The Fibrinogen-binding MSCRAMM (Clumping Factor) of Staphylococcus aureus Has a Ca 2 1 -dependent Inhibitory Site*

The clumping factor (ClfA) is a cell surface-associated protein of Staphylococcus aureus that promotes binding of fibrinogen or fibrin to the bacterial cell. Previous studies have shown that ClfA and the platelet integrin a IIb b 3 recognize the same domain at the extreme C ter- minus of the fibrinogen g -chain. a IIb b 3 interaction with this domain is known to occur in close proximity to a Ca 2 1 -binding EF-hand structure in the a -subunit. Analysis of the primary structure of ClfA indicated the presence of a potential Ca 2 1 -binding EF-hand-like motif at residues 310–321 within the fibrinogen-binding domain. Deletion mutagenesis and site-directed mutagenesis of this EF-hand in recombinant truncated ClfA proteins (Clf40, residues 40–559; and Clf41, residues 221–559) resulted in a significant reduction of affinity for native fibrinogen and a fibrinogen g -chain peptide. Further-more, Ca 2 1 (or Mn 2 1 ) could inhibit the binding of the fibrinogen g -chain peptide to Clf40-(40–559) and the adhesion of S. aureus cells to immobilized fibrinogen with an IC 50 of 2–3 m M . In contrast, Mg A mutants with native fibrinogen. Purified wild-type ( W.T ) and mutant proteins were incubated with immobilized fibrinogen followed by anti-Clf41-(221–559) antibodies. Binding of the primary antibody was detected using horseradish per-oxidase-labeled protein A and quantitated by absorbance at 450 nm. Binding of Clf41-(221–559) to fibrinogen was assigned a value of 100. of Clf41-(221-559)

Staphylococcus aureus causes a wide range of opportunistic infections that range from superficial skin infections to lifethreatening diseases including endocarditis, pneumonia, and septicemia. Adherence of bacteria to host matrix components that is mediated by bacterial surface adhesins is the initial critical event in the pathogenesis of most infections. The extracellular matrix (ECM) 1 contains numerous glycoproteins and proteoglycans assembled into insoluble matrices that serve as substrata for the adhesion and migration of tissue cells. These processes involve integrins, a family of heterodimeric (␣␤) cellsurface receptors that recognize specific ECM proteins. It has become increasingly evident that bacteria, including S. aureus, also utilize the ECM as substrata for their adhesion by way of a family of adhesins called MSCRAMM (microbial surface components recognizing adhesive matrix molecules) (1) that specifically recognize host matrix components.
One important component of the ECM, also occurring in soluble form in blood plasma, is fibrinogen, a 340-kDa hexamer composed of 2␣-, 2␤-, and 2␥-chains linked by disulfide bonds. This protein is recognized by several integrins including the platelet integrin ␣ IIb ␤ 3 . Activation of platelets and integrin ␣ IIb ␤ 3 results in fibrinogen-dependent aggregation in vitro and the formation of platelet-fibrin thrombi in vivo.
S. aureus contains several fibrinogen-binding proteins, one of which (clumping factor, ClfA) is primarily responsible for the clumping of bacteria in fibrinogen solutions and bacterial adherence to fibrinogen substrata (2). The gene encoding the fibrinogen-binding protein of S. aureus has been cloned, sequenced, and characterized in our laboratory (2). The clfA gene encodes a 933-amino acid protein that contains structural features characteristic of many cell surface-associated proteins from Gram-positive bacteria including a typical cell wall attachment region comprising an LPDTG motif, a hydrophobic transmembrane sequence, and a positively