Molecular Interactions of Human Plasminogen with Fibronectin-binding Protein B (FnBPB), a Fibrinogen/Fibronectin-binding Protein from Staphylococcus aureus*

Staphylococcus aureus is a commensal bacterium that has the ability to cause superficial and deep-seated infections. Like several other invasive pathogens, S. aureus can capture plasminogen from the human host where it can be converted to plasmin by host plasminogen activators or by endogenously expressed staphylokinase. This study demonstrates that sortase-anchored cell wall-associated proteins are responsible for capturing the bulk of bound plasminogen. Two cell wall-associated proteins, the fibrinogen- and fibronectin-binding proteins A and B, were found to bind plasminogen, and one of them, FnBPB, was studied in detail. Plasminogen captured on the surface of S. aureus- or Lactococcus lactis-expressing FnBPB could be activated to the potent serine protease plasmin by staphylokinase and tissue plasminogen activator. Plasminogen bound to recombinant FnBPB with a KD of 0.532 μm as determined by surface plasmon resonance. Plasminogen binding did not to occur by the same mechanism through which FnBPB binds to fibrinogen. Indeed, FnBPB could bind both ligands simultaneously indicating that their binding sites do not overlap. The N3 subdomain of FnBPB contains the full plasminogen-binding site, and this includes, at least in part, two conserved patches of surface-located lysine residues that were recognized by kringle 4 of the host protein.

Staphylococcus aureus colonizes the anterior nares of ϳ25% of the healthy human population (1,2). This commensal Grampositive bacterium has the ability to cause a plethora of infections ranging from superficial skin abscesses to serious and potentially life-threatening invasive diseases such as osteomyelitis, endocarditis, and septic arthritis. Strains that are resistant to multiple antibiotics are associated with infections in hospitals. These are referred to as hospital or healthcare-associated methicillin-resistant S. aureus (HA-MRSA), 4 which have a propensity to cause bacteremia often associated with biofilm formation on indwelling medical devices (3,4). Recently a global epidemic of MRSA has occurred in the community (community-associated MRSA (CA-MRSA)) exemplified by USA300 strains such as LAC (5). CA-MRSA strains have fewer antibiotic resistance determinants than HA-MRSA; they express a lower level of resistance to ␤-lactams, and they can survive on human skin and cause serious skin and soft tissue infections often requiring hospitalization.
The surface of S. aureus is decorated with proteins that are covalently anchored to the cell wall by sortases. During the process of secretion and anchoring to the cell wall peptidoglycan, the pre-proteins undergo post-translational changes both at the N terminus to remove the secretory signal sequence and at the C terminus where sortase recognizes the LPXTG motif, and it covalently links the COOH of the Thr to the amino group of the fifth Gly in the nascent cross-bridge of the peptidoglycan precursor lipid II. Transglycosylation joins the cell wall-associated (CWA) protein-linked precursor to peptidoglycan (6). Fig. 1 shows the structural organization of two important CWA proteins, the fibronectin-binding proteins A and B (FnBPA and FnBPB), which are members of the microbial surface component recognizing adhesive matrix molecules (MSCRAMM) family. FnBPs are multifunctional proteins. The C-terminal fibronectin binding repeats are unstructured and form a flexible stalk-like region that projects the A domain away from the cell surface. The N-terminal A domain binds to fibrinogen by the "dock, lock, and latch" mechanism (7,8) in a similar fashion to the archetypal MSCRAMM clumping factor A (9). A hydrophobic trench located between subdomains N2 and N3 accepts the C-terminal peptide of the fibrinogen ␥-chain. A flexible region at the C terminus of N3 (the locking and latching peptide) undergoes a conformational change that locks the ligand in the trench and forms an additional ␤-strand in a * This work was supported by Fondazione CARIPLO Grant Vaccines 2009-␤-sheet in subdomain N2. The A domains of FnBPA and FnBPB both bind to elastin, most probably involving dock lock and latch. They can also engage in homophilic interactions to form dimers, and when two FnBP molecules on neighboring cells interact, this can lead to cell accumulation during biofilm formation (10). The HA-MRSA strain BH1CC and the CA-MRSA strain LAC form biofilm that is dependent on FnBPs (11,12).
Plasminogen is a 92-kDa zymogen that includes a PNA-apple N-terminal domain, five kringle domains (K1-5), and a serine proteinase catalytic domain (23). The kringle domains mediate interactions with fibrin clots and surface receptors from both eukaryotic and bacterial cells (24,25). Proteolytic activation of PLG to plasmin occurs through cleavage of the Arg 561 -Val 562 peptide bond in the catalytic domain by the tissue plasminogen activator (t-PA) and urokinase plasminogen activator. The cleavage results in the formation of active plasmin enzyme that contains a serine protease active site in the C-terminal region (23,26,27).
S. aureus can capture PLG on its cell surface where the zymogen can be activated by host t-PA and urokinase, or by staphylokinase, a zymogen activator encoded by a gene located on lysogenic bacteriophages in S. aureus (28). Surprisingly little is known about the surface-located proteins of S. aureus that capture PLG. The second immunoglobulin-binding protein Sbi and the extracellular fibrinogen-binding protein Efb both occur in the culture supernatant and are associated non-covalently with the cell wall and both can bind PLG (29). The membranebound manganese transporter MntC and the moonlighting cytoplasmic proteins enolase and triose-phosphate isomerase can also bind PLG (30 -32). However, the biological relevance of these PLG-binding proteins has never been elucidated.
