Identification of a Novel Maturation Mechanism and Restricted Substrate Specificity for the SspB Cysteine Protease ofStaphylococcus aureus *

The SspB cysteine protease ofStaphylococcus aureus is expressed in an operon, flanked by the sspA serine protease, and sspC, encoding a 12.9-kDa protein of unknown function. SspB was expressed as a 40-kDa prepropeptide pSspB, which did not undergo autocatalytic maturation. Activity of pSspB was reduced compared with 22-kDa mature SspB, but it was equivalent to mature SspB after incubation with SspA, which specifically removed the pSspB N-terminal propeptide. SspC abrogated the activity of pSspB when incubated in a 1:1 complex but had no effect on SspA or papain. Activity of the pSspB·SspC complex was restored when incubated with SspA, and SspC was cleaved by SspA but not pSspB. Thus, SspC maintains pSspB as an inert zymogen, and SspA is required for removal of the propeptide and inactivation of SspC. Like the papain protease family, SspB cleaved substrates with a hydrophobic amino acid at P2 but had a strong preference for arginine at P1. It did not cleave casein, serum albumin, IgG, or IgA, but it promoted detachment of cultured keratinocytes and cleaved fibronectin and fibrinogen at sites recognized by urokinase plasminogen activator and plasmin, respectively. It also processed high molecular weight kininogen in a manner resembling plasma kallikrein. Thus, SspB exhibits a novel maturation mechanism and mimics the specificity of plasma serine proteases.

Staphylococcus aureus is a leading cause of nosocomial bacteremia and the overall leading cause of nosocomial infections of all kinds (1). Its ability to colonize and infect virtually every tissue and organ system of the body distinguishes it from many other microbial pathogens (2) and is testament of an adaptive capacity that allows it to thrive or persist in the diverse condi-tions encountered at different infection sites. Although S. aureus is an opportunistic pathogen, it is unusually well endowed with a broad spectrum of secreted proteins, and individual strains can differ significantly in their exoprotein profiles (3). This is especially apparent with toxin expression, and genomics applications have established that many toxin-related functions are associated with variable regions of the chromosome (4). Although expression of specific toxins is frequently associated with distinct genetic backgrounds and disease syndromes (5), the majority of S. aureus infections are not of an obviously toxigenic nature. In these situations, a core complement of secreted proteins common to all strains may represent the minimal pathogenic unit that defines the majority of suppurative tissue infections. However, the contributions of these proteins to the growth and survival of S. aureus remain poorly understood in relation to the well characterized toxins.
In this context, interest in the functions of secreted proteases has recently been stimulated through the discovery by signature-tagged mutagenesis that the Staphylococcus serine protease (SspA 1 ; V8 protease) contributes to growth and survival in each of three infection models (6). This followed a report from our laboratory that the cell surface fibronectin-binding protein adhesin of S. aureus was sensitive to degradation by SspA, and its stability was enhanced by supplementing cultures with ␣ 2 -macroglobulin, a protease inhibitor present in plasma (7). Inactivation of SspA promotes enhanced stability of the cell surface fibronectin-binding protein and protein A adhesins (8), and the metalloprotease aureolysin has also been shown to inactivate cell surface clumping factor ClfB by cleaving at a single site (9). These studies have implicated a role for metalloprotease and serine protease activity in controlling S. aureus adhesion functions.
The metalloprotease is also essential for activation of a precursor form of the serine protease (10), such that inactivation of aureolysin results in a culture supernatant that is devoid of proteolytic activity (9). Furthermore, we have discovered that SspA is expressed as the first gene of an operon, which encodes also a cysteine protease SspB, and a third protein of unknown function, SspC (11). Nonpolar inactivation of sspA resulted in the SspB cysteine protease being expressed and secreted in a 40-kDa prepropeptide form pSspB, suggesting that maturation of SspB is dependent on the sequential activity of the metalloprotease and serine protease in a three-step cascade pathway. Herein, we present an analysis of the function of SspB and the roles of SspA and SspC in promoting maturation of the SspB cysteine protease. Our data establish that the Ssp operon promotes a unique mechanism of maturation for the SspB cysteine protease, which exhibits a limited substrate specificity for cleavage of specific plasma proteins.

MATERIALS AND METHODS
Strains and Culture Conditions-S. aureus RN6390 and RN4220 were obtained from Dr. Richard Novick (Skirball Institute, New York). S. aureus SP6391, an isogenic derivative of RN6390 possessing an sspA::erm allelic replacement mutation has been described by us previously (11). Cultures were maintained in brain heart infusion broth (Difco) supplemented with 15 g/liter agar when required. All cultures were grown at 37°C, and culture media were supplemented with 10 g/ml erythromycin or chloramphenicol, when required for plasmid propagation or maintenance of mutant phenotypes. For expression and purification of proteases, cultures were grown in medium optimized for protease expression (12). For expression and purification of SspA or SspB, 2-liter flasks containing 250 ml of protease expression medium were inoculated to achieve an initial absorbance of 0.1 at 600 nm. Cultures were then grown for 18 h with orbital shaking at 250 rpm, and the cells were removed by centrifugation.
