Molecular Analyses of the Interaction of Borrelia hermsii FhbA with the Complement Regulatory Proteins Factor H and Factor H-Like Protein 1
http://www.100md.com
感染与免疫杂志 2006年第4期
Department of Microbiology and Immunology
Center for the Study of Biological Complexity, Medical College of Virginia at Virginia Commonwealth University, Richmond, Virginia 23298-0678
Department of Microbiology and Infectious Diseases, Flinders Medical Centre, Bedford Park, South Australia, Australia
ABSTRACT
Borrelia hermsii, the primary etiological agent of tick-borne relapsing fever in North America, binds the complement regulatory protein factor H (FH) as a means of evading opsonophagocytosis and the alternative complement pathway. The ability of FH-binding protein A (FhbA) to bind FH-like protein 1 (FHL-1) has not been assessed previously. In this study, using a whole-cell absorption assay, we demonstrated that B. hermsii absorbs both FH and FHL-1 from human serum. Consistent with this, affinity ligand binding immunoblot analyses revealed that FH constructs spanning short consensus repeats 1 to 7 and 16 to 20 bind to FhbA. To investigate the molecular basis of the interaction of FhbA with FH/FHL-1, recombinant FhbA truncated proteins were generated and tested for FH/FHL-1 binding. Binding required determinants located in both the N- and C-terminal domains of FhbA, suggesting that long-range intramolecular interactions are involved in the formation and presentation of the FH/FHL-1-binding pocket. To identify specific FhbA residues involved in binding, random mutagenesis was performed. These analyses identified a loop region of FhbA that may serve as a contact point for FH/FHL-1. The data presented here expand our understanding of the pathogenic mechanisms of the relapsing fever spirochetes and of the molecular nature of the interaction between FH/FHL-1 and FhbA.
INTRODUCTION
In North America tick-borne relapsing fever is caused by Borrelia hermsii, Borrelia parkeri, and Borrelia turicatae (5). These pathogens are transmitted to humans through the bites of infected Ornithodoros ticks (5). The hallmark features of tick-borne relapsing fever are cyclic, high-level spirochetemias (107 spirochetes ml blood–1) and a relapsing fever (14). In the United States, tick-borne relapsing fever occurs primarily in the West, and in this region several recent outbreaks have been reported (38). In other parts of the world the impact of tick-borne relapsing fever is staggering. In some regions of Tanzania the incidence in children who are less than 1 year old is 40% (10). In regions where it is endemic, tick-borne relapsing fever is an important human health concern.
B. hermsii employs several mechanisms to persist in the blood and evade the immune response. Antigenic variation mediated by the Vmp proteins is clearly important and has been studied in detail (2-4, 6, 39). Interestingly, Connolly and colleagues have demonstrated that antibody-mediated killing of the relapsing fever spirochetes occurs primarily through a complement-independent mechanism (8, 9). Complement may be more important in opsonization and phagocytosis. We recently demonstrated that B. hermsii and B. parkeri bind the complement regulatory protein factor H (FH) (22, 27). FH is a very abundant serum protein that contributes to regulation of the alternative complement pathway by serving as a cofactor for factor I-mediated cleavage of C3b (36, 37). It also inhibits the initial formation and accelerates the dissociation of the alternative pathway C3 convertase by competing with factor B. FH-like protein 1 (FHL-1), which is derived from FH mRNA by alternative splicing, exhibits similar regulatory activity (16). Binding of FH and FHL-1 to several important human pathogens has been demonstrated (11, 12, 15, 17, 19, 21, 29-35). Surface-bound FH/FHL-1 is thought to locally increase the efficiency of C3b cleavage and thereby inhibit C3b opsonization and subsequent phagocytosis. FH/FHL-1 binding may also facilitate adherence and intracellular localization of some pathogens (33). Treponema denticola, an oral spirochete associated with periodontal disease, binds FHL-1 through a 14-kDa protein designated FHL-1-binding protein B (FhbB) (26). FhbB is unique among bacterial complement regulatory binding proteins in that it binds FHL-1 exclusively and not FH. Interestingly, FHL-1 bound to the surface of T. denticola does not promote or enhance C3b cleavage (26). The ability to bind FHL-1 might be more important in facilitating the interaction of T. denticola with FHL-1 present on the surface of anchorage-dependent cell types. We recently identified an FH-binding protein of B. hermsii and designated it FH-binding protein A (FhbA) (22). FhbA is a 20.5-kDa lipoprotein that is surface exposed and antigenic during infection. It is encoded by a gene located on a linear plasmid estimated to be 200 kb long. FhbA does not exhibit significant sequence identity or similarity with any known FH/FHL-1-binding proteins.
The goals of this study were to determine if FhbA can bind to FHL-1 and to identify the molecular determinants of FhbA and FH/FHL-1 that are involved in the FhbA-FH/FHL-1 interaction. Using FH fragments spanning different short consensus repeats (SCR), coupled with whole-cell absorption assays, we found that FhbA binds both FH and FHL-1. To investigate the molecular nature of the interaction of FH/FHL-1 with FhbA, a series of N- and C-terminal truncations of FhbA were generated and tested for the ability to bind ligand. These analyses demonstrated that determinants located in widely separated domains of FhbA are required for FH/FHL-1 binding. Random mutagenesis was also performed, and the results demonstrated that a loop domain of FhbA may serve as a key contact point for FH/FHL-1. Our hypothesis is that the binding of complement regulatory proteins by the B. hermsii FhbA protein may influence the dissemination characteristics of the tick-borne relapsing fever spirochetes.
MATERIALS AND METHODS
Bacterial isolates and cultivation. The B. hermsii YOR and MAN isolates (provided by Tom Schwan, Rocky Mountain Laboratories) were cultivated in BSK-H complete medium (Sigma) supplemented with 12% rabbit serum (Sigma) at 33°C. The B. hermsii isolates were originally recovered from human relapsing fever patients in the United States (23).
FH/FHL-1 whole-cell absorption assay. To determine if B. hermsii can bind both FH and FHL-1, a whole-cell FH/FHL-1 absorption assay was performed as previously described (26), with some modifications. In brief, B. hermsii YOR, Escherichia coli NovaBlue (negative control), and Borrelia burgdorferi B31MI (positive control) (109 cells) were recovered by centrifugation, washed, and resuspended either in phosphate-buffered saline (PBS) containing purified FH/FHL-1 (0.525 mg ml–1) or in human serum (Valley Biomedical) diluted 1:1 with PBS. The samples were incubated for 1 h at room temperature, recovered by centrifugation, gently washed twice with cold PBS, solubilized in sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) sample buffer, electrophoresed, and immunoblotted. The membranes were then screened for FH or FHL-1 that bound to the surface of the bacteria by using methods described below.
FH/FHL-1 ALBI assays. Recombinant proteins or cell lysates were separated by SDS-PAGE using 12% acrylamide gels (Bio-Rad) and were transferred onto membranes for immunoblot and binding analyses. FH/FHL-1 affinity ligand binding immunoblot (ALBI) assays were conducted as previously described (32), using purified human FH/FHL-1 (Calbiochem). Bound FH/FHL-1 was detected using goat anti-human FH/FHL-1 antiserum (1:800; Calbiochem). Short consensus repeat constructs were prepared as previously described (12). Briefly, the SCR constructs were generated by PCR and cloned into the yeast expression vector pPICZ (Invitrogen). The proteins were expressed in Pichia pastoris and purified by immunoaffinity chromatography. The concentrations of proteins were determined by using an FH standard curve generated by an enzyme-linked immunosorbent assay (ELISA).
Construction of truncated FhbA proteins and other recombinant proteins. To generate FhbA proteins with progressive N- and C-terminal truncations and control proteins, a PCR-based strategy was used. The PCR was performed with Expand High Fidelity polymerase (Roche) and standard cycling conditions. The template for PCR was a previously described plasmid that carries the fhbA gene from the B. hermsii YOR isolate or isolated DNA (22). Primers (Table 1) designed to amplify fragments of fhbA were constructed with tail sequences that allowed annealing into the pET-32 Ek/LIC vector, a procedure referred to as ligase-independent cloning (Invitrogen). The pET-32 EK/LIC vector allowed subsequent expression of the inserts as N-terminal poly(His)- and S-tagged fusion proteins. The fusion added 17 kDa to the mass of each recombinant protein. All cloning methodologies and procedures related to the generation of recombinant proteins have been described previously (13, 42). Expression of the recombinant proteins was confirmed by SDS-PAGE and immunoblotting using horseradish peroxidase (HRP)-conjugated S-protein (1:40,000 dilution; Novagen) and mouse anti-FhbA antiserum (1:1,000 dilution), as previously described (22).
Random mutation of fhbA. fhbA was randomly mutated by PCR by using a GeneMorph II random mutagenesis kit (Stratagene) with primers designed to amplify the entire coding region of fhbA (minus the signal peptide), which spans residues 20 to 192. The primers had tail sequences that allowed annealing of the amplicons into the pET-32 Ek/LIC vector (Novagen). The resulting recombinant plasmid was used to transform E. coli NovaBlue(DE3) cells (Novagen). All methods associated with random mutagenesis of fhbA and generation of the resulting recombinant proteins have been described previously (28). Expression of a recombinant protein was confirmed by screening immunoblots with HRP-conjugated S-protein (1:40,000 dilution; Novagen) and with mouse anti-FhbA antiserum (1:1,000 dilution) (22). The sequences of mutated amplicons were determined by DNA sequencing (MWG Biotech). The possible effects of amino acid substitutions on protein structure were assessed using programs available on the Protein Sequence Analysis (PSA) server (40, 41).