Surprisingly, the ability of CWA proteins to bind PLG has never been examined. Here, for the first time we have tested the ability of CWA proteins to bind PLG and allow it to be activated to form plasmin. Initially, a sortase A mutant was compared with the wild type and was found to bind much less PLG. The FnBPs were found to contribute to the PLG-binding phenotype and thereafter a detailed analysis of one of these, FnBPB was undertaken.

S. aureus Cell Wall-anchored Proteins Bind PLG in Human
Plasma-S. aureus cells are known to bind to human PLG, but the bacterial surface components responsible have not been identified. To determine whether cell wall-anchored proteins on the surface of S. aureus contribute to PLG binding, the wildtype strain LAC and a sortase A-deficient mutant were compared. Bacteria were incubated with different concentrations of human plasma, and proteins that were bound non-covalently to the cell surface were dissociated by addition of extraction buffer, separated by SDS-PAGE under non-reducing conditions, and analyzed by Western immunoblot probing with anti-PLG IgG. Bacterial cells captured a 90-kDa immunoreactive protein in a dose-dependent manner ( Fig. 2A). Densitometric analysis of the blots showed that the wild-type strain captured at least 10-fold more PLG than the sortase A mutant (Fig. 2B). To exclude the possibility that the markedly reduced capture of PLG from plasma by the sortase A mutant is due to higher proteolytic activity, LAC and its srtA mutant were grown to exponential phase and were incubated with purified PLG, and the integrity of the protein in the supernatants was then examined by SDS-PAGE. As expected, the amounts of unbound PLG in the supernatant from srtA mutant were even higher than that from the parental strain (data not shown). It can be concluded The primary translation products of FnBPA and FnBPB proteins contains a signal sequence (S) at the N terminus and a wall-spanning region (M) and sorting signal (W) at the C-terminal end. The N-terminal A region of FnBPA and FnBPB contains three separately folded subdomains, which are known as N1, N2, and N3. Structurally, N2 and N3 form IgG-like folds that combined bind ligands such FBG by the dock, lock, and latch mechanism. Located distal to the A domain is an unfolded region that contains multiple tandemly arranged motifs (11 in FnBPA and 10 in FnBPB) that bind to the N-terminal type I modules of fibronectin and a proline-rich region. FIGURE 2. Capture of PLG from human plasma by S. aureus. A, S. aureus strain LAC and LAC srtA were mixed with different concentrations of human plasma for 60 min. Proteins that were bound to the cell surface were released by extraction buffer and separated by SDS-PAGE under non-reducing conditions and analyzed by Western immunoblotting. The membranes were probed with rabbit anti-human PLG followed by HRP-conjugated mouse antirabbit IgG and developed with the ECL Western blotting detection kit. B, densitometric analysis of PLG bound to S. aureus LAC and the sortase A mutant as reported in A. The band intensity was quantified relative to a sample of pure human PLG. The reported data are the mean values Ϯ S.D. from three independent experiments. Statistically significant differences are indicated (Student's t test; *, p Ͻ 0.05; **, p Ͻ 0.01).
that CWA proteins are the dominant PLG-binding proteins on the bacterial cell surface and that other surface-located proteins contribute minimally. To compare the plasmin and plasminogen binding activity of S. aureus, equal quantities of the purified PLG and plasmin were incubated with LAC cells. After washing, the bacteria-bound plasmin(ogen) was extracted from the staphylococcal surface, and the PLG and plasmin in the extracts were evaluated by ELISA. The results show that bacteria captured the same level of PLG and plasmin (data not shown).
PLG Bound to the Surface of S. aureus Can Be Activated-The ability of PLG bound to the surface of S. aureus LAC to be activated by exogenous or endogenous PLG activators was tested (Fig. 3). S. aureus LAC is lysogenized by a bacteriophage that carries the gene for SAK, a known activator of the zymogen PLG. Human tissue PLG activator was also tested. Bacterial cells (LAC and LAC srtA) were immobilized on the surface of microtiter wells and incubated with PLG and then incubated for 8 h with exogenously purified recombinant SAK (rSAK) or tPA and the chromogenic plasmin substrate S-2251. The substrate was consumed following incubation with the wild type, but minimal substrate consumption was seen with LAC srtA (Fig. 3, A and C). Controls where PLG was omitted did not have the ability to cleave the substrate. Interestingly, incubation of wild-type LAC with PLG and S-2251 alone showed a significantly higher cleavage of the substrate as compared with the controls. Conversely, although the level of substrate consump-tion was significantly lower than that observed with samples with exogenous SAK or tPA added, the assays reached the same level if left for longer periods (data not shown).
These data show that PLG bound to the cell surface by CWA proteins can be activated efficiently by exogenously added rSAK or tPA (Fig. 3, A and C, respectively), and furthermore, the immobilized cells can release an endogenous PLG activator, presumably SAK. Plasmin is a central component of the fibrinolytic system and can actively cleave fibrin and FBG. To determine whether PLG captured by CWA proteins on the surface of S. aureus can be activated by SAK and cleave FBG, immobilized S. aureus LAC cells were incubated with PLG. After washing, rSAK and human FBG were added. The integrity of FBG was assessed by SDS-PAGE and staining with Coomassie Blue. After 2 and 4 h of incubation, FBG was progressively cleaved (Fig. 3B). Samples where rSAK was omitted also cleaved FBG, which was presumably due to secretion of endogenous SAK (Fig. 3B, tracks D and F). When cells were incubated with tPA rather than rSAK, FBG was digested more efficiently. In fact, all three FBG chains were hydrolyzed compared with the mild hydrolysis observed when PLG was activated by SAK. This is probably due to the different mechanisms of PLG activation promoted by tPA and SAK, respectively ( Fig. 3D) (28).