PCR and Recombinant DNA Procedures-S. aureus genomic and plasmid DNA were isolated using Qiagen 100/G genomic tips and plasmid midi kits (Qiagen Inc., Valencia, CA), following the manufacturer's protocol for Gram-positive bacteria. Restriction enzymes, T4 DNA ligase, and calf intestinal alkaline phosphatase were purchased from New England BioLabs and used in accordance with manufacturer's recommendations. Agarose gel electrophoresis, restriction endonuclease digestion, vector dephosphorylation, ligation reactions, phenol chloroform extraction, and ethanol precipitation of ligation mixtures were all conducted according to standard protocols. PCR was conducted with AmpliTaq DNA polymerase (Roche Canada) and reagents, employing standard conditions. The PCR amplicons were treated with proteinase K before digestion with restriction endonuclease. DNA ligation reactions were electroporated into S. aureus RN4220 following standard protocols (13).
Construction of SspA Complementation Vector-Plasmid pRN5548, kindly provided by Dr. Richard Novick (14), was digested with HindIII to remove a blaZ promoter fragment, creating pRN5548a. A DNA segment encompassing the sspA promoter and complete open reading frame was amplified from genomic DNA of S. aureus RN6390 with oligonucleotide primers sspA-F4 and sspA-R1, corresponding to nucleotides 120 -140 and 1573-1552, respectively, of the ssp operon (11). A PstI site was incorporated into the sspA-F4 primer, allowing the 1.45-kb amplicon to be digested with PstI and XbaI, the latter enzyme cleaving at nucleotide 1383 of the ssp operon, 19 nucleotides from the translation stop codon of sspA. The 1.3-kb product was ligated into complementary sites of pRN5548a and electroporated into S. aureus RN4220, followed by selection for chloramphenicol-resistant colonies. The resulting plasmid, p5548-SspA, was then electroporated into S. aureus SP6391, which possess the nonpolar sspA::erm allelic replacement mutation (11).
Expression and Purification of SspA, SspB, pSspB, and SspC-The 40-kDa pSspB was purified from culture supernatant of S. aureus SP6391, whereas SspA and mature SspB were purified from culture supernatant of S. aureus RN6390. Proteins were precipitated from culture supernatant by treatment with 80% saturation of ammonium sulfate, collected by centrifugation, dialyzed versus 20 mM Tris-HCl, pH 7.4, and passed through a 0.45-m syringe filter. Protein was applied to a Hi Prep 16/10 Q-Sepharose Fast Flow column at a flow rate of 2.5 ml/min using the Akta Prime chromatography system (Amersham Biosciences). The column was then washed with 50 ml of equilibration buffer followed by the application of a linear gradient from 0.0 to 0.5 M NaCl applied over a volume of 150 ml, collecting 2-ml fractions. Column fractions containing SspA or SspB as determined by SDS-PAGE and gelatin zymography were pooled and mixed with an equal volume of 4.0 M ammonium sulfate in 50 mM sodium phosphate buffer, pH 7.0, then applied to a 5-ml Hi-Trap phenyl-Sepharose column equilibrated in 2.0 M ammonium sulfate and 50 mM sodium phosphate buffer, pH 7.0. Nonbound proteins were removed by washing with equilibration buffer followed by elution with a linear descending gradient of 2.0 -0.0 M ammonium sulfate applied over a volume of 25 ml. Fractions containing purified SspA or SspB were pooled and desalted on a Hi Prep 26/10 desalting column equilibrated in 20 mM Tris-HCl, pH 7.4.
For expression of recombinant SspC in Escherichia coli, forward primer sspB2601-2620 was paired with reverse primer sspC2991-2968, to PCR amplify a fragment spanning the last 8 codons of SspB and ending with the C-terminal amino acid codon of sspC. The 390-bp amplicon was ligated to the pBAD-Topo TA vector (Invitrogen) creating pBAD-SspC, in which the arabinose-induced pBAD promoter is used in conjunction with the internal sspC translation initiation signals to promote expression of SspC with a C-terminal His 6 tag in E. coli TOP10 cells. For expression of recombinant protein, a culture of E. coli TOP10 pBAD-SspC was grown in LB broth and induced with arabinose for induction of fusion protein expression followed by preparation of cell lysate, following recommendations provided with the pBAD-Topo TA cloning and expression system. The His 6 SspC fusion protein was then affinity purified from the cell lysate using an Akta Prime chromatography system and the Hi-Trap chelating Sepharose high performance column and buffers provided with the Hi-Trap purification kit (Amersham Pharmacia), following the manufacturer's guidelines. His 6 -SspC was desalted employing the Hi Prep 26/10 desalting column equilibrated in 20 mM Tris-HCl, pH 7.4, and purified further by anion exchange chromatography.