Sequence determination and analysis of the fhbA gene of the B. hermsii MAN isolate. The fhbA gene of the B. hermsii MAN isolate was amplified by PCR using primers listed in Table 1 (22). The amplicon was purified by gel extraction, cloned into the pET-32 Ek/LIC vector, and propagated in E. coli NovaBlue(DE3) cells. The recombinant plasmid was purified and isolated, and the insert was sequenced (MWG Biotech). Sequence analyses were conducted using several different programs, as indicated below.
RESULTS
Demonstration that FhbA binds both FH and FHL-1 using whole-cell absorption assays. A whole-cell FH/FHL-1 absorption assay was performed to determine if FhbA can bind both FH and FHL-1. Human serum served as the source of the FH/FHL-1. The cells were incubated with human serum diluted 1:1 with PBS, washed to remove unbound FH/FHL-1, fractionated by SDS-PAGE, immunoblotted, and screened with anti-FH/FHL-1 antiserum. The antiserum detected 150-kDa and 49-kDa proteins that were absorbed by B. hermsii (Fig. 1). Based on the size and immunoreactivity of these proteins with the anti-human FH/FHL-1 antiserum, we concluded that the proteins were FH and FHL-1, respectively. While the amount of FHL-1 bound by the cells was significantly smaller than the amount of FH bound, this was expected in light of the greater abundance of FH than of FHL-1 in serum (500 μg ml–1 and 50 μg ml–1, respectively). When purified FH/FHL-1 was used in the absorption assays, an additional unidentified 75-kDa protein was also absorbed at a low level by both B. hermsii and B. burgdorferi (positive control) and was detected by the anti-FH/FHL-1 antiserum (data not shown). This protein was presumably an FH breakdown product. The negative control for the absorption analyses, E. coli NovaBlue, did not absorb FH or FHL-1, indicating that the absorption of FH and FHL-1 by B. hermsii is specific. These analyses demonstrated that B. hermsii binds both FH and FHL-1.
Identification of the short consensus repeats of FH/FHL-1 that bind to FhbA. FH and FHL-1 are comprised of short consensus repeats (20 and 7 SCRs, respectively) that are 50 to 60 residues long. To determine which SCRs bind to FhbA, recombinant FhbA (rFhbA) was immunoblotted and tested for binding to FH fragments spanning SCRs 1 to 7, 8 to 15, and 16 to 20. First, the integrity of the constructs and their immunoreactivity with the anti-human FH/FHL-1 antiserum were demonstrated by immunoblot analyses (Fig. 2A). Screening with the SCR constructs revealed that FhbA bound SCRs 1 to 7 and 16 to 20 but not SCRs 8 to 15 (Fig. 2B). A truncated form of recombinant BBN39 (encoded by a member of paralogous gene family 163 of B. burgdorferi B31MI) served as a negative control in these analyses and, as expected, was not bound by any SCR construct. Since FHL-1 is a truncated form of FH generated by alternative splicing and contains only SCRs 1 to 7 (plus a unique 4-amino-acid C-terminal tail) (16), the binding of FhbA to SCRs 16 to 20 specifically indicated that binding to FH occurred. The binding of SCRs 1 to 7 to FhbA could indicate binding to FH and/or FHL-1 since both complement regulatory proteins contain this domain. Collectively, the SCR binding analyses and the absorption assays described above indicated that FhbA has the potential to bind both FH and FHL-1.
Localization of the FH/FHL-1-binding domains of FhbA. To localize the binding sites for FH/FHL-1 on B. hermsii FhbA, a series of FhbA proteins with N- and C-terminal truncations were constructed (Fig. 3A). The expression of each protein was confirmed by immunoblot analysis using HRP-conjugated S-protein and anti-FhbA antiserum (Fig. 3B). All of the recombinant proteins were expressed at similar levels, were the predicted size, and were immunoreactive with the anti-FhbA antiserum. The abilities of the truncated constructs to bind FH/FHL-1 were assessed using the ALBI assay. The full-length recombinant protein (residues 20 to 192) and the fragment spanning residues 48 to 192 exhibited similar levels of FH/FHL-1 binding (Fig. 3B). The fragments spanning residues 79 to 192 and 106 to 192 bound FH/FHL-1, but at a significantly attenuated level. No FH/FHL-1 binding to fragments spanning residues 20 to 165 and 20 to 134 was observed. As an additional control for these analyses, a blot was screened with anti-FH/FHL-1 antiserum with no exogenous FH/FHL-1 added (data not shown). No binding of antibody alone to the recombinant proteins was observed, indicating that the methods employed for detecting FH/FHL-1 binding were specific. The data indicate that widely separated domains of FhbA are required for maximal binding to FH/FHL-1. This observation suggests either that FH/FHL-1 has multiple, widely distributed contact points on FhbA or that long-range intramolecular interactions are required to form the FH/FHL-1-binding pocket.
Generation of random FhbA mutants and analysis of their abilities to bind FH/FHL-1. To identify specific domains of FhbA involved in FH/FHL-1 binding, random mutagenesis was performed. Random mutations were introduced using a PCR-based approach with a low-fidelity polymerase. The primers employed amplified the region of fhbA that encodes the mature protein (residues 20 to 192) and were constructed with tails to allow annealing of the mutated amplicons into the pET-32 Ek/LIC vector. After transformation of the recombinant plasmids into E. coli NovaBlue(DE3) cells, expression of the recombinant proteins was induced by overnight cultivation, and the E. coli cell lysates were fractionated by SDS-PAGE. Note that expression did not require isopropyl--D-thiogalactopyranoside (IPTG) induction. The proteins were transferred onto membranes and screened using HRP-conjugated S-protein, which detected the S tag of the fusion proteins (Fig. 4). Recombinant proteins that were the appropriate size were then tested for immunoreactivity with anti-FhbA antiserum and for FH/FHL-1 binding. Three of the E. coli clones, designated clones 63, 85, and 90, produced recombinant proteins that did not bind FH/FHL-1. Other recombinant proteins produced by clones 194, 196, and (most notably) 198 exhibited significantly attenuated binding. Several other clones produced FhbA proteins that exhibited full FH/FHL-1 binding ability. The mutated rFhbA proteins were designated by using the number assigned to the E. coli clone. For example, FhbA from clone 90 was designated FhbA90.
DNA sequence analysis of the random mutants of FhbA and computer-based structural analyses. To determine the molecular basis for the loss of FH/FHL-1 binding observed with some of the mutated rFhbA proteins, the corresponding plasmid inserts were sequenced. For comparative purposes, the inserts of plasmids from several E. coli clones that produced recombinant proteins that exhibited FH/FHL-1 binding were also sequenced (Fig. 5). FhbA63, FhbA85, and FhbA90 were not able to bind FH/FHL-1, while FhbA194, FhbA196, and FhbA198 exhibited significantly attenuated binding. To identify the mutations that affected FH/FHL-1 binding, we first sought to exclude noncontributory mutations. For example, mutations in the N-terminal domain of FhbA (up to residue 47), as well as mutations in the last 10 residues of the C terminus (as observed with FhbA123), were excluded from consideration since the domains harboring the residues could be deleted with no effect on FH/FHL-1 binding. Analysis of the remaining mutations revealed that FH/FHL-1 binding was particularly sensitive to substitutions introduced in a region of FhbA spanning residues 135 to 162. FhbA198 had a single amino acid change (Phe139 to Val) that resulted in significantly attenuated FH/FHL-1 binding. The FhbA90 mutant, which completely lacked the ability to bind FH/FHL-1, had two amino acid substitutions, Asn143 to Asp and Asp146 to Asn. The first of these residues is predicted to be in a loop domain, and the second is predicted to be in a flanking coiled-coil element (Fig. 5 and 6). FhbA85 also had two substitutions in the loop domain (Thr135 to Met and Asn143 to Lys), and it also had a third substitution (Phe182 to Cys) that mapped outside the loop region. Mutant FhbA63, which had two amino acid substitutions (Pro86 to Leu and Asn162 to Thr) also lacked the ability to bind FH/FHL-1. Computer-based structural analyses predicted that the substitutions in this mutant which involved replacement of a Pro residue significantly altered the FhbA structure. Hence, structural alteration was most likely the basis for the loss of FH/FHL-1 binding to FhbA63. Collectively, the data suggest that the loop region may be a key contact point for FH/FHL-1 and that the presentation of the binding site is dependent on specific structural determinants of the protein.
Determination of the fhbA sequence of the B. hermsii MAN isolate and computer-assisted structural analysis of FhbA. The analyses described above suggested that a defined loop region of FhbA serves as a contact point for FH/FHL-1. If this domain is important in FH/FHL-1 binding, then it would likely be conserved in other strains. As a first step toward assessing this possibility, we verified the production of a FH/FHL-1-binding protein in the B. hermsii MAN and YOR isolates using the ALBI assay (Fig. 6A). The MAN isolate produced an FH/FHL-1-binding protein with a molecular mass slightly less than the molecular mass of the protein in the YOR isolate. The antigenic relatedness of FhbAMAN to FhbAYOR was demonstrated by screening an immunoblot of cell lysates with anti-FhbA antiserum. This antiserum was generated using recombinant FhbA derived from the YOR isolate. The fhbAMAN gene was then amplified and annealed into the pET-32 EK/LIC vector, and a recombinant protein was generated. Expression of the recombinant protein was confirmed by screening immunoblots with HRP-conjugated S-protein and with anti-FhbA antiserum. Finally, the ability of the recombinant protein to bind FH/FHL-1 was demonstrated using the ALBI assay as described above. The fhbAMAN gene sequence was then determined by sequence analysis of the insert carried by the corresponding recombinant plasmid. The FhbAYOR and FhbAMAN sequences are aligned and are shown in Fig. 6B. The amino acid sequences exhibited 81.5% identity and 89.6% similarity, and most of the sequence differences, including a 7-amino-acid insertion in FhbAYOR, were located in the N-terminal half of FhbA.
