBI assays were performed as previously described (30). Briefly, immunoblots were incubated in blocking buffer (0.2% Tween 20, 5% nonfat dry milk in PBS) and then with purified human fH (10 ng/μl in blocking buffer; Calbiochem, San Diego, CA) for 1 h at room temperature. The blot was washed extensively with wash buffer (0.2% Tween 20 in PBS) and then incubated with either polyclonal goat anti-human fH antiserum or mAb against human fH (1/5 dilution; Serotec, Oxford, U.K.) in blocking buffer. Primary Ab binding was detected using the appropriate secondary Ab (1/40,000 in blocking buffer; Pierce) and the Pierce Super Signal Substrate.
DNA sequence analysis
To verify the sequence of all constructs, recombinant plasmids were purified from E. coli NovaBlue cells using Qiagen kits, and the inserts were sequenced (MWG Biotech, High Point, NC). The deduced amino acid sequence of each construct was determined, and the sequences were aligned using the PILEUP program with some manual refinement. Secondary structure predictions were obtained using the GOR4 secondary structure prediction method (42), and CC formation probability was assessed using the COILS program (weighted and unweighted) (43).
Internal deletions in the OspE paralog, L39, abolish fH and iAb binding
Previous studies demonstrated that the N and C termini of OspE are required for ligand binding (15, 30). The requirement for both ends of the protein for ligand binding was interpreted to indicate that protein conformation was involved. To determine whether internal domains of L39 are required for ligand binding, OspE deletions were generated using a PCR-based approach (Fig. 1, A and B). Expression of the constructs was confirmed by immunoblot analysis using the S-protein-HRP conjugate (Fig. 1,C) and anti-OspE antiserum (data not shown). To assess the ability of the deletion variants to bind fH and iAb, fH-ALBI and immunoblot assays were performed. The internal deletions resulted in the complete loss of fH and iAb binding. The loss of binding presumably results from the partial or complete deletion of an extended α helix, designated as h3 (see Fig. 3). The control construct, which carries the 4-aa insertion introduced into all deletion variants as part of the construction scheme, retained fH- and iAb-binding ability. The impact of the internal deletions on ligand binding, coupled with the requirement for the N- and C-terminal domains (30), indicates that discontinuous and or conformationally defined determinants are involved in these interactions and that disruption of different regions of the protein interferes with the formation of these important binding determinants.
Analysis of fH and iAb binding to L39 mutants generated through random mutagenesis
To identify L39 residues involved in ligand binding, random mutagenesis was performed, and the mutated amplicons were cloned and expressed as S-tag fusion proteins. E. coli clones expressing recombinant protein of the correct size were identified by immunoblotting with S-protein-HRP conjugate, and 100 of these clones were then tested for immunoreactivity with anti-OspE antiserum (data not shown) and for fH binding using the fH ALBI assay (Fig. 4 depicts representative results). The binding of iAb was assessed by immunoblotting. Approximately 5% of the recombinant proteins tested lost the ability to bind fH, whereas most retained the ability to bind iAb. The identity of the mutations responsible for eliminating fH and or iAb binding was determined through DNA sequence analysis and are discussed in detail below.
Identification of L39 mutations that abolished fH or iAb binding
To identify the molecular basis for the observed loss of fH and or iAb binding, DNA sequence analyses were performed (Fig. 3). The number of amino acid sequence changes introduced ranged from one to six (an average of 2 amino acid sequence changes per 100 residues). In mutants L39-8 and L39-9, a single amino acid change abolished fH binding. L39-8 had a G61 to R change and L39-9 had a S82 to L change. L39-9 also lost iAb binding ability, whereas L39-8 exhibited attenuated binding. These mutants are discussed in more detail below. In other mutants, loss of binding could not be mapped to a specific residue because more than one residue was mutated. Nonetheless, it can be concluded from analysis of the mutant sequences that widely separated domains of L39 are involved in ligand binding. Regarding the involvement of specific structural elements in ligand binding, analysis of the sequence of L39-88 suggests that the centrally located α helix, h3, is important. L39-88 has four sequence changes, three of which are conservative and are not predicted to impact on protein structure or properties. However, one of the four changes resulted in the replacement of A99 with proline. Proline, a helix-breaking residue, would disrupt h3, a helix that may be important in formation of the proper conformational determinants required for ligand binding. Further evidence for the importance of h3 in ligand binding was obtained in the analysis of the internal deletion variants described above. Most of h3 was deleted in the L39Δ89-102 mutant. This mutant did not exhibit fH or iAb binding.
