INTRODUCTION

The extracellular matrix (ECM)¹ is a ubiquitous structure that contributes to the architecture, elasticity, and rigidity of virtually all vertebrate tissues and organs. Within the last several decades, additional biological activities of the ECM have been identified. Distinct components of the ECM have been found to mediate one or several cellular events such as adhesion, proliferation, and regulation of gene expression (1, 2, 3, 4). These cell-ECM interactions, in turn, direct many physiological and pathological processes including development, wound healing, and tumor cell metastasis (5, 6, 7). It is now known that cell surface ECM receptors are key mediators of these biological events. Many ECM receptors belong to a family of dimeric receptor complexes called integrins (8, 9), although nonintegrin ECM receptors have been identified (10). In addition to eukaryotic cells, various pathogenic bacteria also interact specifically with the host ECM through cell surface ECM-binding molecules. ECM-binding molecules of pathogenic bacteria belong to a group of proteins known collectively as adhesins or microbial surface components recognizing adhesive matrix molecules (MSCRAMMs) and are widely believed to play important roles in key steps of disease pathogenesis (11, 12).

Among the many important pathogenic bacteria, few are as efficient in developing multiple resistance to antibiotics and causing a wide spectrum of diseases as Staphylococcus aureus. S. aureus is one of the causative agents of diseases such as infective endocarditis, osteomyelitis, aortitis, pneumonia, and scalded skin syndrome (13, 14, 15). Furthermore, several strains of S. aureus tend to extravasate into the circulation to cause bacteremia and subsequent formation of metastatic abscesses. These properties imply that S. aureus is capable of interacting with host tissue components. Consistent with this notion is the finding that S. aureus binds specifically to major ECM components such as collagen (16), fibronectin (17), laminin (18), proteoglycans (19), and elastin (20). Importantly, there is increasing evidence supporting a relationship between the ability of S. aureus to bind to ECM and its pathogenicity. For example, most clinical isolates of S. aureus demonstrate binding to fibronectin, and mutant strains defective in fibronectin binding have a decreased ability to colonize damaged heart valves in animal models of endocarditis (21). S. aureus binding to collagen has been implicated in osteomyelitis and septic arthritis (22), in which expression of the collagen adhesin has been found to be both necessary and sufficient for bacterial attachment to the type II collagen-rich cartilage. It has also been demonstrated in a murine experimental septic arthritis model that greater than 70% of animals injected with collagen adhesin-positive strains developed septic arthritis, whereas less than 25% of animals challenged with isogenic mutant strains lacking the collagen adhesin developed clinical symptoms of the disease (23). In addition, invasive S. aureus interacts with the basement membrane component laminin, whereas noninvasive Staphylococcus epidermidis shows no binding (18). Taken together, these observations indicate that S. aureus-ECM interactions are playing critical roles in targeting host tissues for attachment, colonization, and invasion.

Elastin is an important structural protein whose primary physiological role is to confer the property of reversible elasticity to tissues and organs (24). Elastin expression is highest in the lung, skin, and blood vessels, but elastin is widely expressed at lower levels in most mammalian tissues. In a previous study we showed that S. aureus binds to elastin with properties of reversibility, saturability, and specificity that suggested the presence of an elastin-binding protein on the bacterial surface (20). Using affinity chromatography techniques we were able to isolate a 25-kDa cell surface elastin-binding protein (named EbpS for lastin-inding rotein of aureus) that has been proposed to facilitate binding of the bacteria to elastin-rich host ECM. EbpS is structurally distinct from the mammalian cell surface elastin-binding protein and exhibits different binding specificity in terms of sequence recognition. Whereas the mammalian elastin-binding protein recognizes the hexapeptide sequence VGVAPG located in the C-terminal half of elastin (25), EbpS binds to a region in the N-terminal one-third of the molecule. In this study, we report the cloning, sequencing, and expression of ebpS. Our results confirm that EbpS is the cell surface molecule mediating binding of S. aureus cells to elastin and that EbpS is a novel protein. Characterization of elastin binding activity by various constructs of EbpS suggests that the N-terminal 59 amino acids of the molecule play an important role in elastin recognition.