Captured PLG Can Be Activated by Endogenously Expressed SAK-To determine whether cleavage of plasmin substrates S-2251 and whole FBG in the experiments described in Figs. 2 FIGURE 3. PLG captured by S. aureus LAC is functionally active. S. aureus LAC and the srtA mutant were immobilized on the surface of microtiter plates and then incubated with human PLG for 1 h at 37°C. After washing, cells were incubated with recombinant SAK (A) or tPA (C) and S-2251 for 8 h, and the activity of plasmin was determined measuring the absorbance at 415 nm. Controls lacking PLG or activator were included. Data points are the means of three independent experiments each performed in triplicate Ϯ S.D. Statistically significant differences are indicated (Student's t test; *, p Ͻ 0.05). S. aureus LAC wild-type cells were also immobilized on microtiter plates and incubated with human PLG for 1 h at 37°C. After washing, PLG was activated by addition of rSAK (B) or tPA (D) and the plasmin substrate FBG. Samples were incubated for 2 or 4 h, and FBG in the fluid phase was subjected to SDS-PAGE and stained with Coomassie Blue. The figure is the representative of three independent experiments. and 3, where exogenous rSAK or tPA was not present, was due to SAK expressed by the bacterium, a null mutation in the chromosomal sak gene of LAC was constructed (LAC sak). Culture supernatants and cell surface extracts were examined by SDS-PAGE and Western immunoblotting using anti-SAK IgG. SAK was present in the supernatant and is associated with the cell surface of LAC, and no SAK was detected in fractions from LAC sak (Fig. 4A). To determine whether endogenously expressed SAK is responsible for activating captured PLG, wild-type and mutant cells were compared (Fig. 4B). Cleavage of the substrate S-2251 was evident with wild-type cells, whereas no cleavage occurred with the sak mutant. Similarly, wild-type cells could capture and activate PLG and cleave FBG (Fig. 4C), whereas the sak mutant was defective. To demonstrate directly that LAC sak could still capture PLG, bacteria were incubated with dif-ferent concentrations of plasma, and the amount of PLG captured was quantified by Western immunoblotting and densitometry (Fig. 4D). No difference in PLG bound to the wild-type or sak mutant cells was observed. Thus, it can be concluded that SAK secreted by S. aureus LAC in the supernatant or bound to the cell surface is capable of activating captured PLG.
Fibronectin-binding proteins A and B Bind PLG-The inability of the srtA mutant of S. aureus LAC to capture PLG indicates that one or more sortase-anchored cell wall-associated protein(s) is involved. To begin the process of identifying which of the plethora of CWA proteins can bind PLG, purified recombinant ligand binding domains of several CWA proteins were tested by far Western blotting. Proteins were subjected to SDS-PAGE, and Western blottings were probed with purified human PLG and anti-PLG IgG. Subdomains N2 and N3 of the

. S. aureus cells-bound PLG can be activated by endogenous SAK.
A, TCA-precipitated culture supernatants and proteins extracted from the cell surface were subjected to SDS-PAGE and Western immunoblot probing with mouse anti-rSAK IgG and HRP-conjugated rabbit anti-mouse IgG. B, S. aureus LAC and the sak mutant were immobilized on the surface of microtiter plates and then incubated with human PLG for 1 h at 37°C. After washing, cells were incubated with S-2251 for 8 h, and the activity of plasmin was determined by measuring the absorbance at 415 nm. Controls lacking PLG were included. Data points are the means of three independent experiments each performed in triplicate Ϯ the S.D. C, S. aureus LAC wild-type and sak mutant cells were immobilized on microtiter plates and incubated with human PLG for 1 h at 37°C. After washing, the plasmin substrate FBG was added. Samples were incubated for 2 or 4 h, and FBG in the fluid phase was subjected to SDS-PAGE and stained with Coomassie Blue. The figure is the representative of three independent experiments. D, S. aureus strain LAC and sak mutant cells were mixed with different concentrations of human plasma for 60 min. Upper panel, proteins that were bound to the cell surface were released with extraction buffer and separated by SDS-PAGE under non-reducing conditions and analyzed by Western immunoblotting. The membranes were probed with rabbit anti-human PLG followed by HRP-conjugated mouse anti-rabbit IgG. Lower panel, densitometric analysis of released PLG. The band intensities were quantified relative to a sample of pure human PLG. The reported data are the mean values from three independent experiments.

