Molecular Co-operation between Protein PAM and Streptokinase for Plasmin Acquisition by Streptococcus pyogenes *

Bacterial surface-associated plasmin formation is believed to contribute to invasion, although the underlying molecular mechanisms are poorly understood. To define the components necessary for plasmin generation on group A streptococci we used strain AP53 which exposes an M-like protein (“PAM”) that contains a plasminogen-binding sequence with two 13-amino acid residues long tandem repeats (a1 and a2). Utilizing an Escherichia coli-streptococcal shuttle vector, we replaced a 29-residue long sequence segment of Arp4, an M-like protein that does not bind plasminogen, with a single (a1) or the combined a1a2 repeats of PAM. When expressed in E. coli, the purified chimeric Arp/PAM proteins both bound plasminogen, as well as plasmin, and when used to transform group A streptococcal strains lacking the plasminogen-binding ability, transformants with the Arp/PAM constructs efficiently bound plasminogen. Moreover, when grown in the presence of plasminogen, both Arp/PAM- and PAM-expressing streptococci acquired surface-bound plasmin. In contrast, plasminogen activation failed to occur on PAM- and Arp/PAM-expressing streptococci carrying an inactivated streptokinase gene: this block was overcome by exogenous streptokinase. Together, these results provide evidence for an unusual co-operation between a surface-bound protein, PAM, and a secreted protein, streptokinase, resulting in bacterial acquisition of a host protease that is likely to spur parasite invasion of host tissues.

Bacterial surface-associated plasmin formation is believed to contribute to invasion, although the underlying molecular mechanisms are poorly understood. To define the components necessary for plasmin generation on group A streptococci we used strain AP53 which exposes an M-like protein ("PAM") that contains a plasminogen-binding sequence with two 13-amino acid residues long tandem repeats (a1 and a2). Utilizing an Escherichia coli-streptococcal shuttle vector, we replaced a 29-residue long sequence segment of Arp4, an M-like protein that does not bind plasminogen, with a single (a1) or the combined a1a2 repeats of PAM. When expressed in E. coli, the purified chimeric Arp/PAM proteins both bound plasminogen, as well as plasmin, and when used to transform group A streptococcal strains lacking the plasminogen-binding ability, transformants with the Arp/PAM constructs efficiently bound plasminogen. Moreover, when grown in the presence of plasminogen, both Arp/PAM-and PAM-expressing streptococci acquired surface-bound plasmin. In contrast, plasminogen activation failed to occur on PAM-and Arp/PAMexpressing streptococci carrying an inactivated streptokinase gene: this block was overcome by exogenous streptokinase. Together, these results provide evidence for an unusual co-operation between a surface-bound protein, PAM, and a secreted protein, streptokinase, resulting in bacterial acquisition of a host protease that is likely to spur parasite invasion of host tissues.
Plasminogen is a single chain 92-kDa glycoprotein present in plasma (ϳ200 mg/liter) and in extracellular fluids. The protein contains five characteristic kringle domains that mediate interactions with multiple ligands, such as fibrin(ogen), ␣ 2 -antiplasmin (␣ 2 AP), 1 and cellular plasminogen receptors (1)(2)(3). In vivo, two specific activators, the tissue-type plasminogen activator and urokinase, convert the single-chain zymogen to a two-chain enzyme, plasmin, by cleavage at a single site (4). Plasmin efficiently degrades fibrin and probably also participates in the breakdown of soft tissue glycoproteins (5). The plasma concentration of active plasmin is normally low, due to its rapid inactivation by proteinase inhibitors, primarily ␣ 2 AP. Under pathophysiological conditions some pathogens, including group A, C, and G streptococci and Staphylococcus aureus, secrete highly specific and potent plasminogen activators such as streptokinase and staphylokinase, respectively (1,2). Other bacteria, including Escherichia coli and Yersinia pestis, express surface-bound plasminogen activators (6,7). Evidence suggest that plasmin generated by bacterial plasminogen activators contributes to virulence, possibly by facilitating bacterial penetration through host barriers (7).