EXPERIMENTAL PROCEDURES

Restriction endonucleases, calf intestinal alkaline phosphatase, T4 DNA ligase, T4 polynucleotide kinase, isopropyl--D-thiogalactopyranoside, 5-bromo-4-chloro-3-indolyl--D-galactoside, Wizard Miniprep plasmid purification kits, and HindIII-digested  DNA markers were purchased from Promega (Madison, WI). DNase-free RNase was obtained from Boehringer Mannheim. Luria-Bertani (LB) medium and LB agar medium capsules were from BIO 101 (La Jolla, CA). Tryptic soy broth (TSB) was obtained from Remel (Lenexa, KS). Na¹²⁵I, [-³²P]ATP, and [-³²P]CTP were from ICN (Costa Mesa, CA). Papain and protein A, immobilized to cross-linked agarose, and IODO-GEN were purchased from Pierce. Rapid-hyb buffer and Rediprime DNA labeling system were obtained from Amersham Corp. Chroma Spin-10 columns were purchased from Clontech (Palo Alto, CA). QIAexpress vector kit type IV and the Midi-Prep plasmid purification kit were obtained from Qiagen (Chatsworth, CA). Affi-Gel 10 affinity support was from Bio-Rad (Melville, NY). All other materials were purchased from Sigma.

S. aureus strain 12598 (Cowan) was purchased from the American Type Culture Collection (Rockville, MD). Escherichia coli strains DH5 (MAX Efficiency) and M15 (pREP4) were from Life Technologies, Inc. and Qiagen, respectively. M15 cells contain the plasmid pREP4, which constitutively expresses the Lac repressor from the lacI gene. S. aureus cells were grown in TSB, and E. coli strains were grown in LB media supplemented with appropriate antibiotics as described below.

The low copy number cloning plasmid, pHSG575 (26), was kindly provided by Dr. Michael Caparon (Department of Molecular Microbiology, Washington University School of Medicine). The plasmid pBluescript KS⁺ was purchased from Stratagene (La Jolla, CA) and used for subcloning and sequencing purposes. The expression plasmid pQE-30 was obtained from Qiagen. All of these plasmids were propagated in DH5 cells and purified using the Qiagen Plasmid Midi-Prep kit for further applications.

High molecular weight genomic DNA was isolated from 400 ml of an overnight culture of S. aureus strain 12598 cells by lysostaphin lysis, followed by treatment with DNase-free RNase and subsequent purification by phenol/chloroform and chloroform extractions. After the final chloroform extraction, DNA in the aqueous layer was precipitated with ethanol and dried.

A degenerate 30-mer oligonucleotide probe corresponding to the amino acid sequence NNFKDDFEKN was generated by chemical synthesis. The oligonucleotide was end-labeled with T4 polynucleotide kinase and [-³²P]ATP, and the radiolabeled oligonucleotide was separated from unincorporated ³²P by Chroma Spin-10 spin chromatography. The specific activity was approximately 5 × 10⁸ cpm/µg of oligonucleotide. The 2.6-kb HindIII/HincII probe was generated as described below and radiolabeled with [-³²P]CTP using the Rediprime DNA labeling system.

Genomic and plasmid DNAs were digested to completion with restriction endonucleases. Restriction endonuclease-cleaved DNAs were separated by TAE-agarose gel electrophoresis and Southern blotted to nitrocellulose membranes. The membranes were baked at 80 °C for 2 h under vacuum, and prehybridization, hybridization, and washing of the membranes were performed according to instructions supplied with the Rapid-hyb buffer. Washed blots were air-dried and exposed to Kodak XAR-5 films at 70 °C with intensifying screens for 0.5-2 days.