N-terminal A regions of both fibronectin-binding proteins
FnBPA and FnBPB bound PLG, whereas the A domains of clumping factors A and B (ClfA and ClfB), the bone sialoprotein-binding protein, and the plasmin-sensitive protein did not (Fig. 5B). To investigate whether FnBPA and FnBPB are the sole PLG-binding proteins of S. aureus LAC, a mutant that lacks both proteins was tested for PLG capture. No statistically significant difference between the wild type and the fnbA fnbB mutant was seen (Fig. 5C), which indicates that one or more CWA protein(s) in addition to FnBPA and FnBPB can bind PLG. Strain LAC expresses low levels of FnBPs compared with the HA-associated strain BH1CC (12), so we decided to determine whether FnBPs contribute to PLG capture when FnBPs are highly expressed. Strain BH1CC showed higher PLG binding than LAC (Fig. 5D), and BH1CC-bound PLG that was acti-vated by either SAK or tPA cleaved more efficiently the artificial substrate S2251 compared with the LAC strain (data not shown). The fnbA fnbB mutant of BH1CC captured significantly less PLG than the parental strain ( Fig. 5D) showing the importance of FnBPs to the phenotype. Likewise, fnbA fnbB mutants of 8325-4 and LS1 showed a significantly lower ability to capture PLG from plasma. Furthermore, the fnbA fnbB deletion mutant of LAC that was transformed with a plasmid bearing either the fnbA or fnbB gene, which overexpressed FnBPA or FnBPB, respectively, showed a much higher potential to capture plasma PLG that the parental strain ( Table 1). All together these data indicate that FnBPA and FnBPB represent critical PLG receptors in S. aureus. In contrast, as reported for LAC, FnBPA and FnBPB apparently do not seem to be the predominant PLG-binding proteins in FIGURE 5. Recombinant cell wall-anchored proteins binding to PLG. Purified recombinant A domains of several CWA proteins of S. aureus were subjected to SDS-PAGE and either stained with Coomassie Blue (A) or transferred to a nitrocellulose membrane (B) and probed with human PLG followed by rabbit anti-PLG serum and then HRP-conjugated goat anti-rabbit IgG. C, S. aureus strains LAC and its fnbA/fnbB double mutant cells (upper panel) were mixed with different concentrations of human plasma for 60 min. Proteins that were bound to the cell surface were released by extraction buffer and separated by SDS-PAGE under non-reducing conditions and analyzed by Western immunoblotting. The membranes were probed with rabbit anti-human PLG followed by HRP-conjugated mouse anti-rabbit IgG and developed with the ECL Western blotting detection kit. D, similar experiment was performed with strain BH1CC and its fnbA/fnbB double mutant (upper panel). Densitometric analysis of PLG bound to the strains (lower panels of C and D). The band intensity was quantified relative to a sample of pure human PLG. The reported data are the mean values Ϯ S.D. from three independent experiments. In the insets the material released by treatment of bacteria with lysostaphin was subjected to Western blotting and probed with FN. Statistically significant differences are indicated (Student's t test; *, p Ͻ 0.05; **, p Ͻ 0.01).
strains P1 and SH1000, suggesting that additional CWA proteins are involved (Table 1).
Lactococcus lactis-expressing FnBPB Can Capture PLG-To study FnBPB in isolation from other S. aureus CWA proteins, a strain of L. lactis-expressing FnBPB from a gene cloned into the plasmid vector pNZ8037 was used. L. lactis pNZ8037::fnbB and the same strain carrying the empty vector were immobilized in microtiter wells and incubated with PLG, and following the washing steps, the chromogenic plasmin substrates S-2251 and rSAK (Fig. 6A) or tPA (Fig. 6C) were added. Cleavage of the substrate was observed, whereas the cells bearing the empty vector had no activity. It should be noted that this system was somewhat less active than S. aureus LAC cells. L. lactis (pNZ8037::fnbB) cells were also incubated with FBG and rSAK ( Fig. 6B) or tPA (Fig. 6D). SDS-PAGE analysis showed that FBG was cleaved by the FnBPB-expressing cells and not by those carrying the empty vector. These experiments show that L. lactis cells normally lack the ability to capture PLG and that FnBPB expressed on the bacterial cell surface can capture PLG in a form that can be activated by exogenous PLG activators.
PLG and FBG Bind FnBPB at Distinct Sites-The A domain of FnBPB binds to FBG by the dock, lock, and latch mechanism. To determine whether PLG and FBG bind FnBPB at the same or distinct sites, rFnBPB(163-480) was immobilized onto the surface of microtiter wells and tested for binding to saturating concentrations of PLG in the presence of increasing amounts of FBG by ELISA-type assay. Bound PLG was detected with specific antibodies, and the amount of PLG bound did not change as the concentration of FBG added increased (Fig. 7A). A similar result was obtained when saturating concentrations of FBG were tested in the presence of increasing amounts of PLG (Fig.  7B). In both panels, dose-dependent binding of increasing amounts of either FBG or PLG to FnBPB was also reported. Together, these data indicate that PLG and FBG bind to distinct sites on the A domain of FnBPB without steric hindrance. The failure of FBG to interfere with rFnBPB binding to PLG suggested that the ligands bind to different sites within the A domain of the protein. To investigate whether the dock, lock, and latch mechanism is involved in PLG binding and to further localize its binding site, variants of the recombinant A domain were tested for ligand binding in solid phase ELISA-type assays. A truncate lacking 17 residues at the C terminus of subdomain N3 that is involved in the locking and latching steps of the dock, lock, and latch mechanism (rFnBPB(163-463)) and a protein carrying amino acid substitutions in residues Asn-312 and Phe-314 of the FnBPB fibrinogen-binding trench (rFnBPB(163-480)-N312A/F314A) were tested for binding to FBG and PLG (Fig. 8, A-D). Neither the C-terminal truncate nor the trench mutant could bind FBG, but each bound PLG dose-dependently and saturably in a manner that was indistinguishable from rFnBPB(163-480). This confirms that PLG does not bind to the FnBPB A domain using the dock, lock, and latch mechanism. Finally, subdomains N2 and N3 were expressed separately (N2 subdomain rFnBPB(163-308) and N3 subdomain rFnBPB(309 -480)) and tested for binding to PLG and FBG (Fig. 8, E and F). Neither subdomain N2 nor N3 bound FBG detectably, whereas subdomain N3 bound PLG dosedependently and saturably with the same profile as N2N3 (rFnBPB(163-480)), which is also consistent with the two ligands binding to FnBPB at different non-overlapping sites. This suggests that the PLG-binding site of FnBPB is contained within the N3 subdomain.
