Plasminogen Substrate Recognition by the Streptokinase-Plasminogen Catalytic Complex Is Facilitated by Arg253, Lys256, and Lys257 in the Streptokinase β-Domain and Kringle 5 of the Substrate*

Streptokinase (SK) conformationally activates the central zymogen of the fibrinolytic system, plasminogen (Pg). The SK·Pg* catalytic complex binds Pg as a specific substrate and cleaves it into plasmin (Pm), which binds SK to form the SK·Pm complex that propagates Pm generation. Catalytic complex formation is dependent on lysine-binding site (LBS) interactions between a Pg/Pm kringle and the SK COOH-terminal Lys414. Pg substrate recognition is also LBS-dependent, but the kringle and SK structural element(s) responsible have not been identified. SK mutants lacking Lys414 with Ala substitutions of charged residues in the SK β-domain 250-loop were evaluated in kinetic studies that resolved conformational and proteolytic Pg activation. Activation of [Lys]Pg and mini-Pg (containing only kringle 5 of Pg) by SK with Ala substitutions of Arg253, Lys256, and Lys257 showed decreases in the bimolecular rate constant for Pm generation, with nearly total inhibition for the SK Lys256/Lys257 double mutant. Binding of bovine Pg (BPg) to the SK·Pm complex containing fluorescently labeled Pm demonstrated LBS-dependent assembly of a SK·labeled Pm·BPg ternary complex, whereas BPg did not bind to the complex containing the SK Lys256/Lys257 mutant. BPg was activated by SK·Pm with a Km indistinguishable from the KD for BPg binding to form the ternary complex, whereas the SK Lys256/Lys257 mutant did not support BPg activation. We conclude that SK residues Arg253, Lys256, and Lys257 mediate Pg substrate recognition through kringle 5 of the [Lys]Pg and mini-Pg substrates. A molecular model of the SK·kringle 5 complex identifies the putative interactions involved in LBS-dependent Pg substrate recognition.

Streptokinase (SK) 6 activates the human fibrinolytic system by activating plasminogen (Pg) through a unique mechanism that is responsible for the use of SK as a thrombolytic drug and its role as a key pathogenicity factor in Group A streptococcal infection (1,2). The crystal structure of SK bound to the catalytic domain of plasmin (Pm) shows that SK consists of three ␤-grasp, tightly folded domains, ␣, ␤, and ␥, linked by flexible segments (3). In solution, SK is highly flexible and behaves hydrodynamically like three beads on a string (4). When bound to Pm, SK assumes a highly ordered structure resembling a three-sided crater surrounding the catalytic site that provides an exosite(s) for binding the catalytic domain of Pg as a substrate (3,5). In the first step of the SK-mediated Pg activation pathway, SK binds the catalytic domain of the Pg zymogen in a rapid equilibrium process and inserts its NH 2 -terminal Ile 1 residue into the NH 2 -terminal binding cleft of Pg, activating the catalytic site nonproteolytically (6 -10). Although structural proof is lacking, SK Ile 1 presumably forms a critical salt bridge with Asp 740(194) (plasminogen numbering; chymotrypsinogen numbering is in parentheses) that initiates conformational activation of the substrate binding site and oxyanion hole required for proteolytic activity (6, 8 -10). The activated SK⅐Pg* complex binds a second molecule of Pg as a specific substrate and cleaves it at Arg 561(15) -Val 562 (16) to form the fibrin-degrading proteinase, plasmin (Pm) (10 -14). Proteolytic generation of Pm is propagated by formation of a high affinity SK⅐Pm complex that converts the remaining free Pg into Pm (5,11).
[Glu]Pg, the full-length form of Pg circulating in blood, consists of an NH 2 -terminal PAN (Pg/Apple/Nematode (15,16)) module, followed by five kringle domains (K1-K5), and the trypsin-like serine proteinase catalytic domain (17). Formation of the SK⅐Pg* and SK⅐Pm catalytic complexes and Pg substrate binding are inhibited by the lysine analog, 6-aminohexanoic acid (6-AHA), which binds to lysine-binding sites (LBS) located primarily in kringles K1, K4, and K5 of Pg and Pm (10, 11, 18 -23). Cleavage of the Lys 77 -Lys 78 peptide bond in [Glu]Pg by Pm releases the PAN module and generates the truncated form, [Lys]Pg. Formation of [Lys]Pg is accompanied by a conformational change of [Glu]Pg from a compact, closed ␣-conformation to a partially extended ␤-conformation with expression of higher affinity LBS for 6-AHA (24,25). The fourth kringle module mediates a second conformational change, from the ␤-conformation to the extended ␥-conformation (25).