In other Borrelia FH/FHL-1-binding proteins, the FH/FHL-1-binding pocket is thought to be formed by long-range intramolecular interactions that present a discontinuous binding pocket. Coiled-coil elements were demonstrated to be involved in presentation of the FH/FHL-1-binding sites of BBA68 and OspE (25, 28). To determine if there are structural features that are present in both FhbAMAN and FhbAYOR that might contribute to FH/FHL-1 binding, computer-assisted structural analyses were performed. The predicted probability of coiled-coil formation in each protein was assessed using the COILS program (24). These analyses indicated that there was a high probability of coiled-coil formation for domains in both the FhbAYOR and FhbAMAN proteins (Fig. 6C). However, only the two most C-terminal coiled coils (cc1 and cc2) were conserved. The N-terminal domain coiled-coil elements that were predicted with high probability in the FhbAYOR sequence were predicted with only low probability in FhbAMAN. This lack of conservation of the N-terminal coiled coils, coupled with the truncation analyses that demonstrated that the N terminus of the protein could be deleted with only minimal impact on FH/FHL-1 binding, indicated that the N-terminal coiled coils are not required for ligand binding. The intervening sequence between cc1 and cc2 that forms the putative FH/FHL-1-binding loop of FhbAYOR is conserved at both the sequence and structure levels in FhbAMAN. The random mutagenesis analyses coupled with the conservation of the loop strongly suggested that this specific domain is a critical interaction site for FH/FHL-1.
DISCUSSION
A unique aspect of the relapsing fever spirochetes is their ability to achieve remarkably high cell densities during infection in the blood of mammals (5). It is well documented that these bacteria, particularly B. hermsii, employ an elaborate system of antigenic variation to evade the host antibody response (2). Efficient persistence in the blood is also likely to depend on the ability to evade complement-mediated killing. We recently demonstrated that some species of relapsing fever spirochetes bind the complement regulatory protein FH and possibly FHL-1, which participates in factor I-mediated cleavage of C3b (22, 27). This suggests that FH/FHL-1 binding is biologically relevant, contributes to evasion of the alternative complement pathway, and may decrease the efficiency of opsonization by C3b and subsequent phagocytosis. The goals of this study were to determine if B. hermsii can bind both FH and FHL-1 and to determine the molecular determinants of both FH/FHL-1 and FhbA that are involved in the interaction.
While it has been demonstrated previously that B. hermsii binds FH through its FhbA protein, the ability of B. hermsii to bind FHL-1 was not specifically addressed previously (22). To assess this, a whole-cell-human serum absorption assay was employed. B. hermsii YOR cells incubated with human serum were found to specifically absorb both FH and FHL-1. To further assess the interaction of FhbA with FH/FHL-1, ALBI assays were performed, in which rFhbA was tested to determine its ability to bind recombinant proteins consisting of different FH/FHL-1 short consensus repeats. Three SCR constructs spanning SCRs 1 to 7, 8 to 15, and 16 to 20 were tested. SCRs 1 to 7 and 16 to 20, but not SCRs 8 to 15, bound FhbA. Binding to the SCR 16 to 20 fragment is indicative of binding to FH since these SCRs are not present in FHL-1. SCRs 19 and 20 have been demonstrated to be the interaction point for FH-binding proteins produced by Onchocerca, B. burgdorferi, Neisseria gonorrhoeae, Candida albicans, and other pathogens (reviewed in references 31, 35, and 43). The binding of SCRs 1 to 7 to FhbA suggests that in addition to FH, FhbA can also bind FHL-1. In most cases, both FH and FHL-1 retain their cofactor activities when they are bound to the surfaces of pathogens (43). These proteins also have the potential to function in adherence since they bind to surface glycosaminoglycans and host cell receptors. FH or FHL-1 bound to glycosaminoglycans or receptors may still be able to bind to microbial cell surfaces and thereby serve as a bridge molecule that facilitates the interaction of microbes with host cells. An RGD motif in SCR 4 appears to be particularly important in binding to host cell receptors (20). As described above, T. denticola is unique among the complement regulatory binding pathogens in that it specifically binds FHL-1 (and not FH), and the bound FHL-1 does not exhibit detectable cofactor activity. This suggests that the biological function of FHL-1 binding by this spirochete may involve adherence. It is tempting to speculate that B. hermsii might also utilize host cell-bound FH or FHL-1 binding to allow adherence to the endothelium or other cells and tissues and thereby facilitate dissemination and invasiveness. It has recently been demonstrated that at least one species of relapsing fever spirochete, Borrelia crocidurae, can interact with red blood cells (7). The molecular basis of this interaction has not been defined yet.
To localize the binding determinants of FhbA required for FH/FHL-1 binding, FhbAYOR proteins with N- and C-terminal truncations were constructed. Proteins with truncations from the N terminus up to residue 48 (r48-192) exhibited full FH/FHL-1 binding, while the r79-192 and r106-192 truncated proteins exhibited significantly attenuated binding. Hence, we concluded that the first 47 residues of FhbA are not required for binding and that determinants located between positions 48 and 105 are required for maximal binding activity. Determinants in the C-terminal domain were also found to be required for FH/FHL-1 binding. A C-terminal 27-residue truncation (r20-165) led to complete loss of FH/FHL-1 binding, while a protein with a C-terminal 7-residue truncation generated by introduction of a premature stop codon by random mutagenesis (FhbA123) exhibited FH/FHL-1-binding ability. The three most C-terminal residues of this construct were also altered by disruption of the reading frame. Hence, while determinants in the C-terminal domain of FhbA are important in FH/FHL-1 binding, the last 10 residues are dispensable. We concluded from these analyses that the FH/FHL-1-binding site is not a simple contiguous linear element since independent truncations in different domains eliminated FH/FHL-1 binding. The apparent discontinuous nature of the FH-binding site is consistent with that reported for the OspE and BBA68 FH-binding proteins of B. burgdorferi (1, 25, 28).
The random mutagenesis analyses led to identification of a loop domain in FhbA that appears to be one of the key binding elements involved in the FH/FHL-1 interaction. After noncontributory mutations were excluded (using criteria described above), the remaining mutations that led to decreased binding clustered within or immediately next to a loop that spans residues 130 through 143. The importance of this loop is highlighted by the fact that FhbA198 and FhbA90, which exhibited reduced or no FH/FHL-1 binding, had only one and two amino acid changes, respectively, that mapped in the loop or the immediately flanking C-terminal coiled coil. With the exception of one residue, the sequence of the loop domain was found to be conserved in a second B. hermsii isolate, B. hermsii MAN. The MAN and YOR isolates belong to two distinct genomic groups (18), and hence the conservation of the loop in the different genomic groups suggests that this domain is functionally significant. Ram and colleagues also demonstrated the importance of a loop domain in the porin protein PorA1 of N. gonorrhoeae in FH binding (34). The loop domain of FhbA is flanked by the alpha-helical domains that carry the coiled-coil heptad repeat sequence. This heptad repeat (a-gn) harbors hydrophobic nonpolar residues at the a and d positions. The importance of coiled coils in the presentation of the FH/FHL-1-binding site in other FH/FHL-1-binding proteins has been firmly established (15, 25, 28). We previously demonstrated that replacement of nonpolar, hydrophobic a and d position residues in the coiled-coil heptad repeat sequences of the BBA68 and OspE proteins affects FH/FHL-1 binding (25, 28). Based on the data presented here, it can be hypothesized that cc1 and cc2 of FhbA interact in an antiparallel fashion to form and present the loop domain which forms, at least in part, the binding site for FH/FHL-1.
In summary, the analyses described here were the first examination of the interaction between FH/FHL-1 and the newly characterized FH/FHL-1-binding protein of B. hermsii, FhbA. The data support the emerging concept that FH/FHL-1 binding is not mediated by a simple linear sequence element but instead involves a conformational and possibly discontinuous binding pocket that is formed by long-range intramolecular interactions. The ability of some relapsing fever spirochete species to circumvent the alternative complement pathway provides a second powerful mechanism of immune evasion for these pathogens. Enhanced understanding of the virulence mechanisms of the relapsing fever spirochetes may allow development of intervention strategies, which could play a significant role in reducing the morbidity and mortality associated with the relapsing fever disease in regions where it is endemic around the globe.
REFERENCES
1. Alitalo, A., T. Meri, T. Chen, H. Lankinen, Z.-Z. Cheng, T. Sakari Jokiranta, I. J. T. Seppala, P. Lahdenne, P. S. Hefty, D. R. Akins, and S. Meri. 2004. Lysine-dependent multipoint binding of the Borrelia burgdorferi virulence factor outer surface protein E to the C terminus of factor H. J. Immunol. 172:6195-6201.
2. Barbour, A. G. 1990. Antigenic variation of a relapsing fever Borrelia species. Annu. Rev. Microbiol. 44:155-171.
3. Barbour, A. G., N. Byrman, C. J. Carter, T. Kitten, and S. Berstrom. 1991. Variable antigen genes of the replasing fever agent Borrelia hermsii are activated by promoter addition. Mol. Microbiol. 5:489-493.
4. Barbour, A. G., C. J. Carter, N. Burman, C. S. Freitag, C. F. Garon, and S. Bergstom. 1991. Tandem insertion sequence-like elements define the expression site for variable antigen genes of Borrelia hermsii. Infect. Immun. 59:390-397.