Measurement of the Dissociation Constant of FnBPB Binding to PLG by Surface Plasmon Resonance-The affinity of the FnBPB for PLG was measured by surface plasmon resonance. Purified rFnBPB(163-480) was immobilized onto the surface of a dextran chip, and PLG was passed over the chip in concentrations ranging from 0.78 to 5 M (Fig. 9). The K D for the interaction was 0.532 Ϯ 0.028 M, an affinity that is nearly 4-fold higher than the interaction between FnBPB and FBG (33).
Localization of the Binding Sites within PLG and FnBPB-PLG binds to the N3 subdomain of FnBPB that was cloned from S. aureus 8325-4. There are seven isoforms of FnBPB N2N3 subdomains with amino acid sequence identities ranging from 61 to 85% (33). FnBPBs from strains 8325-4, LAC, and BH1CC have identical amino acid sequences and belong to isoform I. Each of the seven isoforms retains the ability to bind FBG, fibronectin, and elastin (33). The recombinant N2N3 subdomains of isoforms II-VII were purified and tested for their ability to bind to PLG in an ELISA-type assay. Each isoform bound PLG similarly to isoform I (Fig. 10A).
PLG binding to its natural ligands is attributed to strong affinity of kringle domain(s) for lysine-rich regions of target proteins. To investigate whether lysine residues are important in binding to FnBPB isoform I, PLG binding assays were performed with the recombinant bacterial protein in the presence of lysine and the lysine analog ⑀-aminocaproic acid (EACA). Both molecules caused dose-dependent and complete inhibition of the interaction, with EACA being far more potent (Fig 10B).
The structure of the N3 subdomain of FnBPB isoform I was modeled on the x-ray crystal structure of the related protein S. aureus strains and their fnbA/fnbB mutants were mixed with 5% diluted human plasma for 60 min. Proteins that were bound to the cell surface were released by extraction buffer and separated by SDS-PAGE under non-reducing conditions and analyzed by Western immunoblotting. The membranes were probed with rabbit anti-human PLG followed by HRP-conjugated mouse anti-rabbit IgG and developed with the ECL Western blotting detection kit. The densitometric analysis of PLG bound to S. aureus strains was performed as reported under "Experimental Procedures." The band intensity was quantified relative to a sample of pure PLG (5 g, 100% intensity). Statistically significant differences between WT strains and their corresponding mutants are indicated (Student's t test; **, p Ͻ 0.01). FnBPA. Amino acid sequence alignments of isotypes I-VII of FnBPB were used to identify conserved lysine residues in subdomain N3. The conserved lysine residues were visualized using the molecular model, and two conserved lysine-rich surface regions were identified. Alanine substitutions were isolated in site 1 (Lys-330, Lys-334, and Lys-362) and site 2 (Lys-342, Lys-374, and Lys-423) and then combined to form a double site 1-site 2 mutant. The variants were tested for their ability to bind to PLG and FBG. Both site 1 and site 2 mutants showed a significant reduction in binding to PLG, a difference that was even greater when site 1 and site 2 substitutions were combined. In contrast, FBG binding was not reduced significantly (Fig. 10C). This is consistent with PLG and FBG binding to FnBPB A domain at different sites.

JOURNAL OF BIOLOGICAL CHEMISTRY 18155
which is composed of kringle 5 along with C-terminal residues, were tested for binding to the bacterial protein. Kringles 1-3 and kringle 5 failed to bind FnBPB, whereas kringles 1-4 bound similarly to full-length PLG. Thus, kringle 4 seems to be the sole binding domain within PLG for FnBPB (Fig. 10D).

Discussion
S. aureus, like other invasive pathogens such as S. pneumoniae (13,14), S. pyogenes (15,16), and S. agalactiae (17,34,35) can capture PLG from human plasma. Staphylococcal proteins that bind the human protein have been described previously, the secreted and wall-associated protein Sbi (29), the MntC (30), and the moonlighting cytoplasmic protein enolase (31), but the biological significance of these interactions is not clear because experiments were performed either with purified protein in vitro or with cell lysates.
Here, we have shown for the first time that sortase A-anchored cell wall-associated proteins are the dominant PLGbinding proteins on the S. aureus cell surface. A sortase A mutant of S. aureus LAC bound at least a 10-fold lower level of PLG than the wild type. It is noteworthy that PLG captured from human plasma could be activated by exogenously added tissue PLG activator t-PA and SAK.
Two CWA proteins, FnBPA and FnBPB, were shown to bind PLG in vitro and in the case of FnBPB when expressed on the surface of the surrogate host L. lactis. However, deletion mutants of S. aureus USA300 LAC, P1, and SH1000 lacking FnBPs bound the same level of PLG as the parental strain, although mutants of BH1CC, 8325-4, and LS1 showed at least a 50% reduction in PLG binding, indicating that one or more of the other CWA proteins are also important. Thus, the contribution of FnBPs to PLG capture will vary depending on strain background and the level of fnbA fnbB gene expression, something that is influenced by the growth conditions of the bacterium. To identify the repertoire of PLG binding CWA proteins and to determine their relative importance in the capacity of the bacterium to bind PLG, it would be necessary to express recombinantly all 23 sortase-anchored proteins and to test them by far Western blotting and/or ELISA. This would need to be backed up by the isolation of mutants defective in the candidate proteins, both singly and in combination.