Binding of SK to [Glu]Pg is independent of LBS, with a dissociation constant of 100 -150 nM, whereas formation of SK⅐[Lys]Pg is LBS-dependent with a 13-20-fold higher affinity that is reduced to that of [Glu]Pg by saturating concentrations of 6-AHA (10,21). Activation of the catalytic domain in [Lys]Pm increases affinity for SK about 830-fold, which is reduced 11-20-fold by 6-AHA (5,21). Interaction of the COOH-terminal Lys 414 residue of SK with a Pg/Pm kringle domain is responsible for the LBS-dependent enhancement of the affinity of SK⅐[Lys]Pg* and SK⅐Pm catalytic complex formation (22). Recent rapid reaction kinetic studies of the SK⅐Pm binding pathway demonstrated that interaction of Lys 414 with a Pm kringle enhances formation of an initial rapid equilibrium SK⅐Pm encounter complex, succeeded by two sequential, tightening conformational changes, to achieve an overall dissociation constant of ϳ12 pM (26). The Pg/Pm kringle domain responsible for the enhancement of SK⅐Pg* and SK⅐Pm complex formation is not known. Productive interaction of Pg as a substrate of the SK⅐Pg*/Pm complexes is also greatly inhibited by saturating 6-AHA (11). Kinetic and equilibrium binding studies of SK-mediated Pm formation resolved the conformational activation process from the coupled proteolytic generation of Pm (10,11). The kinetic approach demonstrated that Lys 414 deletion reduced the affinity of formation of the SK⅐Pg* catalytic complex specifically, whereas the subsequent LBS-dependent proteolytic formation of Pm was unaffected, indicating that Pg substrate recognition is mediated by a structurally distinct region of SK and an unknown kringle (22).
Previous structure-function studies have yielded diverse interpretations and conclusions regarding the structural basis of LBS-dependent Pg substrate recognition (23,(27)(28)(29)(30)(31)(32)(33)(34). Each of the three domains of SK has been implicated in this regard (29,30,35,36), and binding of two Pg molecules to the residue 1-59 sequence of the ␣-domain has been reported (36). In particular, segments 16 -36, 41-48, 48 -59, and 88 -97 of the SK ␣-domain have been concluded to play a role in Pg substrate recognition (32,33,37,38). For several SK mutants, a complex mixture of functional effects on their binding to [Glu]Pg and its conformational and proteolytic activation has been reported (28,31,33). Some of these effects may result from the inherent flexibility of SK when bound to Pg or Pm (39), and others may be due to the use of kinetic approaches that do not clearly discriminate between conformational and proteolytic activation.
Some observations implicate a protruding hairpin loop called the 250-loop (residues Ala 251 -Ile 264 ) in the SK ␤-domain in Pg substrate recognition (27,28,31,34). This loop is disordered in the structure of the SK⅐Pm complex but is ordered in the structure of the isolated ␤-domain (3,40). Deletion of the 250loop, Ala substitution of Lys 256 and Lys 257 at the apex of the loop, and substitution of multiple residues near and within the loop resulted in disparate effects on K m and k cat for [Glu]Pg activation (27,28,31). The conclusions of these studies were that Lys 256 and Lys 257 are involved in SK binding and conformational activation of [Glu]Pg in addition to proteolytic processing of Pg as a substrate. Some of these studies are problematic because the natural NH 2 -terminal Ile 1 residue necessary for conformational activation is preceded either by an additional methionine (27,31) or maltose-binding protein (28) in the recombinant SK species used.
Because of the diverse conclusions regarding the functional properties of the 250-loop mutations and the possibility of other potential Pg substrate binding sites, the present studies were undertaken to resolve the function of residues in the 250-loop in LBS-dependent Pg substrate recognition by the SK⅐Pg* complex. The kringle domain of Pg involved in Pg substrate recognition has not been clearly identified but has been suggested to be K5 (27) on the basis that the isolated ␤-domain bound Pg (30) and K5 (29) in an LBS-dependent manner. Given the general specificity of Pg kringles for COOH-terminal Lys residues and zwitterionic ligands, such as 6-AHA, and the internal sequence of the 250-loop, it appeared possible that a pseudolysine motif on SK was involved. In the binding of a 30-residue peptide from plasminogen binding Group A streptococcal M-like protein (PAM), VEK-30, to K2 of Pg, Castellino and co-workers (41,42) showed by crystallography and mutagenesis that residues with cationic (Arg and His) and anionic side chains (Glu) arranged spatially on a helix constituted a pseudolysine structure similar to 6-AHA that binds specifically to the LBS of K2. Additional evidence for pseudolysine structures in Pg binding comes from studies of ␣-enolase from Streptococcus pneumoniae, which has a 9-residue internal binding site for Pg containing essential basic (two Lys residues) and acidic (Asp and Glu residues) located on a surface loop (43,44).
To determine whether a similar SK structure is involved in [Lys]Pg substrate recognition, anionic and cationic residues in the 250-loop were substituted with Ala and characterized in kinetic studies using methods that resolve conformational and proteolytic activation. Studies with [Lys]Pg and mini-Pg, which contains only K5 and the catalytic domain, showed that Arg 253 , Lys 256 , and Lys 257 facilitate LBS-dependent substrate recognition through interactions with K5. The absence of evidence for a pseudolysine structure in the 250loop is compatible with the established atypical specificity of K5 for cationic ligands, such as benzamidine, N ␣ -acetyl-Lysmethyl ester, 6-aminohexane, and 5-aminopentane, in addition to zwitterionic ligands (19,(45)(46)(47). The studies resolve for the first time the structural features of SK that mediate the LBS-dependent interactions that enhance affinity of SK⅐Pg* and SK⅐Pm catalytic complex formation and those that facilitate binding of Pg as a substrate of these complexes.