Based on the observation that the 30-mer oligonucleotide probe hybridized to a single 4.2-kb EcoRI-genomic DNA fragment (Fig. 1, lane A), a size-selected genomic library in the 4.2-kb region was generated. Genomic DNA from S. aureus strain 12598 was digested with EcoRI and fractionated by 1% low melting agarose electrophoresis. The 4.2-kb region was excised from the gel and melted at 68 °C for 15 min. DNA in the melted agarose was ligated in situ with pHSG575 treated with EcoRI and alkaline phosphatase according to instructions provided by FMC Products. Competent DH5 cells were transformed with the ligated material, and different dilutions were plated out on LB agar medium plates supplemented with chloramphenicol (20 µg/ml), isopropyl--D-thiogalactopyranoside (0.5 mM), and 5-bromo-4-chloro-3-indolyl--D-galactoside (40 µg/ml) for antibiotic and blue/white selections. White colonies were collected and propagated overnight, and the Wizard plasmid Mini-prep was used to isolate plasmid DNA from cells. Purified plasmids were digested with EcoRI and screened by Southern blotting using the radiolabeled oligonucleotide probe.

The cloned 4.2-kb fragment was digested with HindIII and HincII, yielding a 2.6-kb fragment, which was subcloned into pBluescript KS⁺ and pUC19. The 2.6-kb fragment was also used as a probe in Southern analyses with S. aureus genomic DNA. The insert was digested using the ExoIII/mung bean nuclease system (Stratagene, La Jolla, CA) to generate two sets of nested deletions. Multiple clones covering both strands in their entirety were sequenced by the Sanger dideoxynucleotide chain termination method as modified for Taq polymerase cycle sequencing using an ABI 373A automated DNA sequencer. Sequence data were assembled and discrepancies resolved using the Wisconsin Package (Genetics Computer Group, Madison, WI). The primary sequence of ebpS as shown in Fig. 2 has been assigned the GenBank^TM . 

A 2.6-kb HindIII/HincII fragment in pBluescript KS⁺ (30 ng) served as the template, and PCR reactions were performed with a Perkin-Elmer thermocycler using standard reagents. The open reading frame of ebpS was PCR-amplified using the sense oligonucleotide, 5-TGTATAGAAAGGAAGGTGGCTGTG-3, and the antisense oligonucleotide, 5-GCAGCTGTACCAGCACCAATT-3. The sense oligonucleotide contained a BamHI site (underlined), and A of the two ATG codons was changed to G (in boldface lettering) to avoid internal initiation of translation as recommended by Qiagen. The antisense oligonucleotide contained a HindIII cleavage site (underlined). The exact conditions for amplification were 94 °C for 1 min, followed by 30 cycles of 94 °C for 30 s, 53 °C for 30 s, and 72 °C for 60 s. The PCR product was digested with BamHI and HindIII and gel-purified. This material was ligated to pQE-30 that had been digested with BamHI and HindIII and treated with calf intestinal alkaline phosphatase. Competent M15 cells were transformed with the ligation product and selected by ampicillin (100 µg/ml) and kanamycin (20 µg/ml), and antibiotic-resistant cells were screened for recombinant protein expression. After obtaining several positive clones, ideal conditions for maximum expression were examined.

Based on results from these studies, the following protocol was used routinely for medium scale purification of recombinant EbpS (rEbpS). A stock culture of the clone was grown overnight in 10 ml of LB media supplemented with ampicillin and kanamycin. On the following day, this culture was added to 100 ml of fresh LB media with antibiotics. Cells were allowed to regrow until the A₆₀₀ value reached 0.8 (~3 h). Expression was then induced with 1 mM isopropyl--D-thiogalactopyranoside for 4 h at 37 °C. The cells were pelleted by centrifugation (5000 × g), resuspended in 15 ml of buffer A (8 M urea, 100 mM NaH₂PO₄, 10 mM Tris-HCl, pH 8), and vortexed gently for 15 min. The lysed cells were centrifuged at 15,000 × g for 20 min at 4 °C, and the supernatant was transferred to a tube containing 4 ml of nickel nitriloacetic acid resin (Ni²⁺-NTA) pre-equilibrated with buffer A. The mixture was incubated for 30 min at room temperature with gentle agitation, transferred to a disposable polypropylene column, and washed consecutively with 100 ml of buffer A and 100 ml of buffer B (same as buffer A except pH 6). The tightly bound recombinant protein was eluted with 10 ml of buffer C (same as buffer A, but pH 4). The eluted material was dialyzed twice against 4 liters of 10 mM Tris-HCl, pH 7.5, and the concentration of protein in the dialysate was determined by UV spectrophotometry based on the number of Tyr and Trp residues in rEbpS (1 × A₂₈₀ = 2.68 mg/ml). The yield of purified rEbpS under these conditions was approximately 5 mg/100 ml of induced culture.