It has been known for some time that many virulent bacterial species express surface proteins capable of binding plasmin(ogen) without causing its activation (8). Some of these can also contribute to bacterial dissemination (9). For Streptococcus pyogenes (group A streptococci), a major causative agent of both mucosal and skin infection in humans, four major types of surface proteins have been forwarded as candidate plasmin(ogen)-binding proteins. Of these the dehydrogenase-like Plr protein (10 -12) binds primarily plasmin, whereas a 45-kDa protein (13) also binds plasminogen. Some members of the M protein family, a well known class of surface-exposed streptococcal virulence determinants (14), have been shown to capture plasmin(ogen) indirectly, through fibrinogen (15,16). Finally, a set of group A streptococcal strains, associated with skin infection (17)(18)(19), expose M-like proteins that bind plasminogen as well as plasmin, directly and with high affinity (20,21). For one of these proteins, PAM (plasminogen-binding group A streptococcal M-like protein) from type 53 group A streptococci, a plasminogen-binding sequence characterized by two 13-amino acid residues long repeats (a1 and a2), has been identified in its surface-exposed variable NH 2 -terminal domain. Similar motifs are also present in M-like proteins from the other "skin" strains (21).
Here we have set out to study the molecular requirements for the binding and activation of plasminogen by PAM-expressing bacteria. We demonstrate a joint role for the plasminogenbinding motif in PAM and streptokinase in the acquisition and exposure of plasmin at the surface of group A streptococci.

MATERIALS AND METHODS
Bacterial Strains and Culture Conditions-Streptococcal strains, their M or M-like proteins and known ligands for these proteins are listed in Table I. AP4, a type M4 group A streptococcal strain expressing Arp4 (22) and the PAM-expressing type 53 group A streptococcal strain AP53 were obtained from the Institute of Hygiene and Epidemiology, Prague, Czech Republic. JRS4, an M6 protein-expressing group A streptococcal strain (23,24), and JRS145 (25), an M6-deficient derivative of JRS4, were from Dr. June Scott, Emory University, Atlanta, GA. JRS145(Arp), a JRS145 strain transformed with the shuttle plasmid pJRS264 (26) that encodes the arp4 gene of AP4, was a kind gift from Dr. Gunnar Lindahl. Streptococci were grown in Todd-Hewitt broth (Difco, Detroit, MI), or in Todd-Hewitt broth supplemented with 30% (v/v) fresh human heparinated plasma, in 5% CO 2 at 37°C for 16 h.
Streptococci transformed with plasmids carrying resistance markers for antibiotics were grown in medium supplemented with 15 mg liter Ϫ1 , 200 mg liter Ϫ1 , or 70 mg liter Ϫ1 erythromycin, kanamycin, or spectinomycin, respectively. The streptococci were harvested by centrifugation, washed twice in 0.15 M NaCl, 0.03 M phosphate, pH 7.2 (PBS), and were finally resuspended in PBS. Transformed E. coli LE392 was grown in LB medium supplemented with erythromycin (750 mg liter Ϫ1 ), kanamycin (200 mg liter Ϫ1 ), or spectinomycin (20 mg liter Ϫ1 ) depending on the selection marker of the plasmid.
Binding of Radiolabeled Ligands to Streptococci-Binding of [ 125 I]labeled plasminogen, IgA, or fibrinogen to bacteria was measured in a total volume of 250 l of PBS containing 0.1% (v/v) Tween 20. Following incubation for 60 min at 20°C, 2 ml of PBS containing 0.1% Tween 20 was added, and the bacteria were centrifuged at 4,000 ϫ g for 5 min. The supernatant was discarded and the radioactivity associated with the pellet was measured in a ␥-counter. The number of plasminogenbinding proteins expressed on streptococci was determined from experiments in which 125 I-labeled plasminogen (0.02 nM) was allowed to compete with various concentrations of unlabeled plasminogen (50 -0.025 nM) for the binding to streptococci (1 ϫ 10 9 bacteria ml Ϫ1 ). The number of receptors was then estimated from the intercept with the abscissa of a curve generated by plotting bound (abscissa) versus bound/ free (ordinate) plasminogen.
Plasma Absorption Experiments-Streptococci (10 10 cells) were incubated with 1.0 ml of fresh human plasma for 60 min at 20°C. Bacterial cells were pelleted and washed three times with PBS. To elute absorbed proteins, the bacteria were incubated with 0.1 M glycine, pH 2.0, for 10 min at 20°C. The bacteria were removed by centrifugation (4000 ϫ g, 5 min), and the proteins of the supernatant were analyzed by SDS-PAGE and immunoblotting (see below).