Enzyme Assays, Substrates, and Antibodies-Purified human plasma fibronectin was a generous gift from Dr. Kenneth Ingham. Resorufinlabeled casein was purchased from Roche Diagnostics (Laval, Quebec). High molecular weight single-chain and two-chain kininogen, and plasminogen-depleted fibrinogen were purchased from Enzyme Research Laboratories (South Bend, IN). Human serum albumin, IgA, and IgG were purchased from Sigma. Monoclonal antibodies toward different domains of kininogen have been described previously (15). Antibody MBK3 is specific for the nonapeptide bradykinin (D4 domain), HKH4 is specific for amino acids 1-123 in the D1 domain of the heavy chain, and HKL1 is directed toward amino acids 543-554 in the D6 domain of the light chain. Synthetic chromogenic protease substrates Bz-Pro-Phe-Arg-pNA, Bz-Val-Gly-Arg-pNA, Bz-L-Arg-pNA, and pyroglutamyl-Phe-Leu-pNA were purchased from Sigma. Tosyl-Gly-Pro-Arg-pNA was obtained from Roche. Z-Phe-Arg-pNA and Suc-Ala-Ala-Pro-Glu-pNA were obtained from Bachem (Torrance, CA).
All assays for proteolytic activity were conducted at 37°C, and unless otherwise indicated, they contained 10 mM cysteine and 5 mM EDTA for assay of SspB. Assays with resorufin-labeled casein were conducted as described previously (11), with buffer consisting of 50 mM Tris, pH 7.8, and 5 mM CaCl 2 for SspA; for SspB, the buffer was 50 mM HEPES, pH 6.4. For synthetic pNA substrates, assays were conducted in triplicate wells of microtiter plates containing 1 mM substrate in 100 l of 50 mM HEPES, pH 6.4. Plates were read at defined time points using a Bio-Rad model 3550 microplate reader equipped with a 405-nm filter. For specific activity determinations, a standard plot of absorbance at 405 nm versus the concentration of pNA was constructed and employed to quantify the amount of substrate hydrolyzed at defined time points. For pH optimum determinations, assays were conducted with McIlvaine's citrate-phosphate buffer, with the proportions of citric acid and disodium phosphate adjusted to obtain pH values ranging from 4.0 to 8.0. To assay for the effect of cysteine and EDTA, assays were conducted in 50 mM HEPES buffer, pH 6.4, containing specific supplements as indicated. Where indicated, the cysteine protease inhibitor E-64 was included in assays at a concentration of 28 M, and the serine protease inhibitor 3,4-dichloroisocoumarin was included at 1 mM for inhibition of SspA where required or 28 M for effect on SspB.
For proteolysis of protein substrates, samples were incubated with SspB at the indicated molar ratios, and proteolysis was stopped at defined time points by the addition of 28 M E-64. The samples were then precipitated by mixing with an equal volume of ice-cold 20% trichloroacetic acid; after centrifugation, precipitated proteins were solubilized by boiling for 5 min in either reducing or nonreducing SDS-PAGE sample buffer. The integrity of the protein substrate was then visualized by SDS-PAGE and staining with Coomassie Blue. The specificity of kininogen proteolysis was assessed further by Western blotting, in which case replicate gels were transferred to Immobilon-P (Millipore; Bedford, MA) membrane and probed with specific antibodies. Blots were developed using goat anti-mouse IgG alkaline phosphatase-conjugated secondary antibody (Jackson ImmunoResearch; West Grove, PA), with nitro blue tetrazolium and 5-bromo-4-chloro-3indolyl phosphate substrate (Bio-Rad). Fibrinolytic activity was assayed by the fibrin plate method (16).
Keratinocyte Cell Culture-Normal human keratinocytes were obtained from neonatal foreskin and maintained in serum-free keratinocyte growth medium (K-SFM) supplemented with bovine pituitary ex-tract and recombinant epidermal growth factor, as described previously (17). For determination of the effect of SspB, keratinocytes at 80 -90% confluence were treated with 0.05% trypsin for 2-3 min at 37°C followed by 1:5 dilution of trypsin with fresh K-SFM and collection of cells by centrifugation. Cells were resuspended in fresh K-SFM followed by quantification of viable cells in a hemocytometer and plating at a density of 2.5 ϫ 10 4 in individual wells of 24-well cell culture plates. After growth to 80 -90% confluence, cells were washed twice with sterile phosphate-buffered saline, followed by the addition of fresh K-SFM containing 10 mM cysteine, and 10 g/ml mSspB or pre-SspB. As a control, mSspB was preincubated with 28 M E-64 at 37°C for 30 min before the addition to cell culture plates. Cells were then visualized microscopically at hourly intervals until changes in cell morphology were observed.