Substitution of specific residues demonstrated to impact on fH binding
The random mutagenesis studies led to the identification of two mutants (L39-8 and L39-9) with single amino acid sequence changes (positions 61 or 82, respectively) that lost the ability to bind fH and exhibited attenuated iAb binding. To investigate the basis for the requirement of specific residues at these positions, amino acids of different physical properties were inserted by site-directed mutagenesis into the L39-8 and L39-9 mutants (not the wt). L39-8 has an R (strongly basic) at position 61 instead of the G (neutral) present in the wild type. R61 of the mutant was replaced with A (hydrophobic), T (neutral with a polar secondary hydroxyl group), or D (acidic) and the resulting recombinant proteins were tested for fH and iAb binding (Fig. 5). Replacement with an A, a relatively inert residue, completely restored fH binding. In contrast, introduction of D, an acidic and polar residue, allowed for only partial binding. No fH binding was detected when T was introduced into position 81. Hence, the introduction of charged and polar residues with extended side chains at position 61 appears to sterically hinder binding, disrupt a potential binding pocket, or perturb long range intramolecular interactions that are required for fH binding. In contrast to that observed for the OspE-fH interaction, the introduction of A, D, and T into position 61 restored iAb binding (data not shown), providing further indication that the determinants involved in fH and iAb binding differ. Site-directed mutagenesis was also performed for position 82 of L39-9. L82 of the mutant was replaced with K (positively charged polar residue) or A (hydrophobic and nonpolar). The A restored fH binding to wt (S82) level, whereas the K did not. These analyses demonstrate that fH binding is not strictly dependent on the identity of the residue at each of these sites but rather on the properties of the residue.
Mutation analysis of putative CC motifs in OspE
Previous studies from our laboratory and data contained within this report suggest that higher order structural elements are involved in ligand binding. Computer-based structural analysis of OspE using the COILS program identified three α helices (h3, h4, and h5) with the potential to form CC motifs (Fig. 6). CCs have been demonstrated to mediate intra- and intermolecular protein interactions and have been identified in proteins involved in transcriptional regulation, cell to cell interactions, and viral-cell fusion events (44). CCs consist of two or more right handed α helices that wrap around each other with a left handed superhelical twist. CCs have a heptad repeat (abcdefg)n that is spread along two turns of the helix. Residues a and d are hydrophobic, and residues e and g are generally charged. To assess the potential involvement of these putative CC motifs in ligand binding, mutations were introduced that either decreased the predicted probability of CC formation or had no effect (Fig. 6). The specific mutations that were introduced are shown in Fig. 2,A. After confirming expression of the OspE mutants using S-protein-HRP conjugate, each was tested for fH-binding ability using the fH ALBI assay (Fig. 2 B). In L39cc1m1 and L39cc1m2, CC1 residues at the a and d positions were replaced with residues with polar side chains. This resulted in complete loss or attenuation of fH binding. In contrast, replacement of the b and e polar side chain residues with amino acids of similar properties had no impact on binding. The strict requirement for the amino acids of specific physical properties within CC1 is consistent with the observed genetic conservation of this domain among OspE sequences (32). Mutation of residues within CC2 attenuated fH binding but did not eliminate it, indicating that this domain is involved but is less important than CC1. Consistent with this, the sequence of CC2 is less conserved than the other CC domains of OspE. The most exhaustive mutational analyses were performed on CC3 which is located at the C terminus of OspE. As with the CC1 mutants, replacement of the residues at the a and d positions with residues with polar side chains completely abolished fH binding. In contrast, when the a and d residues were replaced with amino acids of like properties, no impact on fH binding was observed. Similarly, nonconservative substitution of positions b, c, e, or g also did not impact on fH binding. In summary, all mutations that were predicted to disrupt CC structure impacted on fH binding. This striking correlation establishes the importance of widely separated domains of OspE in fH binding and demonstrates that fH binding requires higher order structural motifs, specifically CCs.
The ability of the CC mutants to bind iAb was also tested. Although some of the mutants exhibited slightly attenuated iAb binding, none completely lost the ability to bind iAb (Fig. 2 B). The different impact that mutations had on fH as compared with iAb binding provides additional evidence that each of these ligands interacts with different domains or structural elements of L39. This conclusion is supported by earlier studies from our laboratory that demonstrated that fH and iAb do not compete for binding (30).