charged C terminus ( Fig. 1). In addition, the protein contains a repeat sequence (region R) of 308 alternating aspartate and serine residues located just outside the cell wall attachment region. Region R is required for the surface display of the 520-amino acid-long region A, which contains the fibrinogen-binding domain (3,4). Recombinant region A bound fibrinogen and strongly inhibited bacteria-fibrinogen interactions, as did anti-region A antibodies. Analysis of PCR-generated truncated proteins localized the binding domain to between residues 221 and 559 (3).
In earlier studies, Hawiger and co-workers (5,6) showed that a synthetic peptide mimicking the extreme C terminus of the fibrinogen ␥-chain inhibited fibrinogen-induced clumping of S. aureus cells. Recently, it was shown that purified recombinant ClfA protein specifically recognized these amino acid residues and that a synthetic peptide corresponding to this domain effectively inhibited binding of fibrinogen to recombinant ClfA (7). Interestingly, this same synthetic peptide also interacts with the ␣-subunit of the platelet integrin ␣ IIb ␤ 3 (8,9). The ␥-chain-binding site has been mapped to a region of the ␣ IIb polypeptide that contains a sequence motif resembling the Ca 2ϩ -binding EF-hand motif found in many eukaryotic Ca 2ϩbinding proteins (10). The EF-hand motif consists of 13 residues, with coordination typically supplied by oxygenated resi-dues at positions 1, 3, 5, 7, and 12 and by a solvent molecule hydrogen-bonded to residue 9 ( Fig. 2A) (11). These residues form a coordination sphere for the cation and are flanked by ␣-helices. Cooperative binding of multiple Ca 2ϩ ions is not unusual, and more than one Ca 2ϩ -binding motif often can be found within the same protein. Analysis of the ␣-subunit of ␣ IIb ␤ 3 revealed the presence of four functional motifs similar to EF-hands, although they lack an oxygenated residue at position 12 ( Fig. 2B) (31,39). Chemical cross-linking experiments provided direct evidence for the role of an ␣ IIb EF-hand-like sequence in fibrinogen binding (10) (Fig. 2B). In addition, a peptide corresponding to this EF-hand sequence bound to fibrinogen in a divalent cation-dependent manner (12). Taken together, these data provide evidence for the involvement of EF-hand-like sequences in the divalent cation-dependent binding of the fibrinogen C-terminal ␥-chain peptide. Further biochemical characterization of integrin ␣ IIb ␤ 3 has demonstrated that Ca 2ϩ binds to two distinct classes of sites: high affinity binding sites that promote ligand binding and low affinity binding sites that inhibit ligand binding (13).
In this report, we propose a mode of interaction between ClfA and the fibrinogen ␥-chain peptide that exhibits some similarities to ␣ IIb ␤ 3 -ligand interactions. A potential divalent cationbinding EF-hand motif was identified in ClfA (Figs. 1 and 2C). It differs from the EF-hand consensus (Fig. 2, A and B) at only one residue, a non-cation-coordinating residue. We demonstrate that region A of ClfA can bind Ca 2ϩ and that the interaction between region A and fibrinogen is inhibited by millimolar concentrations of Ca 2ϩ . In addition, we show using far-UV spectroscopy that Ca 2ϩ ions affect the secondary structure of the fibrinogen-binding region of ClfA. Site-specific mutants of ClfA with a modified EF-hand were generated and shown to have a lower affinity for fibrinogen compared with the wild type. In addition, the effects of Ca 2ϩ were reduced in these mutant proteins. Together, these studies indicate that Ca 2ϩ plays a regulatory role in the interaction of fibrinogen and region A of ClfA.