To generate CNBr-cleaved fragments, 500 µg of rEbpS was incubated in the dark for 24 h at room temperature with 1 mg of CNBr in 200 µl of 70% formic acid. At the end of incubation, the sample was diluted with 14 ml of deionized H₂O and Speed-vac dried. The dried material was resuspended in 10 ml of deionized H₂O and redried in 100-µg aliquots.

Preimmune sera were collected, and New England White rabbits were injected with 1 ml of purified rEbpS (20 µg) mixed 1:1 with complete Freund's adjuvant. Booster injections (20 µg) mixed 1:1 with incomplete Freund's adjuvant were given at 5, 7, 10, 14, and 19 weeks. Sera were tested by Western immunoblotting using rEbpS.

IgG fractions were purified from immune and preimmune sera by either caprylic acid precipitation (27) or protein A affinity chromatography. For generation of an antibody affinity resin, approximately 100 mg of anti-rEbpS IgG were covalently coupled to 5 ml of Affi-Gel 10 according to the manufacturer's instructions. To generate anti-rEbpS Fab fragments, 50 mg of lyophilized IgG was reacted overnight at 37 °C with 2 ml of immobilized papain in 5 ml of papain digestion buffer (20 mM NaH₂PO₄, 20 mM cysteine-HCl, 10 mM EDTA, pH 6.5). Fab fragments were separated from undigested IgGs and free Fc fragments by protein A affinity chromatography.

Preparation and coupling of elastin peptides to Affi-Gel 10 were as described previously (20). Both rEbpS (20 µg) and CNBr-cleaved rEbpS (80 µg) were iodinated with 300 µCi of Na¹²⁵I by the IODO-GEN method. The specific activities were approximately 2.3 × 10⁴ and 1.2 × 10⁴ cpm/ng protein for rEbpS and CNBr-cleaved rEbpS fragments, respectively. Radiolabeled rEbpS (45 ng) in 1.5 ml of binding buffer (50 mM Tris, 500 mM NaCl, 2 mM CaCl₂, 0.1 mg/ml bovine serum albumin, pH 7.5) was incubated with 1 ml of the elastin peptide affinity resin for 2 h at room temperature in the absence or presence of 2 mg of unlabeled elastin peptides. The mixture was transferred to disposable polypropylene columns and washed with binding buffer by gravity flow until radioactivity of the flow-through reached background. Bound rEbpS was eluted with 3 ml of 1% SDS buffer, spin-concentrated, and analyzed by 10% SDS-PAGE and autoradiography. Binding of radiolabeled CNBr-cleaved rEbpS to immobilized elastin was assessed similarly, except 80 ng of the starting material was used, and bound material was visualized by 12% SDS-PAGE and autoradiography.

Surface-labeled extracts from S. aureus cells were prepared by lysostaphin digestion as described previously (20). Approximately 10⁷ cpm of surface-labeled extract was first absorbed with 3 ml of pig IgG-Affi-Gel 10 resin for 2 h at room temperature. The unbound supernatant was collected and incubated with 1 ml of the anti-rEbpS IgG affinity resin in the absence or presence of 2 mg of unlabeled rEbpS for 2 h at room temperature in 2 ml of binding buffer. The mixtures were transferred to disposable columns and washed with binding buffer until flow-through reached background radioactive levels. Bound cell surface-labeled molecules were eluted from the column by 3 ml of 1% SDS buffer, spin-concentrated, and analyzed by 15% SDS-PAGE and autoradiography.