In light of the widespread use of fH binding as a virulence mechanism by human pathogens, determination of the molecular basis of the interaction between fH and fHBP is an important undertaking. Studies along these lines will increase our understanding of this important virulence mechanism and facilitate efforts to use fHBPs in vaccine development. fHBP-based vaccines may provide two distinctly different but synergistic correlates of protection. In addition to the Ab-mediated killing induced by vaccination, Ab to the fHBPs may block fH binding by pathogens and render them more susceptible to opsonophagocytosis. Earlier studies from our laboratory suggested that the binding of both iAb and fH to OspE involves conformational or structurally defined determinants (30). Metts et al. (30) demonstrated that truncation of either the N or C terminus abolished the binding of ligand to the OspE paralogs, L39 and N38. Kraiczy et al. (15) also demonstrated that the C terminus of OspE is required for fH binding but concluded that interaction was mediated by a specific C-terminal primary amino acid sequence. Most recently, Alitalo et al. (37) demonstrated through alanine substitution of OspE peptides that multiple K residues located primarily in the C terminus of OspE are involved in the OspE-fH interaction. However, Alitalo et al. did not investigate the effects of alanine substitution on the ability of full length forms of OspE to bind fH. As a result, the impact that the substitutions could have on structural elements that might be involved in forming the fH binding site was not assessed. The goals of this study were to further define the molecular basis of the interaction between fH and iAb with OspE using multiple mutation strategies.
To identify the OspE determinants involved in ligand binding, deletion, random, and site-directed mutagenesis approaches were applied. Focusing first on the random mutational analyses, this unbiased screening approach identified several residues distributed throughout the protein that are required for ligand binding. Approximately 5% of the clones obtained by this procedure lost the ability to bind fH and or iAb. Two of these (L39-8 and L39-9) lost fH binding ability as a result of single amino acid sequence changes that occurred at position 61 or 82. Analysis of OspE sequences revealed that positions 61 and 82 are highly conserved and that when sequence differences do occur at these positions, they are conservative (32). In fact, most of the mutations associated with the loss of fH binding mapped to conserved residues (11, 32). This serves as an indication of the potential importance of these residues in OspE structure and or function. Positions 61 and 82 reside within predicted turns in the protein and immediately precede α helices, h2 and h3. Hence, mutations at these positions could affect protein folding and conformation. Site-directed mutagenesis of these positions revealed that there is not a strict requirement for the wt sequence and that replacement with a variety of residues still allows for ligand binding. It is interesting that the introduction of a K into position 82 did not restore ligand binding. Alitalo et al. recently demonstrated using peptides that K residues of the OspE ortholog, P21, may mediate a charge-based interaction between fH and OspE. Although charge is clearly important in fH binding, the data presented here provide a further indication that fH binding is not simply a nonspecific charge interaction and that the specific K residues involved must be presented in the proper context to facilitate binding.
In addition to the mutants discussed above, several additional mutants were identified in the random mutagenesis and deletion analyses that lost fH and or iAb binding. Several of the randomly generated mutants had more than one amino acid change, and as a result the specific residues that are required for ligand binding could not be unambiguously identified. Nonetheless, the observation that these sequence changes occurred in regions distal to positions 61 and 82 indicates that the OspE-fH and OspE-iAb interaction involves residues located in different domains of OspE. The sensitivity of ligand binding to perturbation of OspE structure was also evident in the deletion analyses. Deletion, in whole or in part, of h2 resulted in the complete elimination of fH and iAb binding. Most of the mutations and deletions that disrupted fH binding occurred outside of the 5 putative fH binding domains identified by Alitalo et al. (37) in the OspE ortholog, P21. However, this does not indicate a disagreement in results, nor does it suggest that the 5 putative fH binding domains are not involved in the fH interaction. In fact, the data presented here and in earlier studies strongly support the hypothesis that fH binding does not occur through a single contiguous linear sequence determinant but instead involves multiple contact points that require structural or conformational determinants for proper presentation.