SDS-PAGE, Gelatin Zymography, Western Blotting, and N-terminal Sequence Analyses-SDS-PAGE was conducted using the electrophoresis and sample buffer system as described by Laemmli (18) and the Bio-Rad Mini-Protean 3 apparatus. After electrophoresis, proteins were visualized by staining with Coomassie Brilliant Blue R-250. For Western blotting, proteins were transferred to Immobilon-P membrane in standard transfer buffer (19); samples for N-terminal sequence analyses were transferred to membrane using CAPS transfer buffer following established protocols (20). Protein bands on membranes for N-terminal sequence analysis were visualized by brief staining with 0.1% Coomassie Blue in 40% methanol, after which the protein bands were excised with a scalpel and submitted to the University of Toronto HSC Biotechnology Center for N-terminal sequencing. For gelatin zymography, electrophoresis was conducted in SDS-polyacrylamide gels that were copolymerized with 1 mg/ml gelatin as described previously (11).

Maturation of the SspB Prepropeptide Requires SspA-As
reported previously (11), when compared with the profile of secreted proteins from S. aureus RN6390 (Fig. 1A, lane 1), culture supernatant of S. aureus SP6391 harboring the defective sspA::erm allele exhibits loss of the SspA serine protease and a 22-kDa peptide believed to represent mature SspB, accompanied by the appearance of the 40-kDa pSspB (Fig. 1A, lane 2). Complementation of the sspA defect with p5548-SspA restored the normal profile of secreted proteins (Fig. 1A, lane 4), whereas pRN5548a alone had no effect (Fig. 1A, lane 3), and these changes were duplicated in a zymogram for detection of proteolytic activity (Fig. 1B). The 22-kDa polypeptide purified from S. aureus RN6390 (Fig. 2A, lane 2) possessed an Nterminal sequence spanning amino acids 184 -194 of pSspB (Fig. 2B), indicating that processing occurred at Glu 183 , consistent with it having been processed by SspA, which is a member of the glutamyl endopeptidase family (21).
Confirmation of processing by SspA was obtained with pSspB purified from culture supernatant of S. aureus SP6391 ( Fig. 2A, lane 4), which possessed an N-terminal sequence DSHSKQLEIN, corresponding to pSspB after processing by signal peptidase (Fig. 2B). When pSspB was incubated with SspA in a 100:1 w/w ratio for 2 h, it was converted to polypeptides of 18 and 22 kDa (Fig. 2C, lane 2). Complete conversion occurred within 15 min, and in a 60-min incubation, unprocessed pSspB was not apparent until the ratio exceeded 500:1 (data not shown). The N terminus of the 22-kDa cleavage product indicated that processing occurred at Glu 177 (Fig. 2B), compared with Glu 183 for mature SspB from culture supernatant. Because the N terminus of the mature protease is preceded by a sequence that is enriched in glutamic acid (Fig. 2B), this may allow for some variability in the site of processing, in response to differences in pH and ionic strength. The N terminus of the lower mass cleavage product corresponds to the N-terminal propeptide after cleavage at Glu 8 (Fig. 2B), and the mass of this fragment correlates with a predicted 19.1-kDa mass after cleavage by SspA at Glu 8 and Glu 177 . Therefore, although the propeptide contains 16 glutamate residues (at positions 8,47,50,60,61,94,126,132,133,151,153,165,176,177,179,183), processing by SspA is limited to specific sites.
When the activity of the matured cysteine protease was activated by addition of cysteine and EDTA, the liberated propeptide remained stable during an additional 2-h incubation (Fig. 2C, lane 3). When pSspB alone was incubated in buffer containing cysteine and EDTA, no autocatalytic processing occurred (Fig. 2C, lane 4); and when incubated with mature SspB at a 100:1 ratio, pSspB also remained largely unprocessed (Fig. 2C, lane 5). Therefore, there is an absolute requirement for SspA in maturation of pSspB. This differentiates pSspB from other microbial proteases, including subtilisin of Bacillus subtilus and the SpeB cysteine protease of Streptococcus pyogenes, which undergo autocatalytic maturation (22,23).
SspC Is an Inhibitor of the SspB Prepropeptide and Is Inactivated by SspA-In the established paradigm for protease maturation exemplified by subtilisin of B. subtilus, the prepropeptide is an inert zymogen, and the N-terminal propeptide plays a key role in maturation. Initially, it functions as an intramolecular chaperone to facilitate folding of the precursor. Then, upon autocatalytic processing, it forms a complex with the mature protease, inhibiting its activity until degraded by the mature protease in the final step of maturation (23)(24)(25). Based on this example, we anticipated that pSspB would represent an inert zymogen. Unexpectedly, when assayed with Bz-Pro-Phe-Arg-pNA substrate (Fig. 3A), it exhibited a constant rate of hydrolysis over a period of 3 h, but this was consistently 3-4-fold less than that of an equimolar amount of mature SspB. However, when the same amount of pSspB was assayed after treatment with SspA to remove the N-terminal propeptide, its hydrolysis kinetics were indistinguishable from mature SspB. Therefore, although pSspB did not represent an inert zymogen, its catalytic activity was enhanced after maturation with SspA.