Plasminogen Activation Assays-Heat-killed streptococci (10 10 cells) were incubated with 1.0 ml of human plasma for 60 min at 20°C. The bacteria were washed three times with 10 ml of PBS containing 0.1% Tween 20. The bacteria were resuspended in blocking buffer containing 100 units of group C streptococcal streptokinase (Chromogenix, Mölndal, Sweden) and incubated again for 60 min at 20°C, followed by centrifugation and two washes with PBS containing 0.1% Tween 20. The pellet was incubated in 2.5 ml of PBS containing 0.4 M NaCl and 80 M of the chromogenic substrate H-D-Val-Leu-Lys-p-nitroanilide (S2251; Chromogenix) for 60 min at 37°C. The bacteria were pelleted, and the absorbance of the supernatant was measured at 405 nm. Alternatively, 1 ϫ 10 10 cells from an overnight culture were washed three times in PBS, and subsequently added to a mixture of 2.8 ml of Todd-Hewitt broth and 1.2 ml of plasma (for some experiments supplemented with 100 units of group C streptococcal streptokinase). The cultures were incubated at 37°C in 5% CO 2 ; at varying time points the cultures were terminated by centrifugation at 4°C. The bacterial pellets were washed three times in ice-cold PBS containing 0.1% Tween 20 and placed on ice. Plasmin activity associated with the bacteria was measured as described above.
Construction of Arp/PAM Chimeric Proteins-The Arp4-encoding plasmid, pJRS264, contains unique SacI and BstBI sites at positions 315 and 403 of the arp4 gene. For construction of the Arp/a1a2 construct, a 110-base pair DNA segment encoding the a1a2 repeat unit of PAM was generated by PCR using chromosomal DNA from AP53 streptococci as template and the synthetic oligonucleotides 5Ј-GAGGAGCT-CAAAGATGATGTTGAGAAGCTTACC-3Ј and 5Ј-GCTCCGGAGATC-ATGTCTCTCGCTTTTAAGTCG-3Ј as primers, thereby introducing SacI and HpaII recognition sites, respectively. The 110-base pair DNA fragment was ligated with SacI/BstBI-digested pJRS264 DNA to yield the Arp/a1a2 construct. The Arp/a1 construct was generated by ligating the annealed synthetic phosphorylated oligonucleotides 5Ј-CAAGGTA-GATGCTGAGTTGCAACGACTTAAAAACGAGAGACATGAAGGTG-AAAATCAAGATCTT-3Ј and 5ЈCGAAGATCTTGATTTTCACCTTCAT- Inactivation of the Streptokinase Gene (ska)-Chromosomal DNA from AP53 or JRS4 streptococci was used as templates for PCR to amplify a 1180-base pair ska fragment with primers SK3 (5Ј-GACGTC GACACTTGCATCTCTGGAAAATAGTC-3Ј) and SK4 (5Ј-GACGGATC-CATGAGTGACGATTGAGGAGTCAC-3Ј), or to amplify a 950-base pair ska fragment with primers SK6 (5Ј-GACGGATCCTTTCTGAGAAATA-TTACGTCC-3Ј) and SK7 (5Ј-GACGAGCTCGGTACCTTTCTATTGAT-GGGAAAATTGC-3Ј); the relative locations of the SK primers in the ska sequence are indicated in Fig. 1C. The 1180-and 950-base pair fragments were digested with SalI/BamHI or BamHI/SacI, respectively, and were then ligated with SalI/SacI-digested pUC18 DNA. Following transformation and amplification in E. coli, the resultant plasmid DNA was made linear with BamHI and ligated with the BamHI-digested kanamycin-resistance cassette ⍀Km-2 (25). Plasmid DNA was prepared from a kanamycin-resistant transformant and digested with SalI and SacI. The resulting SalI/SacI fragment, containing ⍀Km-2 flanked by ska fragments, was blunted and ligated with the SalI-digested and blunted plasmid pJRS233. Plasmid pJRS233 contains an erythromycin resistance marker (erm), a temperature-insensitive E. coli replicon, and a promiscuous replicon that is sensitive to temperatures above 35°C in both E. coli and S. pyogenes (27). Following transformation and subsequent propagation in E. coli at 37°C, the resulting plasmid (pJRS233kana-ska) was used to electroporate S. pyogenes. Kanamycin-resistant and erythromycin-sensitive isolates were selected following sequential culturing of the transformed bacteria at 30 and 37°C, respectively. Integration of ⍀Km-2 into the ska genes of AP53 and JRS145, respectively, was monitored by PCR: the oligonucleotides SK9 (5Ј-ACTCATC-CAATCTTTATCTCC-3Ј) and SK10 (5Ј-TGCAACGGGTCAAAGAG-GCC-3Ј), located on opposite strands immediately outside the ska fragment defined by SK3 and SK6, were used in combination with the ⍀Km-2-derived outwardly directed primers KM1 (5Ј-CTCTTCATC-CTCTTCGTCTTG-3Ј) and KM2 (5Ј-ATCCGCAACTGTCC ATACTCT-3Ј) (Fig. 1C). PCR-positive clones failed to produce active streptokinase as determined by analysis of growth supernatants and streptococcal whole cell lysates by the S-2251 assay. No difference in the growth rates was observed between the ska positive and the ska negative streptococcal isolates.