5. Barbour, A. G., and S. F. Hayes. 1986. Biology of Borrelia species. Microbiol. Rev. 50:381-400.
6. Barbour, A. G., S. L. Tessier, and H. G. Stoenner. 1982. Variable major proteins of Borrellia hermsii. J. Exp. Med. 156:1312-1324.
7. Burman, N., A. Shamaei-Tousi, and S. Bergstrom. 1998. The spirochete Borrelia corcidurae causes erythrocyte rosetting during relapsing fever. Infect. Immun. 66:815-819.
8. Connolly, S. E., and J. L. Benach. 2001. Cutting edge: the spirochetemia of murine relapsing fever is cleared by complement-independent bactericidal antibodies. J. Immunol. 167:3029-3032.
9. Connolly, S. E., D. G. Thanassi, and J. L. Benach. 2004. Generation of a complement-independent bactericidal IgM against a relapsing fever Borrelia. J. Immunol. 172:1191-1197.
10. Cutler, S., and A. Talbert. 2003. Tick-borne relapsing fever in Tanzania—a forgotten problem ASM News 69:542-543.
11. Dave, S., S. Carmicle, S. Hammerschmidt, M. K. Pangburn, and L. S. McDaniel. 2004. Dual roles of PspC, a surface protein of Streptococcus pneumoniae, in binding human secretory IgA and factor H. J. Immunol. 173:471-477.
12. Duthy, T. G., R. J. Ormsby, E. Giannakis, A. D. Ogunniyi, U. H. Stroeher, J. C. Paton, and D. L. Gordon. 2002. The human complement regulator factor H binds pneumococcal surface protein PspC via short consensus repeats 13 to 15. Infect. Immun. 70:5604-5611.
13. Earnhart, C. G., E. L. Buckles, J. S. Dumler, and R. T. Marconi. 2005. Demonstration of OspC type diversity in invasive human Lyme disease isolates and identification of previously uncharacterized epitopes that define the specificity of the OspC antibody response. Infect. Immun. 73:7869-7877.
14. Felsenfeld, O. 1971. Borrelia. Strains, vectors, human and animal borreliosis. Warren H. Green, Inc., St. Louis, MO.
15. Fischetti, V. A., R. D. Horstmann, and V. Pancholi. 1995. Location of the complement factor H binding site on streptococcal M6 protein. Infect. Immun. 63:149-154.
16. Friese, M. A., J. Hellwage, T. S. Jokiranta, S. Meri, H. H. Peter, H. Eibel, and P. F. Zipfel. 1999. FHL-1/reconection and factor H: two human complement regulators which are encoded by the same gene are differentially expressed and regulated. Mol. Immunol. 36:809-818.
17. Giannakis, E., T. S. Jokiranta, D. A. Male, S. Ranganathan, R. J. Ormsby, V. A. Fischetti, C. Mold, and D. L. Gordon. 2003. A common site within factor H SCR7 responsible for binding heparin, C-reactive protein and streptococcal M protein. Eur. J. Immunol. 33:962-969.
18. Guyard, C., E. M. Chester, S. J. Raffel, M. E. Schrumpf, P. F. Policastro, S. F. Porcella, J. M. Leong, and T. G. Schwan. 2005. Relapsing fever spirochetes contain chromosomal genes with unique direct tandemly repeated sequences. Infect. Immun. 73:3025-3037.
19. Hellwage, J., T. Meri, T. Heikkila, A. Alitalo, J. Panelius, P. Lahdenne, I. J. T. Seppala, and S. Meri. 2001. The complement regulator factor H binds to the surface protein OspE of Borrelia burgdorferi. J. Biol. Chem. 276:8427-8435.
20. Hellwage, J., C. Skerka, and P. F. ZIpfel. 1997. Biochemical and functional characterization of the factor-H-related protein 4 (FHR-4). Immunopharmacology 38:149-157.
21. Horstmann, R. D., H. J. Sievertsen, J. Knobloch, and V. A. Fischetti. 1988. Antiphagocytic activity of streptococcal M protein: selective binding of complement controls protein factor H. Proc. Natl. Acad. Sci. USA 85:1657-1661.
22. Hovis, K., J. V. McDowell, L. Griffin, and R. T. Marconi. 2004. Identification and characterization of a linear plasmid-encoded factor H-binding protein (FhbA) of the relapsing fever spirochete, Borrelia hermsii. J. Bacteriol. 186:2612-2618.
23. Kurashige, S., M. Bissett, and L. Oshiro. 1990. Characterization of a tick isolate of Borrelia burgdorferi that possesses a major low-molecular-weight surface protein. J. Clin. Microbiol. 28:1362-1366.
24. Lupas, A., M. Van Dyke, and J. Stock. 1991. Predicting coiled coils from protein sequences. Science 252:1162-1164.
25. McDowell, J. V., M. E. Harlin, E. Rogers, and R. T. Marconi. 2005. Putative coiled-coil structural elements of the BBA68 protein of the Lyme disease spirochetes are required for formation of its factor H binding site. J. Bacteriol. 187:1317-1323.
26. McDowell, J. V., J. Lankford, L. Stamm, T. Sadlon, D. L. Gordon, and R. T. Marconi. 2005. Demonstration of factor H-like protein 1 binding to Treponema denticola, a pathogen associated with periodontal disease in humans. Infect. Immun. 73:7126-7132.
27. McDowell, J. V., E. Tran, D. Hamilton, J. Wolfgang, K. Miller, and R. T. Marconi. 2003. Analysis of the ability of spirochete species associated with relapsing fever, avian borreliosis, and epizootic bovine abortion to bind factor H and cleave C3b. J. Clin. Microbiol. 41:3905-3910.
28. McDowell, J. V., J. Wolfgang, L. Senty, C. M. Sundy, M. J. Noto, and R. T. Marconi. 2004. Demonstration of the involvement of outer surface protein E coiled-coil structural domains and higher order structural elements in the binding of infection-induced antibody and the complement-regulatory protein, factor H. J. Immunol. 173:7471-7480.
29. McDowell, J. V., J. Wolfgang, E. Tran, M. S. Metts, D. Hamilton, and R. T. Marconi. 2003. Comprehensive analysis of the factor H binding capabilities of Borrelia species associated with Lyme disease: delineation of two distinct classes of factor H binding proteins. Infect. Immun. 71:3597-3602.
30. Meri, T., A. Hartmann, D. Lenk, R. Eck, R. Wurzner, J. Hellwage, S. Meri, and P. F. Zipfel. 2002. The yeast Candida albicans binds complement regulators factor H and FHL-1. Infect. Immun. 70:5185-5192.
31. Meri, T., T. S. Jokiranta, J. Hellwage, A. Bialonski, P. F. Zipfel, and S. Meri. 2002. Onchocerca volvulus microfilariae avoid complement attack by direct binding of factor H. J. Infect. Dis. 185:1786-1793.
32. Metts, S., J. V. McDowell, M. Theisen, P. R. Hansen, and R. T. Marconi. 2003. Analysis of the OspE determinants involved in the binding of factor H and OspE targeting antibodies elicited during infection in mice. Infect. Immun. 71:3587-3596.
33. Pandiripally, V., L. Wei, C. Skerka, P. F. Zipfel, and D. Cue. 2003. Recruitment of complement factor H-like protein 1 promotes intracellular invasion by group A streptococci. Infect. Immun. 71:7119-7128.
34. Ram, S., D. P. McQuillen, S. Gulati, C. Elkins, M. K. Pangburn, and P. A. Rice. 1998. Binding of complement factor H to loop 5 of porin protein1A: a molecular mechanism of serum resistance of nonsialyated Neiserria gonorrhoeae. J. Exp. Med. 188:671-680.
35. Ram, S., A. K. Sharma, and S. D. Simpson. 1998. A novel sialic acid binding site on factor H mediates serum resistance of non-sialylated Neisseria gonorrhoeae. J. Exp. Med. 187:743-752.
36. Ruddy, S., and K. F. Austen. 1969. C3 inactivator of man. I. Hemolytic measurement by the inactivation of cell-bound C3. J. Immunol. 102:533-543.
37. Ruddy, S., and K. F. Austen. 1971. C3b inactivator of man. II. Fragments produced by C3b inactivator cleavage of cell-bound or fluid phase C3b. J. Immunol. 107:742-750.
38. Schwan, T. G., P. F. Policastro, Z. Miller, R. L. Thompson, T. Damrow, and J. E. Keirans. 2003. Tick-borne relapsing fever caused by Borrelia hermsii in Montana. Emerg. Infect. Dis. 9:1151-1154.
39. Stoenner, H. G., T. Dodd, and C. Larsen. 1982. Antigenic variation of Borrelia hermsii. J. Exp. Med. 156:1297-1311.
40. Stultz, C. M., J. V. White, and T. F. Smith. 1993. Structural analysis based on state-space modeling. Protein Sci. 2:305-314.
41. White, J. V., C. M. Stultz, and T. F. Smith. 1994. Protein classification by stochastic modeling and optimal filtering of amino-acid sequences. Math. Biosci. 119:35-75.
42. Zhang, H., and R. T. Marconi. 2005. Demonstration of cotranscription and 1-methyl-3-nitroso-nitroguanidine induction of a 30-gene operon of Borrelia burgdorferi: evidence that the 32-kilobase circular plasmids are prophages. J. Bacteriol. 187:7985-7995.