Because the N-terminal propeptide is not degraded by SspA or SspB (Fig. 2C), these data indicate that the propeptide does not function as an inhibitor of enzyme activity. Therefore, we focused our attention on SspC, the third component of the Ssp operon, which is a predicted 12.9-kDa acidic protein with no obvious signal peptide or transmembrane domains (11). When pSspB was preincubated with an equimolar amount of purified His 6 -SspC, its activity was abrogated over a 3-h incubation (Fig. 3A). However, when SspA was added to achieve a 1:1:1 ratio of each protein, activity was restored, and the rate of hydrolysis approached that of purified mature SspB. These data suggest that SspC is a specific inhibitor of SspB and is inactivated by SspA. Accordingly, SspC also inhibited mature SspB but exhibited negligible inhibition of SspA or the nonrelated cysteine protease, papain (Fig. 3B). Furthermore, when SspC was incubated with SspA at a 10:1 molar ratio, it was cleaved rapidly to a lower molecular weight product (Fig. 4A), whereas pSspB did not cleave SspC at the same 10:1 ratio (Fig.  4B). Cumulatively, these data establish that SspC functions to maintain pSspB as an inert zymogen, with maturation of pSspB being dependent on SspA to cleave the N-terminal propeptide and to inactivate SspC.
Biochemical Characterization of SspB-When assayed for activity toward Bz-Pro-Phe-Arg-pNA in McIlvaine's citratephosphate buffer at pH values ranging from 4.0 to 8.0, activity peaked between pH 6.0 and 6.4 (97 and 100% activity) and declined at pH 5.4 (69%) and 7.4 (71%). Additional assays for effect of buffer composition and substrate specificity were conducted in 50 mM HEPES, pH 6.4, as the basal buffer (Table I). SspB was inactive in the absence of cysteine and stimulated by EDTA, but not unless cysteine was also present. SspB was inactivated by the cysteine protease inhibitor E-64 but exhibited 85% activity in presence of 3,4-dichloroisocoumarin, which is a serine protease inhibitor. Together with our previous report that SspB possesses a eukaryotic thiol protease histidine active site consensus pattern (11), these data are consistent with SspB being a cysteine protease, and as with papain, it is also stimulated by EDTA. Additional assays of substrate specificity were conducted in buffer containing 10 mM cysteine and 5 mM EDTA (Table I).
The papain cysteine protease family exhibits preferential cleavage of substrates with a hydrophobic amino acid at the P2 position, with P1 being a less critical determinant of specificity (26,27). Hydrolysis of Bz-Pro-Phe-Arg-pNA is consistent with a requirement for a hydrophobic amino acid at P2, and SspB also cleaved Z-Phe-Arg-pNA, but with reduced efficiency compared with Pro-Phe-Arg-pNA. Compared with its specific activity toward Bz-Pro-Phe-Arg-pNA, SspB was 30-fold less active when assayed with pyroglutamyl-Phe-Leu-pNA, which is a general substrate for the papain-like cysteine protease family. This observation suggests a preference for arginine at P1. The occurrence of arginine at P1 as the exclusive determinant of specificity was excluded because SspB did not cleave substrates in which arginine was not preceded by a hydrophobic amino acid, including Bz-Val-Gly-Arg, tosyl-Gly-Pro-Arg, or Bz-L-Arg. Therefore, SspB exhibits a strong preference for substrates where arginine is preceded by a hydrophobic amino acid. This hypothesis was tested further by assay of SspB for activity toward different protein substrates.
Specificity of SspB for Protein Substrates-Purified SspB displayed no activity toward resorufin-labeled casein, a sensitive substrate for detection of proteolytic activity, whereas SspA exhibited a concentration-dependent response (Fig. 5). When assayed for the ability to cleave a variety of human plasma proteins, SspB also exhibited no activity toward IgG, IgA, or serum albumin even after 24 h of incubation (data not shown). Therefore, SspB is not a general proteinase that functions in protein degradation. However, fibronectin, fibrinogen, and kininogen were all cleaved by SspB, and these interactions were studied in greater detail.
Specificity of Fibronectin Proteolysis-When plasma fibronectin was incubated with SspB at a 20:1 molar ratio (Fig.  6), intact fibronectin was not detected after 6 h. Fragments  (28). The N terminus of the 27.5-kDa fragment liberated by SspB was blocked to N-terminal sequencing, which is characteristic of the N terminus of fibronectin. This same fragment could also be affinity purified using a Sepharose matrix containing a conjugated synthetic peptide representing the D3 motif of the S. aureus fibronectinbinding protein adhesin (data not shown), which binds the N-terminal fragment of fibronectin with high specificity (29).
Under nonreducing conditions, the 200-and 192-kDa fragments showed a sharp decrease in mobility, indicative of their retaining the C-terminal disulfides that promote fibronectin dimerization. Therefore, these fragments appear to result from selective removal of the N-terminal domain from dimeric fibronectin. The size of the 165-kDa polypeptide lacking the N-terminal domain is consistent with it also being cleaved by SspB to remove the C-terminal heparin and fibrin binding domains of fibronectin. Therefore, SspB appears to remove the N-and C-terminal domains of fibronectin selectively, with a preference for cleavage at the N terminus.