Recombinant DNA Techniques-Standard ligation and DNA isolation procedures were employed (28). Ligase, T4 DNA polymerase, and restriction enzymes were from Promega (Madison, WI). PCR was performed (29) using Amplitaq (Perkin-Elmer, Wilton, CT). The sequences of the SK primers were derived from the published S. pyogenes streptokinase gene sequence (30,31). Chimeric Arp/PAM genes were sequenced using the dideoxy nucleotide chain termination method (32). Electroporation of E. coli and streptococci was performed as described (25).
Protein Analysis and Preparation of Affinity Matrices-Recombinant PAM, Arp4, and Arp/PAM proteins were purified from whole cell lysates of overnight cultures of E. coli LE392 transformed with the corresponding genes cloned in pJRS264 or derivatives thereof. PAM and Arp/PAM hybrid proteins were purified on plasminogen agarose, whereas Arp4 was purified on an agarose column coupled with human IgA. Proteins bound to the affinity columns were eluted using 0.1 M glycine-HCl, pH 2.0, and the effluent was dialyzed against PBS. Plasminogen was purified on a column with lysine-Sepharose 4B (Pharmacia, Uppsala, Sweden). Proteins bound to the column were eluted with 0.2 M ⑀-aminocaproic acid. Fractions containing plasminogen were pooled and dialyzed against PBS. Plasmin was generated by incubation of 0.8 mg of plasminogen with 80 g of tissue-type plasminogen activator at 37°C, for 45 min. Activation was interrupted by incubation (20°C, 10 min) with a 100-fold molar excess of aprotinin. The efficiency of plasminogen activation was monitored by SDS-PAGE (33) and im-  (34), with 125 I-labeled anti-plasminogen purified from rabbit plasma by immunoaffinity chromatography. For electroblot analysis, proteins were transferred to polyvinylidene difluoride membranes. The membranes were blocked in PBS containing 0.25% Tween 20 and 0.25% gelatine (blocking buffer), incubated with radiolabeled proteins in blocking buffer, and finally washed in blocking buffer containing 0.5 M NaCl. Ligand blotting was done with 125 I-labeled plasminogen, IgA, or fibrinogen. Iodination of proteins was done using the chloramine T method (35). Coupling of proteins to agarose columns with preactivated N-hydroxysuccinimide ester (HiTrap) was done according to the manufacturer's instructions (Pharmacia, Uppsala, Sweden). For binding analysis in microtiter plates (Microtest III, Becton-Dickinson, Cockeysville, MD) wells were coated with 100 l of the protein solution (10 g ml Ϫ1 ) in PBS and incubated overnight at 4°C. The following day, 100 l of blocking buffer was added and the plates were again incubated overnight at 4°C. On day 3 the plates were washed with blocking buffer. A mixture containing the radiolabeled ligand ( 125 I-labeled plasminogen or plasmin) and the unlabeled competitor (plasminogen, plasmin, or ⑀-aminocaproic acid) was then added in a total volume of 150 l in blocking buffer. For experiments with plasmin, a 100-fold molar excess of aprotinin was included in each incubation. After incubation at 20°C for 4 h, the plates were washed four times in blocking buffer and the radioactivity in the wells was counted. For affinity calculations, 125 I-plasminogen (0.1 nM) or 125 I-plasmin (0.1 nM) was mixed with increasing concentrations of unlabeled plasminogen or plasmin, in serial 2-fold dilutions (0.05 to 50 nM for plasminogen and 6 nM to 0.8 M for plasmin). Calculation of affinity constants were made using the formula of Scatchard (36).

