BBK32, a fibronectin binding MSCRAMM from Borrelia burgdorferi, contains a disordered region that undergoes a conformational change on ligand binding.

BBK32 is a fibronectin-binding lipoprotein on Borrelia burgdorferi, the causative agent of Lyme disease. Analysis using secondary structure prediction programs suggested that BBK32 is composed of two domains, an N-terminal segment lacking well defined secondary structure and a C-terminal segment composed largely of alpha-helices. Analysis of purified recombinant forms of the two domains by circular dichroism spectroscopy, gel permeation chromatography, and intrinsic viscosity determination were consistent with an N-terminal-extended, unstructured segment and a C-terminal globular domain in BBK32. Solid phase binding experiments suggest that the unstructured N-terminal domain binds fibronectin. Analysis of changes in circular dichroism spectra of the N-terminal segment of BBK32 upon binding of the N-terminal domain of fibronectin revealed an increase in beta-sheet content in the complex. Hence, BBK32, which belongs to a different family of proteins and shows no overall sequence similarity with the fibronectin binding MSCRAMMs (microbial surface components recognizing adhesive matrix molecules) of Gram-positive bacteria, binds fibronectin by a mechanism that is reminiscent of the "tandem beta-zipper" previously demonstrated for the fibronectin binding of streptococcal adhesins.

Lyme disease is caused by the spirochete Borrelia burgdorferi that can be transmitted to a mammalian host during the blood meal of infected ticks (1). The molecular pathogenesis of the disease is poorly understood, and the microbial virulence factors involved remain largely unknown. B. burgdorferi has a small genome composed of a 0.9-megabase chromosome and multiple linear and circular plasmids totaling 0.6 megabases (2). Sequence analysis of the B. burgdorferi genome did not reveal any obvious virulence factors such as toxins. Instead, components exposed on the surface of the organism that are capable of interacting with targets in the host may play key roles in the disease process.
Lyme disease progresses in several discrete steps. Initially bacteria need to travel against the flow of blood from the tick salivary gland to the host and colonize the dermis at the site of the tick bite (3). This initial infection, which often is accompanied by a local inflammatory response called erythema migrans, is followed by a dissemination step. Through hematogenous spread, the spirochetes can find their way to and colonize essentially all tissues in infected animals. Disseminated Borrelia infections can cause neurological, ocular, cutaneous, cardiac, and arthritic disease (4,5). Untreated infection can persist for months or years, and these extracellular pathogens apparently survive the attacks of the host defense systems (4). Even after appropriate antibiotic treatment of Lyme disease, a small percentage of patients have "chronic Lyme disease," which often is manifested by musculoskeletal pain, neurocognitive difficulties, or fatigue that can continue for years (5).
Presumably the different stages of Lyme disease involve different sets of virulence factors. The transition of the spirochetes from the tick to mammals involves exposure to dramatically different environments with significant changes in temperature, pH, and composition of nutrients (6 -8). Genome-wide transcriptional analysis of B. burgdorferi, cultured under different in vitro conditions designed to mimic such a transition, revealed significant regulation of subsets of genes (9). The transcription of a subfamily of genes encoding putative lipoproteins potentially exposed at the surface of the spirochete is dramatically up-regulated as the bacteria are grown in conditions mimicking the environment encountered in the mammalian host (9,10). These up-regulated lipoproteins, which are putative virulence factors in Lyme disease, include the decorin binding adhesins DbpA and DbpB (11)(12)(13)(14) and the fibronectin (Fn) 1 binding adhesin BBK32 (15).
Fn in the host is exploited by many pathogenic bacteria as a target for microbial adhesion and colonization during the early stage of the infection process (for reviews, see Refs. 16 and 17). Fn is a dimeric glycoprotein present in a soluble form in body fluids such as blood plasma and in a fibrillar form in the extracellular matrix of loose connective tissue. Fn participates in many physiological processes such as cell proliferation, migration, and survival through interactions with a variety of extracellular macromolecules and cellular receptors of the integrin family (18,19). Bacteria expressing Fn binding MSCRAMMs (microbial surface components recognizing adhesive matrix molecules) have been shown to interact not only with fibrillar Fn during bacterial adhesion but also to bind soluble Fn to their surfaces. The bacteria-bound Fn can then be targeted by appropriate integrins in a process leading to microbial invasion of the host cell (20 -22). Fn binding MSCRAMMS from pathogenic staphylococci and streptococci have been shown to bind to the N-terminal domain (NTD) of Fn, which contains a string of five Fn type 1 ( 1-5 F1) modules (23).