The A domain of FnBPB was chosen for detailed analysis of PLG binding. This domain can bind to FBG by the dock, lock, and latch mechanism (8). We have shown that PLG and FBG bind FnBPB at distinct non-overlapping sites and that PLG FIGURE 10. A, recombinant N2N3 subdomains of isoforms I-VII of FnBPB were immobilized in microtiter plate wells and tested for binding to PLG. Bound PLG was detected with rabbit antibodies to human PLG followed by HRP-conjugated goat anti-rabbit IgG. The data points are the means Ϯ S.D. of three independent experiments each performed in triplicate. B, rFnBPB(163-480) was immobilized onto microtiter plates and incubated with PLG in the presence of increasing concentrations of lysine and the lysine analog ⑀-aminocaproic acid. PLG bound to FnBPB was detected with a mouse polyclonal PLG antiserum and HRP-conjugated anti-rabbit IgG. Binding of PLG by FnBPB in the absence of inhibitor was set to 100%. Values represent the means of three separate experiments performed in triplicate, and error bars show S.D. C, microtiter plates coated with FnBPB(163-480) and its isoforms mutated at site 1, site 2, or both sites were incubated with PLG or FBG. Bound PLG was detected with rabbit antibodies followed by HRP-conjugated goat anti-rabbit IgG. Bound FBG was detected with mouse antibody followed by HRP-conjugated rabbit anti-mouse IgG. Binding of each ligand to the variants was expressed as percentage of the controls where binding of PLG or FBG to FnBPB(163-480) was set to 100%. The data points are the means Ϯ S.D. of three independent experiments each performed in triplicate. Statistically significant differences are indicated (Student's t test; *, p Ͻ 0.05; **, p Ͻ 0.01). D, microtiter plates coated with FnBPB(163-480) were incubated with equimolar concentrations of PLG or PLG fragments K1-K3, K1-K4, and mini-PLG. PLG and the fragments bound to FnBPB were detected by incubation of the plates with a rabbit PLG antibody followed by goat HRP-conjugated secondary IgG. Three independent experiments were conducted in triplicate, and graphs represent means Ϯ S.D.
binding does not involve the dock, lock, and latch mechanism. This is important because PLG and FBG are both components of blood plasma. FnBPB can simultaneously capture both PLG and FBG (a substrate for plasmin, once plasminogen becomes activated).
The PLG-binding site is confined to subdomain N3, and it does not overlap the FBG-binding site, and variants of the recombinant A domain with substitutions at two distinct conserved lysine-rich sites in subdomain N3 bound PLG with reduced affinity. X-ray crystal structure analysis of the FnBPB-PLG complex will provide molecular details of the interaction. The seven distinct isoforms of FnBPB each bound PLG with similar affinities in solid phase assays. This attests to the idea that PLG binding is an important property that has been retained during diversification of the N3 subdomain surface residues. It will be interesting to compare PLG binding by FnBPA and its seven isoforms.
Our data indicate that kringle 4 probably comprises the only binding domain within PLG for FnBPB and that this binds the two lysine-rich patches within subdomain N3 of FnBPB. Structural analysis of streptokinase binding to PLG showed that the bacterial zymogen activator bound in such a way as to avoid the kringle domains. SAK and streptokinase belong to the same protein superfamily even though they share little amino acid sequence similarity (36), so it seems likely that they interact with PLG at the same or a similar site leaving kringle 4 free to bind to FnBPB. It would be interesting to determine whether SAK can bind to and displace PLG from the FnBPB and stimulate the release of active plasmin or whether the host protease remains attached to the cell-bound MSCRAMM.
The presence of active plasmin on the surface of S. aureus cells most likely contributes to the pathogenesis of different infections. For example, degradation of host-or SAK-promoted fibrin clots could help bacteria to spread (37). Damage to subcutaneous tissue could contribute to the pathogenesis of skin soft tissue infections caused by CA-MRSA (38). Degradation of opsonins could contribute to the avoidance of neutro-phil-mediated phagocytosis and killing (20 -22). To investigate this further requires the employment of appropriate animal models, and to test the role of SAK in these processes requires the use of a transgenic mouse expressing human PLG.
In summary, we have shown that CWA proteins are the dominant bacterial receptors for capturing host PLG by S. aureus, and we have investigated in detail the mechanism of PLG binding by one such CWA protein FnBPB.

Experimental Procedures
Bacterial Strains and Culture Conditions-All strains are listed in Table 2 (39 -47). S. aureus was grown in brain heart infusion broth (VWR International Srl, Milan, Italy) at 37°C with shaking. L. lactis cells carrying the expression vector (pNZ8037) alone or harboring the fnbB gene (pNZ8037::fnbB) were grown overnight in M17 medium (Difco, Detroit, MI) supplemented with 10% lactose, 0.5% glucose, and chloramphenicol (10 g/ml) at 30°C without shaking. Cultures were diluted 1:100 in the above medium and allowed to reach exponential phase. Nisin (6.4 ng/ml) was added, and cultures were allowed to grow overnight at 30°C without shaking. In those experiments, where a defined number of cells were used, bacteria were harvested from the cultures by centrifugation, washed and suspended in PBS, and counted in a Petroff-Hausser chamber. Escherichia coli TOPP3 transformed with vector pQE30 (Stratagene, La Jolla, CA) or derivatives were grown in Luria agar and Luria broth (VWR) containing 100 g/ml ampicillin.