RESULTS
Activation of Plasminogen Bound to the Surface of S. pyogenes-We followed the time course of plasminogen/plasmin binding to AP53 streptococci expressing the PAM protein on their surface. To this end the bacteria were grown in medium containing 30% human plasma. Following centrifugation, the plasma proteins bound to the bacterial surface were eluted, separated by SDS-PAGE, and visualized by immunoblotting using a 125 I-labeled specific antibody recognizing both human plasmin and plasminogen (Fig. 1, panel A, top). The plasmin/ plasminogen ratio increased with time of incubation (Fig. 1A). Using a chromogenic substrate we were able to demonstrate that at least part of the bacteria-associated plasmin was present in its active form (Fig. 1A, bottom). These results indicate that efficient plasminogen-binding and activation mechanism(s), which are not blocked by the proteinase inhibitors present in human plasma, must exist at the streptococcal surface.
The a1-repeat of PAM Is Sufficient to Endow Group A Streptococci with Plasminogen Binding-Initially we focused on the role of PAM for the streptococcal capture of plasmin(ogen). However, previous experiments with recombinant proteins and synthetic peptides (21) have suggested that a major plasminogen-binding site is located in a 30-amino acid residue long segment containing a sequence with two characteristic tandem repeats (a-repeats) in the NH 2 -terminal variable region of PAM. To test the notion that a-repeats confer plasminogen binding properties, we used another M-like protein, Arp4, that lacks a plasminogen-binding site, but exposes a well defined IgA-binding site (37), as a recipient for the a-repeats of PAM (Fig. 2). Manipulations were done in the E. coli-streptococcal shuttle vector pJRS264 that encodes the complete Arp4 protein. We first replaced 29 residues in the IgA-binding region of Arp4 (positions 29 -57 of the Arp4 sequence) by a segment of 36 residues from PAM which holds both a1 and a2 (positions 53-88 of the PAM sequence; see Ref. 20), or by a segment of 13 residues that comprises only a1 (positions 62 to 74). We then transformed E. coli with the corresponding plasmids and isolated the chimeric constructs Arp/a1a2 and Arp/a1. Western blots of the chimeric proteins probed by 125 I-plasminogen or IgA revealed that both Arp/a1a2 and Arp/a1 bound plasminogen, but not IgA (Fig. 3). In competitive binding experiments, FIG. 1. Acquisition of plasmin by streptokinase producing and non-producing PAM-expressing S. pyogenes. Protein PAM-expressing AP53 streptococci from an overnight culture were washed three times with PBS and then used to inoculate fresh culture medium containing 30% human plasma. The cultures were terminated at the indicated time points and the bacteria were harvested. Following repeated washing, the bacteria were divided in two equal parts. The bacterial-bound proteins from one aliquot were eluted with 0.1 M glycine-HCl, pH 2.0, and analyzed by Western blotting with 125 I-antiplasminogen as the probe using ska positive (panel A, top) and ska negative AP53 streptococci (panel B, top). The other aliquot was resuspended in buffer containing the chromogenic substrate S2251 and incubated for 60 min at 37°C. The bacteria were then pelleted and the absorbance of the supernatant was measured at 405 nm (A and B,  bottom). The data are representative of five experiments with single samples. Insertional inactivation of the ska gene in AP53 and JRS145 streptococci was accomplished with the kanamycin-resistance cassette ⍀Km-2, flanked ska fragments generated by PCR using the primer combinations SK3/SK4 and SK6/SK7 (panel C). The external ska primers SK9 or SK10 were used in combination with primers KM1 or KM2 from the ⍀Km-2 sequence to verify the chromosomal integration of the cassette by PCR. unlabeled plasminogen inhibited the binding of 125 I-plasminogen to Arp/a1a2 and Arp/a1 that had been immobilized in microtiter plates (Fig. 4A). Similarly, unlabeled plasmin inhibited the binding of radiolabeled plasmin to the two chimeric proteins (Fig. 4B). The affinity constants (K a ) for the interactions between plasminogen and plasmin and the two chimeras were calculated by Scatchard plots from the data obtained in the competitive binding experiments. For the interaction with plasminogen, the K a values were calculated to be 6 ϫ 10 8 M Ϫ1 for both chimeras (Fig. 4A, insets), a value close to that (8 ϫ 10 8 M Ϫ1 ) previously determined for the interaction between radiolabeled plasminogen and immobilized PAM. The corresponding K a values for the interactions between plasmin and the chimeric proteins (Fig. 4B, insets) were 1 order of magnitude lower (3 ϫ 10 7 M Ϫ1 and 6 ϫ 10 7 M Ϫ1 , for Arp/a1a2 and Arpa1, respectively). As was the case for the interaction between 125 I-plasminogen and PAM (20), the binding of plasminogen to the Arp/PAM constructs could be inhibited by the lysine analogue ⑀-aminocaproic acid at low concentrations, indicating that the interaction involves lysine-binding site(s) in plasminogen (data not shown). Hence, the results show that the a-repeat motif(s) can confer the plasminogen-binding properties of PAM to a plasminogen non-binding member of the M protein family and that the a1-repeat contains the major plasminogen-binding site of PAM.