Specificity of Fibrinogen Cleavage by SspB-Fibrin clotting is initiated when thrombin cleaves the ␣-chain of fibrinogen after arginine in the sequence Gly-Val-Arg 16 , releasing fibrinopeptide A (FbpA1-16) and exposing a new N-terminal Gly-Pro-Arg sequence, which forms a knob-like structure that fits into a complementary pocket elsewhere in fibrinogen to initiate polymerization (30). Because human plasma fibronectin was cleaved at Val 258 -Arg 259 2Ala 260 , the Val-Arg sequence in the fibrinogen ␣-chain is also a potential cleavage site for SspB. However, when human fibrinogen was incubated with SspB at a 20:1 molar ratio, the 66-kDa ␣-chain was no longer detected after 30 min (Fig. 7). This was accompanied by the appearance of a 30-kDa fragment possessing an N-terminal sequence identical to the intact ␣-chain. Therefore, SspB did not exert a procoagulant activity by releasing the fibrinopeptide A fragment. The 54-kDa ␤-chain and 48-kDa ␥-chain were unaffected by SspB after 30 min, but upon an extended 3-h incubation, the 54-kDa ␤-chain gradually disappeared, and the intensity of the ␥-chain band thickened. N-terminal sequence analysis of the 48 kDa band after 3 h of proteolysis provided two amino acid residues at each cycle, from which two sequences could be discerned: Tyr-Val-Ala-Thr-Arg-Asp-Asn, corresponding to the N terminus of the intact ␥-chain, and Ala-Arg-Pro-Ala-?-Ala-Ala-Ala-Thr-Gln, representing amino acids 43-52 of the ␤-chain. Therefore, the ␥-chain remained intact, whereas the ␤-chain was cleaved at Tyr-Arg 42 -2-Ala 43 , which is identical to the site cleaved by plasmin in the initiation of fibrinolysis (30). An assay for fibrin degradation employing the fibrin plate method established that SspB exhibited a limited capacity to promote fibrin lysis compared with an equimolar amount of plasmin (data not shown).
Specificity of Kininogen Cleavage by SspB-High molecular weight plasma kininogen is a single-chain protein, with the Nand C-terminal ends joined by a disulfide bond. At sites of vascular damage, it is cleaved by the serine protease plasma kallikrein at Leu-Met-Lys 363 -2-Arg 364 and Pro-Phe-Arg-2-Ser 372 to release the proinflammatory vasoactive nonapeptide bradykinin (31), converting single-chain kininogen to a 62-kDa N-terminal heavy chain and a 52-kDa light chain. Thus, the Bz-Pro-Phe-Arg-pNA substrate preferred by SspB is also a substrate for plasma kallikrein, suggesting that SspB may mimic this activity. Accordingly, when single-chain kininogen (116 kDa) was incubated with SspB at a 20:1 ratio (Fig. 8),   30 min (lane 2), 1 h (lane 3), 2 h (lane 4), or 3 h  (lane 5). Proteins were resolved by SDS-PAGE employing a 5-15% acrylamide gradient gel.
intact kininogen was no longer evident after 2 h, and fragments of 65, 51, and 43 kDa were detected, together with a cluster of smaller fragments centered at 23 kDa. The higher mass fragments comigrated with the respective 63-and 52-kDa heavy and light chains of two-chain kininogen, which has been processed correctly by plasma kallikrein (Fig. 8, lane 4). The higher mass fragments in the SspB digest of kininogen were less evident at 6 h, with a greater proportion of lower mass fragments, indicative of additional proteolysis. Attempts on two occasions to obtain N-terminal sequence from the 65, 51, and 43 kDa bands were not successful, which was expected for the putative 65-kDa heavy chain fragment because the N terminus of single-chain kininogen contains a modified pyrrolidone carboxylic acid. Failure to obtain a sequence from the latter fragments could be the result of their sensitivity to proteolysis and the possibility of producing overlapping fragments of similar size.
In a Western blot analysis (Fig. 9), monoclonal antibody HKH4, which is specific for an epitope in the N-terminal portion of the heavy chain, detected fragments of ϳ62 and 40 kDa in the SspB digest, and the higher mass fragment comigrated with the kininogen heavy chain. A similar pattern was revealed with MBK3, which is specific for bradykinin, and an ϳ20-kDa fragment was also detected. MBK3 did not detect the heavy or light chain of two-chain kininogen, in which bradykinin is excised by plasma kallikrein. Antibody HKL1, specific for an epitope in the kininogen light chain, recognized one fragment in the SspB digest, which migrated closely with the kininogen light chain. Identical results were obtained when kininogen was treated with SspB at an 80:1 ratio, except that HKH4 and MBK3 also detected a fragment that was slightly larger than the expected size of the heavy chain. Therefore, SspB cleaves single-chain kininogen, producing fragments similar in size to the correctly processed heavy and light chains. However, bradykinin is not excised from the putative heavy chain, and there appear to be multiple cleavage sites.