Purification and radiolabeling of full-length recombinant human elastin and cellular elastin binding assays were performed as described previously (20). Automated amino acid sequence and composition analyses were carried out in our laboratory with the Applied Biosystems 473A protein sequencer and Beckman System 6300 High Performance Analyzer, respectively. Electron spray mass spectrometry was performed by the Protein Chemistry Laboratory at Washington University School of Medicine (St. Louis, MO).

RESULTS

The N-terminal sequence of native EbpS expressed on the cell surface of S. aureus was previously determined to be ANNFKDDFEKNRQ (20). A degenerate oligonucleotide corresponding to residues 2-11 of this N-terminal sequence was generated and used as a probe. Southern blot analysis was first performed with S. aureus strain 12598 genomic DNA digested with restriction endonucleases to identify the genomic fragment containing ebpS. As shown in Fig. 1, the oligonucleotide probe hybridized to a 4.2-kb EcoRI fragment (lane A). On the basis of this observation, a size-selected genomic plasmid library in the 4.2-kb region was constructed from EcoRI-digested S. aureus genomic DNA and was screened with the oligonucleotide probe by Southern blotting. Of 120 colonies screened, two positive clones with identical restriction enzyme digestion patterns were isolated. One of these clones, pEBPS-1, was used for further analysis.

To verify that the correct 4.2-kb fragment was selected, the radiolabeled oligonucleotide was hybridized to the pEBPS-1 insert, and the cloned insert itself was used as a probe for Southern analyses with EcoRI- and EcoRI/HindIII/HincIIdigested genomic DNA. The oligonucleotide probe hybridized to the 4.2-kb pEBPS-1 insert (Fig. 1, lane B), and the insert recognized a 4.2-kb EcoRI genomic fragment (Fig. 1, lane C). With EcoRI/HindIII/HincII-digested genomic DNA, the radiolabeled pEBPS-1 insert hybridized to a 2.6-kb fragment (Fig. 1 lane D). The oligonucleotide and cloned insert probes consistently detected single fragments of identical size in Southern analyses using genomic DNA digested with various restriction endonucleases, indicating that ebpS is present as a single copy gene.

pEBPS-1 was digested with HindIII and HincII to yield a 2.6-kb fragment. This fragment was subcloned into pBluescript II KS⁺ to generate pKS-2.6 and was sequenced. A 606-base pair open reading frame that starts with an ATG codon was identified about 1.2-kb 3 of the HindIII site. The primary sequence of the open reading frame with up- and downstream sequences is shown in Fig. 2. Putative 10 and 35 hexamers were identified at positions 31 and 54, with a spacing of 17 base pairs. A third A/T-rich promoter sequence has been proposed recently to exist in a region about 20 base pairs upstream of the 35 hexamer in E. coli (28), and this region for ebpS was 75% A/T. A potential ribosome binding sequence, which complemented perfectly the extreme 3 region of Bacillus subtilis 16 S RNA (UCUUUCCUCC) (29), was found at position 7. Overall, ebpS was 64% A/T and 36% C/G. Although two ATG codons were found in the correct reading frame of ebpS, we have designated the second ATG as the initiation codon based on the location of the putative ribosome binding site. The N-terminal sequence of cell surface EbpS determined from peptide sequencing was found to start at the second residue of the predicted sequence, suggesting that the initial Met residue is cleaved. The deduced sequence matched perfectly with the determined sequence of cell surface EbpS except for the first amino acid (Ala in native, Ser in deduced). Because Ser residues produce a small peak in peptide sequencing chromatograms and are often misread, we reexamined the original sequencing chromatogram of cell surface EbpS and have identified clear Ser and Ser peaks, indicating that the residue is indeed a Ser and not Ala.

The mature protein has a predicted molecular mass of 23,344.7 daltons and an acidic pI of 4.9. Accordingly, the protein has a preponderance of acidic amino acids Asp (10.9%) and Glu (11.9%) and is devoid of Cys residues. Garnier analysis predicted a secondary structure that is 58.4% -helical and 23.8% coiled-coil. The BLAST network service of the National Institutes of Health was used to search for sequence homologies. The December 1, 1995 releases of the Brookhaven Protein Data Bank, GenBank^TM, EMBL Data Library, and SWISS-PROT protein sequence data base and the translated coding sequence of GenBank^TM were used for comparison. No significant homologies were found between reported sequences in these data bases and the primary sequence of ebpS.