The plasmids encoding Arp/a1a2 and Arp/a1 were then used to transform the group A streptococcal strain JRS145, an isogenic M-protein-deficient (emm Ϫ ) derivative of the M6-expressing strain JRS4. Using radiolabeled ligands, streptococci expressing the Arp/PAM constructs were shown to bind plasminogen but failed to bind IgA (Fig. 5). Furthermore, heatkilled Arp/a1a2-and Arp/a1-expressing streptococci absorbed plasminogen from plasma, whereas the Arp4-expressing strain or JRS145 did not (Fig. 6). The finding that JRS145(Arpa1) showed a reduced capacity to bind 125 I-plasminogen compared with JRS145(Arpa1a2), although the recombinant chimeric proteins have the similar affinity for the zymogen, is likely to be attributable to different surface expression levels. Thus, estimation of the numbers of receptors from Scatchard analysis of experiments where 125 I-plasminogen and different concentrations of plasminogen were allowed to compete for the binding to JRS145 transformed with the two PAM/Arp hybrid genes, showed that JRS145(Arpa1a2) expressed 30 times as many chimeric molecules as compared with JRS145(Arpa1) (data not shown). Because of this difference, JRS145(Arpa1a2) was used in further experiments.
We also transformed strain JRS4, the parent strain of JRS145 that expresses the fibrinogen-binding M6 protein, with the plasmid encoding the hybrid Arp/a1a2 protein, and thereby converted the phenotype of JRS4 from plasminogen non-binding to plasminogen binding while retaining its fibrinogen bind-ing capacity (Fig. 7, inset). When grown in medium containing plasma, significant surface acquisition of plasmin was observed with transformed JRS4, but not with the wild-type strain (Fig. 7).

Co-operation between Streptokinase and the a-repeats of PAM in the Generation of Streptococcal-associated Plasmin-Our
initial experiments with the AP53 strain demonstrated a conversion of plasminogen to plasmin after Ͼ30 min of incubation of the bacteria with plasma-containing medium. Because of the presence of a large molar excess of proteinase inhibitors in plasma which likely inactivate any exogenous plasminogen activators, we speculated that activators of bacterial origin, such as streptokinase, could be involved in plasmin generation. Indeed, streptokinase added to plasminogen-loaded heat-inactivated AP53 cells converted plasminogen to plasmin in a dosedependent manner (data not shown). To rigorously test the hypothesis that streptokinase contributes to the activation of surface-associated streptococcal plasminogen, we inactivated the streptokinase gene (ska) of strain AP53 by homologous recombination (Fig. 1C). Ska Ϫ AP53 streptococci bound 125 Iplasminogen at a level (65% of the total radioactivity) comparable to the parent strain (63%), and demonstrated a plasma absorption pattern similar to that of the original isolate (data not shown). However, little if any conversion of plasminogen was observed when the ska Ϫ variant was incubated with human plasma (Fig. 1B). The inability of ska Ϫ AP53 streptococci to acquire surface-associated plasmin could be reversed by addition of streptokinase (data not shown).