Bacteria and Growth Conditions
Escherichia coli XL1-Blue (14) was used as the bacterial host for plasmid cloning and protein expression. E. coli cells harboring plasmids were routinely grown in L-broth, Terrific broth, and L-agar (15). Ampicillin (100 g/ml) was incorporated as appropriate. S. aureus strain Newman was grown in Trypticase soy broth or agar.

Manipulation of DNA
Restriction and DNA modification enzymes were purchased from New England Biolabs Inc. or Promega and were used according to the manufacturers' instructions. DNA manipulations were performed using standard procedures (15).

Amplification of clfA Gene Fragments
PCR, with the oligonucleotides listed in Table I, was used to amplify specific clfA fragments from chromosomal DNA of S. aureus strain Newman. Genomic DNA was isolated as described previously (16). The oligonucleotides contained restriction enzyme cleavage sites at their 5Ј-ends to facilitate directional cloning. PCR was performed with a Perkin-Elmer DNA thermocycler. Reaction mixtures contained 50 ng of target DNA, 100 pmol of forward and reverse primers, 10 mM KCl, 10 mM (NH 4 ) 2 SO 4 , 20 mM Tris-HCl (pH 8.0), 2 mM MgSO 4 , 250 M each dNTP, 0.1% Triton X-100, and 2 units of Vent DNA polymerase (New England Biolabs Inc.). The reaction mixtures were overlaid with 100 l of mineral oil and amplified for 25 cycles consisting of 1 min of denaturation at 94°C, a 1-min annealing period at a temperature depending on the primers used, and an extension of 1 min, 30 s at 72°C. On completion of the 25 cycles, the reaction mixture was incubated at 72°C for 10 min. After amplification, the PCR products were treated with the Wizard PCR-preps DNA purification system (Promega) and analyzed by agarose gel electrophoresis.

Construction of Expression Plasmids
Amplified fragments of the clfA gene were cloned into the expression plasmid pQE30 (QIAGEN Inc.) to generate the constructs pCF40-(40 -559) and pCF41-(221-559). Recombinant protein expressed from this vector contains an N-terminal extension of six histidine residues (His tag). These fragments of the clfA gene were also cloned into the expression plasmid pGEX-KG such that an in-frame fusion was formed with glutathione S-transferase (17).