The Fn binding adhesin BBK32 was originally identified by Probert and Johnson (15) by probing lysates of B. burgdorferi strain B31 isolate with Fn in Western-ligand blots. Furthermore, BBK32 was localized to the surface of the spirochete, and the attachment of B. burgdorferi to an Fn substrate was blocked by the addition of soluble recombinant BBK32. We previously localized an Fn-binding site in BBK32 to a 32-amino acid-long segment in the protein (24). This ligand binding segment was shown to share 81-91% amino acid sequence identity with the homologous proteins encoded by bbk32 genes of B. burgdorferi sensu stricto, Borrelia garinii, and Borrelia afzelii. We also identified an amino acid sequence homology in this ligand binding region of BBK32 to the so-called upstream region of protein F1, an Fn binding MSCRAMM from Streptococcus pyogenes (24). The upstream region segment is one of the Fn binding motifs in protein F1, and it specifically binds to the collagen binding domain of Fn (25). From the observation that gelatin can partially inhibit the binding of Fn to BBK32, the collagen binding domain of Fn was suggested to contain a binding site for BBK32 (15).
Antisera raised against BBK32 have been reported to protect naive mice against tick-transmitted B. burgdorferi infection and reduce spirochetal transmission from mammalian host to engorging ticks (26,27). A continuous spirochetal expression of BBK32 in the mammalian host from the early to the late stage of the infection suggests a role for BBK32 throughout the Lyme disease process (28).
In this study we have examined the structural organization of BBK32 by using a number of biophysical techniques to characterize recombinant proteins representing predicted subdomains comprised of the N-and C-terminal halves of BBK32. We also have characterized conformational changes in the Nterminal subdomain of BBK32 upon Fn binding. Overall, this study demonstrates that BBK32, the spirochetal Fn binding MSCRAMM (FnBM), shows similarities to Gram-positive bacterial FnBMs in their structural organization and Fn binding mechanisms. This finding is notable since BBK32 is a surface lipoprotein oriented in the opposite direction compared with that of FnBM from the Gram-positive bacteria. Furthermore, BBK32 does not contain Fn binding repeating motifs that are common in FnBMs from Gram-positive bacteria.

EXPERIMENTAL PROCEDURES
Secondary Structure Prediction-Initially, amino acid sequences representing BBK32 and the Borrelia surface proteins OspA and VlsE were analyzed using eight secondary structure prediction programs (SOPM, HNN, DPM, DSC, GORIV, PHD, PREDATOR, and SIMPA96). Consensus secondary structure predictions were provided from Pôle Bio-Informatique Lyonnais (npsa-pbil.ibcp.fr/cgi-bin/npsa_automat.pl?pageϭ/ NPSA/npsa_seccons.html). Preliminary analysis showed that a combination of SOPM, DSC, and PHD most accurately predicted the composition of secondary structure in OspA and VlsE for which protein crystal structures have been reported (29,30). We, therefore, relied on the combination of these three programs for the secondary structure predictions of BBK32.
Construction and Expression of Recombinant Fusion Proteins-Amplified DNA segments encoding BBK32 (21-354), BBK32-(56 -354), BBK32-(56 -205), and BBK32-(206 -354) of strain B31 of B. burgdorferi sensu stricto were cloned into the BamH1 and Hind III sites of pQE30 (Qiagen Inc.) (15,31). The resulting pQE30-based plasmids were introduced into Escherichia coli strain JM101 made competent through heat-shock treatment. Transformed JM101 were grown for 12-17 h on LB containing ampicillin (100 g/ml) plates. Bacteria in colonies were screened for the presence of the expression plasmid with inserts by susceptibility to restriction enzyme digestion and sequencing of inserts in both directions. The constructs were transformed into E. coli strain TOPP 3 for expression of recombinant proteins.