[Glu]Pg 2 (30 mg) was diluted to a final concentration of 30 M in 12 ml of 50 mM HEPES, 125 mM NaCl, 1 mM EDTA, pH 7.4. Elastase and bovine pancreatic trypsin inhibitor (Sigma) were added to final concentrations of 6 M and 0.1 mg/ml, respectively. Digestion was performed at 25°C for 60 min before inactivation of elastase by the addition of 120 M N-(methoxysuccinyl)-Ala-Ala-Pro-Val-CH 2 Cl (Calbiochem) and further incubation for 45 min. D-Phe-Phe-Arg-CH 2 Cl (FFR-CH 2 Cl) and D-Phe-Pro-Arg-CH 2 Cl (FPR-CH 2 Cl) were added to final concentrations of 1 M and phenylmethylsufonyl fluoride to 1 mM. The digest was concentrated to ϳ10 mg/ml by YM-10 ultrafiltration and dialyzed at 4°C against 8 liters of lysine-Sepharose equilibration buffer (150 mM NaH 2 PO 4 , 3 mM EDTA, 0.1 M FFR-CH 2 Cl and FPR-CH 2 Cl, 0.02% NaN 3 , pH 7.4). Mini-Pg appeared in the flow-through of a lysine-Sepharose column (30 ml) at 22°C, with subsequent elution of kringles 1-4 with 3 mM 6-AHA. Mini-Pg was dialyzed at 4°C against 8 liters of 20 mM MES, 25 mM NaCl, 0.02% NaN 3 , pH 6.0, and chromatographed on Resource S (1 ml) (GE Healthcare) in the above buffer. Mini-Pg was pooled from the major peak eluted at ϳ110 mM NaCl from the column developed with a 50-ml gradient up to 110 mM NaCl in the equilibration buffer. Purified mini-Pg was dialyzed against three changes of 4 liters of 50 mM HEPES, 125 mM NaCl, pH 7.4.
Cloning, Expression, and Purification of Wild-type SK and SK Mutants-Wild-type SK (wtSK) and SK⌬K414 constructs were cloned into the pET-30b(ϩ) vector as fusion proteins with a tobacco etch virus proteinase cleavage site, as described previously (22). Deletion of the Arg 253 -Leu 260 segment and Ala substitutions for Glu 249 , Arg 253 , Lys 256 , Lys 257 , Glu 262 , and Glu 263 in the ␤-domain of SK were performed on the SK⌬K414 background. Deletion of Arg 253 -Leu 260 was performed by Quik-Change (Stratagene) mutagenesis with the sense strand primer 5Ј-CGGGAACAAGCTTATAATGAAGAAATAAACAACA-CTGACCTG-3Ј and antisense primer 5Ј-CAGGTCAGTGT-TGTTTATTTCTTCATTATAAGCTTGTTCCCG-3Ј. Confirmed mutants were transformed in Escherichia coli Rosetta 2 (DE3) pLysS cells, and the fusion proteins were expressed and purified by Ni 2ϩ -iminodiacetic acid-Sepharose chromatography, cleaved with tobacco etch virus protease, and subjected to chromatography on Ni 2ϩ -iminodiacetic acid-Sepharose to remove tobacco etch virus protease and noncleaved protein, as described previously (22). When necessary to remove minor impurities, SK mutants were chromatographed on Q-Sepharose (21 ml) equilibrated with 50 mM HEPES, 50 mM NaCl, 1 mm EDTA, pH 7.4, and eluted with a 400-ml gradient of 50 -500 mM NaCl in the equilibration buffer.
Met-SK(K256A/K257A)-The construct for this mutant was cloned into the pET-3 vector and expressed in BL21 (DE3) pLysS cells (Novagen). The cells were grown to an optical density of 0.5 at 600 nm and induced by 0.8 mM isopropyl-␤-Dthiogalactopyranoside for 4 h at 37°C. Lysozyme (Sigma) was added to a final concentration of 1 mg/ml, and cells were lysed by three cycles of freeze-thawing followed by sonication to reduce DNA-associated viscosity. The lysate was centrifuged at 43,000 ϫ g for 30 min, and the supernatant was loaded onto a Pm-SulfoLink column (16 ml) (18,48)  Active Site Labeling of Pm-The active site specific inhibitor, N ␣ -[(acetylthio)acetyl]-(D-Phe)-Pro-Arg-CH 2 Cl, prepared by published procedures (52)(53)(54), was incubated at a 10 -15-fold molar excess with 10 M [Lys]Pm 2 in 0.1 M HEPES, 0.3 M NaCl, 1 mM EDTA, 10 mM 6-AHA, 1 mg/ml PEG, pH 7.0, at 25°C for 60 min. Inactivation was monitored from the initial rate of VLK-pNA hydrolysis until complete (Ͻ0.1% active). Excess inhibitor was removed by chromatography on a 27-ml column of Sephadex G25 equilibrated with the above pH 7.0 buffer. N a -[(acetylthio)acetyl]-(D-Phe)-Pro-Arg-Pm was labeled with tetramethylrhodamine-5-iodoacetamide (Molecular Probes) in 10-fold molar excess initiated by the subsequent addition of 1 M NH 2 OH to a final concentration of 0.1 M and incubation in the dark at 25°C for 1 h. Excess probe was removed by chromatography on Sephadex G25 (39 ml) equilibrated with 50 mM HEPES, 125 mM NaCl, 5 mM 6-AHA, 1 mM EDTA, pH 7.4, and dialysis against the same buffer. Probe incorporation and protein concentration were calculated from the absorbance at 280 nm and the probe absorbance in 6 M guanidine buffer as described (52), with an absorption coefficient of 90,527 M Ϫ1 cm Ϫ1 at 557 nm and an absorbance ratio A 280 nm /A 557 nm of 0.215. Probe incorporation was 0.9 mol of probe/mol of Pm.
