Variable region in streptococcal M-proteins provides stable binding with host fibrinogen for plasminogen-mediated bacterial invasion

Dimeric M-proteins (M-Prt) in group A Streptococcus pyogenes (GAS) are surface-expressed virulence factors implicated in processes that contribute to the pathogenicity of infection. Sequence analyses of various GAS M-Prts have shown that they contain a highly conserved sortase A-dependent cell wall-anchored C terminus, whereas the surface-exposed N terminus is highly variable, a feature used for identification and serotyping of various GAS strains. This variability also allows for strain-specific responses that suppress host defenses. Previous studies have indeed identified the N-terminal M-Prt B-domain as the site interacting with antiphagocytotic human-host fibrinogen (hFg). Herein, we show that hFg strongly interacts with M-Prts containing highly variable B-domains. We further demonstrate that specific GAS clinical isolates display high affinity for the D-domain of hFg, and this interaction allowed for subsequent surface binding of human-host plasminogen (hPg) to the E-domain of hFg. This GAS surface-bound hPg is then activated by GAS-secreted streptokinase, leading to the generation of an invasive proteolytic bacterial surface. Our results underscore the importance of the human fibrinolytic system in host-pathogen interactions in invasive GAS infections.

The Gram-positive bacterium, Streptococcus pyogenes, or group A Streptococcus (GAS) 2 is responsible for wide-ranging pathological outcomes in its human hosts. Infections are commonly manifested as mild skin or upper respiratory tract infections, i.e. impetigo and "strep throat." However, the World Health Organization estimates over 18 million severe complications annually, including necrotizing fasciitis, toxic shock syndrome, and secondary sequelae rheumatic fever (1). Critical to the survival and dissemination of these bacteria are their interactions with host systems. Of particular note is the human fibrinolytic system, which has previously been implicated in multiple bacterial infections, including enhanced GAS virulence (2,3). The major bacterial virulence factors responsible for this interaction are surface-expressed M-proteins (M-Prt) and the secreted protein streptokinase (SK).
M-Prt is a sortase A-dependent cell wall (CW)-anchored protein. Although there are very little rigorous structural data available on the majority of M-Prts that have been isolated, it is believed that these proteins exist as amphipathic ␣-helical parallel coiled-coils (4). This dimeric structure is composed of a series of distinct amino acid repeat regions, designated A-D (Fig. 1A), which increase in their level of conservation among strains from the N to C termini. In the past, GAS strains have been identified via serological sensitivity against the extreme highly variable N terminus of these proteins. More recently, nucleotide and amino acid sequencing have allowed for more stringent classification of these proteins, resulting in over 250 identified strains (5,6). Furthermore, GAS strains can also be classified by chromosomal arrangement of M-like proteins (ML-Prts) in the multiple gene activator regulon (the emm pattern). Patterns A-E are assigned based on the presence and arrangement of specific M-Prts and ML-Prts (7).
During the course of an infection, M-Prts play various roles in maintaining bacterial survival against host defenses. The exact nature of its interactions with host proteins is specific to various anatomical niches in which the bacteria are present. M-Prts and ML-Prts can function as epithelial and endothelial cell adherence mediators, as suppressors of the innate immune response in tissues, and as receptors for surface protease recruitment, all of which enhance GAS strain-specific interactions with host components and, ultimately, bacterial virulence. The multiplicities of functions that M-Prts possess also render this class of proteins as important targets for GAS virulence.
SK solely functions to convert human host plasminogen (hPg) to the serine protease, plasmin (hPm). SK consists of three distinct structural domains designated ␣, ␤, and ␥. Whereas the ␣ and ␥ chains of SK are homologous among GAS strains, the ␤-domain shows the most variation (8 -10). Based on sequence analyses of the ␤-region of SK, GAS strains may be classified into two main groups, cluster 1 and cluster 2, the latter of which is further subdivided into clusters 2a and 2b (Fig.  1B). Functional consequences accompany differences in the ␤-domains of SKs that relate closely to emm patterns. Cluster 2b SKs, virtually exclusively present in pattern D strains, only effectively activate hPg that is bound to the GAS surface, whereas Cluster 1 SKs do not have cell surface dependence and activate hPg equally well in solution and on the cell surface (11)(12)(13). Cluster 2a SKs, commonly associated with pattern A-C strains, are intermediate, viz. they function in solution but are enhanced in activation capabilities when hPg is bound to GAS.
Consequently, two distinct mechanisms have been proposed in the recruitment of hPg/hPm to the surface of GAS, viz. (a) direct binding of hPg to its GAS M-Prt receptor (M53; PAM) initially found on strain 53 pattern D GAS (14). Homologous PAM-like M-Prts are exclusively found on all pattern D strains, and (b) indirect binding, with fibrinogen (hFg) acting as an anchor for hPg/hPm binding in some pattern A-C and E strains. Although the tight binding of hFg to M-Prts has been generally reported to occur through the B-region, no singular binding motif has been identified, in contrast to the case of hPg binding to PAM, where a small repeat region within the A-domain of PAM is essential for its tight interaction with hPg/hPm (15,16). Previous detailed structural and functional data on hFg binding to M-Prts have primarily focused on M1 and M5 proteins (17)(18)(19). Therefore, it is important to determine the nature of hFg binding to other clinically relevant strains of GAS and to assess how the dramatic sequence variability of the M-Prt B-domain affects the binding capabilities of these proteins to hFg and subsequently to hPg/hPm, a process that is critical to GAS virulence.