Starter cultures of E. coli TOPP 3 containing the recombinant plasmids were diluted 1:25 in LB containing ampicillin (100 g/ml) and incubated with shaking until the optical density at 600 nm reached 0.7-0.9. Protein expression was induced by the addition of isopropyl ␤-D-thiogalactopyranoside (Invitrogen) to a final concentration of 0.1 mM, and the bacterial suspension was incubated for an additional 4 h. Bacterial cells were harvested by centrifugation, resuspended in a minimal volume of phosphate-buffered saline (PBS; 137 mM NaCl, 3 mM KCl, 10 mM Na 2 HPO 4 , 2 mM KH 2 PO 4 , pH 7.4), and frozen at Ϫ20°C.
Purification of Recombinant Proteins-Cells were lysed by passage through a French press at 1100 p.s.i. (SLM Instruments Inc., Rochester, NY). Cellular debris and insoluble proteins were removed by centrifugation at 20,000 ϫ g for 20 min. The cleared, lysed cell supernatant was filtered through a 0.45-m membrane and applied to a 5-ml HiTrap chelating column (Amersham Biosciences) of nickel-charged iminodiacetic acid-Sepharose. Bound proteins were eluted with a continuous linear gradient of imidazole (0 -200 mM; total volume 200 ml) in 20 mM Tris-HCl, 100 mM NaCl, pH 7.9. In the cases of BBK32-(56 -354) and BBK32-(56 -205), the concentration of NaCl in the buffer was increased to 500 mM to facilitate binding of the His tag of the protein to the column. EDTA (0.5 M) was added to the eluant to a final concentration of 5 mM. Eluted proteins were monitored by analyzing the absorbance at 280 nm and by SDS-PAGE. Fractions containing the recombinant proteins were pooled, dialyzed overnight against 25 mM Tris-HCl, pH 7.4, and applied to a 5-ml HiTrap Q-Sepharose column (Amersham Biosciences). Bound proteins were eluted with a continuous linear gradient of NaCl (50 -500 mM; total volume 100 ml) in 25 mM Tris-HCl, 1 mM EDTA, pH 7.9). The protein concentration in the eluted fractions was monitored by measuring the absorbance at 280 nm. Fractions with high absorbance were analyzed by SDS-PAGE, pooled, and dialyzed against PBS, 1 mM EDTA. The amino acid composition of each of the recombinant proteins was used to determine molar extinction coefficients at 280 nm (32), and the protein concentrations were determined by measuring the absorbance at 280 nm. All samples were frozen at Ϫ20°C or used immediately. The molecular masses of BBK32-(56 -205) and BBK32-(206 -354) were determined by mass spectrometry (MALDI-TOF mass spectrometer, Department of Biochemistry and Biophysics, Texas A&M University).
Circular Dichroism (CD)-The secondary structure composition of the recombinant BBK32 proteins was examined by CD spectroscopy. Far UV CD spectra were recorded on a Jasco J720 spectropolarimeter calibrated with d-10-camphorsulfonic acid employing a band pass of 1 nm and integrating over 4 s at 0.2-nm intervals. All samples were dialyzed against 10 mM KH 2 PO 4 /K 2 HPO 4 (1:4), pH 7.6, and diluted to a concentration of 10 M in the same buffer. Spectra were recorded at ambient temperature in a cylindrical cuvette with a 0.2-mm path length. Twenty scans were averaged for each spectrum, and the contribution from the buffer was subtracted. The measured CD (ellipticity, ) was converted into a mean residue weight ellipticity [], expressed in deg⅐cm 2 /dmol.
To quantify secondary structure elements, the measured ellipticity over 260ϳ180 nm was converted into a CD absorption value using the SOFTSEC TM file conversion program. The CD absorbance was analyzed using five deconvolution software programs: CD Estima, Contin, Neural Network, Selcon, and Varslc1 (provided by University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School (Piscataway, NJ) and D. Greenwood (Softwood Co., Brooksfield, CT)). The convoluted data were collected as three groups of secondary structure, ␣-helix, ␤-sheet, and random coil including ␤-turns, and expressed as an average percentage of each element with S.E.
Structural changes in BBK32-(56 -254), BBK32-(205-354), and Fn-bpA-(D123) on binding of the NTD of Fn were examined by analyzing changes in the CD. The recombinant MSCRAMM proteins were preincubated with or without the NTD of Fn for 1 h at room temperature, and the far-UV CD spectra were recorded as described above. All proteins, including free NTD, were prepared at a concentration of 10 M so that the molar ratio of protein to ligand in the mixture was 1:1. The changes in the generated CD spectra upon NTD-MSCRAMM interaction were determined by computationally subtracting the ellipticity of the NTD-MSCRAMM complex from the added ellipticities of the free forms of the two interacting proteins. The deconvolution of the resulting spectrum was performed as described above.