Construction of a Deletion Mutant Plasmid
Residues 310 -321 of ClfA corresponding to a putative Ca 2ϩ -binding EF-hand were deleted in Clf40-(40 -559) and Clf41-(221-559). The deletion mutant plasmids were designated pCF51 and pCF64, respectively. Construction of pCF51 involved two separate PCRs. The first reaction amplified DNA encoding residues 40 -309 using a reverse primer (DOCR3) with an XbaI site incorporated at the 5Ј-end and a forward primer (DOCF1) incorporating a BamHI site. The second reaction amplified DNA encoding residues 332-559 using a forward primer (DOCF2) incorporating an XbaI site and a reverse primer (DOCR1) incorporating a HindIII site. Following cleavage by XbaI and either BamHI or HindIII (depending on the reaction), the two products were ligated together and cloned into the expression vector pQE30 as de-  (11,38). Acceptable residues are in parentheses; unacceptable residues are in braces. X indicates any residue. The boldface letters indicate cation-coordinating residues. B, metal ion-binding sites in the ␣ IIb subunit of platelet integrin (adapted from Gulino et al. (39)). The asterisk indicates sequence identified as the fibrinogen Cterminal ␥-chain-binding site in ␣ IIb (10). C, divalent cation-binding motif in region A of ClfA. The boldface letters indicate putative cationcoordinating residues. scribed above. Creation of the XbaI site introduced a serine and an arginine residue at the site of deletion. pCF64 was constructed in the same way using the appropriate forward primer (F5) in the first reaction.

Site-directed Mutagenesis
Plasmid DNA purified from E. coli XL1-Blue cells containing pCF40-(40 -559) and pCF41-(221-559) served as the template in the mutagenesis studies. The residues were mutated to alanine as single, double, or quadruple mutations (Table I). A novel method of mutagenesis was developed that involved two separate PCRs. 2 The first reaction employed a flanking reverse primer (incorporating a HindIII site) and a forward primer that introduced the nucleotide mismatch required for the desired mutation and, in addition, incorporated a novel restriction site at the 5Ј-end of the oligonucleotide that would not affect the amino acid sequence of the final gene product. In the second reaction, a reverse primer incorporating the same silent restriction site mutation was used in conjunction with a flanking forward primer (with a BamHI site). Both gene fragments were digested at the common restriction site, ligated, digested with BamHI and HindIII, cloned into the expression vector pQE30, and transformed into E. coli XL1-Blue cells. Transformants were screened for the proper plasmid construction. The DNA sequence was verified by the dideoxy termination method (15) using ␣-35 S-dATP (Amersham Corp.) and Sequenase 4.0 (U. S. Biochemical Corp.). The oligonucleotides used to construct the site-directed mutants are listed in Table I.

Expression and Purification of Recombinant Proteins
Recombinant plasmids were transformed into E. coli XL1-Blue cells and expressed as described previously (3). Fusion proteins containing the His tag were purified by immobilized metal chelate affinity chromatography. A iminodiacetic acid-Sepharose 6B Fast Flow column (10 ϫ 1 cm; Sigma) was charged with 150 mM Ni 2ϩ and equilibrated with binding buffer (5 mM imidazole, 0.5 M NaCl, and 20 mM Tris-HCl (pH 7.9)). The cleared lysed cell supernatant was applied to the column and washed with binding buffer until the absorbance at 280 nm of the eluate was Ͻ0.001. Bound protein was eluted with a continuous linear gradient of imidazole (5-100 mM; total volume of 200 ml) in 0.5 M NaCl and 20 mM Tris-HCl (pH 7.9). Protein-containing fractions were dialyzed against 50 mM NaCl, 2 mM EDTA, and 20 mM Tris-HCl (pH 7.4) and applied to a Q-Sepharose column pre-equilibrated with the same buffer. Bound protein was eluted with a continuous linear gradient of NaCl (50 -500 mM; total volume of 200 ml) in 20 mM Tris-HCl and 2 mM EDTA (pH 7.9). Eluted fractions were monitored by absorbance at 280 nm, and peak fractions were analyzed by SDS-polyacrylamide gel electrophoresis and Western immunoblotting. The glutathione S-transferase-Clf41-(221-559) fusion protein was purified by glutathione-Sepharose (Pharmacia Biotech Inc.) affinity chromatography and cleaved with bovine thrombin as described previously (3). The recombinant MSCRAMM fragment was isolated after passing the digest through a glutathione-Sepharose column, followed by ion-exchange chromatography on a Q-sepharose column as described above.