Results
It has been established that the M-Prt, PAM, present in strains that bind hPg directly, is co-inherited with the SK2b subtype (11,20,21). Although PAM-containing strains are more common in invasive skin infections, strains implicated in recent streptococcal outbreaks in the United States and Europe express other M-Prt types (19,22). These include M1, M3, and M18 strains, which are more associated with nasopharynx-associated invasive infections and rheumatic heart disease and secrete the SK2a subtype (11,21). One goal of this study was to establish whether SK2a-containing strains display similar functional co-inheritances as observed in SK2b strains (21). Herein, we document that M-Prts associated with these strains are dependent upon hFg to recruit hPg to their surface and utilize SK2a for hPg activation. For this study, M-Prts were first cloned and expressed using the primers listed in Table 1 to investigate their binding to hFg. Strains used in this work have been studied, and their genomes have been sequenced and annotated (23)(24)(25)(26)(27)(28)(29). The M-Prts used are illustrated in Fig. 1A and listed in Table 2. To confirm their identities, the calculated and MALDIdetermined molecular weights of the particular constructs of these proteins are also compared in Table 2. The experimental values obtained are consistent with the calculated values of the constructs.
Previous studies have highlighted the importance of the coiled-coil structure of M-Prts in its functions, particularly in their abilities to bind ligands, such as hFg (30). However, this generalization is based on a crystal structure of a central region of one M-Prt, M1, and likely is not applicable to other M-Prts containing variable N termini. Based upon X-ray crystallographic data from M1 binding to Fg, it has been postulated that irregularities in the heptad repeats of these proteins, and consequent kinks in the ␣-helix, at least in the case of M1, have been implicated in disrupting its regular coiled-coil structure, thus allowing ligands to bind (17).
Because of the high sequence variabilities in different M-Prts, a number of these proteins that interact with Fg were analyzed by CD to determine the extent of their ␣-helicities. As anticipated, two local minima in the CD spectrum at wavelengths of 208 and 220 nm were observed (Fig. 2), which are the hallmarks of ␣-helical structure (31). Thus, all recombinant M-Prts maintained the expected 2°structure at least to some small extent. It was observed that these M-Prts varied considerably in their helical contents, with M1 and M3 containing the highest ␣-helical content, and M5, M6, M18, and M23 containing low to very low levels of overall ␣-helix. Not only do the N-terminal sequence variabilities of MPrts argue against a universal M-Prt helical model, but the C-terminal regions of all M-Prts contain numerous conserved Gly and Pro residues, which would also reduce their helical contents. Nonetheless, whereas the CD spectra showed major differences in the helical contents of the r-M-Prts, analytical ultracentrifugation data demonstrated that all of the r-M-Prts assembled as dimers with only a single molecular weight species present throughout the concentration gradient in the cell ( Table 2).