Plasminogen Activation Kinetics-Continuous assays of conformational and proteolytic activation of [Lys]Pg by SK species were performed as described previously (10,11,22). Parabolic progress curves of hydrolysis of 200 M VLK-pNA were obtained in the presence of 15 nM [Lys]Pg as a function of SK concentration.
The first order term in the second order polynomial (Equation 1) fit to the progress curves represents the initial rate (v 1 ) of VLK-pNA hydrolysis by the conformationally activated SK⅐Pg* complex, and the rate of the activity increase (v 2 ) obtained from the second order term reflects the initial rate of Pm generation. The kinetics were analyzed based on the simplified mechanism (Reactions 1 and 2), which assumes a negligible contribution from Pm generated by the SK⅐Pm complex (10,11).  (10).
For analysis of v 1 as a function of SK species concentration, K m was fixed at its determined value of 3000 M (10), and K A and k cat were fitted parameters. The stoichiometric factor (n) was also fit for tight binding of wild-type and native SK and otherwise fixed at 1. v 2 is given by Equation 3, where K I is the empirically determined competitive inhibition constant (330 Ϯ 80 M) of the bimolecular generation of Pm by the chromogenic substrate (11).
[Pg] free is given by the quadratic Equation 4.
For analysis of v 2 , K I , and KЈ A were fixed at their previously determined values (5,11), and k Pg and K A were the fitted parameters.
Kinetics of BPg Activation by the SK⅐Pm Complex-The SK⅐Pm complex was formed by incubation of 0.1 nM Pm with 1.2 nM native SK or Met-SK(K256A/K257A) for 12 min before initiation of the reactions by the addition of 200 M VLK-pNA and BPg. The rate of Pm formation as a function of BPg concentration, obtained from the parabolic progress curves as described above, was fit by the Michaelis-Menten equation to obtain k cat and K m .
, and stoichiometric factor (n) the fitted parameters for tight binding. For weaker binding, n was fixed at 1. Nonlinear least-squares fitting was performed with SCIENTIST (MicroMath). Error estimates represent the 95% confidence interval.
Molecular Modeling-X-ray coordinates for the isolated SK ␤-domain (Protein Data Bank entry 1QQR) 7 were superimposed on the structure of the SK⅐Pm complex (1BML) (3). The 250-loop was reconstructed from the superimposed structure of the free ␤-domain, and the resulting model was minimized. The coordinates for human Pg K5 were taken from PDB entry 5HPG (47). Docking of the kringle module to the SK⅐Pm model with a restored 250-loop was performed with ZDOCK (60). Structure figures were generated with PyMOL (68).

RESULTS
To evaluate the possibility that the 250-loop of the SK ␤-domain contained a pseudolysine motif, basic (Arg 253 , Lys 256 , and Lys 257 ) and acidic residues (Glu 249 , Glu 262 , and Glu 263 ) were substituted with Ala. The mutants were made on the backbone of SK⌬K414 to eliminate the interaction of Lys 414 with a Pg kringle that enhances formation of the SK⅐Pg* and SK⅐Pm catalytic complexes (22). Purified SK mutants were evaluated in kinetic studies of [Lys]Pg and mini-Pg activation measured in continuous assays collected as a function of SK concentration. Fitting of the parabolic rate equation (Equation 1) to the progress curves yielded v 1 from the first order term as described under "Experimental Procedures." Analysis of the hyperbolic dependence of v 1 on SK concentration gave the dissociation constant for the conformationally activated SK⅐Pg* complex (K A ) and the specificity constant (k cat /K m ) for chromogenic substrate hydrolysis by the complex (Equation 2). The rate of Pm formation was obtained from the second order term (v 2 ), the SK dependence of which increases at approximate concentration ratios Ͻ0. 5 (Equation 3). This is due to Pg acting both as the catalyst in the SK⅐Pg* complex and as the substrate of this complex, resulting in inhibition of the rate at SK concentrations higher than the optimum due to depletion of free Pg (11). Analysis of v 2 yielded a second estimate of K A because of the coupling of the conformational and proteolytic activation reactions and the bimolecular rate constant for proteolytic generation of Pm (k Pg ) (11). Values of K A derived from v 2 are typically smaller and less well determined than those from v 1 because of the contribution of the SK⅐Pm catalytic complex to Pm generation, which is not included in the simplified mechanism (Reactions 1 and 2) (11). Only values of K A from v 1 are reported for this reason.