Analysis of Fibrinogen-binding Activity of Recombinant Proteins
Western Ligand Blot Assay-Fibrinogen-binding proteins fractionated by SDS-polyacrylamide gel electrophoresis and transferred to a nitrocellulose membrane were detected using fibrinogen conjugated to horseradish peroxidase and enhanced chemiluminescence (Amersham Corp.) as described previously (3).
Bacterial Adherence Assay-The adherence of S. aureus Newman cells expressing ClfA to microtiter wells coated with fibrinogen was assayed as described previously (6).
Enzyme-linked Immunosorbent Assay-The ability of recombinant protein to bind to fibrinogen was analyzed using the enzyme-linked immunosorbent assay. Microtiter wells (Sarstedt, Inc.) were coated with 5 g/ml fibrinogen (Kabi Pharmacia/Chromogenix) in coating solution (0.02% sodium carbonate buffer (pH 9.6)) for 18 h at room temperature. The plates were washed three times with PBS, 0.05% Tween 20, and 0.1% bovine serum albumin (PBS-TB). A solution of 2.5% bovine serum albumin and 0.05% Tween 20 in PBS was added to the wells to block any remaining protein-binding sites. After 1 h at 37°C, the wells were washed again three times with PBS-TB, and purified recombinant protein in PBS was added and incubated for 2 h at 37°C. The wells were again washed with PBS-TB and incubated with polyclonal antiserum raised against Clf41-(221-559), diluted 1:800, for 1 h. After further washing, 100 l of horseradish peroxidase-labeled protein A (1:1000; Sigma) was added. Following incubation for 1 h at 37°C and washing with PBS-TB, 100 l of chromogenic substrate (580 g/ml tetramethylbenzidine and 0.0001% H 2 O 2 in 0.1 M sodium acetate buffer (pH 5.0)) was added per well and developed for 10 min, and the reaction was stopped by the addition of 50 l of 2 M H 2 SO 4 . Plates were read at 450 nm in an enzyme-linked immunosorbent assay plate reader (Lab-

GCCGACGCCATTTGCCAATACTTGATCTCCAGC
An Inhibitory Ca 2ϩ -binding Site in S. aureus Clumping Factor systems Multiskan Plus). Fluorescence Polarization-A fluorescence polarization assay was developed to determine the equilibrium constants for the interaction of the recombinant proteins with a synthetic peptide conjugated with a fluorescent probe. This peptide consisted of the 17 C-terminal residues of the ␥-chain of fibrinogen (i.e. GEGQQHHLGGAKQAGDV) and was synthesized by a solid-phase method on a p-benzyloxybenzyl alcohol resin using Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry and a Model 396 multiple peptide synthesizer (Advanced ChemTech Inc.). The peptide was labeled with fluorescein as follows. The peptide (1 mg in H 2 0) was incubated with fluorescein succinimidyl ester (1 mg dissolved in 100 l of dimethyl sulfoxide) at 37°C for 60 min in the presence of coupling buffer (100 mM KH 2 PO 4 (pH 7.0)). The reaction was quenched by the addition of 100 l of 1 M Tris-HCl (pH 8.0), vortexed, and left at room temperature for 30 min. The fluoresceinated peptide was fractionated by reverse-phase chromatography on a Delta-Pak C 18 Radial-Pak cartridge HPLC column connected to a Waters 486 multi-wavelength detector. The labeling procedure can yield three forms of fluoresceinated peptide with the probe attached to 1) the N-terminal amino group, 2) the amino group of the internal lysine residue, and 3) the amino groups in both positions. Fractionation by HPLC yielded two major and one minor fluorescein-containing peak. Substitutions at the internal lysine residue should yield a peptide resistant to trypsin digestion. The separated peptides were incubated with trypsin, followed by HPLC analysis. Only one of the original major peaks contained peptide susceptible to trypsin (data not shown). This peptide, which contains the fluorescein probe linked to the N-terminal amino group, was used in the fluorescence polarization studies. A scrambled peptide consisting of the 17 C-terminal residues of the ␥-chain in a random sequence (i.e. GHEHGLQGQGAVKDGAQ) was used as a control.
Recombinant protein in 10 mM Tris-HCl (pH 7.4) was incubated with 10 nM peptide in the dark at 20°C for 3 h. Samples were then analyzed using a Beacon fluorescence polarization system. Binding curves were analyzed by nonlinear regression used to fit a binding function defined as in Equation 1, where ⌬P corresponds to the change in fluorescence polarization, ⌬P max is the maximum fluorescence polarization change, and K D is the dissociation constant of the interaction. A single ligand-binding site was assumed in this analysis.

Analysis of Structure by Circular Dichroism Spectroscopy
The average secondary structures of recombinant proteins, in the presence and absence of metal ions, were monitored by CD spectroscopy on a Jasco J720 spectropolarimeter calibrated with 0.1% (w/v) 10camphorsulfonic acid-d solution. CD spectra were measured at 25°C in a 0.2-mm path length quartz cell, and six scans from 250 to 180 nm (far-UV) were generated and averaged. Protein concentrations, determined using an extinction coefficient at 280 nm (31,320 M Ϫ1 cm Ϫ1 ), were typically 20 M in 10 mM Tris-HCl (pH 7.4). Molar ellipticity is expressed in degrees⅐cm 2 ⅐mol Ϫ1 .