Plasmid and DNA Manipulation-Plasmid DNA (Table 3) was isolated using the WizardPlus SV miniprep kit (Promega, Madison, WI), according to the manufacturer's instructions, and finally transformed into E. coli TOPP3 cells using standard procedures (48). Transformants were screened by restriction analysis and verified by DNA sequencing (Eurofins Genomics, Milan, Italy). Chromosomal DNA was extracted using the bacterial genomic DNA purification kit (Edge Biosystems, Gaithersburg, MD). DNA encoding residues 163-480 of FnBPB incorporating alanine substitutions in site 1 (residues Lys-330,  (Table 4) were used to amplify the sequence for cloning into pQE30. Restriction digests and ligations were carried out using enzymes from New England Biolabs (Ipswich, MA) according to the manufacturer's protocols. Oligonucleotides were purchased from Integrated DNA Technologies (Leuven, Belgium) and are listed in Table 4. DNA purification was carried out using the WizardSV gel and PCR clean-up system (Promega).  (52), and plasmin-sensitive protein(49 -694) (53), were each previously expressed with His 6 N-terminal affinity tags using E. coli vectors and purified on a HiTrap chelating column (GE Healthcare) by Ni 2ϩ -chelate chromatography as described above. The E. coli clone expressing recombinant SAK protein was generously provided by Dr. S. H. M. Rooijakkers, University Medical Center Utrecht, The Netherlands, and the SAK protein was purified according to the methods described for isolation of CHIPS by De Haas et al. (54). Protein purity was assessed to be 98% by SDS-PAGE, Coomassie Brilliant Blue staining, and den-sitometry analysis. A bicinchoninic acid protein assay (Pierce) was used to measure concentrations of purified proteins.

Expression and Purification of Recombinant
Proteins-Human PLG was purified from plasma by affinity chromatography on a Lys-Sepharose column (55). Human FBG was obtained from Calbiochem. FBG was further purified on a gelatin-Sepharose column to remove contaminating fibronectin (FN). Human FN was purified from plasma by a combination of gelatin-and arginine-Sepharose affinity chromatography. To exclude the presence of trace amounts of contaminants, affinity-purified FN was spotted onto nitrocellulose membranes at different concentrations and overlaid with anti-FBG and anti-PLG antibodies (56). The purity of the proteins was assessed by 7.5-12.5% SDS-PAGE and Coomassie Brilliant Blue staining. Kringle 1-3 and Kringle 1-4 were purchased from Sigma and MyBiosource (San Diego), respectively. The mini-PLG (residues Val 442 -Asn 790 ) was obtained by digestion of PLG with porcine pancreatic elastase (Sigma), as described previously (57,58). Unless stated otherwise, all other reagents were purchased from Sigma.
Human Plasma-Normal human plasma was prepared from freshly drawn blood obtained from healthy volunteers with informed consent and permission of the ethical board of the University of Pavia (permit no. 19092013). After centrifugation, the plasma fraction was frozen in aliquots and stored at Ϫ20°C.
Antibodies-Polyclonal antiserum against purified PLG was raised in a rabbit by routine immunization procedures using purified human PLG as antigen. Polyclonal antisera against human FBG or recombinant SAK were raised in mice using purified or recombinant SAK as antigens. Purification of rabbit or mouse antibodies from sera was performed by affinity chromatography using protein G-Sepharose columns (GE Healthcare, Buckinghamshire, UK). The anti-human FN rabbit polyclonal IgG was purchased from Pierce. Rabbit anti-mouse or goat anti-rabbit horseradish peroxidase (HRP)-conjugated secondary antibodies were purchased from Dako Cytomation (Glostrup, Denmark).
ELISA-type Solid-phase Binding Assays-The ability of immobilized proteins (FnBPB(163-480) and its derivatives) to The effect of lysine and lysine analog EACA on FnBPB(163-480)/PLG interaction was examined incubating recombinant FnBPB(163-480) immobilized onto microtiter plates (0.5 g/well) with 1 g of purified PLG in the presence of increasing concentrations of lysine or EACA. PLG bound to FnBPB was detected with mouse polyclonal PLG antiserum and HRP-conjugated anti-rabbit IgG.
SDS-PAGE and Western Immunoblotting and Far Western Immunoblotting-Samples for analysis by SDS-PAGE were boiled for 3 min in sample buffer (0.125 M Tris-HCl, 4% (w/v) SDS, 20% (v/v) glycerol, 10% (v/v) ␤-mercaptoethanol, 0.002% (w/v) bromphenol blue) and separated by 10 -15% (w/v) PAGE. The gels were stained with Coomassie Brilliant Blue (Bio-Rad). For Western immunoblotting, proteins were subjected to SDS-PAGE and electroblotted onto a nitrocellulose membrane (GE Healthcare), and the membrane was blocked overnight at 4°C with 5% (w/v) skim milk (Sigma) in PBS. Blotted proteins were probed with rabbit polyclonal antibody against human PLG (1:5000) or mouse polyclonal antibody against SAK (5 g/ml) for 1 h at 22°C. Following washes with PBST, the membrane was incubated for 1 h with either HRP-conjugated goat antirabbit IgG (1:10,000) or rabbit anti-mouse IgG (1:10,000). Finally, blots were developed using the ECL Advance Western blotting detection kit (GE Healthcare), and an ImageQuant TM LAS 4000 mini-biomolecular imager (GE Healthcare) was used to capture images of the bands. When recombinant staphylococcal proteins were subjected to SDS-PAGE and far Western blotting, membranes were incubated with 1 g/ml PLG in PBST for 1 h at 22°C and washed, and the complexes were detected with anti-PLG antibody as reported above.
Preparation of Cell Surface-bound Proteins and TCA-precipitated Culture Supernatants-S. aureus USA300 was grown to exponential phase and washed, and ϳ1 ϫ 10 8 cfu were incubated for 1 h with the indicated concentrations of human plasma. Cells were harvested by centrifugation, washed with PBS, treated with extraction buffer (125 mM Tris-HCl, pH 7.0, containing 2% SDS) (10 l/mg wet weight of pellet) (59), for 3 min at 95°C, and then centrifuged at 10,000 ϫ g for 3 min. The supernatants were subjected to 10% PAGE under non-reducing conditions and transferred to a nitrocellulose membrane. The membrane was incubated with rabbit anti-PLG IgG followed by HRP-conjugated secondary antibody. The band intensities were quantified relative to the PLG sample (5 g, 100% intensity) with the Quantity One software (Bio Rad, Milan, Italy).