Kinetics of [Lys]Pg Activation by SK and SK Mutants-Results of the kinetic analysis for [Lys]Pg activation by native SK and wtSK and those for SK mutants that showed significant decreases in k Pg , including the loop deletion mutant SK⌬(R253-L260), as well as SK(R253A), SK(K256A), SK(K257A), and SK(K256A/K257A) are shown in Fig. 1. The kinetic parameters for all of the mutants are listed in Table 1. The kinetic studies were done in the absence and presence of 10 mM 6-AHA to identify LBS-dependent reactions ( Fig. 1 and Table 1). wtSK and native SK gave the same results, with K A (v 1 ) values of 2 Ϯ 2 and 1.4 Ϯ 0.8 nM, respectively, that were weakened 18-and 28-fold, respectively, by 6-AHA. k Pg in the absence of 6-AHA was reduced to 1-3% of the wild-type and native SK values in the presence of the lysine analog. There is some bias in K A (v 1 ) when reactions are performed at Pg concentrations greater than 5-10 nM, as done in these experiments (15 nM [Lys]Pg), which results in slightly lower K A values. The dissociation constant for SK⅐Pg* measured from equilibrium binding studies is 6 Ϯ 2 nM in the absence of 6-AHA and 115 Ϯ 32 nM in 10 mM 6-AHA (10), whereas the kinetic values were ϳ2 and ϳ40 nM, both ϳ3-fold lower (Table 1). Repeating the kinetic analysis of wtSK and native SK at 5 nM [Lys]Pg gave K A (v 1 ) values of 5 Ϯ 1 and 5 Ϯ 7 nM, respectively (not shown). This presumably also reflects a limitation of the simplified mechanism. On the basis that these differences were fairly small for such tight binding, all of the experiments were conducted at 15 nM [Lys]Pg to maintain internal consistency. Reliable values of k Pg from v 2 could not be obtained in the presence of 10 mM 6-AHA or for SK⌬(R253-L260) or SK(K256A/K257A), because these were essentially zero, reflecting 1-3% residual Pg activator activity relative to wtSK ( Fig. 1 and Table 1).
[Lys]Pg activation by SK⌬K414 was characterized by an 11-fold decrease in affinity of SK⅐Pg* formation in the absence of 6-AHA that was reduced only 1.7-fold in the presence of 6-AHA (Fig. 1A), whereas k Pg was indistinguishable from that of wild-type and native SK and nearly totally inhibited by 6-AHA ( Fig. 1B and Table 1). These results were consistent with previous equilibrium binding and kinetic experiments (10,11,22). SK⌬(R253-L260) and the substitution mutants SK(R253A), SK(K256A), SK(K257A), and SK(K256A/K257A) showed no significant difference from SK⌬K414 in K A (v 1 ), with a mean and range of 24 Ϯ 13 nM, indicating no significant effect of the mutations on SK⅐Pg* formation. In contrast, in the absence of 6-AHA, certain SK mutants showed decreases in k Pg compared with wtSK of 2.1-fold (R253A), 4.3-fold (K256A), and 2.0-fold (K257A), whereas k Pg for SK⌬(R253-L260) and SK(K256A/K257A) was below the limit of detection (Fig. 1, D  and F). The rates of Pm formation were inhibited totally by 10  Table 1. Kinetic data were collected and analyzed as described under "Experimental Procedures." JULY 17, 2009 • VOLUME 284 • NUMBER 29 mM 6-AHA (Fig. 1, B, D, and F, and Table 1). Substitution of acidic residues in SK(E249A), SK(E262A), and SK(E262A/ E263A) produced no significant effects on the dissociation constant and substrate hydrolysis specificity constant for the SK⅐Pg* complex, except for SK(E262A), where the substrate specificity constant was 1.4-fold higher relative to SK⌬K414 (Table 1). k Pg was increased by substitution of the acidic residues: 1.3-fold SK(E249A), 2.0-fold SK(E262A), and 1.8-fold SK(E262A/E263A).

Streptokinase-Plasminogen Substrate Recognition
Kinetics of Mini-Pg Activation by Native SK, wtSK, and SK Mutants-The ability of the same SK species to activate mini-Pg, which lacks kringles 1-4, was also evaluated in kinetic studies of mini-Pg activation ( Fig. 2 and Table 1). K A (v 1 ) was indistinguishable from that of wtSK, with a mean and range of 38 Ϯ 14 nM for native, wild type, and all of the SK mutants, and was affected Յ1.4-fold by 6-AHA (Fig. 2, A, C, and E). The observation that the affinity of SK⅐mini-Pg* was independent of Lys 414 deletion, and 6-AHA demonstrated that Lys 414 does not interact with K5, the only kringle present in mini-Pg, to enhance SK⅐mini-Pg* catalytic complex formation (Fig. 2A). The specificity constant for chromogenic substrate hydrolysis by wtSK⅐mini-Pg* of 16 Ϯ 1 nM Ϫ1 s Ϫ1 was similar to that of wildtype and native SK⅐[Lys]Pg* of 12.6 Ϯ 0.3 and 12.8 Ϯ 0.4 nM Ϫ1 s Ϫ1 , respectively, and was also independent of Lys 414 and 6-AHA.
In contrast, ␤-domain substitutions and 250-loop deletion in SK greatly affected recognition of mini-Pg as a substrate. The value of k Pg for mini-Pg decreased 4.0-fold for the SK mutant R253A, 6.9-fold for K256A, and 4.3-fold for K257A and was essentially abolished for SK⌬(R253-L260) and SK(K256A/ K257A). The value of k Pg was affected no more than 1.4-fold for the other mutants (Fig. 2, B, D, and F, Table 1. Kinetic data were collected and analyzed as described under "Experimental Procedures."