Effects of Cations on the Interaction between ClfA and the
Fibrinogen ␥-Chain Peptide (Residues 396 -412)-To examine a possible effect of divalent cations on the ClfA-fibrinogen interaction, a fluorescence polarization assay was developed using recombinant forms of ClfA encompassing full-length region A (Clf40, residues 40 -559) or the smallest truncated protein that maintained fibrinogen-binding activity (Clf41, residues 221-559) and the fluorescently labeled C-terminal ␥-chain peptide. Since fibrinogen is known to be a Ca 2ϩ -binding protein (23), we chose to use the ␥-chain peptide as a ligand and assumed that the observed effects of metal ions would reflect interactions with ClfA. We have previously shown that the fibrinogen ␥-chain peptide effectively inhibits interaction between ClfA and native fibrinogen (7).
In the absence of added metal ions, Clf40-(40 -559) bound the ␥-chain peptide with a dissociation constant of 20.8 Ϯ 2.5 M (Fig. 3). An unlabeled ␥-chain peptide inhibited the binding of the fluorescent peptide to ClfA in a concentration-dependent manner, whereas a scrambled version of the peptide had no effect at similar concentrations (Fig. 3, inset). Inclusion of Ca 2ϩ (5 mM) significantly reduced the binding affinity for the peptide (K D Ͼ 200 M) (Fig. 3). Incubation of Clf40-(40 -559) with 5 mM Mn 2ϩ caused complete loss of ␥-chain peptide-binding activity. On the other hand, Mg 2ϩ or Na ϩ at similar concentrations had no effect on protein-peptide interaction (Fig. 4A). In addition, Mg 2ϩ (5 mM) did not counteract the inhibitory effect of Ca 2ϩ (data not shown). Incubation of Clf40-(40 -559) with EDTA (5 mM) increased the affinity of MSCRAMM for the ␥-chain peptide by 2-fold (K D ϭ 11.5 Ϯ 0.6 M) compared with non-EDTAtreated protein (K D ϭ 20.8 Ϯ 2.5 M). It is likely that contaminating Ca 2ϩ and Mn 2ϩ in the assay solutions and protein preparations contribute to inhibition of the protein-peptide interaction. Inclusion of EDTA removes these ions. Furthermore, these data show that divalent cations are not required to promote ␥-chain peptide binding. Results similar to those described for Clf40-(40 -559) were obtained when these experiments were repeated with the smaller construct, Clf41-(221-559). Thus, in the absence of added metal ions, Clf41-(221-559) bound the ␥-chain peptide with a dissociation constant of 15.0 Ϯ 1.1 M (Fig. 3), and the addition of 5 mM EDTA slightly reduced the K D to 11.1 Ϯ 0.4 M. When these experiments were repeated with Clf41-(221-559) without an N-terminal extension of histidine residues, identical results were obtained, indicating that the effects of cations on region A function are independent of the purification tag (data not shown).
The interaction of Clf40-(40 -559) with a set amount of fluoresceinated ␥-chain peptide across a concentration range of Ca 2ϩ and Mg 2ϩ was also measured. With Mg 2ϩ , the binding of the ␥-chain peptide to Clf40-(40 -559) was unaffected across the entire concentration range tested. In contrast, concentrations of Ca 2ϩ above 2 mM inhibited the interaction (Fig. 4A). At concentrations greater than 5 mM, very little binding of the ␥-chain peptide to region A was observed. Very similar results were obtained when the effects of different cations on S. aureus cell adhesion to immobilized fibrinogen were measured (Fig.  4B). Concentrations of Ca 2ϩ above 2 mM inhibited bacterial adhesion to fibrinogen, whereas Mg 2ϩ did not affect cell adhesion throughout the concentration range analyzed. These results indicate the existence of an inhibitory Ca 2ϩ -binding site in region A of ClfA with an apparent K D of 2.5 mM, a value similar to the physiological concentration of Ca 2ϩ (ϳ2.5 mM) present in normal human sera. Thus, under some physiological conditions, this site may be occupied. Inclusion of Mg 2ϩ had no effect on Ca 2ϩ -induced inhibition (data not shown), suggesting that Mg 2ϩ is unable to compete with Ca 2ϩ for binding to the inhibitory binding site and that there is a certain degree of specificity for Ca 2ϩ .
Effect of Ca 2ϩ Ions on the Structure of Region A: Circular Dichroism Analysis-Circular dichroism spectroscopy was used to investigate if the binding of Ca 2ϩ ions affects the secondary structures of the Clf40-(40 -559) and Clf41-(221-559) proteins. Cation binding by integrins has been shown to be associated with conformational changes. For example, expression of the epitope recognized by monoclonal antibody 24 on ␣ L ␤ 2 was dependent on Mg 2ϩ , and monoclonal antibody 24 expression correlates with the ability to bind ligand (24,25). The CD spectra of Clf41-(221-559) were dominated by a large minimum at 215 nm. Inclusion of Ca 2ϩ and Mn 2ϩ had a reproducible effect on the protein far-UV CD spectra at ϳ200 nm (Fig. 5). This effect on secondary structure was dependent on the concentration of divalent cation and on the particular cation used. Ca 2ϩ altered the CD spectra of Clf41-(221-559) by reducing the signal at 200 nm in a concentration-dependent manner (Fig. 5). The presence of Mn 2ϩ resulted in a qualitatively similar change, although the effect was more pronounced (data not shown). Identical results were obtained with Clf41-(221-559) without an N-terminal extension of six histidine residues, indicating that the effects of Ca 2ϩ and Mn 2ϩ on the secondary structure of region A are independent of the purification tag (data not shown).
The effect of Ca 2ϩ and Mn 2ϩ on secondary structure correlates with the effect that these metal ions have on the ability of the protein to bind the C-terminal ␥-chain peptide. These data indicate that when Ca 2ϩ and Mn 2ϩ bind to region A of ClfA, alterations to the secondary structure of the protein occur, with Mn 2ϩ having the largest effect. These structural alterations may be responsible for the inhibition of ligand binding. ␥-Chain Peptide-The role of the putative Ca 2ϩ -binding EFhand at residues 310 -321 in ligand binding was investigated using deletion mutagenesis and site-directed mutagenesis. Deletion of amino acids 310 -321 from Clf40-(40 -559) and Clf41-(221-559) resulted in complete loss of fibrinogen-binding activity. These mutant proteins did not bind fibrinogen in a Western ligand blot assay (data not shown). They also failed to bind to the fibrinogen C-terminal ␥-chain peptide in a fluorescence polarization assay (Fig. 7A).