Gel Permeation Chromatography-The recombinant BBK32 proteins were analyzed for their effective radii (Stokes radii, R s ) in solution using a Superdex 75 HR 10/30 column (Amersham Biosciences) attached to an fast protein liquid chromatography (Amersham Biosciences) system. Protein samples were preequilibrated with PBS, 1 mM EDTA, applied to the column, and eluted with the same buffer at a flow rate of 0.2 ml/min. The column was calibrated using a low molecular weight gel filtration calibration kit (Amersham Biosciences). The standard globular proteins contained in the kit were ribonuclease A (13,700 Da), chymotrypsinogen (25,000 Da), ovalbumin (43,000 Da), and albumin (67,000 Da). Blue dextran 2000 (2,000,000 Da) (Amersham Biosciences) and tyrosine (181 Da) (Sigma) were used to indicate the void volume (V o ) and the total volume (V t ), respectively. The elution volume (V e ) of each sample was measured. To define the relationships between the elution volumes of protein samples and their respective molecular weight, the K av value for each protein was calculated using the equation where V o and V t for the column used were 9.5 and 24.0 ml, respectively. K av values of standard and BBK32 proteins were plotted against the logarithm of the protein molecular weights. R s of the proteins were determined using sample elution volumes and standard curves as described in the calibration kit. Changes in elution profiles of BBK32 proteins upon increasing concentrations (0 -6 M) of guanidine hydrochloride (GnHCl) were monitored and plotted as described above. Before analyzing the elution volumes of proteins under different denaturing conditions, each protein and the column were equilibrated with the appropriate buffer containing the indicated concentrations of GnHCl.
Intrinsic Viscosity-The viscosities of the BBK32 proteins were measured using a Cannon-Ubbelohde Semi-Micro Dilution Viscometer  Table III. concentrations was dialyzed overnight against PBS with or without 4 M urea, and the same buffer was used as a reference solution. The specific viscosity ( sp ) was determined as described previously (33). The intrinsic viscosity [] was calculated by using the equation where c is protein concentration, and k is a dimensionless constant. Purification of Fn and N-terminal Fragment of Fn-Human plasma was purchased from the Gulf Coast Regional Blood Center (Houston, TX). Fn was purified from plasma by affinity chromatography on a gelatin-Sepharose column as described previously (34) with some modifications. Gelatin (Bio-Rad) was coupled to CNBr-activated Sepharose 4B (Amersham Biosciences) according to the manufacturer's instruction. The plasma was treated with EDTA (5 mM) and phenylmethylsulfonyl fluoride (2 mM) and was passed through a 0.45-m filter. The filtered plasma was applied to a Sepharose 4B column equilibrated with PBS containing 1 mM EDTA, pH 7.4 (EB) to remove components that nonspecifically bind to the Sepharose column. The flow-through from the Sepharose 4B column was applied to a gelatin-Sepharose column equilibrated with the same buffer. The column was washed with 10 column volumes of EB, 10 column volumes of EB with 1 M NaCl, and 5 column volumes of EB containing 0.2 M arginine. Fn was finally eluted using EB containing 4 M urea and then dialyzed against PBS containing 1 mM EDTA.
The dialyzed Fn was further purified by applying the Fn solution to a heparin-Sepharose column (Amersham Biosciences) connected to a fast protein liquid chromatography system. The column was equilibrated and washed before and after applying the sample with 50 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, pH 7.4. Fn was eluted with an NaCl gradient (150 -500 mM) in the same buffer. The peak fractions were analyzed by measuring the absorbance at 280 nm and by SDS-PAGE, pooled, and dialyzed against PBS containing 1 mM EDTA, pH 7.4. Fn was used immediately or stored at 4°C.