43. Zipfel, P. F., C. Skerka, J. Hellwage, S. T. Jokiranta, S. Meri, V. Brade, P. Kraiczy, M. Noris, and G. Remuzzi. 2002. Structure-function studies of the complement system. Biochem. Soc. Trans. 30:971-978.(Kelley M. Hovis, Janice P)
Center for the Study of Biological Complexity, Medical College of Virginia at Virginia Commonwealth University, Richmond, Virginia 23298-0678
Department of Microbiology and Infectious Diseases, Flinders Medical Centre, Bedford Park, South Australia, Australia
ABSTRACT
Borrelia hermsii, the primary etiological agent of tick-borne relapsing fever in North America, binds the complement regulatory protein factor H (FH) as a means of evading opsonophagocytosis and the alternative complement pathway. The ability of FH-binding protein A (FhbA) to bind FH-like protein 1 (FHL-1) has not been assessed previously. In this study, using a whole-cell absorption assay, we demonstrated that B. hermsii absorbs both FH and FHL-1 from human serum. Consistent with this, affinity ligand binding immunoblot analyses revealed that FH constructs spanning short consensus repeats 1 to 7 and 16 to 20 bind to FhbA. To investigate the molecular basis of the interaction of FhbA with FH/FHL-1, recombinant FhbA truncated proteins were generated and tested for FH/FHL-1 binding. Binding required determinants located in both the N- and C-terminal domains of FhbA, suggesting that long-range intramolecular interactions are involved in the formation and presentation of the FH/FHL-1-binding pocket. To identify specific FhbA residues involved in binding, random mutagenesis was performed. These analyses identified a loop region of FhbA that may serve as a contact point for FH/FHL-1. The data presented here expand our understanding of the pathogenic mechanisms of the relapsing fever spirochetes and of the molecular nature of the interaction between FH/FHL-1 and FhbA.
INTRODUCTION
In North America tick-borne relapsing fever is caused by Borrelia hermsii, Borrelia parkeri, and Borrelia turicatae (5). These pathogens are transmitted to humans through the bites of infected Ornithodoros ticks (5). The hallmark features of tick-borne relapsing fever are cyclic, high-level spirochetemias (107 spirochetes ml blood–1) and a relapsing fever (14). In the United States, tick-borne relapsing fever occurs primarily in the West, and in this region several recent outbreaks have been reported (38). In other parts of the world the impact of tick-borne relapsing fever is staggering. In some regions of Tanzania the incidence in children who are less than 1 year old is 40% (10). In regions where it is endemic, tick-borne relapsing fever is an important human health concern.
B. hermsii employs several mechanisms to persist in the blood and evade the immune response. Antigenic variation mediated by the Vmp proteins is clearly important and has been studied in detail (2-4, 6, 39). Interestingly, Connolly and colleagues have demonstrated that antibody-mediated killing of the relapsing fever spirochetes occurs primarily through a complement-independent mechanism (8, 9). Complement may be more important in opsonization and phagocytosis. We recently demonstrated that B. hermsii and B. parkeri bind the complement regulatory protein factor H (FH) (22, 27). FH is a very abundant serum protein that contributes to regulation of the alternative complement pathway by serving as a cofactor for factor I-mediated cleavage of C3b (36, 37). It also inhibits the initial formation and accelerates the dissociation of the alternative pathway C3 convertase by competing with factor B. FH-like protein 1 (FHL-1), which is derived from FH mRNA by alternative splicing, exhibits similar regulatory activity (16). Binding of FH and FHL-1 to several important human pathogens has been demonstrated (11, 12, 15, 17, 19, 21, 29-35). Surface-bound FH/FHL-1 is thought to locally increase the efficiency of C3b cleavage and thereby inhibit C3b opsonization and subsequent phagocytosis. FH/FHL-1 binding may also facilitate adherence and intracellular localization of some pathogens (33). Treponema denticola, an oral spirochete associated with periodontal disease, binds FHL-1 through a 14-kDa protein designated FHL-1-binding protein B (FhbB) (26). FhbB is unique among bacterial complement regulatory binding proteins in that it binds FHL-1 exclusively and not FH. Interestingly, FHL-1 bound to the surface of T. denticola does not promote or enhance C3b cleavage (26). The ability to bind FHL-1 might be more important in facilitating the interaction of T. denticola with FHL-1 present on the surface of anchorage-dependent cell types. We recently identified an FH-binding protein of B. hermsii and designated it FH-binding protein A (FhbA) (22). FhbA is a 20.5-kDa lipoprotein that is surface exposed and antigenic during infection. It is encoded by a gene located on a linear plasmid estimated to be 200 kb long. FhbA does not exhibit significant sequence identity or similarity with any known FH/FHL-1-binding proteins.
The goals of this study were to determine if FhbA can bind to FHL-1 and to identify the molecular determinants of FhbA and FH/FHL-1 that are involved in the FhbA-FH/FHL-1 interaction. Using FH fragments spanning different short consensus repeats (SCR), coupled with whole-cell absorption assays, we found that FhbA binds both FH and FHL-1. To investigate the molecular nature of the interaction of FH/FHL-1 with FhbA, a series of N- and C-terminal truncations of FhbA were generated and tested for the ability to bind ligand. These analyses demonstrated that determinants located in widely separated domains of FhbA are required for FH/FHL-1 binding. Random mutagenesis was also performed, and the results demonstrated that a loop domain of FhbA may serve as a key contact point for FH/FHL-1. Our hypothesis is that the binding of complement regulatory proteins by the B. hermsii FhbA protein may influence the dissemination characteristics of the tick-borne relapsing fever spirochetes.
MATERIALS AND METHODS
Bacterial isolates and cultivation. The B. hermsii YOR and MAN isolates (provided by Tom Schwan, Rocky Mountain Laboratories) were cultivated in BSK-H complete medium (Sigma) supplemented with 12% rabbit serum (Sigma) at 33°C. The B. hermsii isolates were originally recovered from human relapsing fever patients in the United States (23).
FH/FHL-1 whole-cell absorption assay. To determine if B. hermsii can bind both FH and FHL-1, a whole-cell FH/FHL-1 absorption assay was performed as previously described (26), with some modifications. In brief, B. hermsii YOR, Escherichia coli NovaBlue (negative control), and Borrelia burgdorferi B31MI (positive control) (109 cells) were recovered by centrifugation, washed, and resuspended either in phosphate-buffered saline (PBS) containing purified FH/FHL-1 (0.525 mg ml–1) or in human serum (Valley Biomedical) diluted 1:1 with PBS. The samples were incubated for 1 h at room temperature, recovered by centrifugation, gently washed twice with cold PBS, solubilized in sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) sample buffer, electrophoresed, and immunoblotted. The membranes were then screened for FH or FHL-1 that bound to the surface of the bacteria by using methods described below.
FH/FHL-1 ALBI assays. Recombinant proteins or cell lysates were separated by SDS-PAGE using 12% acrylamide gels (Bio-Rad) and were transferred onto membranes for immunoblot and binding analyses. FH/FHL-1 affinity ligand binding immunoblot (ALBI) assays were conducted as previously described (32), using purified human FH/FHL-1 (Calbiochem). Bound FH/FHL-1 was detected using goat anti-human FH/FHL-1 antiserum (1:800; Calbiochem). Short consensus repeat constructs were prepared as previously described (12). Briefly, the SCR constructs were generated by PCR and cloned into the yeast expression vector pPICZ (Invitrogen). The proteins were expressed in Pichia pastoris and purified by immunoaffinity chromatography. The concentrations of proteins were determined by using an FH standard curve generated by an enzyme-linked immunosorbent assay (ELISA).
Construction of truncated FhbA proteins and other recombinant proteins. To generate FhbA proteins with progressive N- and C-terminal truncations and control proteins, a PCR-based strategy was used. The PCR was performed with Expand High Fidelity polymerase (Roche) and standard cycling conditions. The template for PCR was a previously described plasmid that carries the fhbA gene from the B. hermsii YOR isolate or isolated DNA (22). Primers (Table 1) designed to amplify fragments of fhbA were constructed with tail sequences that allowed annealing into the pET-32 Ek/LIC vector, a procedure referred to as ligase-independent cloning (Invitrogen). The pET-32 EK/LIC vector allowed subsequent expression of the inserts as N-terminal poly(His)- and S-tagged fusion proteins. The fusion added 17 kDa to the mass of each recombinant protein. All cloning methodologies and procedures related to the generation of recombinant proteins have been described previously (13, 42). Expression of the recombinant proteins was confirmed by SDS-PAGE and immunoblotting using horseradish peroxidase (HRP)-conjugated S-protein (1:40,000 dilution; Novagen) and mouse anti-FhbA antiserum (1:1,000 dilution), as previously described (22).
Random mutation of fhbA. fhbA was randomly mutated by PCR by using a GeneMorph II random mutagenesis kit (Stratagene) with primers designed to amplify the entire coding region of fhbA (minus the signal peptide), which spans residues 20 to 192. The primers had tail sequences that allowed annealing of the amplicons into the pET-32 Ek/LIC vector (Novagen). The resulting recombinant plasmid was used to transform E. coli NovaBlue(DE3) cells (Novagen). All methods associated with random mutagenesis of fhbA and generation of the resulting recombinant proteins have been described previously (28). Expression of a recombinant protein was confirmed by screening immunoblots with HRP-conjugated S-protein (1:40,000 dilution; Novagen) and with mouse anti-FhbA antiserum (1:1,000 dilution) (22). The sequences of mutated amplicons were determined by DNA sequencing (MWG Biotech). The possible effects of amino acid substitutions on protein structure were assessed using programs available on the Protein Sequence Analysis (PSA) server (40, 41).