SspB Promotes Detachment of Primary Human Keratinocytes-Because SspB appears to cleave only a limited number of host proteins, assays were conducted to evaluate the effect of SspB on human primary keratinocyte cell culture (Fig. 10). When medium from confluent keratinocyte cell culture was exchanged with medium supplemented with mature SspB, the cells maintained a typical adherent morphology for 6 h and then rapidly began to round up and detach from the culture vessel, such that after 7 h, few adherent cells remained. This effect was prevented by preincubation of SspB with E-64. Although the keratinocyte cells were detached from the culture vessel after 7 h of incubation, they remained viable as evident from trypan blue staining, even after 24 h of exposure to ma-ture SspB (data not shown). Therefore, SspB did not exert a direct cytotoxic effect on the cell monolayer. DISCUSSION Data presented in this study establish that the Ssp operon of S. aureus facilitates a novel pathway for protease maturation, whereby maturation of the SspB cysteine protease is dependent on the SspA serine protease for removal of the N-terminal propeptide and for relief of the SspC-mediated inhibition. The SspA serine protease is itself expressed as an inactive zymogen that is activated by a metalloprotease aureolysin (10), which also exhibits an N-terminal propeptide and C-terminal catalytic domain (32). Although the mechanism for aureolysin maturation has not been elucidated, these steps define a pathway in which maturation of the SspB cysteine protease requires the sequential activity of metalloproteases and serine proteases. To our knowledge, a similar operon structure and sequential cascade pathway for protease maturation have not been described in other prokaryotic organisms, and several observations implicate an important role for this proteolytic cascade in promoting the metastasis of S. aureus infections.
In addition to activating a precursor form of SspA, aureolysin cleaves fibrinogen-binding protein ClfB at a single site, rendering it unable to bind fibrinogen (9). SspA exhibits a potent activity in degrading cell surface fibronectin-binding protein adhesins of S. aureus (7,8), and our present data also support a role for SspB in controlling adhesive functions. The fibronectin-binding protein adhesins of S. aureus bind the N-terminal fragment of fibronectin with high affinity and specificity (29,33). Although fibronectin possesses several domains that are liberated by a number of different proteases, SspB preferentially cleaved fibronectin to release the N-terminal fragment and to a lesser extent, probably also the C-terminal heparin and fibrin binding domains. Cleavage occurred at Val 258 -Arg 259 -2-Ala 260 , and this same site is cleaved by urokinase plasminogen activator (28), which cleaves fibronectin at just one other site inside of the C-terminal disulfide bonds to generate fragments of 210, 200, 25, and 6 kDa. Thus, treatment of fibronectin with urokinase preferentially releases the 25-kDa N-terminal fragment, and SspB mimics this specificity by generating primarily high molecular mass fragments lacking the N-terminal domain.
SspB also cleaved the ␤-chain of fibrinogen at a single site, Tyr-Arg 42 -2-Ala 43 , which is the same site cleaved by plasmin in both fibrinogen and fibrin to promote an anticoagulant or fibrinolytic activity (34). Residues 15-42 of the ␤-chain are also implicated in promoting platelet spreading on fibrin matrices (35), and removal of this sequence may therefore interfere with wound healing. Whereas the ␤-chain of fibrinogen was cleaved slowly, SspB rapidly removed the C terminus of the ␣-chain. During coagulation, the N terminus of fibronectin is crosslinked to the C terminus of the fibrinogen ␣-chain by coagulation factor XIII (36,37), and fibrin matrices that contain fibronectin are much better substrates for fibroblast adhesion and spreading than those lacking fibronectin (38). Therefore, by removing both the N-terminal fibronectin fragment and C-terminal ␣-chain fragment of fibrinogen, SspB has the capacity to affect the function and integrity of fibrin clots. This may be of particular relevance to infective endocarditis, where the bacteria are enmeshed within platelet-fibrin vegetations on traumatized heart valves.
Cleavage of kininogen by SspB could also play an important role in promoting the dissemination of infection. Cleavage of kininogen to release the vasoactive peptide bradykinin has been reported for a number of proteases secreted by microbial and parasitic pathogens (39,40), potentially promoting a proinflammatory response (41). The enhanced vascular permeability Lane 4 contains 10 g of two-chain high molecular weight kininogen that has been processed with plasma kallikrein. Proteins were resolved by SDS-PAGE utilizing a 5-15% acrylamide gradient gel.