We studied whether the cloned gene encodes an elastin-binding protein by expressing ebpS in E. coli. The PCR-amplified ebpS open reading frame was expressed in E. coli as a fusion protein containing six His residues attached to the N terminus. Recombinant EbpS (rEbpS) was purified from E. coli extracts by Ni²⁺-NTA chromatography based on the high affinity binding interaction between Ni²⁺ and His residues. As can be seen in Fig. 3, the three positive clones expressed large amounts of homogeneous rEbpS. Interestingly, purified rEbpS from all three clones migrated as a 45-kDa protein when fractionated on reducing SDS-PAGE, which was a significant deviation from its predicted molecular mass of 26 kDa.

To evaluate the integrity of rEbpS, the N-terminal sequence of full-length rEbpS, as well as internal sequences from a degradation product and two fragments generated by CNBr cleavage, were determined by protein microsequencing. Altogether, unambiguous sequences were obtained for 58 residues, and they matched perfectly with the predicted sequences (Fig. 2, underlined sequences). Furthermore, amino acid and mass spectrometry analyses indicated that the composition of rEbpS and actual molecular mass of rEbpS agree with the predicted data. These results indicate that the correct protein has been expressed and that overestimation of the molecular mass is due to aberrant migration in SDS-PAGE.

To investigate whether rEbpS interacts specifically with elastin, elastin peptide affinity chromatography was performed with radiolabeled rEbpS. Iodinated rEbpS was incubated with the elastin peptide affinity resin for 2 h at room temperature in the absence or presence of excess unlabeled elastin peptides. The mixture was then washed extensively with buffer until radioactivity in the wash reached background levels. Bound material was eluted with 1% SDS buffer and analyzed by SDS-PAGE and autoradiography. The starting material for this experiment had been stored for 1 week at 4 °C after purification with Ni²⁺-NTA chromatography and, as can be seen in Fig. 4, was partially degraded (lane B). This turned out to be fortuitous, however, because it showed that full-length rEbpS, but not the lower molecular weight degradation products, bound to elastin (Fig. 4, lane C). Importantly, binding of the full-length protein was inhibited by unlabeled elastin peptides (lane D). Sequence analysis showed that the amino terminus of the 40-kDa major degradation product began at position 60 (see Fig. 2). These results suggest that the first 59 amino acids in EbpS play a critical role in elastin recognition.

To determine whether the N-terminal region of EbpS contains the elastin binding site, elastin binding properties of CNBr-cleaved rEbpS fragments were examined. EbpS contains a single internal Met residue at position 125 such that cleavage with CNBr would generate two fragments. In agreement with the predicted sequence, two dominant bands were detected in CNBr-cleaved rEbpS. Peptide microsequencing was employed to verify correct cleavage and to identify which band corresponded to the N- and C-terminal fragments (Fig. 2, underlined). When elastin binding activity of these fragments was assayed with elastin peptide affinity chromatography, only the N-terminal fragment bound to the elastin peptide affinity resin, supporting the conjecture that the elastin binding site is contained in the first 59 amino acids of EbpS.

If EbpS is the cell surface molecule responsible for elastin binding at the cellular level, then an active form of soluble EbpS should interfere with S. aureus binding to elastin. This hypothesis was tested by incubating radiolabeled elastin with S. aureus cells in the absence or presence of various concentrations of unlabeled rEbpS. As can be seen in Fig. 5, rEbpS inhibited binding of labeled elastin in a concentration-dependent manner. S. aureus binding to radiolabeled elastin was abrogated at the highest concentration of rEbpS (19 µM). The control polyhistidine fusion protein mouse dihydrofolate reductase did not influence binding at 26 µM. These results demonstrate that rEbpS inhibition of cellular elastin binding is specific and that the polyhistidine domain of rEbpS does not affect elastin binding.