M-Prts bind to hFg and hFg degradation products
Subsequent to structural characterization by CD, surface plasmon resonance (SPR) was used to observe the binding properties of various M-Prts with hFg and its associated hFg degradation products. Although not rigorously quantitated, hFg binding had previously been reported for both M1 and M5 proteins (32). Illustrating one example with r-M1 (Fig. 3A), we show that tight hFg binding, with K D values ranging from 2.3 to 22 nM, occurred with r-M1, r-M3, r-M5, r-M6, r-M18, and r-M23 (Table 3). We note that these values are empirical estimates of the binding interactions in these cases, because the fit of the curves (Fig. 3) suggests a more complex binding/dissociation event, perhaps more than one binding site within the B-repeats. Of particular note, these tight interactions were predominantly attributed to slow k off rates of hFg on the various M-Prts. A notable exception was observed for M5, which possessed faster k on and k off values that nonetheless combined to present a similar K D value for Fg binding. These kinetic binding features were also shown by mouse Fg, which interacted with r-M1 with a K D similar to that of hFg, within experimental error. These data clearly demonstrate the ability of these M-Prts to bind to hFg with high affinity, despite their sequence diversities in the variable regions. However, as anticipated, PAM, present on the pattern D skin-tropic GAS strain, AP53, did not show direct binding to hFg.
hPm degradation of hFg results in two major fragments: the central fragment-E (FgE) domain, and two larger fragments corresponding to the distal fragment-D (FgD) domains. FgD has been previously implicated as the major component of hFg responsible for hPm accumulation by GAS (33). To directly and Figure 1. A, schematic of M-Prts used in this study. Modular structures of the M-Prts from the N to C termini. The conservation in amino acid sequences is illustrated as light (low conservation) to dark (high conservation). The A and B repeats share very little homology among strains and serve as the major sites of interaction with host proteins. The C and D repeats are generally well conserved, whereas the glycine/proline-rich region (labeled Pro) and the LPSTG cell wall-anchoring region at the extreme C terminus are virtually identical in all M-Prts. B, phylogenic analysis of SK ␤-domains. The ␤-domains of SKs were aligned using the BLAST sequence alignment tool (NCBI). Phylogenic analyses were then carried out using Molecular Evolutionary Genetic Analysis software (MEGA 6.0) (56). Based on these alignments, three distinct branches emerge, designated cluster (C)-1, cluster-2a, and cluster-2b, respectively. The tree shown here is based on previous analyses (21).

Fibrinogen binding to S. pyogenes M-proteins
quantitatively determine the portion of hFg that interacts with M-Prts, FgD and FgE were prepared by plasmin digestion of hFg, and SPR binding studies were carried out. It was demonstrated that all M-Prts studied here, except PAM, displayed comparably high affinity for FgD as seen with full-length hFg ( Table 4 and Fig. 3B). Again, the high affinity observed in FgD binding is mainly attributed to the slow k off of binding, except for that of r-M6 that dissociated with higher k off and k on values than the other M-Prts studied, again combining to yield a K D value suggestive of strong affinity. In a similar set of experiments, the ability of M-Prts to interact with the FgE domain of hFg was determined. FgE binding to M-Prts was not detected, with the lone exception being r-M18 protein. r-M18 binding to FgE was observed with K D of 0.83 M (Fig. 3C). In contrast, M18 binding affinity to FgD was ϳ100-fold tighter, with a K D of ϳ7.9 nM. It was therefore concluded that FgD is the major constituent contributing to M-Prt binding to Fg, rather than FgE.

hFg binding to GAS cells
To further investigate the ability of GAS to recruit hFg and, in turn, hPg to the cell surface, binding analyses were carried out on mid-exponential growth phase (EP) cultures using flow cytometric analysis (FCA). The binding of FITC-labeled hFg to the M1-containing GAS strain, 5448, was arbitrarily assigned a value of 100%, and all other strains were compared with this isolate. Using this approach, most strains, including M5, M6, M18, and M23 GAS, showed high levels of binding to hFg, at 60 -110% (Fig. 4A). However, the M3 strain MGAS315, as well as PAM-containing AP53, showed very low levels of hFg binding compared with 5448. The lack of appreciable binding of hFg to AP53 cells is consistent with the fact that this strain contains PAM as its M-Prt and therefore binds hPg, and not hFg, directly. In the case of MGAS315, because isolated r-M3 binds tightly to this strain (Table 2), it is hypothesized that M3 expression is highly attenuated in this MGAS315, accounting for the lack of hFg binding. Indeed, Western blotting analysis of cell wall extracts (CWE) confirmed the lack of robust M3 expression in the MGAS315 strain (Fig. 4B).