TABLE 1 Kinetic parameters determined for conformational and proteolytic [Lys]Pg and mini-Pg activation
Values of K A (v 1 ) and k cat /K m (v 1 ) (for VLK-pNA hydrolysis by the SK⅐Pg* complexes) derived from analysis of v 1 , and k Pg (v 2 ) from analysis of v 2 are listed for wtSK, native SK, SK⌬K414, the indicated SK deletion mutant, and Ala substitution mutants. All of the mutants were made on the background of SK⌬K414. The parameters determined for ͓Lys͔Pg and mini-Pg activation are listed for reactions in the absence of 6-AHA (No 6-AHA) and presence of 10 mM 6-AHA (10 mM 6-AHA). Experimental error represents Ϯ2 S.D. NM, not measurable. Kinetics experiments were performed, and the progress curves were analyzed as described under "Experimental Procedures." No 6-AHA 10 mM 6-AHA exhibited the largest decrease in K A (v 1 ) of 1.5-fold and the largest increase in k Pg of 1.4-fold (Table 1). Similar to the results with [Lys]Pg, k Pg for wild-type SK, native SK, and the SK mutants was fully inhibited (1-3% activity) by 6-AHA (Fig. 2, B,  D, and F). The results demonstrated that substrate recognition of mini-Pg by the SK⅐mini-Pg* catalytic complex was facilitated by interaction of residues Arg 253 , Lys 256 , and Lys 257 through K5 of mini-Pg.
Recognition of BPg as a Substrate of the SK⅐Pm Complex-Equilibrium binding and kinetic studies of BPg were performed to evaluate the role of Lys 256 and Lys 257 in substrate binding to the SK⅐Pm complex. BPg binds SK weakly in the catalytic mode and is not thought to be conformationally activated by SK but acts as a specific substrate of the SK⅐human Pm catalytic complex, allowing a clear distinction between conformational activation and substrate recognition (12,13,61). 8 For these experiments, we used a full-length SK construct (i.e. with the COOH-terminal Lys 414 ) containing a Met residue preceding Ile 1 . Blockage of the natural NH 2 terminus of SK prevents conformational activation of Pg but is inconsequential in studies of the SK⅐Pm complex, where the NH 2 terminus of Pm is inserted, as shown in the SK⅐Pm structure (3). Pm was labeled specifically at the active site with tetramethylrhodamine-5-iodoacetamide linked by  Table 2). These values were 24-and 11-fold weaker than determined previously for native Pm (12 pM) (5) and Pm labeled at the active site with fluorescein (11-19 pM) (5,21,22). The presence of 40 mM 6-AHA reduced the affinity of native SK and Met-SK(K256A/K257A) 5-and 10-fold, respectively ( Fig. 3A and Table 2). The loop deletion mutant, SK⌬(R253-L260), also lacking Lys 414 , bound to [TMR]FPR-Pm with a K D of 1.2 Ϯ 0.2 nM, essentially the same as Met-SK(K256A/K257A) in the presence of 6-AHA (K D ϭ 1.3 nM), suggesting that there were no major global structural changes in SK accompanying deletion of the loop ( Table 2).
Titrations of the fluorescence of [TMR]FPR-Pm complexes formed in the presence of saturating SK with BPg resulted in an additional 114 Ϯ 5% fluorescence enhancement with a dissociation constant of 990 Ϯ 130 nM (Fig. 3B). By contrast, BPg produced only a ϳ3% maximum fluorescence change in the Met-SK(K256A/K257A)⅐[TMR]FPR-Pm complex over the same concentration range, indicating greatly reduced affinity of the mutant for BPg (Fig. 3B). Steady-state kinetics of BPg activation by the SK⅐Pm complex gave a K m of 1120 Ϯ 140 nM and k cat of 2.4 Ϯ 0.1 s Ϫ2 (Fig. 3C and Table 2). The K m was indistinguishable from the K D for BPg binding in the fluorescence experiments. Titrations of SK⅐[TMR]FPR-Pm complex with BPg in the presence of 40 mM 6-AHA reduced the fluorescence change to a small linear increase (Fig. 3B). Assuming the same fluorescence maximum gave an estimate of ϳ30-fold decreased affinity (Table 2). Likewise, no significant rate of BPg activation was observed with the Met-SK(K256A/K257A)⅐Pm complex (Fig. 3C). The rate of BPg activation was reduced to an unmea-8 P. E. Bock, unpublished results. surable level by 40 mM 6-AHA (Fig. 3C). The absence of a fluorescence change in BPg titrations of the Met-SK(K256A/ K257A)⅐[TMR]FPR-Pm complex was also unaffected by 6-AHA (Fig. 3B). These results indicated that BPg substrate binding was facilitated by the same LBS-dependent interactions with the principal residues identified kinetically and was independent of the weaker affinity of the fluorescently labeled Pm when it was saturated with SK.