The ability of rEbpS to interact directly with elastin and to inhibit cellular elastin binding strongly suggest that EbpS is the cell surface protein mediating S. aureus binding to elastin. To provide further evidence that EbpS is on the bacterial surface, affinity chromatography was performed with surface-labeled S. aureus extracts and immobilized anti-rEbpS IgG. S. aureus cells were surface-labeled by the IODO-GEN method, and extracts were prepared by lysostaphin digestion. Approximately 10⁷ cpm of this material was exposed to a porcine IgG affinity resin to remove surface-labeled protein A, and the nonbound fraction was incubated with the anti-rEbpS IgG affinity resin for 2 h at 25 °C. After washing extensively with binding buffer, bound cell surface molecules were eluted with 1% SDS buffer and were analyzed by SDS-PAGE and autoradiography. As shown in Fig. 6, preabsorption with the porcine IgG resin removed surface-labeled protein A from the starting material (compare 50 kDa band in lanes A and B). Of the remaining numerous surface-labeled proteins, a 35- and 25-kDa protein associated with the anti-rEbpS IgG affinity resin (lane C). To determine the specificity of binding, the same experiment was performed in the presence of excess unlabeled rEbpS. Binding of the surface 25-kDa protein, but not the 35-kDa protein, to immobilized anti-rEbpS IgG was inhibited by unlabeled rEbpS (not shown). Densitometric scanning of the bands revealed that the band intensity for the 25- and 35-kDa proteins decreased by 64 and 7%, respectively, in the presence of excess unlabeled rEbpS. These results indicate that the 25-kDa protein is cell surface EbpS and that the 35-kDa protein is interacting nonspecifically with the agarose affinity support of the elastin peptide affinity resin.

To test whether antibodies to rEbpS would block binding to elastin, S. aureus cells were incubated with radiolabeled elastin in the absence or presence of immune or preimmune Fab fragments. As shown in Fig. 7, Fab fragments from immune IgGs inhibited binding of S. aureus to radiolabeled elastin in a concentration-dependent manner. In contrast, Fab fragments from preimmune antibodies had no effect on binding at the two concentrations tested.

DISCUSSION

Cell surface components of pathogenic bacteria play important roles in facilitating the organism's survival in the hostile environment of the host. For Gram-positive bacteria, these surface molecules are used in pathogenic processes such as in evading host immune responses (30), digesting host carbohydrates to expose host attachment sites (31, 32), capturing host enzymes to digest host tissues (33), and binding host tissue determinants to establish a firm basis for colonization (34). Cell surface adhesins and MSCRAMMs interact with host ECM components and participate in the colonization of and extravasation through tissues and organs. We have demonstrated previously that S. aureus binds specifically to elastin and have identified a cell surface elastin-binding protein (EbpS) that mediates the S. aureus-elastin interaction. On the basis of these findings, EbpS has been proposed to be the elastin MSCRAMM.

Several independent criteria indicate that EbpS is the surface protein mediating S. aureus binding to elastin. First, rEbpS binds specifically to immobilized elastin and inhibits binding of S. aureus cells in a dose-dependent manner. These results establish that EbpS is an elastin-binding protein that is functionally active in a soluble form. Second, Fab fragments of an antibody raised against rEbpS inhibit binding of S. aureus to elastin. This suggests that the topology of EbpS is such that the elastin binding site is accessible to interact with the ligand and is not embedded in the cell wall or membrane domains. Third, immunochemical analysis found that the antibody to rEbpS recognizes the native, 25-kDa protein expressed on the surface of S. aureus cells.

Cloning and detailed characterization of ebpS show that it exists as a single copy gene in the S. aureus genome. The 606-base pair gene encodes a protein with a predicted molecular mass of 23 kDa that is highly acidic at neutral pH. Expression of ebpS in E. coli produces a protein of 26 kDa, as determined by mass spectrometry, that migrates as a 45-kDa protein on SDS-PAGE. The exact cause of this aberrant migration is unknown, although anomalous migration in SDS-PAGE is frequently observed for Gram-positive cell surface proteins and for polyhistidine fusion proteins (35, 36, 37, 38, 39). In some cases, the anomalous behavior of Gram-positive surface proteins on SDS-PAGE has been attributed to the presence in the protein of multiple repetitive domains (35, 36) or to a high proline content (37, 38). Other factors must account for the abnormal migration of EbpS, however, since it contains no repetitive domains and proline makes up only about 4% of the total amino acid residues.