To detect SAK on the surface of S. aureus USA300 LAC and LAC sak, bacteria were treated with extraction buffer, heated at 95°C for 3 min, and subjected to 15% SDS-PAGE and Western blotting. SAK was detected by incubation with mouse anti-SAK IgG followed by a rabbit HRP-conjugated secondary antibody. Bacteria grown to exponential phase were removed from cultures by centrifugation at 4000 ϫ g for 5 min, and the supernatant was concentrated by treatment with 20% trichloroacetic acid (TCA) (v/v) at 4°C for 30 min. The precipitates were collected by centrifugation at 10,000 ϫ g for 10 min, washed with acetone, dissolved in the sample buffer, boiled for 3 min and finally subjected to 15% SDS-PAGE and Western blotting. SAK was detected as reported above.
Release of Cell Wall-anchored Proteins from S. aureus and Detection of Fibronectin Binding Activity-To release cell wallanchored proteins from S. aureus, bacteria were grown to an A 600 of 0.4 to 0.6, harvested by centrifugation at 7000 ϫ g at 4°C for 15 min, washed three times with PBS, and resuspended to an A 600 of 40 in lysis buffer (50 mM Tris-HCl, 20 mM MgCl 2 , pH 7.5) supplemented with 30% raffinose. Cell wall proteins were solubilized from S. aureus by incubation with lysostaphin (200 g/ml) at 37°C for 20 min in the presence of protease inhibitors (Complete Mini; Roche Applied Science). Protoplasts were recovered by centrifugation at 6000 ϫ g for 20 min, and the supernatants were taken as the wall fraction. The material was adsorbed on IgG-Sepharose column to remove protein A and subjected to SDS-PAGE and Western blotting incubating the membrane with human FN (5 g/ml) and a rabbit anti-FN antibody (2 g/ml). Surface Plasmon Resonance Analysis of PLG Binding to FnBPB(163-480)-To estimate the affinity of the interaction between PLG and FnBPB(163-480), surface plasmon resonance was conducted using a BIAcore X-100 instrument (GE Healthcare). FnBPB(163-480) was covalently immobilized on dextran matrix CM5 sensor chip surface by using an FnBPB(163-480) solution (30 g/ml in 50 mM sodium acetate buffer, pH 5) in a 1:1 dilution with N-hydroxysuccinimide and 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride. The excess of active groups on the dextran matrix was blocked using 1 M ethanolamine, pH 8.5. On another flow cell, the dextran matrix was treated as described above but without any ligand to provide an uncoated reference flow cell. The running buffer used was PBS containing 0.005% (v/v) Tween 20. A 2-fold linear dilution series (0.078 -5 M) of PLG in running buffer was passed over the ligand at the flow rate of 10 l/min, and all the sensorgrams were recorded at 22°C. Assay channel data were subtracted from reference flow cell data. The response units at steady state were plotted as a function of PLG concentration and fitted to the Langmuir equation to yield the K D of the PLG-FnBPB(163-480) interaction.
Effect of Bacteria Bound PLG on the Cleavage of Chromogenic Substrate S-2251 and Degradation of Human Fibrinogen-S. aureus LAC (5 ϫ 10 7 cells) isolated from exponential phase or L. lactis-expressing FnBPB were suspended in PBS and immobilized onto microtiter wells at 37°C overnight. After several washings with PBST, the cells were incubated for 1 h with human PLG (10 g/well) in PBS at 37°C. After washing with PBST, bacterially bound PLG was activated by addition of recombinant SAK (15 g/ml) or 27 nM tPA, incubated with 0.6 mM chromogenic substrate S-2251 (H-D-valyl-L-leucyl-L-lysine-p-nitroaniline dihydrochloride; Chromogenix), dissolved in 100 mM HEPES, pH 7.4, 100 mM NaCl, 1 mM EDTA, 1 mg/ml PEG 8000 for 8 h, and the reaction followed spectrophotometrically at 415 nm.
Alternatively, surface-coated cells of S. aureus LAC or L. lactis were incubated for 90 min at 37°C with PLG (1 g/well) diluted in PBS. After several washings with PBST, FBG (10 g/well) together with tPA (27 nM) or rSAK (15 g/ml) in 100 mM HEPES, pH 7.4, 100 mM NaCl, 1 mM EDTA, 1 mg/ml PEG 8000 were added to the wells, and the mixtures were incubated at 22°C for the indicated times. The reaction was stopped by the addition of reducing sample buffer and boiling for 3 min. Thereafter, the samples were separated by 12.5% SDS-PAGE, and degradation of FBG was evaluated by Coomassie Blue staining.
Statistical Methods-Continuous data were expressed as means and standard deviations. Two group comparisons were performed by Student's t test. One-way analysis of variance, followed by Bonferroni's post hoc tests, was exploited for comparison of three or more groups. Analyses were performed using Prism 4.0 (GraphPad). Two-tailed p values of 0.05 were considered statistically significant.
Author Contributions-G. P., T. J. F., J. A. G., and P. S. designed the study; G. P., G. N., V. G., and M. Z., performed the research; G. P., G. N., V. G., M. Z., T. J. F., J. A. G., and P. S., analyzed the data; G. P., T. J. F., J. A. G., and P. S., wrote the paper. All authors reviewed the results and approved the final version of the manuscript.