DISCUSSION
The results of kinetics and equilibrium binding studies support the conclusion that residues Arg 253 , Lys 256 , and Lys 257 in the SK ␤-domain facilitate LBS-dependent Pg substrate recognition through interaction with kringle 5 of the substrate. The increases in [Lys]Pg processing to Pm accompanying alanine substitution of acidic residues in and near the 250-loop argues against a pseudolysine structure in the 250-loop. Application of kinetic approaches that resolve Pg conformational activation and proteolytic generation of Pm made it possible to demonstrate a specific function of residues in the 250-loop in Pg substrate recognition, which was independent of conformational Pg activation. This is apparently not in agreement with the results of some previous studies indicating that both conformational and proteolytic activation are affected by mutations in the 250-loop (28,31). The results also extend previous work by Sahni and co-workers (27,31,34), from which it was concluded that the 250-loop was involved in Pg substrate interactions by the SK⅐Pm complex. Our studies and those of others have also shown that two molecules bind to SK in LBS-dependent interactions through different sites to assemble the product-forming SK⅐Pg*/Pm⅐Pg ternary complexes (10,11,13,35). The results obtained here with mini-Pg activation are the first to identify kringle 5 as the site of LBS-dependent interactions with SK residues Arg 253 , Lys 256 , and Lys 257 in the SK 250-loop that facilitate Pg substrate recognition.
Binding of the COOH-terminal Lys 414 residue of SK is responsible for LBS-dependent enhanced affinity of formation of the SK⅐Pg* and SK⅐Pm catalytic complexes (22). Recent rapid reaction kinetic studies of the SK⅐Pm binding pathway and the effects of 6-AHA and benzamidine on the elementary steps sug-gested that Lys 414 probably binds to K4 (26). The results of the present studies with mini-Pg show that Lys 414 does not bind to K5 in this substrate. Remarkably, mini-Pg had a K A (v 1 ) of 36 Ϯ 12 nM with SK⌬K414, whereas [Lys]Pg had only a 1.6-fold higher affinity of 22 Ϯ 7 nM, and the specificity constants for chromogenic substrate hydrolysis by SK⌬K414⅐[Lys]Pg* and SK⌬K414⅐mini-Pg* were similar at 12.6 Ϯ 0.3 and 16 Ϯ 1 nM Ϫ1 s Ϫ1 , respectively. It appears that in the absence of Lys 414 , kringles 1-4 only have a small effect on conformational activation of Pg, indicating that the affinity and activation of the Pg catalytic domain, when unassisted by Lys 414 , is dominated by catalytic domain interactions. This can also be appreciated by the similar affinities and k cat /K m values of substrate hydrolysis for conformational activation of [Lys]Pg and mini-Pg with wildtype and native SK, compared with SK⌬K414 when LBS interactions are blocked by 6-AHA (Table 1).
K5 binds cationic ligands such as benzamidine, N ␣ -acetyl-Lys-methyl ester, 6-aminohexane, and 5-aminopentane, in addition to zwitterionic lysine analogs (19,20,(45)(46)(47). The specificity of K5 might be explained in principle by the absence of basic residues that interact with the anionic groups of lysine analogs, the so-called cationic center residues Arg 117 and/or Arg 153 in K1 (62), Arg 234 in K2 (63), and Lys 392 and/or Arg 426 in K4 (64). By contrast, the topologically equivalent residues in human Pg kringle 5 possess aliphatic/aromatic side chains (Ile 496 , Phe 497 , and in particular Leu 532 ) (47). This specificity is compatible with the absence of a pseudolysine structure in the 250-loop of SK. However, we stress that most of the information regarding LBS specificity has been derived from studies with low molecular weight compounds, which might not accurately reproduce the binding mode of kringle-interacting macromolecules. For instance, binding of an internal lysine residue to the LBS of human K4 has been reported (65) (Protein Data Bank code 1PML). Although the physiological significance of the trimer of kringles trapped in this crystal structure has not been clarified, it highlights the possibility of "noncanonical" binding to a kringle module, even in those specific for small zwitterionic compounds. In the structures of the internal PAM peptide, VEK-30, bound to either a triple mutant of human K2    (41) or to recombinant human angiostatin (Pg kringles K1-K3 (42)), both anionic and cationic centers of the kringle engage in electrostatic interactions with basic (Arg 101 and His 102 ) and acidic ligand residues (Glu 104 ; PAM numbering system). However, a recent study identified two residues that do not belong to the "canonical" LBS of K2 as responsible for the specific binding of VEK-30 to human but not murine Pg (Arg 220 and Leu 222 in the human protein) (66). These observations, among others, prompt further structural and functional investigations of kringle interactions with macromolecular ligands. The SK ␤-domain, and in particular its 250-loop, is one of the most variable regions in SK from different strains (67) (Fig. S1). In one of the deposited crystal structures of the ␤-domain (1QQR), 7 the sequence of the 250-loop corresponds to the protein used in the current investigation. Interestingly, this loop adopts essentially the same conformation in all four crystallographically independent molecules, with Arg 253 forming a hydrogen-bonded salt bridge with Glu 262 . In the other reported structure of the SK ␤-domain (1C4P) (40), Arg 253 is replaced by a Glu residue found in a different streptococcus species (40) (Fig. S1). In this case, the 250-loop presents slightly different conformations in all four independent molecules, with side chains of Glu 253 and Glu 262 rotated away to minimize electrostatic repulsion. In all of these eight independent molecules, Glu 262 is oriented with its carboxylic acid located at least ϳ9 Å from the ⑀-amino group of Lys 257 and between ϳ10 and 14 Å from the ⑀-amino group of Lys 256 , whereas the carboxylic acid of Glu 263 points in the opposite direction. Although the high flexibility of the 250-loop could theoretically allow the lysine residues to approach the carboxylates from Glu 262 /Glu 263 , it is important to notice that the overall extended (hairpin-like) conformation of this loop is maintained in all independent copies of the SK ␤-domain, which is stabilized by a large number of mostly main chain to main chain hydrogen bonds. These interactions extend to the very tip of the 250-loop, with, for example, the Asn 255 carboxamide accepting hydrogen bonds from the backbone nitrogen atoms of residue 257, 258, and/or 259. Accordingly, alanine substitution of both Glu residues produced a modest 1.8-fold increase in k Pg , indicating that they do not participate in a pseudolysine structure. The increase in k Pg produced by mutating these acidic residues to alanine suggests that the negative charge of these residues inhibits the Lys 256 and Lys 257 interaction with substrate Pg or alters the conformation of the 250-loop in a favorable way for Pm generation.