The aberrant migration of rEbpS on SDS-PAGE suggests an answer to a question raised during our initial characterization of EbpS. In that study, we found two forms of the protein: a functionally active 40-kDa form of EbpS that was only detected intracellularly and the 25-kDa form present on the cell surface. Our current findings suggest that full-length, native EbpS, with a predicted size of 23 kDa, may be migrating in SDS-PAGE as the 40-kDa intracellular precursor, and that the 25-kDa surface form of EbpS is actually a smaller form of the molecule processed at the C terminus. This conclusion is supported by amino acid sequencing showing that both the 25- and 40-kDa proteins have identical amino acid sequences over 20 residues at their amino terminus, by the presence of a single gene for EbpS in the S. aureus genome, and by the observation that the 26-kDa rEbpS migrates anomalously as a 45-kDa protein on SDS-PAGE. The significance of C-terminal processing of EbpS is unknown. One possibility, however, is that the processing event may be important for targeting EbpS to the cell surface. Because EbpS lacks an N-terminal signal peptide and other known sorting and anchoring signals, this proposed intracellular processing event may facilitate surface targeting. In fact, C-terminal signal peptides have been identified in several bacterial proteins (40), and alternative means of anchoring proteins to the cell surface have been reported in Gram-positive bacteria (41).

The mechanism of EbpS expression on the bacterial cell surface is still unclear. Several surface proteins of Gram-positive bacteria have been found to share common motifs involved in sorting, transporting, and anchoring to the cell surface (42). These motifs include a cleaved signal peptide, which is followed by the ligand binding extracellular N-terminal domain, a Pro-rich region thought to span the cell wall, a conserved LPXTGX hexapeptide sequence, a hydrophobic membrane-spanning domain, and a charged C-terminal tail. A recent study by Schneewind et al. (43) has shown that protein A of S. aureus is cleaved after the threonine residue of the LPXTGX sequence and is anchored to the cell wall via amide linkage of the carboxyl group of threonine to a free amino group on the pentaglycine peptide moiety of the staphylococcal peptidoglycan. Apart from the putative localization of the elastin binding site to the extracellular N-terminal domain and identification of a charged C-terminal tail, EbpS contains none of the other common surface protein motifs. EbpS is not unique in this regard, however, since several other Gram-positive surface proteins lack conserved structures. Examples include streptococcal proteins such as the fibronectin/fibrinogen-binding protein (44), albumin-binding protein (38), and the plasmin receptor (45). Like EbpS, these proteins are all smaller than the majority of Gram-positive surface proteins with common structural motifs, and in the case of the streptococcal plasmin receptor, the initial methionine residue is cleaved in the mature protein (45), similar to what is found with EbpS.

It is now apparent that interactions between pathogenic bacteria and host ECM components play an important role in disease pathogenesis. However, molecular structure-function analyses for most ECM adhesins have not been performed despite the obvious potential of developing effective prophylactic and therapeutic agents based on information derived from these studies. In cases where information is available, the primary ligand binding site has been found to be contained in the N-terminal extracellular domain (36, 46, 47). Consistent with this observation is our finding that truncated fragments of rEbpS lacking the first 59 amino acids do not bind elastin, confirming that the elastin binding site in EbpS is also contained within the extracellular N-terminal domain. Furthermore, preliminary studies show that a recombinant construct of EbpS containing only this 59-amino acid domain directly binds to elastin and inhibits binding of S. aureus to elastin.² Similar to other domains in EbpS, this putative elastin binding region lacks homology with sequences reported to various data bases. Additional studies using recombinant constructs, synthetic peptides, and domain-specific antibodies are in progress to further define the elastin binding site in EbpS.