An N-terminal proteolytically derived fragment of human Fn (NTD) was either purchased from Sigma or purified from proteolytically digested Fn. To purify the NTD proteolytic fragment, Fn was digested with thermolysin as described previously (35). The digestion was stopped by adding EDTA to a final concentration of 5 mM and verified by SDS-PAGE. The NTD was purified by affinity chromatography on the B3 repeat region of FnbB (an Fn binding MSCRAMM from Streptococcus dysgalactiae (36)) immobilized on a CNBr-activated Sepharose column (Amersham Biosciences). Digested Fn was loaded onto the EB-equilibrated B3 repeat column and washed with 10 column volumes of EB and EB containing 1 M NaCl. Bound NTD was eluted with 4 M urea in EB. The purified NTD was dialyzed against EB and kept at Ϫ20°C until used. To confirm the sequence identity and the molecular mass of the NTD, MALDI-TOF mass spectrometry and N-terminal sequencing were carried out at the Baylor College of Medicine Core Laboratory, Protein Chemistry Core Facility (Houston, TX).
SDS-PAGE-Proteins were fractionated by SDS-PAGE (37) using a 4% stacking gel and a 12% resolving gel. The gels were stained with Coomassie Brilliant Blue R250 (Amersham Biosciences). Prestained high molecular weight standards (Bio-Rad) were used for the estimation of molecular mass.
Solid Phase Binding Assay-An ELISA based binding assay was used to examine MSCRAMM ligand interactions. The washing buffer consisted of PBS containing 0.05% (v/v) Tween 20 (PBST), the blocking solution consisted of 2% (w/v) ovalbumin in PBS, and the dilution buffer consisted of PBS containing 1 mM EDTA. In every assay microtiter plates (Immunolon-1 microtiter plates, Dynatech Laboratories Inc., Alexandria, Va) were coated with 1 g of ligand or antigen diluted in 50 l of PBS, 1 mM EDTA per well and incubated overnight at 4°C.
In the Fn binding assay, Fn-coated microtiter wells were washed, blocked, and incubated with increasing concentrations of His-tag fusion BBK32 polypeptides in 100 l of dilution buffer for 60 min at room temperature. After washing, bound recombinant polypeptides were detected with a 1:3000 dilution of anti-His monoclonal antibodies (Amersham Biosciences) in PBST and a secondary incubation with a 1:3000 dilution of AP-conjugated goat anti-mouse antibodies (Sigma) in PBST. The wells were incubated for 1 h at room temperature, washed, and incubated with 100 l of a 1 mg/ml Sigma 104 alkaline phosphatase substrate (Sigma) dissolved in 1 M diethanolamine, 0.5 mM MgCl 2 , pH 9.8, for 1-3 h. The absorbance at 405 nm was determined using a microplate reader (Molecular Devices, Menlo Park, CA).

Sequence Analysis Indicates the Presence of Two Subdomains
in BBK32-Secondary structure predictions of BBK32 using a combination of the programs SOPM, DSC, and PHD suggested that the protein can be divided into two structurally distinct subdomains, an N-terminal region (amino acid 21-205) composed largely of random coil structures and a C-terminal region (amino acid 206 -354) rich in ␣-helices. To evaluate the reliability of this prediction, we used the combined programs to predict the secondary structure composition of two structurally different Borrelia lipoproteins, OspA and VlsE, for which the crystal structures have been determined. The SOPM-DSC-PHD combination program predicted the secondary structure compositions of these two known structures reasonably well (Table I). We, therefore, believe that the secondary structure composition predicted for BBK32 is also reliable.
To further examine the putative subdomains of BBK32, we next expressed segments corresponding to these domains as well as full-length BBK32 as recombinant proteins (Fig. 1A). Initial attempts to purify a recombinant form of BBK32-(21-354) revealed that this was readily degraded (Fig. 1B). Mass spectrometry and N-terminal sequence analysis indicated that  the protein was cleaved at amino acid 55 (data not shown). Therefore, we chose to express segments BBK32-(56 -354), BBK32-(56 -205), and BBK32-(206 -354) as recombinant fusion proteins with an N-terminal His tag. The recombinant fusion proteins were purified by a combination of Ni 2ϩ -chelating and ion exchange chromatography. The secondary structure compositions predicted for these segments of BBK32 are presented in Table I, and the molecular weights, calculated for the recombinant fusion proteins from their amino acid composition and determined by mass spectrometry, are presented in Table II.