The notion that streptokinase might be critically involved in plasminogen activation at the streptococcal surface was further investigated in a derivative of the M protein-deficient group A streptococcal strain JRS145 also deleted in the streptokinase gene. As expected, this strain failed to bind plasmin(ogen) at all (data not shown). Transformation with the plasmid encoding Arp/a1a2 conferred plasminogen-binding properties to JRS145(emm Ϫ /ska Ϫ ). However, little if any activation of plasminogen was observed by immunoblotting and by the amidase activity assay (Fig. 8). In contrast, expression of Arp/a1a2 in ska ϩ JRS145 bacteria allowed acquisition of plasminogen from plasma-containing growth medium as well as time-dependent conversion of plasminogen to plasmin (Fig. 8). As with the AP53 ska Ϫ strain, the inability of streptokinase-deficient JRS145(Arp/a1a2) streptococci to capture plasmin could be restored following addition of streptokinase. In accord with these observations we found that heat-killed JRS145(Arp/a1a2) bacteria, unable to secrete streptokinase, bound plasminogen but failed to acquire significant amounts of plasmin (data not shown). Hence, the co-operation of two bacterial effectors, the surface-bound acceptor protein PAM and the secretory activator protein streptokinase, is mandatory for generation of human plasmin at group A streptococcal surfaces.  4) were separated by SDS-PAGE and stained with Coomassie Blue (Stain). The relative molecular masses of standard marker proteins run simultaneously are given on the left. Separated proteins were transferred to polyvinylidene difluoride membranes by electroblotting, and identical replicas were probed with 125 I-plasminogen or 125 I-IgA (Blot).

DISCUSSION
Group A streptococci elaborate at least two classes of plasmin(ogen)-binding proteins, i.e. secretory proteins such as streptokinase and streptococcal cysteine proteinase (38), and surface attached proteins such as PAM (20), Plr (10), and a 45-kDa protein (13). Moreover, group A streptococci can capture plasmin(ogen) indirectly through fibrinogen which in turn binds to M-like proteins (15). Such a multiplicity of interactions with the key player of the fibrinolytic pathway suggests that the plasmin(ogen)-binding property carries a selective advantage to streptococci, motivating an analysis of the relative importance and biological role of the different mechanisms.
In the present work we have focused on the role of protein PAM, a member of the streptococcal M family of surface proteins. To assess the contribution of the PAM a-repeats for in vitro plasmin acquisition by group A streptococci, we grafted them on the plasmin(ogen) non-binding protein Arp4, another member of the M protein family. We chose Arp4 as the "recipient" molecule since this protein and PAM are similar in many ways. Thus, both proteins are anchored to the bacterial cell wall through their highly conserved COOH-terminal segments, they contain three conserved C-repeats in the COOH-terminal half of the protein, and have surface-exposed variable NH 2terminal regions. Moreover, both PAM and Arp can be predicted to form ␣-helical coiled coil dimers as a consequence of their high contents of characteristic heptad repeats (39). Finally, the binding patterns of PAM and Arp resemble one another: (i) both proteins bind C4bp with high affinity through a site in the NH 2 -terminal hypervariable domain (40); (ii) they do not bind fibrinogen; (iii) a unique binding property (i.e. plasminogen-binding for PAM and IgA-binding for Arp4) is for both proteins located in the central region of their NH 2 -terminal domains. The experiments showed that replacement of the IgA-binding site of Arp by the a-repeats of PAM was necessary and sufficient to endow group A streptococci expressing two (JRS145) or three (JRS4) of the other putative plasmin(ogen)- binding surface proteins (10,13,15) with the capacity to bind plasmin and plasminogen, thereby emphasizing the power of the PAM motif in relation to other structures believed to contribute to group A streptococcal plasmin(ogen) acquisition. Moreover, the data establish PAM as the most well characterized cell-surface receptor for plasminogen and the a1 repeat of 13 residues as the smallest and best defined internal plasminogen acceptor site known, making it a candidate for analyses of structural requirements for interactions between kringles and sequences lacking carboxyl-terminal lysines.