Interaction of ClfA Region A Mutants with the Fibrinogen
The EF-hand motif was further investigated by mutating putative cation-coordinating residues in Clf40-(40 -559) and Clf41-(221-559) and characterizing the interactions of these mutant proteins with the fluorescently labeled C-terminal ␥-chain peptide. The amino acid residues investigated were replaced with alanine, a residue that was not expected to interfere with existing secondary structure, but would be unable to coordinate cations. The corresponding base changes were made using a novel PCR mutagenesis technique as described under "Experimental Procedures." The recombinant proteins containing the mutations were purified to homogeneity by metal ion chromatography and anion exchange chromatography. Structural analysis of the isolated mutant proteins by CD spectroscopy indicated differences in secondary structure resulting from the introduction of mutations (Fig. 6). Interestingly, the observed effect of Ca 2ϩ on structure was significantly less in the mutants as compared with the wild-type protein (Fig. 6).
The effects of the mutations on the interaction of Clf40-(40 -559) with the fibrinogen ␥-chain peptide are shown in Fig. 7A. Similar effects were observed when the mutations were introduced into the smaller construct (Clf41-(221-559) (data not shown). Thus, it is apparent that substitutions within the putative EF-hand affect peptide-binding affinity. Substitution of four residues (D310A/D312A/T318A/D321A) exhibited the most dramatic decrease. Because of the very low affinity of this quadruple mutant for the peptide, the dissociation constant could not be measured accurately. The D310A/D312A double mutant bound the peptide over three times more weakly than the wild-type protein (K D ϭ 66.5 Ϯ 7.4 M). Mutation of the first aspartate in the EF-hand (D310A) had no effect on ␥-chain peptide-binding activity.
The effects of Ca 2ϩ on the interaction of the D310A single mutant and the D310A/D312A double mutant with the ␥-chain peptide were also measured. As shown in Fig. 7B, the degree of inhibition induced by the inclusion of Ca 2ϩ is significantly less with the mutants compared with the wild-type protein. These results show that the introduced mutations have (a) reduced the ability of region A of ClfA to bind to the fibrinogen ␥-chain peptide and (b) reduced the ability of Ca 2ϩ to inhibit this interaction. One interpretation of these data is that the inhibitory Ca 2ϩ -binding site and the fibrinogen ␥-chain peptidebinding site share contact points within amino acid sequence 310 -321.
Interaction of Region A Mutants with Fibrinogen-An enzyme-linked immunosorbent assay was developed (as described under "Experimental Procedures") to assess the effects of the mutations on the interaction of Clf41-(221-559) with native fibrinogen (Fig. 8). Deletion of the EF-hand motif or substitution of four residues within it had the severest effect on fibrinogen-binding ability. The D310A/D312A double mutant also displayed significant reduction in the ability to bind fibrinogen. The D310A single mutant reduced binding to 60% of the wild type. These results further implicate amino acids 310 -321 of region A as playing an important part in the ligand-binding function of ClfA. DISCUSSION The adhesion of microorganisms to host tissues is the critical first step in the series of events that lead to clinically manifested infections. It has become evident that eukaryotic adhesive ECM components that support adhesion of host cells also serve as ligands for pathogenic microorganisms (1). Fibrinogen, the blood plasma coagulation protein, is also found in the ECM and plays important roles in wound healing. During coagulation, fibrinogen is proteolytically converted to fibrin, which forms the structure of the blood clot. In addition, fibrinogen is the major blood protein deposited on implanted biomaterial (27). Immobilized fibrin/fibrinogen in a blood clot, in the ECM, or present on the surface of biomaterial can serve as a substrate for the adherence of S. aureus cells (28,29).
S. aureus has long been known to form clumps in the presence of blood plasma. Hawiger et al. (5) identified fibrinogen as the plasma protein responsible for this phenomenon. Further study revealed that the domain of fibrinogen that interacts with the fibrinogen receptor of S. aureus is located at the carboxyl-terminal of the ␥-chain (6,7). Recently, this was confirmed by showing that purified ClfA protein binds to the extreme C terminus of the ␥-chain of fibrinogen and that a synthetic C-terminal ␥-chain peptide fully inhibits this interaction (7). The 12 residues at the C terminus of the fibrinogen ␥-chain also bind to the integrin receptor (␣ IIb ␤ 3 ) on the surface of platelets, resulting in aggregation in vitro and the formation of platelet-fibrin thrombi (9, 18 -20, 30). These residues mediate initial contact with nonstimulated platelets and on activation are sufficient to promote stable adhesion to fibrinogen (21). The mode of interaction between fibrinogen and ␣ IIb ␤ 3 has been studied in some detail. The ␥-chain peptide of fibrinogen binds in a divalent cation-dependent manner to a region of ␣ IIb corresponding to an EF-hand-like sequence (10,12). A requirement of divalent cations for ligand binding to integrins is often observed. However, inhibition of integrin-ligand binding by Ca 2ϩ has also been reported (13,25,(32)(33)(34), although there is no evidence to indicate that Ca 2ϩ inhibits the interaction between ␣ IIb ␤ 3 and the fibrinogen ␥-chain peptide. Recently, Hu et al. (13) identified two classes of cation-binding sites in ␤ 3containing integrins: sites that, when occupied by Ca 2ϩ , promote ligand binding and sites that inhibit ligand binding. The location of the inhibitory cation-binding site(s) has not been identified.
Analysis of the primary structure of region A of ClfA identified a potential divalent cation-binding EF-hand motif at residues 310 -321, which differs from the EF-hand consensus at only one residue, a non-cation-coordinating site ( Fig. 2A). Secondary structure analysis of the primary sequence of ClfA predicts this putative EF-hand to be flanked by ␣-helices, which are required for correct EF-hand conformation (22). The proposed cation-binding EF-hand motif lies within the minimum 329-residue segment of region A that retains fibrinogenbinding activity (residues 221-550) (3). Also present within the fibrinogen-binding domain is a putative MIDAS motif, a cationbinding sequence contained within the I-domain of integrins (35). Analysis of this motif by site-directed mutagenesis indicated that it plays a role in ClfA binding to fibrinogen. 2 The I-domain of an integrin-like protein from Candida albicans, which has homology to the I-domain of the leukocyte integrin ␣ M ␤ 2 , also has homology to the fibrinogen-binding domain of ClfA (36).
We have shown that Ca 2ϩ plays a role in the interaction between ClfA and fibrinogen. Preliminary experiments showed