CD Spectra of Recombinant Proteins Support a Two-domain
Structure-The far UV CD spectra of the different recombinant BBK32 segments were recorded (Fig. 2). All protein samples were at 10 M and extensively dialyzed against a phosphate buffer (2.5 mM KH 2 PO 4, 7.5 mM K 2 HPO 4 , pH 7.6). The spectrum of the buffer alone was also recorded and subtracted from those of the protein samples.
The CD spectrum of BBK32-(206 -354) exhibited a strong positive signal at 195 nm and separated minima of similar magnitude at 222 and 208 nm. These peaks are characteristic of CD spectra of ␣-helices. The CD spectrum of BBK32-(56 -205), with a strong negative signal just below 200 nm, is consistent with the structure of BBK32-(56 -205) being predominantly random coil (39).
To estimate the secondary structure composition of each protein, the recorded CD spectra () were analyzed using a panel of five deconvolution programs. The secondary structure compositions of the putative BBK32 subdomains, calculated from the deconvolution of the recorded CD spectra and predicted from the primary sequences, respectively, are in rela- tively good agreement. BBK32-(56 -205) is an exception; the deconvolution of recorded CD spectra suggests a higher content of random coil and ␤-sheet and lower content of ␣-helix compared with the composition predicted by analysis of the primary sequence ( Table I).
The CD spectrum of BBK32-(56 -354) was identical to the CD spectrum of mixed solutions of BBK32-(56 -205) and BBK32-(206 -354) (Fig. 2). These spectra were also identical to a spectrum derived from the computational addition of spectra recorded for BBK32-(56 -205) and BBK32-(206 -354) (data not shown). These results suggest that the two segments of BBK32 represent independent domains that do not interact with each other to the extent that it affects their secondary structure compositions.
Gel Permeation Chromatography of Recombinant BBK32 Polypeptides-The secondary structure predictions suggest that the N-terminal domain, BBK32-(56 -205), may have an extended structure, whereas the C-terminal domain, BBK32-(206 -354), may adopt a globular shape. Therefore, the recombinant BBK32 polypeptides were analyzed by gel permeation chromatography to compare the effective radii (R s ) of the two putative subdomains. K av values for recombinant BBK32-(56 -354), BBK32-(56 -205), and BBK32-(206 -354) proteins and globular protein standards (13.7 kDa ϳ 67 kDa) were determined using Equation 1 (see "Experimental Procedures"). The K av value for each protein was plotted against the log of its molecular mass (Fig. 3) (40). In this plot, the K av values determined for standard globular proteins form a straight line, whereas the K av value for each BBK32 polypeptide significantly deviates from this line. BBK32-(56 -205) and BBK32-(56 -354) were eluted at smaller K av values than expected for globular proteins of similar molecular masses. The R s of the proteins were calculated by fitting their K av to the line derived from the standard proteins (Table II). The R s calculated for BBK32-(56 -354) and BBK32-(56 -205) were 33.1 and 28.4 Å, respectively, whereas the R s estimated for globular proteins of similar molecular masses were 26.7 and 18.8 Å, respectively. The R s of BBK32-(206 -354) calculated from K av was smaller than the R s expected for a globular protein of similar size. These results suggest that BBK32-(56 -205) and BBK32-(56 -354) have extended structures in solution, whereas BBK32-(206 -354) might interact with the chromatography matrix since the determined K av is larger than expected for a globular protein of similar molecular mass (see below).
We further analyzed the effects of increasing concentrations of the chaotropic agent GnHCl on the R s of the recombinant BBK32 subdomains. When gel permeation chromatography of BBK32-(56 -206) was conducted in the presence of increasing concentrations of GnHCl, we observed a continuous small decrease in the K av value of the eluted protein (Fig. 4). These data suggest that the N-terminal subdomain of BBK32 is largely disordered under physiological conditions and that denaturation of the protein does not dramatically change its R s . From these results we can also exclude the possibility that BBK32-(56 -205) forms a dimer under the conditions examined.
On the other hand the elution behavior of BBK32-(206 -354) was dramatically affected by GnHCl. In 1 M GnHCl, the protein eluted with a K av of 0.3, which is the elution position expected for a globular protein of the size of BBK32-(206 -354) (Fig. 4). Thus, we propose that the reduction in K av at 1 M GnHCl is due largely to a disruption of the hypothetical interaction between the protein and the chromatography matrix under native conditions (see above). Between 3 and 4 M GnHCl, the K av value of the eluted protein dropped dramatically, suggesting an unfolding of the globular structure. At 5 and 6 M GnHCl, the elution positions of the two recombinant subdomains were very similar, which is expected for fully denatured proteins of similar molecular weights.