Several studies have investigated the contribution of the C-terminal domain of OspE in ligand binding, and although differing interpretations of the data have been offered, all are in agreement that there is an absolute requirement for an intact C terminus (15, 30). We previously demonstrated that the N terminus of OspE is also required (30), and the deletion analyses presented above indicate a requirement for internal domains of the protein as well. These analyses strongly support the hypothesis that higher order structural elements are required. Computer-assisted structural analyses of OspE revealed that the protein has three potential CC-forming domains, one of which occurs within the C terminus of OspE. It is well established that CC domains are important in inter- and intramolecular protein interactions (43, 44). To determine whether the CC domains of OspE are required for ligand binding, site-directed mutagenesis of 2 or more amino acids at different positions in each CC domain was performed. The selection of the amino acids used in the replacement, and their positioning was determined by computer modeling. Mutants were then constructed that were predicted to either lose or retain CC formation ability. Mutants in which the nonpolar residues at the a and d positions of the heptad repeats were replaced with polar residues were predicted to lose CC formation ability; consistent with this, these mutants lost the ability to bind fH but retained the ability to bind iAb. Theses analyses revealed that CC1 and CC3 are absolutely required for fH ligand binding. In contrast, CC2 was needed for maximal binding but was not required. To our knowledge, this important observation is the first demonstration of the direct involvement and requirement for CC formation in ligand binding in any spirochetal protein. A possible model for the OspE-fH interaction is that a hydrophobic intramolecular interaction between the CCs of OspE results in the formation of a charged pocket that presents the required residues for fH binding in the correct context. The role of OspE positively charged residues in fH binding is strongly supported by the observation that heparin inhibits the interaction (10, 15).
Analyses of fHBPs of other organisms suggests that the fH binding determinants of these proteins are also conformationally defined (21, 25). Although there is little or no homology among the known fHBPs, there is apparent structural conservation. As a case in point, although the fHBPs OspE and BBA68 of B. burgdorferi (14) and FhbA of B. hermsii (13) exhibit no discernable homology, computer-based structural analysis of the published sequences indicates that they have a high probability of forming CCs. Hence, fH binding may not be dependent on a specific primary sequence but rather on conserved structural motifs and the proper presentation of specific K residues. A similar observation has been noted for viral proteins involved in membrane fusion (44). Despite little or no sequence conservation, all members of this diverse group of proteins form homotrimeric CCs that are required for membrane fusion. The fact that wt OspE retains ligand-binding ability after fractionation by SDS-PAGE and immunoblotting might argue against the involvement of conformational or structural binding elements. However, there is precedent for the retention of structural elements in several proteins even after being subjected to denaturing conditions. Structural elements of the paramyxovirus fusion protein have been demonstrated to be resistant to 2% SDS and 40°C (45) and the fungal hydrophobin proteins retain key structural determinants after heating for 10 min at 100°C in 2% SDS (46, 47). Hence, the apparent conformational nature of the OspE fH-binding site is not surprising and is not without precedent.
Earlier analyses investigated the use of OspE in Lyme disease vaccine development (48, 49). Although rOspE was immunogenic, it did not elicit a protective response, a finding confirmed in our laboratory (unpublished data). Due to the high concentration of fH in serum (∼500 μg/ml) and other body fluids, upon vaccination, OspE would most likely become saturated with fH. As a result, the anti-OspE Ab response would be relatively restricted. This may explain the nonprotective nature of the Ab response. The binding of fH would also block the development of Ab to the surface-exposed, fH-binding site. In support of this, we previously demonstrated that iAb and fH do not compete with each other for binding to OspE (30). The absence of an Ab response to the fH-binding site suggests that lipidated OspE is a B cell mitogen that most likely elicits Ab production in a processing independent, T cell-independent manner. It has been reported that other lipoproteins of the Lyme disease spirochetes also elicit T cell-independent Ab production (50). As we have demonstrated here, minor modification of the OspE protein can result in the complete elimination of fH binding. Because these proteins do not bind fH, it is possible that they may now be able to elicit a less restricted Ab response that would include Ab targeting the fH-binding pocket. Abs that bind to or block this pocket may be able to block fH binding by B. burgdorferi and render the bacteria more susceptible to complement. This, coupled with the possible development of bactericidal Abs, would allow for multiple correlates of protection. Because OspE is also expressed in the tick (48), vaccination may also block or interfere with transmission of the spirochetes from the tick to the mammal.
In summary, we have demonstrated that fH and iAb binding to OspE is highly sensitive to perturbation of protein structure, that it involves multiple interaction sites, and that the proper presentation of the fH-binding determinants requires higher order structural elements including a series of CC domains that may mediate critical intramolecular interactions. Studies are now under way to assess the efficacy of mutated OspE proteins to protect against infection with the Lyme disease spirochetes.
We thank members of the Marconi laboratory for critical review of the manuscript.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported in part by Grants R21-AI059257 and RO1-AI037787 from the National Institutes of Health.
Abbreviations used in this paper: fH, factor H; fHL-1, fH-like protein 1; fHBPs, fH-binding protein; iAb, infection-induced Ab; CC, coiled coil; OspE, outer surface protein E; LB, Luria-Bertani; IPTG, isopropylthiogalactoside; wt, wild type; ALBI, fH affinity ligand-binding immunoblot; Ek/LIC, enterokinase/ligase-independent cloning.