promoted by bradykinin could also contribute to the ability of bacteria to enter into the bloodstream from localized tissue infections. Although bradykinin remains associated with larger proteolytic fragments when kininogen is treated with SspB, others have shown that conversion of single-chain kininogen to the two-chain form promotes enhanced attachment to cell surfaces via the D5 domain of the light chain (42,43). Because the C-terminal D6 domain of the kininogen light chain binds prekallikrein, this also serves to deliver prekallikrein to cell surfaces, facilitating its conversion to active kallikrein, which can then promote release of bradykinin (44). Furthermore, twochain kininogen and the isolated D5 domain exert a potent antiadhesive effect on endothelial cells through the ability of D5 to interfere with vitronectin-dependent adhesion mechanisms (45,46). Therefore, proteolysis of single-chain kininogen by SspB at a site of infection, combined with its activity toward fibronectin and potentially other as yet unidentified substrates, could promote a potent antiadhesive effect, disrupting the integrity of cellular barriers to infection. Thus, our observation that SspB promotes detachment of primary human keratinocytes from cell culture could occur through a number of mechanisms, including also the targeting of integrins, which has been reported for the SpeB cysteine protease of S. pyogenes (47).
Intriguingly, the substrate specificity of SspB most closely mimics the serine protease halystase, present in venom of the common viper, Agkistrodon halys blomhoffii. As with SspB, halystase cleaved the ␤-chain of fibrinogen at the same site as plasmin and also the ␣-chain, generating a 35-and 33-kDa doublet with an intact ␣-chain N terminus (48). Halystase also exhibited fibrinolytic capacity, cleaved Bz-Pro-Phe-Arg-pNA, but not several other arginine containing substrates, and cleaved high molecular weight kininogen to release bradykinin. Thus, SspB duplicates many of these properties. The cleavage specificity of SspB for synthetic substrates, together with cleavage sites identified in fibrinogen and fibronectin, indicate that it maintains the specificity of the papain family at the P2 position but exhibits a strong preference for arginine at P1. This differs from the SpeB cysteine protease of S. pyogenes, which is a more typical member of the papain family, being able to accommodate arginine, lysine, asparagine, alanine, glutamic acid, glycine, and histidine in the P1 position (27). Thus, SpeB has been shown to cleave casein and all classes of immunoglobulins as well as fibronectin, fibrinogen, and kininogen (39, 49 -51), whereas SspB of S. aureus shows no cleavage of casein, IgG, IgA, or serum albumin and exhibits a strict requirement for arginine at P1. To our knowledge, a similar substrate specificity has not been reported for other microbial cysteine proteases.
It is interesting to speculate whether the novel mechanism for protease maturation we have identified for SspB has enabled the evolution of a restricted substrate specificity, or alternatively, whether this specificity has necessitated the evolution of an alternative maturation mechanism. In the conventional pathway for protease maturation, there are three distinct stages, as exemplified by the serine protease subtilisin. These include (i) folding of the protease mediated by its cognate N-terminal propeptide; (ii) autoprocessing of the bond between the propeptide and protease domain, resulting in structural reorganization; and (iii) degradation of the propeptide, which locks the protease into a stable conformation (23,24). The SpeB cysteine protease of S. pyogenes also seems to follow this paradigm, as evident from the identification of five distinct sites in the N-terminal propeptide which are cleaved in a stepwise process during autocatalytic activation (27,52). Remarkably, although the N-terminal propeptide of S. aureus SspB possesses a predicted basic isoelectric point of pI 8.4, it contains 31 lysine residues and just 1 arginine at Arg 138 (11). Because the mature SspB protease appears to be highly selective for cleavage where arginine is preceded by a hydrophobic amino acid, it appears that the propeptide was not designed to accommodate an autocatalytic maturation pathway. Thus, the N-terminal propeptide and mature protease domain are free to evolve independently of a requirement for cleavage and degradation of the propeptide.
In this respect, our study has established that SspC maintains pSspB as an inert zymogen, whereas SspA eliminates a requirement for autocatalytic activation and inactivates the inhibitory activity of SspC. Studies are in progress to determine whether the N-terminal propeptide of SspB has retained the function of an intramolecular chaperone to facilitate folding of SspB or whether this function is also satisfied by SspC. Others have shown that the chaperone and inhibitor functions of the subtilisin propeptide are not obligatorily linked to one another, such that mutations that eliminate inhibitory activity do not interfere with chaperone function (53). Therefore, it is  1 (lane 3) kininogen:SspB ratio. Lane 4 contains two-chain high molecular weight kininogen that has been processed with plasma kallikrein. The amount of protein loaded was 2 g in lanes 1 and 4 and 5 g in lanes 2 and 3. Blots were probed with monoclonal antibodies HKH4, specific for the N-terminal region of the heavy chain; HKL1, specific for the C-terminal region of the light chain; and MBK3, specific for the nonapeptide bradykinin that is excised upon processing with plasma kallikrein. The leftmost lane of each blot contains prestained molecular mass markers (New England BioLabs) with the kDa values indicated on the left. possible that the N-terminal propeptide of pSspB retains a function as an intramolecular chaperone. Alternatively, if this function is also promoted by SspC, then the N-terminal propeptide may possess another function, possibly in delivering SspB to host cell surfaces and the extracellular matrix of traumatized tissues.