hPg binding to GAS cells
The PAM-containing strain, AP53, was used as a positive indicator of hPg binding by FCA. According to our unified approach, it was anticipated that the non-PAM-containing strains would not be able to bind hPg directly. However, it is proposed that hPg binding would be observed in the presence of hFg. As anticipated, most SK2a-expressing strains showed relatively low hPg binding when compared with AP53 ( Fig. 4C), ranging from 3 to 30%. Some low-level binding is probably due to other hPg-binding proteins on GAS, e.g. enolase, a property independent of the major M-Prt. Unexpectedly, the consistent outlier was the M23ND strain, which showed 70% binding relative to AP53 (Fig. 4C). Further protein binding experiments using SPR revealed that the M-Prt of this strain, r-M23, displayed high affinity for hPg (K D ϳ21 nM; Fig. 4D). Sequence alignment of M23 with PAM showed no significant homology within the hPg a1a2 region of PAM that is responsible for hPg binding (Fig. 4E). Thus, the mode of binding of hPg to r-M23 is not similar to the PAM-binding model. Direct binding of hPg to M-Prts is only observed in strains that contain PAM as the serotype determinant M-Prt or PAM homologues that exist in other pattern D GAS strains. Additional hPg receptors in GAS, viz. streptococcal enolase and glyceraldehyde-3-phosphate dehydrogenase (GAPDH/Plr) that serve as essential glycolytic enzymes are common to all GAS strains but have a considerably lower affinity for hPg compared with PAM (34). Here, we report that M23 is a novel high affinity hPg-binding protein, also capable of interacting with hFg. To our knowledge, this dual binding ability has not been reported in any other M-Prt.
The binding of hFg to hPg is attributed to association with the lysine-binding sites of hPg kringles (K) primarily K1 and K4. On the surface of GAS it is proposed that hFg, while in contact with M-Prt, also engages hPg in this fashion, thus acting as an intermediate in hPg acquisition. We clearly demonstrate that in the presence of hFg, SK2a-containing strains are able to bind hPg on their surfaces at levels comparable with that of the PAM-containing strain AP53 (Fig. 5, A and B). It is noted that M5-Manfredo, a cluster 1 SK-containing strain, was also able to employ this mechanism of hPg recruitment. Also, some GAS strains displayed hFg-binding levels at or exceeding that of the GAS 5448 strain. M23 binds considerably more hFg in the presence of hPg than when hPg is omitted. One possible explanation for this is that M23 may have a preference for the hFg⅐hPg complex versus hFg or hPg alone. Again, MGAS315, which showed no appreciable hFg binding, also showed no hPg binding when both proteins were present, undoubtedly due to the near absence of expression of M-Prt in this strain.

hPg activation on GAS cells
The experiments above with several pattern A-C strains demonstrated the ability of these strains to acquire hPg on its surface via hFg. It is therefore important to establish whether such an interaction would result in surface activation of hPg and hPm enzymatic activity. Whole-cell hPg activation assays were performed in which cells were incubated with hPg alone or in the presence of hFg. The cells were washed with PBS, and the hPm activity that was generated after addition of a catalytic

Fibrinogen binding to S. pyogenes M-proteins
amount of SK2a (the SK subtype secreted by these cell lines) was monitored by cleavage of the chromogenic substrate S2251. The SF370-derived SK2a was used as a representative SK for all strains. The washing steps after hPg/hFg incubations were intended to remove excess or non-bound hPg from the reaction. Because hPg does not bind directly to any of the pattern A-C strains (except for M23ND), the cells obviously did not retain hPg after washing, and no activity was observed under these assay conditions (Fig. 6A). However, AP53 did show hPm activity in this assay because its M-Prt (PAM), as anticipated, does retain hPg after washing the cells, showing that SK2a activates GAS-bound hPg.
When this same assay was performed with cells that were incubated with both hFg and hPg prior to washing, activity was observed proportional to the ability of these cells to express M-Prt (Fig. 6B). These data support the observations that these cells interacted with hPg through interaction with hFg. Again, MGAS315 showed no appreciable activity, which also corre-lates with the binding assays that showed an attenuated expression of M3 in this strain. It is especially important to note that although M23ND binds hPg alone considerably, activation on the surface of this strain still requires the presence of hFg, suggesting that the nature of the direct binding of hPg to M23 is not productive.
The ability of GAS to effectively engage host hPg via cellbound hFg has been implicated in the pathogenicity of some GAS strains. For example, cleavage of fibrin nets meant to entrap the bacteria (35), activation of matrix metalloproteinases (36), which allows GAS to penetrate the extracellular matrix, and inhibition of C3b-dependent opsonophagocytosis (37,38) have been shown to be consistent with enhanced dissemination of the bacteria. These data demonstrate that SK2acontaining strains are able to acquire hPm activity only in the presence of hFg.

Enhancement of SK-hPg activation by M-Prts
Previous studies have demonstrated that r-PAM greatly enhances SK-mediated activation of hPg (39). Furthermore, it has also been demonstrated that hFg makes a similar contribution to activation (40). Because M-Prts have been shown to be cleaved from the bacterial surface (41), it was of interest to determine whether the presence of other soluble M-Prts affected the SK2a-mediated activation of hPg in solution. As demonstrated in Fig. 6C, hFg enhanced the SK2a-mediated activation of hPg by ϳ2-fold. However, in the presence of M-Prts, further stimulation was observed, resulting in a 7-9fold enhancement of activation. This stimulatory effect was not observed in the absence of GAS pattern A-C M-Prts, suggesting that the complex with hFg, and not M-Prt alone, is required for enhancement of hPg activation. This is in contrast to PAM, which binds hPg and stimulates activation directly, and Fg does not further stimulate this activation rate.