Sequence determination and analysis of the fhbA gene of the B. hermsii MAN isolate. The fhbA gene of the B. hermsii MAN isolate was amplified by PCR using primers listed in Table 1 (22). The amplicon was purified by gel extraction, cloned into the pET-32 Ek/LIC vector, and propagated in E. coli NovaBlue(DE3) cells. The recombinant plasmid was purified and isolated, and the insert was sequenced (MWG Biotech). Sequence analyses were conducted using several different programs, as indicated below.
RESULTS
Demonstration that FhbA binds both FH and FHL-1 using whole-cell absorption assays. A whole-cell FH/FHL-1 absorption assay was performed to determine if FhbA can bind both FH and FHL-1. Human serum served as the source of the FH/FHL-1. The cells were incubated with human serum diluted 1:1 with PBS, washed to remove unbound FH/FHL-1, fractionated by SDS-PAGE, immunoblotted, and screened with anti-FH/FHL-1 antiserum. The antiserum detected 150-kDa and 49-kDa proteins that were absorbed by B. hermsii (Fig. 1). Based on the size and immunoreactivity of these proteins with the anti-human FH/FHL-1 antiserum, we concluded that the proteins were FH and FHL-1, respectively. While the amount of FHL-1 bound by the cells was significantly smaller than the amount of FH bound, this was expected in light of the greater abundance of FH than of FHL-1 in serum (500 μg ml–1 and 50 μg ml–1, respectively). When purified FH/FHL-1 was used in the absorption assays, an additional unidentified 75-kDa protein was also absorbed at a low level by both B. hermsii and B. burgdorferi (positive control) and was detected by the anti-FH/FHL-1 antiserum (data not shown). This protein was presumably an FH breakdown product. The negative control for the absorption analyses, E. coli NovaBlue, did not absorb FH or FHL-1, indicating that the absorption of FH and FHL-1 by B. hermsii is specific. These analyses demonstrated that B. hermsii binds both FH and FHL-1.
Identification of the short consensus repeats of FH/FHL-1 that bind to FhbA. FH and FHL-1 are comprised of short consensus repeats (20 and 7 SCRs, respectively) that are 50 to 60 residues long. To determine which SCRs bind to FhbA, recombinant FhbA (rFhbA) was immunoblotted and tested for binding to FH fragments spanning SCRs 1 to 7, 8 to 15, and 16 to 20. First, the integrity of the constructs and their immunoreactivity with the anti-human FH/FHL-1 antiserum were demonstrated by immunoblot analyses (Fig. 2A). Screening with the SCR constructs revealed that FhbA bound SCRs 1 to 7 and 16 to 20 but not SCRs 8 to 15 (Fig. 2B). A truncated form of recombinant BBN39 (encoded by a member of paralogous gene family 163 of B. burgdorferi B31MI) served as a negative control in these analyses and, as expected, was not bound by any SCR construct. Since FHL-1 is a truncated form of FH generated by alternative splicing and contains only SCRs 1 to 7 (plus a unique 4-amino-acid C-terminal tail) (16), the binding of FhbA to SCRs 16 to 20 specifically indicated that binding to FH occurred. The binding of SCRs 1 to 7 to FhbA could indicate binding to FH and/or FHL-1 since both complement regulatory proteins contain this domain. Collectively, the SCR binding analyses and the absorption assays described above indicated that FhbA has the potential to bind both FH and FHL-1.
Localization of the FH/FHL-1-binding domains of FhbA. To localize the binding sites for FH/FHL-1 on B. hermsii FhbA, a series of FhbA proteins with N- and C-terminal truncations were constructed (Fig. 3A). The expression of each protein was confirmed by immunoblot analysis using HRP-conjugated S-protein and anti-FhbA antiserum (Fig. 3B). All of the recombinant proteins were expressed at similar levels, were the predicted size, and were immunoreactive with the anti-FhbA antiserum. The abilities of the truncated constructs to bind FH/FHL-1 were assessed using the ALBI assay. The full-length recombinant protein (residues 20 to 192) and the fragment spanning residues 48 to 192 exhibited similar levels of FH/FHL-1 binding (Fig. 3B). The fragments spanning residues 79 to 192 and 106 to 192 bound FH/FHL-1, but at a significantly attenuated level. No FH/FHL-1 binding to fragments spanning residues 20 to 165 and 20 to 134 was observed. As an additional control for these analyses, a blot was screened with anti-FH/FHL-1 antiserum with no exogenous FH/FHL-1 added (data not shown). No binding of antibody alone to the recombinant proteins was observed, indicating that the methods employed for detecting FH/FHL-1 binding were specific. The data indicate that widely separated domains of FhbA are required for maximal binding to FH/FHL-1. This observation suggests either that FH/FHL-1 has multiple, widely distributed contact points on FhbA or that long-range intramolecular interactions are required to form the FH/FHL-1-binding pocket.
Generation of random FhbA mutants and analysis of their abilities to bind FH/FHL-1. To identify specific domains of FhbA involved in FH/FHL-1 binding, random mutagenesis was performed. Random mutations were introduced using a PCR-based approach with a low-fidelity polymerase. The primers employed amplified the region of fhbA that encodes the mature protein (residues 20 to 192) and were constructed with tails to allow annealing of the mutated amplicons into the pET-32 Ek/LIC vector. After transformation of the recombinant plasmids into E. coli NovaBlue(DE3) cells, expression of the recombinant proteins was induced by overnight cultivation, and the E. coli cell lysates were fractionated by SDS-PAGE. Note that expression did not require isopropyl--D-thiogalactopyranoside (IPTG) induction. The proteins were transferred onto membranes and screened using HRP-conjugated S-protein, which detected the S tag of the fusion proteins (Fig. 4). Recombinant proteins that were the appropriate size were then tested for immunoreactivity with anti-FhbA antiserum and for FH/FHL-1 binding. Three of the E. coli clones, designated clones 63, 85, and 90, produced recombinant proteins that did not bind FH/FHL-1. Other recombinant proteins produced by clones 194, 196, and (most notably) 198 exhibited significantly attenuated binding. Several other clones produced FhbA proteins that exhibited full FH/FHL-1 binding ability. The mutated rFhbA proteins were designated by using the number assigned to the E. coli clone. For example, FhbA from clone 90 was designated FhbA90.
DNA sequence analysis of the random mutants of FhbA and computer-based structural analyses. To determine the molecular basis for the loss of FH/FHL-1 binding observed with some of the mutated rFhbA proteins, the corresponding plasmid inserts were sequenced. For comparative purposes, the inserts of plasmids from several E. coli clones that produced recombinant proteins that exhibited FH/FHL-1 binding were also sequenced (Fig. 5). FhbA63, FhbA85, and FhbA90 were not able to bind FH/FHL-1, while FhbA194, FhbA196, and FhbA198 exhibited significantly attenuated binding. To identify the mutations that affected FH/FHL-1 binding, we first sought to exclude noncontributory mutations. For example, mutations in the N-terminal domain of FhbA (up to residue 47), as well as mutations in the last 10 residues of the C terminus (as observed with FhbA123), were excluded from consideration since the domains harboring the residues could be deleted with no effect on FH/FHL-1 binding. Analysis of the remaining mutations revealed that FH/FHL-1 binding was particularly sensitive to substitutions introduced in a region of FhbA spanning residues 135 to 162. FhbA198 had a single amino acid change (Phe139 to Val) that resulted in significantly attenuated FH/FHL-1 binding. The FhbA90 mutant, which completely lacked the ability to bind FH/FHL-1, had two amino acid substitutions, Asn143 to Asp and Asp146 to Asn. The first of these residues is predicted to be in a loop domain, and the second is predicted to be in a flanking coiled-coil element (Fig. 5 and 6). FhbA85 also had two substitutions in the loop domain (Thr135 to Met and Asn143 to Lys), and it also had a third substitution (Phe182 to Cys) that mapped outside the loop region. Mutant FhbA63, which had two amino acid substitutions (Pro86 to Leu and Asn162 to Thr) also lacked the ability to bind FH/FHL-1. Computer-based structural analyses predicted that the substitutions in this mutant which involved replacement of a Pro residue significantly altered the FhbA structure. Hence, structural alteration was most likely the basis for the loss of FH/FHL-1 binding to FhbA63. Collectively, the data suggest that the loop region may be a key contact point for FH/FHL-1 and that the presentation of the binding site is dependent on specific structural determinants of the protein.
Determination of the fhbA sequence of the B. hermsii MAN isolate and computer-assisted structural analysis of FhbA. The analyses described above suggested that a defined loop region of FhbA serves as a contact point for FH/FHL-1. If this domain is important in FH/FHL-1 binding, then it would likely be conserved in other strains. As a first step toward assessing this possibility, we verified the production of a FH/FHL-1-binding protein in the B. hermsii MAN and YOR isolates using the ALBI assay (Fig. 6A). The MAN isolate produced an FH/FHL-1-binding protein with a molecular mass slightly less than the molecular mass of the protein in the YOR isolate. The antigenic relatedness of FhbAMAN to FhbAYOR was demonstrated by screening an immunoblot of cell lysates with anti-FhbA antiserum. This antiserum was generated using recombinant FhbA derived from the YOR isolate. The fhbAMAN gene was then amplified and annealed into the pET-32 EK/LIC vector, and a recombinant protein was generated. Expression of the recombinant protein was confirmed by screening immunoblots with HRP-conjugated S-protein and with anti-FhbA antiserum. Finally, the ability of the recombinant protein to bind FH/FHL-1 was demonstrated using the ALBI assay as described above. The fhbAMAN gene sequence was then determined by sequence analysis of the insert carried by the corresponding recombinant plasmid. The FhbAYOR and FhbAMAN sequences are aligned and are shown in Fig. 6B. The amino acid sequences exhibited 81.5% identity and 89.6% similarity, and most of the sequence differences, including a 7-amino-acid insertion in FhbAYOR, were located in the N-terminal half of FhbA.