The binding to PAM is mediated through the lysine-binding site(s) of a plasminogen fragment comprising its first three kringle domains (20). The principal physiological regulator of plasmin, ␣ 2 AP, binds to related site(s) (41). Given the high affinity for the interaction with plasmin(ogen), this protease is likely to be protected from regulation by ␣ 2 AP when bound by PAM-expressing streptococci. Our finding that absorption and activation of plasminogen on PAM-and Arp/PAM-expressing streptococci can take place in plasma supports such a notion. Since the binding is mediated through lysine-binding site(s), the catalytic domain of plasminogen remains accessible by the major streptococcal plasminogen activator, streptokinase. Indeed, streptokinase binds to PAM-expressing bacteria or to PAM-coated polyacrylamide beads only when they have been pre-absorbed with plasminogen. 2 This suggests that streptokinase might bind and activate PAM-bound plasminogen although our present experiments do not entirely rule out the possibility that a cycle of dissociation from and association with PAM is required to allow plasminogen activation in solution. Unlike plasmin in solution, neither the plasminogen-streptokinase complex nor the plasmin-streptokinase complex can be inactivated by ␣ 2 AP (42). Therefore the finding that PAM-and Arp/PAM-expressing bacteria acquire active plasmin even in the presence of plasma points to the possibility that bacteria gain an unregulatable plasminogen activator due to the deposition of plasmin(ogen)-streptokinase complexes on the streptococcal surfaces.
The key role of plasmin is to degrade fibrin (43,44). However, due to its broad specificity, plasmin can serve a host of other functions, e.g. cleavage of connective tissue proteins and basement membrane components, and activation of latent metalloproteinases. Cells that are able to generate plasmin on their surface may therefore use it to degrade physiological barriers other than fibrin. Indeed, plasmin-induced matrix degradation has been shown to facilitate the tissue penetration of tumor cells exhibiting an increased surface-associated plasminogen activation (45,46). The present results as well as data published by other investigators suggest that bacteria and certain mammalian cells may use similar mechanisms to cause pericellular proteolysis. Recent results using in vivo models suggest that surface-bound plasmin(ogen) can contribute to bacterial invasion. Thus, expression of a surface-bound plasminogen activator vastly increased the virulence of Y. pestis by facilitating their dissemination from the site of infection in the skin of mice (7). Moreover, plasminogen boosted spirochtemia in mice infected with the plasminogen-binding bacterial species Borrelia burgdorferi (9). It is conceivable that plasmin bound to the streptococcal surface may also function as a means to break tissue barriers, thereby promoting the spreading of these bacteria. The results presented here emphasize the unique property of some group A streptococci to expose a surface protein that binds plasminogen with high affinity, and at the same time to secrete a highly potent plasminogen activator that likely activates the surface-bound plasminogen. The joint role for protein PAM and streptokinase in the acquisition of surface-bound plasmin by group A streptococci associated with skin infections is yet another facet of the complex interplay that governs host-parasite relationships. FIG. 7. Binding and activation of plasminogen on M proteinexpressing streptococci transformed with an Arp/PAM construct. The M6 protein expressing wild-type group A streptococcal strain JRS4 was transformed with the Arp/a1a2-encoding plasmid. The 125 I-plasminogen (circles) and fibrinogen (squares) binding capacity of the parent strain and the transformant was measured (inset). The parent and the transformed strains were then grown in culture medium containing 30% human plasma. The cultures were terminated at the indicated time points, and the bacteria were harvested. After repeated washes, the bacterial pellets were resuspended in buffer containing the chromogenic substrate S2251. Following incubation for 60 min at 37°C the bacteria were pelleted, and the absorbance of the supernatant was measured at 405 nm. Parent strain, open bars and symbols; transformed strain, solid bars and symbols.
FIG. 8. Binding and conversion of plasminogen by Arp/PAMexpressing S. pyogenes in the absence or presence of endogenous streptokinase. JRS145 ska ϩ or ska Ϫ isolates that had been transformed with the Arp/a1a2-encoding plasmid were grown overnight. The cultures were used to inoculate fresh culture medium containing 30% human plasma and terminated at the indicated time points. Following centrifugation and washing, the bacterial pellet was divided in two equal parts. The plasma proteins absorbed by the bacteria from one aliquot were eluted and analyzed by Western blot experiments using 125 I-anti-plasminogen as a probe (top). The other aliquot was dissolved in buffer containing the chromogenic substrate S2251 for 60 min at 37°C, the bacteria were pelleted, and the absorbance of the supernatant was measured at 405 nm (bottom). Left panel, streptokinase expressing strain; right panel, streptokinase-deficient strain. The data are representative of five experiments with single samples.