Assembly of the fibrinolytic system on GAS cells
The trimolecular M-Prt⅐Fg⅐hPg complex employs primarily K1 and K4 of hPg, and not K2, to interact with hFg. SPR measurements show that hPg preferentially interacts with the E-domain of hFg (K D ϭ 0.41 Ϯ 0.01 M), rather than the D-domain of hFg (K D ϭ 11.2 Ϯ 0.6 M), leading to the models of assembly of the fibrinolytic system shown in Fig. 7 for two disparate types of M-Prt-based assembly. The secreted subtype of SK then

Discussion
M-Prts are among the most widely studied virulence factors associated with GAS. Previous studies have shown that GAS virulence is significantly reduced in the absence of M-Prt, implicating that they are crucial for GAS infection. The high degree of variability of M-Prts has many functional consequences in subverting the innate immune response. This study focuses on one such aspect of virulence, viz. the ability of specific GAS strains to assemble fibrinolytic components on the bacterial cell surface, which would enhance the invasive properties of this strain. The pattern A-C nasopharynx-tropic strains studied here require hFg-binding M-Prts, and their coevolved SK2a, to engage the host fibrinolytic system and to provide a proteolytic surface for dissemination.
Attempts have previously been made to propose a common binding motif associated with the M-Prt B-domains for hFg binding. However, very little sequence homology exists in the B-domains of Fg-binding M-Prts. Furthermore, as demonstrated here, the number of B-repeats between strains also varies significantly. Despite such diversity, the results provided in this communication show that each of the M-Prts associated with SK2a strains bind hFg with high affinity. It was confirmed that these proteins also bind primarily to the D-domain of hFg. As expected, PAM from AP53 did not bind to either hFg or FgD, confirming previous observations that PAM interacts directly and with high affinity to hPg.
The evidence presented here demonstrates that multiple FgD-binding motifs must exist in M-Prts. The hFg-binding region of M-Prts has been identified as the B-repeats, which are part of the variable N terminus. Recent studies have suggested Although significant amounts of PAM were detected in the cell wall extract of AP53, no M3 was observed from the cell wall extracts of MGAS315. The gels on the left and right were run at different times, and the position of the 50-kDa marker is shown on each. C, hPg binding to the indicated r-M-Prts. Binding was detected using mouse anti-hPg 1°antibody and AlexaFluor 488 goat anti-mouse IgG as the 2°antibody. D, r-M23 protein binding to hPg by SPR. r-M23 was immobilized to a CM5 chip via thiol coupling, utilizing a C-terminal Cys residue engineered into the r-M23 protein construct. This ensures that the majority of r-M23 is immobilized in a uniform manner that reflects the positioning of M-Prt on the cell surface. Curves were fit to a 1:1 Langmuir binding model. k on ϭ 2.9 Ϯ 0.7 x 10 4 M Ϫ1 s Ϫ1 ; k off ϭ 5.8 Ϯ 1.3 ϫ 10 Ϫ4 s Ϫ1 ; K D of 20 Ϯ 0.7 nM. E, amino acid sequence alignment of the M23 and PAM protein ligand binding domains. Top, hPg-binding peptide from the a1a2 region of AP53-derived PAM aligned with the same region of M23 protein (bottom) using the MUSCLE 3.8 multiple sequence alignment tool (57). Residues known to be critical for PAM hPg binding are shown in bold. * indicates identical residues shared between proteins. Ϫ indicates gaps in the sequence alignments. Figure 5. Binding of hPg to hFg on GAS cells by FCA. Cells were grown to mid-EP (OD 600 nm ϳ0.6), washed, and incubated with a mixture of preincubated hPg (20 g/ml) and FITC-hFg (40 g/ml). A, hFg/hPg co-incubation detecting FITC-hFg on the various cell lines indicated. As a control for hFg binding, cells were also incubated with unlabeled hFg. B, hFg/hPg co-incubation detecting hPg using mouse anti-hPg 1°antibody and AlexaFluor 488 goat anti-mouse IgG 2°antibody. The controls for hPg binding included the antibody isotype control (ITC), as well as assaying cells in the absence of protein, 1°, and 2°antibodies (data not shown). Data from 10,000 events were collected using a doublet discrimination mode, gating on side scatter SSC-H and fluorescence amplitude on a log scale. The percent binding was calculated using the median of the area under the curve of FCA histograms produced in FCA Express version 4 software.