In other Borrelia FH/FHL-1-binding proteins, the FH/FHL-1-binding pocket is thought to be formed by long-range intramolecular interactions that present a discontinuous binding pocket. Coiled-coil elements were demonstrated to be involved in presentation of the FH/FHL-1-binding sites of BBA68 and OspE (25, 28). To determine if there are structural features that are present in both FhbAMAN and FhbAYOR that might contribute to FH/FHL-1 binding, computer-assisted structural analyses were performed. The predicted probability of coiled-coil formation in each protein was assessed using the COILS program (24). These analyses indicated that there was a high probability of coiled-coil formation for domains in both the FhbAYOR and FhbAMAN proteins (Fig. 6C). However, only the two most C-terminal coiled coils (cc1 and cc2) were conserved. The N-terminal domain coiled-coil elements that were predicted with high probability in the FhbAYOR sequence were predicted with only low probability in FhbAMAN. This lack of conservation of the N-terminal coiled coils, coupled with the truncation analyses that demonstrated that the N terminus of the protein could be deleted with only minimal impact on FH/FHL-1 binding, indicated that the N-terminal coiled coils are not required for ligand binding. The intervening sequence between cc1 and cc2 that forms the putative FH/FHL-1-binding loop of FhbAYOR is conserved at both the sequence and structure levels in FhbAMAN. The random mutagenesis analyses coupled with the conservation of the loop strongly suggested that this specific domain is a critical interaction site for FH/FHL-1.
DISCUSSION
A unique aspect of the relapsing fever spirochetes is their ability to achieve remarkably high cell densities during infection in the blood of mammals (5). It is well documented that these bacteria, particularly B. hermsii, employ an elaborate system of antigenic variation to evade the host antibody response (2). Efficient persistence in the blood is also likely to depend on the ability to evade complement-mediated killing. We recently demonstrated that some species of relapsing fever spirochetes bind the complement regulatory protein FH and possibly FHL-1, which participates in factor I-mediated cleavage of C3b (22, 27). This suggests that FH/FHL-1 binding is biologically relevant, contributes to evasion of the alternative complement pathway, and may decrease the efficiency of opsonization by C3b and subsequent phagocytosis. The goals of this study were to determine if B. hermsii can bind both FH and FHL-1 and to determine the molecular determinants of both FH/FHL-1 and FhbA that are involved in the interaction.
While it has been demonstrated previously that B. hermsii binds FH through its FhbA protein, the ability of B. hermsii to bind FHL-1 was not specifically addressed previously (22). To assess this, a whole-cell-human serum absorption assay was employed. B. hermsii YOR cells incubated with human serum were found to specifically absorb both FH and FHL-1. To further assess the interaction of FhbA with FH/FHL-1, ALBI assays were performed, in which rFhbA was tested to determine its ability to bind recombinant proteins consisting of different FH/FHL-1 short consensus repeats. Three SCR constructs spanning SCRs 1 to 7, 8 to 15, and 16 to 20 were tested. SCRs 1 to 7 and 16 to 20, but not SCRs 8 to 15, bound FhbA. Binding to the SCR 16 to 20 fragment is indicative of binding to FH since these SCRs are not present in FHL-1. SCRs 19 and 20 have been demonstrated to be the interaction point for FH-binding proteins produced by Onchocerca, B. burgdorferi, Neisseria gonorrhoeae, Candida albicans, and other pathogens (reviewed in references 31, 35, and 43). The binding of SCRs 1 to 7 to FhbA suggests that in addition to FH, FhbA can also bind FHL-1. In most cases, both FH and FHL-1 retain their cofactor activities when they are bound to the surfaces of pathogens (43). These proteins also have the potential to function in adherence since they bind to surface glycosaminoglycans and host cell receptors. FH or FHL-1 bound to glycosaminoglycans or receptors may still be able to bind to microbial cell surfaces and thereby serve as a bridge molecule that facilitates the interaction of microbes with host cells. An RGD motif in SCR 4 appears to be particularly important in binding to host cell receptors (20). As described above, T. denticola is unique among the complement regulatory binding pathogens in that it specifically binds FHL-1 (and not FH), and the bound FHL-1 does not exhibit detectable cofactor activity. This suggests that the biological function of FHL-1 binding by this spirochete may involve adherence. It is tempting to speculate that B. hermsii might also utilize host cell-bound FH or FHL-1 binding to allow adherence to the endothelium or other cells and tissues and thereby facilitate dissemination and invasiveness. It has recently been demonstrated that at least one species of relapsing fever spirochete, Borrelia crocidurae, can interact with red blood cells (7). The molecular basis of this interaction has not been defined yet.
To localize the binding determinants of FhbA required for FH/FHL-1 binding, FhbAYOR proteins with N- and C-terminal truncations were constructed. Proteins with truncations from the N terminus up to residue 48 (r48-192) exhibited full FH/FHL-1 binding, while the r79-192 and r106-192 truncated proteins exhibited significantly attenuated binding. Hence, we concluded that the first 47 residues of FhbA are not required for binding and that determinants located between positions 48 and 105 are required for maximal binding activity. Determinants in the C-terminal domain were also found to be required for FH/FHL-1 binding. A C-terminal 27-residue truncation (r20-165) led to complete loss of FH/FHL-1 binding, while a protein with a C-terminal 7-residue truncation generated by introduction of a premature stop codon by random mutagenesis (FhbA123) exhibited FH/FHL-1-binding ability. The three most C-terminal residues of this construct were also altered by disruption of the reading frame. Hence, while determinants in the C-terminal domain of FhbA are important in FH/FHL-1 binding, the last 10 residues are dispensable. We concluded from these analyses that the FH/FHL-1-binding site is not a simple contiguous linear element since independent truncations in different domains eliminated FH/FHL-1 binding. The apparent discontinuous nature of the FH-binding site is consistent with that reported for the OspE and BBA68 FH-binding proteins of B. burgdorferi (1, 25, 28).
The random mutagenesis analyses led to identification of a loop domain in FhbA that appears to be one of the key binding elements involved in the FH/FHL-1 interaction. After noncontributory mutations were excluded (using criteria described above), the remaining mutations that led to decreased binding clustered within or immediately next to a loop that spans residues 130 through 143. The importance of this loop is highlighted by the fact that FhbA198 and FhbA90, which exhibited reduced or no FH/FHL-1 binding, had only one and two amino acid changes, respectively, that mapped in the loop or the immediately flanking C-terminal coiled coil. With the exception of one residue, the sequence of the loop domain was found to be conserved in a second B. hermsii isolate, B. hermsii MAN. The MAN and YOR isolates belong to two distinct genomic groups (18), and hence the conservation of the loop in the different genomic groups suggests that this domain is functionally significant. Ram and colleagues also demonstrated the importance of a loop domain in the porin protein PorA1 of N. gonorrhoeae in FH binding (34). The loop domain of FhbA is flanked by the alpha-helical domains that carry the coiled-coil heptad repeat sequence. This heptad repeat (a-gn) harbors hydrophobic nonpolar residues at the a and d positions. The importance of coiled coils in the presentation of the FH/FHL-1-binding site in other FH/FHL-1-binding proteins has been firmly established (15, 25, 28). We previously demonstrated that replacement of nonpolar, hydrophobic a and d position residues in the coiled-coil heptad repeat sequences of the BBA68 and OspE proteins affects FH/FHL-1 binding (25, 28). Based on the data presented here, it can be hypothesized that cc1 and cc2 of FhbA interact in an antiparallel fashion to form and present the loop domain which forms, at least in part, the binding site for FH/FHL-1.
In summary, the analyses described here were the first examination of the interaction between FH/FHL-1 and the newly characterized FH/FHL-1-binding protein of B. hermsii, FhbA. The data support the emerging concept that FH/FHL-1 binding is not mediated by a simple linear sequence element but instead involves a conformational and possibly discontinuous binding pocket that is formed by long-range intramolecular interactions. The ability of some relapsing fever spirochete species to circumvent the alternative complement pathway provides a second powerful mechanism of immune evasion for these pathogens. Enhanced understanding of the virulence mechanisms of the relapsing fever spirochetes may allow development of intervention strategies, which could play a significant role in reducing the morbidity and mortality associated with the relapsing fever disease in regions where it is endemic around the globe.
REFERENCES
1. Alitalo, A., T. Meri, T. Chen, H. Lankinen, Z.-Z. Cheng, T. Sakari Jokiranta, I. J. T. Seppala, P. Lahdenne, P. S. Hefty, D. R. Akins, and S. Meri. 2004. Lysine-dependent multipoint binding of the Borrelia burgdorferi virulence factor outer surface protein E to the C terminus of factor H. J. Immunol. 172:6195-6201.
2. Barbour, A. G. 1990. Antigenic variation of a relapsing fever Borrelia species. Annu. Rev. Microbiol. 44:155-171.
3. Barbour, A. G., N. Byrman, C. J. Carter, T. Kitten, and S. Berstrom. 1991. Variable antigen genes of the replasing fever agent Borrelia hermsii are activated by promoter addition. Mol. Microbiol. 5:489-493.
4. Barbour, A. G., C. J. Carter, N. Burman, C. S. Freitag, C. F. Garon, and S. Bergstom. 1991. Tandem insertion sequence-like elements define the expression site for variable antigen genes of Borrelia hermsii. Infect. Immun. 59:390-397.