Comparison (denaturing conditions) (Fig. 5). The intrinsic viscosity was calculated by extrapolating the functions (Equation 2, see "Experimental Procedures") to a zero protein concentration, shown in Fig. 5. The measured reduced viscosities and calculated intrinsic viscosity for BBK32-(56 -205) were essentially identical in the presence and absence of 4 M urea. This result again demonstrates that this domain has a largely disordered structure under native conditions. BBK32-(206 -354), on the other hand, under native conditions behaved as a globular protein with an intrinsic viscosity that was significantly lower compared with the intrinsic viscosity measured under denaturing conditions.
The N-terminal Domain of BBK32 Is Responsible for the MSCRAMM Fibronectin Binding Activity and Changes Its Conformation on Ligand Binding-We examined the ability of the recombinant BBK32 proteins to bind to Fn-coated microtiter wells. The results of these experiments (Fig. 6) showed that BBK32-(56 -354) and BBK32-(56 -205) bound to Fn in a process dependent on the concentration of the MSCRAMM added. The Fn-coated wells were saturated with bound MSCRAMM when the two forms of recombinant BBK32 were added at concentrations of 50 and 500 nM, respectively. Half maximal binding was reached at 10 nM BBK32-(56 -354) and 100 nM BBK32-(56 -205). No Fn binding was observed for BBK32-(206 -354) or Ace19, a collagen binding MSCRAMM from Enterococcus faecalis (41), which was used as a control protein.
The results presented above demonstrate that an unstructured subdomain in BBK32 is responsible for the MSCRAMM Fn binding activity. It is not clear why the full-length protein BBK32-(56 -354) appears to bind more avidly to Fn than the N-terminal region BBK32-(56 -205) in this assay.
Binding of BBK32 to Fn Involves the Formation of Additional ␤-Sheet Structure-The FnBMs from Gram-positive bacteria have been shown in earlier reports to contain a similar unstructured subdomain that binds Fn. In these interactions, ligand binding is associated with a conformational rearrangement of unstructured repeat region in the MSCRAMMs (33,42,43). In attempts to determine whether the unstructured N-terminal subdomain of BBK32 also undergoes conformational change upon ligand binding, we compared the far UV CD spectra of the MSCRAMM subdomains in the presence or absence of Fn ligand. We have found previously 2 that the NTD of Fn, generated by thermolysin digestion of the glycoprotein, binds to BBK32, and this Fn fragment was used in the studies. The far UV spectra of BBK32-(56 -205), BBK32-(206 -354), or FnbpA-(D123) were recorded in the absence or presence of the NTD of Fn (molar ratio 1:1). The far UV CD spectrum of the NTD of Fn in the absence of MSCRAMM was also recorded.
A CD spectrum representing the conformational changes induced in the MSCRAMM on ligand binding was obtained by subtracting the spectra of the free forms of the MSCRAMM and ligand components from the spectrum recorded for the MSCRAMM-ligand complex. Using this approach we found that the spectra of the BBK32-(56 -205)-Fn (NTD) and the FnbpA-(D123)-Fn (NTD) complexes are significantly different from spectra derived by adding the spectra of the individual components (Fig. 7). On the other hand, the far UV spectrum of BBK32-(206 -354)-Fn (NTD) mixture was identical to a spectrum derived by adding the spectra of the individual components (data not shown). These results further demonstrate that binding of BBK32-(56 -205) and FnbpA-(D123) to Fn is associated with a conformational change in the interacting components.