Fibrinogen binding to S. pyogenes M-proteins
that the high sequence variability within B-repeats may be due to immunogenic pressure (42). Our data demonstrate that, despite this pressure, their ability to bind to hFg is maintained. However, it is unlikely that the same site within the Fg D-domain is engaged by all M-types. Given the retention of this property, this further highlights the importance of the recruitment of hFg, and in turn hPg, for invasive GAS infection.
Although others have demonstrated that the M1 protein can be cross-linked to fibrin through the action of FXIIIa (43), it is unclear whether this is a universal mechanism for Fg-binding M-Prts. However, such results highlight the importance of this interaction early in an infection.
Another relevant observation suggests that the binding of hFg to M-Prt does not inhibit complement deposition in M6 strains, unlike M5 strains (18). It is possible that some strains utilize fibrin as an adherence molecule, and therefore they bind tightly during the early stages of infection. Later, bacterial and host protease activity may free the bacteria from clot entrapment. Alternatively, other strains may acquire circulating hFg to prevent complement deposition (44). Evidence suggests that M-Prts may also be cleaved from the bacterial surface by host and endogenous proteases, such as the cysteine protease, SpeB, while remaining biologically active. This may provide a means by which GAS directs immune responses away from the cell surface (45). Thus, it is hypothesized that strains have acquired Fg binding for the benefit of different stages of infection that are likely to be strain-and niche-specific. Nonetheless, one unifying feature highlighted by this study is the ability of these M-Prt-directed Fg-binding bacteria to recruit hPg to its surface. Additionally, we found that all Fg-binding M-Prts studied here further stimulated hPg activation in solution. Our observance of the relatively slow off-rates that govern hFg binding and the subsequent ability of this interaction to influence hPg activity away from the cell surface support previous findings of M-Prts in complex with hFg circulating in the plasma of patients with streptococcal toxic shock syndrome (46). This observation further reinforces the importance of hPm activity in GAS virulence.
Whole-cell binding studies using FCA demonstrated that direct binding of Fg was observed on the surface of GAS cells. In the presence of hPg alone, most pattern A-C GAS strains bound hPg minimally. More importantly, it was demonstrated that for most of these strains that utilize hFg binding for hPg recruitment, hPg was localized to the surface at a level comparable with that of PAM-containing AP53, which binds hPg directly. These data reinforce the idea of two distinct but equally effective mechanisms of hPg acquisition by GAS and refine our knowledge of the indirect mechanism of stimulation of hPg activation by other M-Prt types associated with common virulent strains. One notable exception to this was the MGAS315 strain. This strain unexpectedly showed no binding to hFg or hPg. Furthermore, survival assays utilizing this strain showed ϳ50% survival after 10 days, at the same CFUs that provided 100% lethality after 2 days with AP53 strains. 3 Previous observations using AP53(⌬PAM), which lacks PAM expression, also displayed ϳ50% lethality (47). Western blotting analysis of CWEs and supernatants showed that the M3 protein of this strain was not produced, despite the fact that intact emm3 was fully present in this strain (MGAS315). This suggests that the emm3 gene is present and intact, but expression and or trafficking of the mature M3 protein is defective. Others have demonstrated that other strains of the M3 serotype 3 J. A. Mayfield, unpublished data.

Fibrinogen binding to S. pyogenes M-proteins
may contain mutations within the mga promoter, responsible for regulation of M3 protein (48). It is therefore possible that relative M-Prt levels in these strains may be lower than other more invasive serotypes.
Another unexpected observation was the novel finding that M23 from GAS M23ND bound to both hFg and hPg. No other M-Prt has previously been described that directly binds to both of these plasma proteins. Although a significantly tight binding constant was observed for hPg to r-M23, a 20-fold reduction in affinity was observed when compared with the known direct binding PAM protein. Furthermore, a comparison between the variable regions of these two proteins showed minimal sequence overlap. Residues that have been identified as critical for the interaction of PAM with the kringle-2 region of hPg, viz. Arg 101 /His 102 and Arg 114 /His 115 of the A-domain region (39,40), are not found in M23. It is therefore unlikely that M23 binds to hPg in the same manner as PAM, thus possibly explaining the lack of stimulation of hPg activation by SK2a after binding of hPg to M23. Whole-cell binding assays recapitulated the results of the recombinant protein-protein binding studies in that M23ND cells bound to hPg, despite not expressing a PAM homologous protein. Additionally, further examination of these interactions showed that the activation of hPg on the M23ND surface is observed only in the presence of hFg. This is consistent with the hypothesis that SK2a-containing strains require hFg for hPg activity and likely co-evolved in this manner.
Other pathogenic bacteria, such as Staphylococcus aureus, have also shown variation in the ligand-binding domains of hFg receptors. Recent studies have also demonstrated that S. aureus is able to simultaneously bind hFg and hPg via fibrinogen-/ fibronectin-binding protein B (49). Similarly, it is clear that despite high variability of the M-Prt primary sequence, selective pressure has ensured that hFg binding is maintained. Notably, hFg also plays an important role in the host defense system. Mice lacking hFg were found to be more susceptible to GAS infection in a subcutaneous infection model than those expressing normal levels of hFg (50). The fibrin clot acts as a first line of defense against dissemination, trapping bacteria and bringing them into contact with immune cells to be cleared from the host system. The high mortality of hFg-deficient mice may be due to circumvention of this barrier, allowing for the rapid spread of the invading bacteria. The overlapping functions of hFg between coagulation, fibrinolysis, and the innate immune response make this protein an important consideration for the study of bacterial pathogenesis. The data presented here highlight the role of hFg in enabling fibrinolysis on the GAS surface in strains that are unable to bind hPg directly.
In conclusion, this investigation sought to directly expand our understanding of M-Prt-hFg interactions in the context of hPg acquisition and activation. The data presented provide important insights regarding the molecular mechanisms involved in the indirect pathway that GAS employs to productively bind and activate hPg. We also disambiguate some of the assumptions regarding the nature of hFg binding to M-Prts in that the sequence, length, and number of B-repeats do not affect the ability of M-Prt to strongly interact with hFg.