5. Barbour, A. G., and S. F. Hayes. 1986. Biology of Borrelia species. Microbiol. Rev. 50:381-400.
6. Barbour, A. G., S. L. Tessier, and H. G. Stoenner. 1982. Variable major proteins of Borrellia hermsii. J. Exp. Med. 156:1312-1324.
7. Burman, N., A. Shamaei-Tousi, and S. Bergstrom. 1998. The spirochete Borrelia corcidurae causes erythrocyte rosetting during relapsing fever. Infect. Immun. 66:815-819.
8. Connolly, S. E., and J. L. Benach. 2001. Cutting edge: the spirochetemia of murine relapsing fever is cleared by complement-independent bactericidal antibodies. J. Immunol. 167:3029-3032.
9. Connolly, S. E., D. G. Thanassi, and J. L. Benach. 2004. Generation of a complement-independent bactericidal IgM against a relapsing fever Borrelia. J. Immunol. 172:1191-1197.
10. Cutler, S., and A. Talbert. 2003. Tick-borne relapsing fever in Tanzania—a forgotten problem ASM News 69:542-543.
11. Dave, S., S. Carmicle, S. Hammerschmidt, M. K. Pangburn, and L. S. McDaniel. 2004. Dual roles of PspC, a surface protein of Streptococcus pneumoniae, in binding human secretory IgA and factor H. J. Immunol. 173:471-477.
12. Duthy, T. G., R. J. Ormsby, E. Giannakis, A. D. Ogunniyi, U. H. Stroeher, J. C. Paton, and D. L. Gordon. 2002. The human complement regulator factor H binds pneumococcal surface protein PspC via short consensus repeats 13 to 15. Infect. Immun. 70:5604-5611.
13. Earnhart, C. G., E. L. Buckles, J. S. Dumler, and R. T. Marconi. 2005. Demonstration of OspC type diversity in invasive human Lyme disease isolates and identification of previously uncharacterized epitopes that define the specificity of the OspC antibody response. Infect. Immun. 73:7869-7877.
14. Felsenfeld, O. 1971. Borrelia. Strains, vectors, human and animal borreliosis. Warren H. Green, Inc., St. Louis, MO.
15. Fischetti, V. A., R. D. Horstmann, and V. Pancholi. 1995. Location of the complement factor H binding site on streptococcal M6 protein. Infect. Immun. 63:149-154.
16. Friese, M. A., J. Hellwage, T. S. Jokiranta, S. Meri, H. H. Peter, H. Eibel, and P. F. Zipfel. 1999. FHL-1/reconection and factor H: two human complement regulators which are encoded by the same gene are differentially expressed and regulated. Mol. Immunol. 36:809-818.
17. Giannakis, E., T. S. Jokiranta, D. A. Male, S. Ranganathan, R. J. Ormsby, V. A. Fischetti, C. Mold, and D. L. Gordon. 2003. A common site within factor H SCR7 responsible for binding heparin, C-reactive protein and streptococcal M protein. Eur. J. Immunol. 33:962-969.
18. Guyard, C., E. M. Chester, S. J. Raffel, M. E. Schrumpf, P. F. Policastro, S. F. Porcella, J. M. Leong, and T. G. Schwan. 2005. Relapsing fever spirochetes contain chromosomal genes with unique direct tandemly repeated sequences. Infect. Immun. 73:3025-3037.
19. Hellwage, J., T. Meri, T. Heikkila, A. Alitalo, J. Panelius, P. Lahdenne, I. J. T. Seppala, and S. Meri. 2001. The complement regulator factor H binds to the surface protein OspE of Borrelia burgdorferi. J. Biol. Chem. 276:8427-8435.
20. Hellwage, J., C. Skerka, and P. F. ZIpfel. 1997. Biochemical and functional characterization of the factor-H-related protein 4 (FHR-4). Immunopharmacology 38:149-157.
21. Horstmann, R. D., H. J. Sievertsen, J. Knobloch, and V. A. Fischetti. 1988. Antiphagocytic activity of streptococcal M protein: selective binding of complement controls protein factor H. Proc. Natl. Acad. Sci. USA 85:1657-1661.
22. Hovis, K., J. V. McDowell, L. Griffin, and R. T. Marconi. 2004. Identification and characterization of a linear plasmid-encoded factor H-binding protein (FhbA) of the relapsing fever spirochete, Borrelia hermsii. J. Bacteriol. 186:2612-2618.
23. Kurashige, S., M. Bissett, and L. Oshiro. 1990. Characterization of a tick isolate of Borrelia burgdorferi that possesses a major low-molecular-weight surface protein. J. Clin. Microbiol. 28:1362-1366.
24. Lupas, A., M. Van Dyke, and J. Stock. 1991. Predicting coiled coils from protein sequences. Science 252:1162-1164.
25. McDowell, J. V., M. E. Harlin, E. Rogers, and R. T. Marconi. 2005. Putative coiled-coil structural elements of the BBA68 protein of the Lyme disease spirochetes are required for formation of its factor H binding site. J. Bacteriol. 187:1317-1323.
26. McDowell, J. V., J. Lankford, L. Stamm, T. Sadlon, D. L. Gordon, and R. T. Marconi. 2005. Demonstration of factor H-like protein 1 binding to Treponema denticola, a pathogen associated with periodontal disease in humans. Infect. Immun. 73:7126-7132.
27. McDowell, J. V., E. Tran, D. Hamilton, J. Wolfgang, K. Miller, and R. T. Marconi. 2003. Analysis of the ability of spirochete species associated with relapsing fever, avian borreliosis, and epizootic bovine abortion to bind factor H and cleave C3b. J. Clin. Microbiol. 41:3905-3910.
28. McDowell, J. V., J. Wolfgang, L. Senty, C. M. Sundy, M. J. Noto, and R. T. Marconi. 2004. Demonstration of the involvement of outer surface protein E coiled-coil structural domains and higher order structural elements in the binding of infection-induced antibody and the complement-regulatory protein, factor H. J. Immunol. 173:7471-7480.
29. McDowell, J. V., J. Wolfgang, E. Tran, M. S. Metts, D. Hamilton, and R. T. Marconi. 2003. Comprehensive analysis of the factor H binding capabilities of Borrelia species associated with Lyme disease: delineation of two distinct classes of factor H binding proteins. Infect. Immun. 71:3597-3602.
30. Meri, T., A. Hartmann, D. Lenk, R. Eck, R. Wurzner, J. Hellwage, S. Meri, and P. F. Zipfel. 2002. The yeast Candida albicans binds complement regulators factor H and FHL-1. Infect. Immun. 70:5185-5192.
31. Meri, T., T. S. Jokiranta, J. Hellwage, A. Bialonski, P. F. Zipfel, and S. Meri. 2002. Onchocerca volvulus microfilariae avoid complement attack by direct binding of factor H. J. Infect. Dis. 185:1786-1793.
32. Metts, S., J. V. McDowell, M. Theisen, P. R. Hansen, and R. T. Marconi. 2003. Analysis of the OspE determinants involved in the binding of factor H and OspE targeting antibodies elicited during infection in mice. Infect. Immun. 71:3587-3596.
33. Pandiripally, V., L. Wei, C. Skerka, P. F. Zipfel, and D. Cue. 2003. Recruitment of complement factor H-like protein 1 promotes intracellular invasion by group A streptococci. Infect. Immun. 71:7119-7128.
34. Ram, S., D. P. McQuillen, S. Gulati, C. Elkins, M. K. Pangburn, and P. A. Rice. 1998. Binding of complement factor H to loop 5 of porin protein1A: a molecular mechanism of serum resistance of nonsialyated Neiserria gonorrhoeae. J. Exp. Med. 188:671-680.
35. Ram, S., A. K. Sharma, and S. D. Simpson. 1998. A novel sialic acid binding site on factor H mediates serum resistance of non-sialylated Neisseria gonorrhoeae. J. Exp. Med. 187:743-752.
36. Ruddy, S., and K. F. Austen. 1969. C3 inactivator of man. I. Hemolytic measurement by the inactivation of cell-bound C3. J. Immunol. 102:533-543.
37. Ruddy, S., and K. F. Austen. 1971. C3b inactivator of man. II. Fragments produced by C3b inactivator cleavage of cell-bound or fluid phase C3b. J. Immunol. 107:742-750.
38. Schwan, T. G., P. F. Policastro, Z. Miller, R. L. Thompson, T. Damrow, and J. E. Keirans. 2003. Tick-borne relapsing fever caused by Borrelia hermsii in Montana. Emerg. Infect. Dis. 9:1151-1154.
39. Stoenner, H. G., T. Dodd, and C. Larsen. 1982. Antigenic variation of Borrelia hermsii. J. Exp. Med. 156:1297-1311.
40. Stultz, C. M., J. V. White, and T. F. Smith. 1993. Structural analysis based on state-space modeling. Protein Sci. 2:305-314.
41. White, J. V., C. M. Stultz, and T. F. Smith. 1994. Protein classification by stochastic modeling and optimal filtering of amino-acid sequences. Math. Biosci. 119:35-75.
42. Zhang, H., and R. T. Marconi. 2005. Demonstration of cotranscription and 1-methyl-3-nitroso-nitroguanidine induction of a 30-gene operon of Borrelia burgdorferi: evidence that the 32-kilobase circular plasmids are prophages. J. Bacteriol. 187:7985-7995.
43. Zipfel, P. F., C. Skerka, J. Hellwage, S. T. Jokiranta, S. Meri, V. Brade, P. Kraiczy, M. Noris, and G. Remuzzi. 2002. Structure-function studies of the complement system. Biochem. Soc. Trans. 30:971-978.(Kelley M. Hovis, Janice P)