FIG. 8. Interaction of region A mutants with native fibrinogen.
Purified wild-type (W.T) and mutant proteins were incubated with immobilized fibrinogen followed by anti-Clf41-(221-559) antibodies. Binding of the primary antibody was detected using horseradish peroxidase-labeled protein A and quantitated by absorbance at 450 nm. Binding of Clf41-(221-559) to fibrinogen was assigned a value of 100. Introduction of mutations or deletions did not significantly affect the ability of Clf41-(221-559) to bind the polyclonal antiserum raised against Clf41-(221-559) (as judged by quantitative Western blot analysis; data not shown). Data are expressed as means Ϯ S.E. of two independent experiments. that Ca 2ϩ at high concentrations prevented clumping of S. aureus cells in the presence of fibrinogen. In addition, clumping of bacteria (due to the interaction of ClfA and fibrinogen) could be reversed by the addition of Ca 2ϩ (data not shown). These effects were prevented by the addition of EGTA or EDTA. In the studies reported in this work, we have used purified components and a quantitative binding assay. Ca 2ϩ dramatically inhibits the interaction between the ␥-chain peptide and region A of ClfA. Mn 2ϩ is a more potent inhibitor than Ca 2ϩ , suggesting that Mn 2ϩ has a higher affinity for the inhibitory cationbinding site. Alternatively, Mn 2ϩ may bind to a different cation-binding site. The protein-peptide interaction is not affected by Mg 2ϩ or monovalent cations (Figs. 3 and 4A). Ca 2ϩ also inhibited the adherence of S. aureus cells to immobilized fibrinogen (Fig. 4B) at concentrations similar to those that affect the ␥-chain peptide interaction. These observations are consistent with the presence of an inhibitory site(s) in ClfA that allows Ca 2ϩ and Mn 2ϩ (but not Mg 2ϩ ) to bind. When Clf40-(40 -559) and Clf41-(221-559) were incubated with the concentrations of Ca 2ϩ that inhibited ligand binding, distinct differences in secondary structure were evident (Figs. 5 and 6). Thus, Ca 2ϩ binding to an inhibitory binding site(s) in region A of ClfA induces a conformational change in the ligand-binding site that apparently results in inhibition of the ClfA-ligand interaction. It is noted that inclusion of EDTA/EGTA slightly increased the affinity of the fibrinogen ␥-chain peptide for ClfA, indicating that divalent cations are not required to promote peptide binding. This feature represents a significant difference from the observed metal ion dependence of the ␣ IIb ␤ 3 -fibrinogen interaction (12).
Deletion mutagenesis and site-directed mutagenesis were used to identify the Ca 2ϩ -dependent inhibitory site. Deletion of the EF-hand (residues 310 -321) or substitution of four of the putative cation-coordinating residues resulted in almost complete loss of both fibrinogen-binding activity and ␥-chain peptide-binding activity (Figs. 7A and 8). A single substitution (D310A) and a double substitution (D310A/D312A) caused a significant reduction in ligand-binding affinity (Fig. 8). In addition, the inclusion of Ca 2ϩ with these mutant proteins only marginally affected their binding to the ␥-chain peptide (Fig.  7B). The proteins may have retained some ability to bind Ca 2ϩ , explaining the small reduction in activity in the presence of the ion. Incomplete EF-hand motifs in other proteins can chelate Ca 2ϩ , for example, the platelet integrin ␣ IIb ␤ 3 (31,37). The reduced inhibitory effect of Ca 2ϩ with the D310A and D310A/ D312A mutant proteins compared with the wild type indicates that the ion interacts with the inhibitory site less effectively. This conclusion is supported by the observation that Ca 2ϩinduced changes in secondary structure were significantly less in the mutants compared with the wild-type protein (Fig. 6). Taken together, these results implicate the EF-hand motif at residues 310 -321 as an inhibitory Ca 2ϩ -binding site. Substitution of four cation-coordinating residues or deletion of the EFhand motif should result in complete loss of cation-binding ability, but as these proteins have no fibrinogen-binding activity, inhibition of function by Ca 2ϩ and Mn 2ϩ could not be assessed.
In summary, these findings indicate that changes in amino acid sequence 310 -321 of region A compromise the protein's ability to bind fibrinogen and to bind Ca 2ϩ at an inhibitory cation-binding site. There are several possible explanations for this observation. One interpretation is that the inhibitory binding site and the fibrinogen ␥-chain peptide-binding site share contact points within sequence 310 -321. A model can be proposed in which Ca 2ϩ , at millimolar concentrations, binds to an inhibitory binding site within region A of ClfA that overlaps with the ligand-binding site (Fig. 9A). Occupation of this Ca 2ϩbinding site directly interferes with ligand binding. In an alternative model, the inhibitory site and the fibrinogen-binding site are physically distinct (Fig. 9B). Interaction of Ca 2ϩ with the inhibitory site induces a conformational change in the ligand-binding site that affects the affinity for fibrinogen. Further work is needed to differentiate between these models.
Does Ca 2ϩ have a role in the regulation of ClfA activity in vivo? The IC 50 determined for the interaction of S. aureus cells with fibrinogen is similar to the total concentration of Ca 2ϩ present in normal human sera, although the concentration of ionized Ca 2ϩ is lower (maintained between 1.0 and 1.3 mM) (40). Even at these lower concentrations, Ca 2ϩ may regulate ligand-binding function in a subset of ClfA molecules bound to the cell wall of S. aureus. Indeed, the concentration of Ca 2ϩ in the local environment of ClfA may vary significantly due to the protein's acidic aspartate-serine repeat sequence and the extent of the acidic cell wall and capsule, both of which associate with divalent cations. Thus, the ability of Ca 2ϩ to affect ClfA function may be a direct consequence of the protein's microenvironment on the cell surface, an environment that changes during in vivo growth. One can speculate that Ca 2ϩ -dependent regulation of ClfA activity prevents all of the receptors on intravascular S. aureus cells from being occupied by soluble fibrinogen, thus allowing the bacteria (under the right conditions) to adhere to solid-phase fibrinogen or fibrin clots. In addition, this process may allow cells to detach from the initial vegetation, allowing microbial proliferation.
We have shown that ClfA contains a class of Ca 2ϩ -binding sites that, when occupied, regulate fibrinogen-binding activity. Regulation of integrin-ligand binding by divalent cations is well documented, and this phenomenon represents an intriguing similarity between eukaryotic and prokaryotic adherence proteins. FIG. 9. Model depicting the interaction of region A of ClfA, the C terminus of the ␥-chain of fibrinogen, and Ca 2؉ . Occupation of the inhibitory site(s) by Ca 2ϩ prevents fibrinogen binding by inducing a conformational change in the ligand-binding site. The competitive inhibition of ligand binding by Ca 2ϩ indicates either that ligand (L) and Ca 2ϩ compete for the same site on ClfA (A) or that ligand and Ca 2ϩ bind to distinct sites that are mutually exclusive (B).