Experimental procedures
Plasma proteins and GAS strains hFg was obtained from ERL (South Bend, IN). GAS strains SF370, M5-Manfredo, and AP53 were provided by G. Lindahl (University of Lund), and JRS4, MGAS315, and MGAS8232 were from J. R. Scott (Emory University). M23ND was obtained from M. J. Walker (University of Queensland). Strain SF370 was purchased from the ATCC.

Phylogenic analysis
Phylogenic trees were constructed using the nucleotide region corresponding to the ␤-domain of SK, as described (11,21). In addition to the strains included in this study, other strains were included for statistical rigor.

Cell wall extraction and analysis
Bacterial CWE were isolated as outlined (51) from mid-EP cultures. CW components were subsequently subjected to SDS-PAGE and Western blotting analysis, using an in-house antibody reactive to the conserved region of PAM and to every other M-Prt tested.

Recombinant protein expression
hPg was expressed in Drosophila S2 cells as outlined previously (52,53). S2 cells were transfected utilizing pMT-PURO-hPg plasmid (53) and selected by puromycin resistance. Protein expression was induced by adding 0.6 mM CuSO 4 solution to cultures grown in 1-liter spinner flasks. After 4 -5 days, the cells were pelleted and supernatants removed, filtered, and concentrated using an Amicon 30-kDa cutoff membrane. r-Glu 1 -hPg was purified via affinity chromatography by adsorption to Lys-Sepharose (54) and elution with ⑀-aminocaproic acid. The eluate was dialyzed to remove salts and ⑀-aminocaproic acid. All M-Prts were cloned using the primers listed in Table 1 from their respective GAS genomic DNAs employing standard PCR amplifications and then ligated into pet28a plasmid (21). The recombinant protein constructs contained an N-terminal methionine and a C-terminal 6 -8-residue His tag used for purification. In some constructs, a C-terminal Cys residue was engineered at the end of the His tag for subsequent SPR coupling. The plasmids were transfected into E. coli strain BL21-DE3 and were subsequently cultured in 1 liter LB media. Recombinant protein expression was induced by 1 mM isopropyl ␤-D-1-thiogalactopyranoside.

Analytical ultracentrifugation
Sedimentation equilibrium ultracentrifugation was performed employing a Beckman XL-I analytical ultracentrifuge at 20°C in a six-sector cell in the UV absorption mode at 280 nm. The samples were dialyzed into 100 mM sodium phosphate, pH 7.4 ( ϭ 1.0091 g/ml). The M-Prt samples were centrifuged at 13,000 and 18,000 rpm until equilibrium was achieved as defined by temporal scans of the protein concentration distribution versus radial distances in the cell being invariant. The partial specific volumes were 0.721-0.723 ml/g for each sample. Molecular weights were obtained using the Optima XL-A/ XL-I data analysis software (Beckman Coulter, Brea, CA).

Fibrin(ogen) fragment generation and purification
A solution containing 6 mg/ml hFg (ERL) was dialyzed overnight using a 30-kDa cutoff membrane in cleavage buffer (50 mM Tris, 0.1 mM NaCl, 1 mM CaCl 2 , pH 7.4). Urokinase-activated hPm was then added to a final ratio of 1:100 hPm/hFg, and cleavage was carried out for 2 h at 37°C. The reaction was terminated by addition of PMSF. The solution was then dialyzed overnight at 4°C in 50 mM Tris-Cl, 1 mM CaCl 2 , pH 8.
A DEAE-cellulose column was equilibrated with a solution of 50 mM Tris-Cl, 1 mM CaCl 2 , pH 8. The crude cleavage products, dialyzed against this buffer, were applied to the column. Fragments were eluted with a salt gradient, ranging from 0 to 250 mM NaCl in the same buffer for fragment D and 250 -500 mM NaCl for fragment E. Fractions containing protein were detected by A 280 nm .