It should be appreciated that we have focused on identifying the SK structure responsible for the LBS dependence of Pg substrate recognition. The results do not rule out roles for other structural elements of SK in Pg recognition and processing to Pm. Notably, some studies proposed that a region in the SK ␣-domain (residues 45-70) that is disordered in the SK⅐Pm crystal structure is important for Pg substrate recognition (32,33,36). Although the present results show a nearly total dependence of the bimolecular rate constant for Pm formation on Lys 256 and Lys 257 and 6-AHA, how much of this is due to increases in K m or decreases in k cat remains to be determined.
The results of equilibrium binding studies with BPg as a substrate indicated that loss of binding affinity contributes significantly to the loss of activity in Pm generation by the SK⅐Pg/Pm complexes associated with Lys 256 and Lys 257 substitution. The use of [TMR]FPR-Pm in these experiments allowed both SK binding and formation of a catalytically inactive, SK⅐[TMR]FPR-Pm⅐BPg ternary complex to be quantitated in the absence of proteolysis. Ternary complex formation was inhibited roughly 30-fold by 6-AHA, demonstrating the involvement of LBS interactions. The demonstration that BPg substrate binding was undetectable for the Met-SK(K256A/ K257A)⅐[TMR]FPR-Pm complex shows the importance of Lys 256 and Lys 257 in this interaction. Importantly, these results were quantitatively paralleled by the kinetics of BPg activation catalyzed by the SK⅐Pm complex. The K m and K D for BPg activation and binding were indistinguishable; BPg activation was not supported by Met-SK(K256A/K257A) and was weakened similarly by 6-AHA. The close correspondence between K D and K m indicates that although labeling the active site of Pm reduced its affinity for SK and Met-SK(K256A/K257A) binding, there were no significant effects on BPg binding as a substrate at saturating concentrations of SK.
The critical role played by the triplet of basic residues Arg 253 , Lys 256 , and Lys 257 in Pg substrate recognition and processing can be rationalized on the basis of a model of the human Pg K5 domain docked to the SK⅐Pm structure with an intact 250loop (Fig. 4, A and B; see above). This unique docking solution allows simultaneous interaction of all three basic residues with the kringle module. In this model, both nitrogen atoms of the Arg 253 guanidinium group are within hydrogen bond distance of the Tyr 525 hydroxyl oxygen, and the Arg side chain also engages in close van der Waals contacts with Ile 496 and Leu 532 . Lys 256 occupies the LBS via strong interactions with the Phe 497 aromatic ring and with its ammonium group in position to form either a salt bridge to Asp 516 or a hydrogen bond with the Asn 502 carboxamide oxygen. Finally, Lys 257 is within hydrogen bonding distance of the Glu 500 carboxylate (Fig. 4B). These interactions would help in positioning of the presumably rigidly assembled K5-catalytic domain tandem into the active site of SK-bound Pg*/Pm, in such a manner that the disulfide-constrained activation loop of substrate Pg inserts into the proteinase active site cleft. It is noteworthy that Lys 257 is the only residue within the 253-265 stretch that is strictly conserved among SK from different streptococcal species (Fig. S1). It is not possible to assess whether SK variants with nonconservative variations at positions 253 (Glu, Gln, or Gly) and/or 256 (Pro, Ser, or Asn) would bind in similar conformations to the Pg K5 domain or rely on other surface-exposed loops for substrate recognition. However, it is important to appreciate that over half of the SK sequences deposited to date contain the Lys 256 / Lys 257 pair (74 of 138, or 53.6%), whereas position 253 among "Lys 257 only variants" is usually occupied by a second lysine residue (44 isolates; 31.9%). Thus, it would seem that different combinations of basic residues along the tip of the 250-loop are suitable for Pg K5 recognition. In conclusion, the combined kinetic, equilibrium binding, and molecular modeling results support the conclusion that Pg substrate binding and proteolytic generation of Pm are facilitated primarily by noncanonical interactions between SK residues Arg 253 , Lys 256 , and Lys 257 and K5 of the substrate.