To determine the nature of the conformational change induced in BBK32 on binding to Fn, the resulting spectra were further analyzed using deconvolution programs. The results of the CD analysis suggested that the conformational change involved primarily is an increase in ␤-sheet character (Table  III). The structure of an MSCRAMM peptide in complex with an F1 module pair ( 1-2 F1) from the NTD of Fn has recently been solved (43). The interaction involves the formation of novel ␤-strands in the interacting MSCRAMM, whereas there is no significant change in the secondary structure of the F1 modules. Because the CD spectra recorded for the BBK32-and FnbpA-Fn complexes are consistent with the formation of additional ␤-sheet structure, it is tempting to speculate that these MSCRAMMs bind Fn by a common mechanism. DISCUSSION Many pathogenic microorganisms have been shown to interact with Fn. These interactions have in some cases been characterized, and the participating microbial components have been identified (for reviews, see Refs. 16, 17, and 44 -46). Perhaps the best characterized Fn binding adhesins are FnbpA from Staphylococcus aureus and protein F1 (or SfbI) from S. pyogenes (17). It was recently proposed that the C-terminal region of both proteins is composed of tandemly repeated Fn binding units (43). The proposed organization represents an extension and revision of an earlier model, which was based on sequence comparison rather than the recent functional analysis (38,45,47,48). These Fn binding repeats appear to have a disordered structure in the absence of ligand (33,49). Ligand binding is proposed to proceed by a tandem ␤-zipper mechanism, where short segments in the repeats form ␤-strands, complementing the ␤-sheets in the F1 modules of the NTD of Fn (48). The conformation in FnbpA induced on Fn binding creates neo-epitopes that are recognized by specific monoclonal antibodies and also by antibodies present in the sera of patients previously diagnosed with staphylococcal infections (36,50). These ligand-induced ligand-binding site-specific antibodies do not inhibit Fn binding but appear to stabilize the Fn-MSCRAMM complex.
BBK32 is a lipoprotein located on the surface of B. burgdorferi. The protein contains a "lipo-box" in the N-terminal segment that allows covalent attachment to a triglyceride and insertion into the outer membrane of the spirochete. Thus, the orientation of BBK32 is opposite to that of FnbpA and protein F1 (which are covalently attached to the bacterial cell-wall via an LPXTG motif at the C terminus). The experiments described in this report indicate that BBK32 is composed of two major domains. Immediately after the cell surface-anchoring motif (in the N-terminal region of the protein) follows a segment of BBK32 that has a disordered structure, based on secondary structure predictions. Gel permeation chromatography and intrinsic viscosity measurements under native and denaturing conditions confirmed the proposed disordered conformation for this domain of BBK32. This N-terminal domain of the MSCRAMM is primarily responsible for BBK32 Fn binding activity. A second domain located outside the disordered structure has a globular conformation rich in ␣-helix and does not appear to bind Fn on its own. Binding of the unstructured domains in BBK32 to Fn results in a conformational change in the MSCRAMMs, and the conformational change induced on ligand binding was detected by comparing the CD spectra of the complex with those of the isolated interacting components. Deconvolution of the differentiated spectra suggests that NTD binding to BBK32 (and Fn-bpA) involves the formation of additional ␤-strands. This observation is consistent with the tandem ␤-zipper model of ligand binding proposed for the FnBMs of Gram-positive bacteria (43) and raises the possibility that BBK32 also uses a similar mechanism of Fn binding. However, BBK32 does not contain any obvious repeated elements, and the valence of ligand binding for this Fn binding MSCRAMM has not been determined. Although there are some sequence similarities between the Fn binding motifs in the FnBMs of Gram-positive bacteria and BBK32 (25), the overall amino acid sequences of BBK32 and the FnBMs from Gram-positive bacteria are very different. Nevertheless, the spirochetes and the Gram-positive pathogens appear to have solved their needs to bind Fn by mechanistically related processes.
There is another curious similarity between the spirochete and Gram-positive Fn binding MSCRAMMs. FnbpA from S. aureus contains an N-terminal globular domain (the A domain) that binds fibrinogen with high affinity (31) and is located "outside" the disordered Fn binding region. The N-terminal region of protein F1 from S. pyogenes (M15) has also been shown to have fibrinogen binding activity (51,52). Remarkably, recent studies in our laboratory show that the C-terminal globular domain of BBK32 can also bind fibrinogen. 3 Thus, BBK32, FnbpA, and protein F1 appear not only to bind Fn by mechanisms that show significant similarities but to also contain fibrinogen binding activity in other regions of the protein. It is tempting to speculate that the consequences of borrelial, staphylococcal, and streptococcal Fn binding are similar and that BBK32 also could mediate Fn-dependent mammalian cell invasion, as has previously been demonstrated for FnbpA and protein F1 (16,22,25,53). These hypotheses are currently being examined in our laboratories.