Surface plasmon resonance (SPR)
Binding kinetic analyses were carried out using a Biacore X100. CM5 chips were used to immobilize respective ligands via amine-coupling chemistry. hFg and FDPs were diluted in 10 mM NaOAc, pH 4.5, for immobilization at a final concentration of 20 -40 g/ml. Association (k on ) and dissociation (k off ) rates were measured at 25°C with a continuous flow rate of 30 l min Ϫ1 . M-Prts were serially diluted (0.5-100 nM) in running buffer HBS-EP. The chips were regenerated using 12.5 mM NaOH. Data analyses were carried out using BIAevaluation software (GE Healthcare).

Circular dichroism (CD)
An AVIV (Lakewood, NJ) 202SF spectrometer was utilized for the collection of CD spectra. Spectral scans were obtained at 25°C in 1.0-nm intervals ranging from 190 to 240 nm. A reference buffer scan was subtracted from the average of three runs for each sample. The mean residue ellipticities (MRE) were calculated using the equation, MRE ϭ ( ϫ MRW)/(l ϫ c), where is the CD signal in millidegrees; MRW is the mean residue weight in g/mol; l is the path length in millimeters, and c is the protein concentration in mg/ml.

hPg activation
Chromogenic substrate, S2251 (D-Val-Leu-Lys-p-nitroanilide), was diluted to a final concentration of 0.25 mM in activation buffer (10 mM HEPES, 150 mM NaCl, pH 7.4). r-Glu-hPg was then added to a final concentration of 200 nM. In some instances, hFg and its derivatives were added to a final concentration of 200 nM to observe their role in stimulation of SK-hPm-catalyzed activations. Additionally, M-Prt and M-Prt fragments were also added at concentrations of 200 -400 nM in the presence or absence of hFg to determine their roles in this process. The final solution was then pipetted onto protein-nonbinding 96-well plates, and each set of conditions was carried out in triplicate. To accelerate the activation reaction, a catalytic amount of SK was added to a final concentration of 5 nM. The rate of substrate cleavage was detected as A 405 nm versus time at 37°C for 2 h. Initial velocities (vi) were determined by linear regression by plotting the A 405 nm versus time 2 .

Whole GAS cell FACS binding analyses
GAS strains of interest were prepared as outlined previously (47), with the exception that after the final wash the cells were blocked with 1% ovalbumin in sterilized PBS and resuspended to an OD 600 nm of ϳ1.6 or 5 ϫ 10 7 cells. The cells were then resuspended in 300 l of blocking buffer containing the ligand of interest (FITC-hFg or hPg) at a concentration of 200 nM and incubated at room temperature for 1 h. For hFg/hPg co-incubations, the hPg concentration was in 5-fold excess. The cells were then pelleted by centrifugation and washed three times using 750 l of PBS, repeated twice. hPg binding was detected using an in-house mouse anti-hPg 1°antibody followed by AlexaFluor 488 goat anti-mouse IgG as the 2°antibody. FITClabeled hFg was directly detected. To fix cells, 100 l of prechilled 2% paraformaldehyde, diluted in blocking buffer, was added and gently mixed by pipetting. The cells were then distributed to Falcon tubes for FACS analysis (FACS ARIA III), using the doublet discrimination gating mode.

Whole GAS cell hPg activation assays
Cells were cultured overnight in 5 ml of THY media, and 30 ml were pre-warmed at 37°C in 40-ml Falcon tubes. Three ml of overnight culture was seeded into 30 ml of culture and incubated until the OD 600 nm reached ϳ0.6. The cells were then centrifuged at 2,500 rpm, and the supernatant was discarded. The resulting cells were washed twice by resuspending in 10 ml of sterilized 0.7% saline, or PBS, and then vortexed to dislodge bacterial aggregates. The cells were centrifuged, and the supernatant was discarded. Finally, the cells were resuspended in PBS to a final OD 600 nm of 1.0. An aliquot of 400 l was removed and placed in 1.5-ml Eppendorf tubes. Next, hPg alone, hFg alone, and/or hPg/hFg was added, each to a final concentration of 200 nM. The cells were incubated for 1 h at room temperature. The tubes were then centrifuged at 10,000 rpm for 30 s and the supernatants removed. The resulting pellet was washed with PBS and resuspended in 400 l of activation buffer (10 mM HEPES, 100 mM NaCl, pH 7.4). Next, 100 l of the cell suspension was distributed into a 96-well protein-nonbinding plate. hPg activation was initiated by the addition of 100 l of reaction buffer containing 0.25 mM S2251 substrate and 10 nM SK2a. Substrate cleavage was detected as ⌬A 405 nm versus time using a spectrophotometer plate reader.
Author contributions-K. G., J. B., and C. Q. performed the experiments. Z. L. contributed valuable reagents. F. J. C., S. W. L., and V. A. P. supervised the study. All authors contributed to writing the manuscript.