Rapid Binding of Plasminogen to Streptokinase in a Catalytic Complex Reveals a Three-step Mechanism*

Background: We previously showed that plasmin binding to streptokinase is a three-step mechanism with a slow off-rate. Results: Using rapid kinetics and equilibrium binding, we defined the unknown mechanism of plasminogen binding to streptokinase. Conclusion: Encounter complex formation and conformational tightening are weakened in the three-step binding mechanism. Significance: The results define the molecular basis for plasminogen displacement by plasmin in complexes with streptokinase. Rapid kinetics demonstrate a three-step pathway of streptokinase (SK) binding to plasminogen (Pg), the zymogen of plasmin (Pm). Formation of a fluorescently silent encounter complex is followed by two conformational tightening steps reported by fluorescence quenches. Forward reactions were defined by time courses of biphasic quenching during complex formation between SK or its COOH-terminal Lys414 deletion mutant (SKΔK414) and active site-labeled [Lys]Pg ([5-(acetamido)fluorescein]-d-Phe-Phe-Arg-[Lys]Pg ([5F]FFR-[Lys]Pg)) and by the SK dependences of the quench rates. Active site-blocked Pm rapidly displaced [5F]FFR-[Lys]Pg from the complex. The encounter and final SK·[5F]FFR-[Lys]Pg complexes were weakened similarly by SK Lys414 deletion and blocking of lysine-binding sites (LBSs) on Pg kringles with 6-aminohexanoic acid or benzamidine. Forward and reverse rates for both tightening steps were unaffected by 6-aminohexanoic acid, whereas benzamidine released constraints on the first conformational tightening. This indicated that binding of SK Lys414 to Pg kringle 4 plays a role in recognition of Pg by SK. The substantially lower affinity of the final SK·Pg complex compared with SK·Pm is characterized by a ∼25-fold weaker encounter complex and ∼40-fold faster off-rates for the second conformational step. The results suggest that effective Pg encounter requires SK Lys414 engagement and significant non-LBS interactions with the protease domain, whereas Pm binding additionally requires contributions of other lysines. This difference may be responsible for the lower affinity of the SK·Pg complex and the expression of a weaker “pro”-exosite for binding of a second Pg in the substrate mode compared with SK·Pm.

The serine proteinase plasmin (Pm) 2 is primarily known for its role in dissolving fibrin thrombi (1). It also causes cell surface remodeling, signaling, and cancer progression (2). Proteolytic activation of the zymogen plasminogen (Pg) by tissue plasminogen activator and urokinase-type plasminogen activator differs from conformational activation by the non-enzymatic streptococcal pathogenicity factor streptokinase (SK). We studied SK from Streptococcus dysgalactiae subsp. equisimilis because of its 90% homology with phylogenetic cluster 1SKs from the human host-specific, virulent Streptococcus pyogenes (3). S. dysgalactiae subsp. equisimilis, which is generally opportunistic in horses, also causes severe human infections such as bacteremia, pneumonia, endocarditis, arthritis, and streptococcal toxic shock syndrome (4,5).
The Pg activation mechanism by SK is unique (6 -8). Stoichiometric binding of SK to Pg and Pm forms catalytically active SK⅐Pg* and SK⅐Pm complexes that bind Pg as a substrate in SK⅐Pg*⅐Pg and SK⅐Pm⅐Pg assemblies and cleave Arg 561 -Val 562 in the Pg protease domain to form Pm (6, 8 -14). Conformational activation of Pg in the catalytic SK⅐Pg* complex by the molecular sexuality mechanism involves insertion of the NH 2 -terminal Ile 1 -Ala 2 residues of SK into the binding cleft of the Pg protease domain (9,11,12,(15)(16)(17). Ile 1 binds Pg Asp 194 (chymotrypsinogen numbering), causing expression of the substrate-binding site and formation of the oxyanion hole (15,16,18,19). The mechanism is also valid for conformational prothrombin activation by staphylocoagulase and von Willebrand factor-binding protein from Staphylococcus aureus (20,21). This mechanism allows group A and C streptococci to hijack Pg in the human fibrinolytic system by quorum sensing-induced secretion of SK. This results in localized plasmin generation for dissolution of host fibrin barriers and facilitated bacterial spreading (22)(23)(24)(25).
In our unified model, the conformationally activated SK⅐Pg* complex binds Pg as a substrate and cleaves it to Pm. This is the trigger step in a self-limiting mechanism (6 -8). After 1 SK eq of Pm is formed, it displaces Pg from the SK⅐Pg* complex to form the tight SK⅐Pm catalytic complex (with dissociation constant (K D ) of 12 pM (26,27)) that cleaves the remaining free Pg to Pm in a second catalytic cycle, the bullet cycle (6). In the SK⅐Pm complex, the three SK ␤-grasp domains rearrange from a beads-on-a-string conformation (28) to a crater surrounding the Pm active site (19). This forms a novel exosite for substrate Pg binding (19,26).
Allosteric linkage between the protease active site and its exosite(s) allows investigating equilibrium binding of ligands to serine proteases labeled at their active sites with fluorescent probes (54 -56). Active site labeling of the conformationally activated zymogens plasminogen and prothrombin (27,57) provides the advantage of studying ligand binding uncoupled from catalytic activity. Introducing the fluorescent label 5-fluorescein ([5F]) and a tripeptide chloromethyl ketone in the Pm active site (FFR-Pm) does not affect the affinity for SK, whereas labeled [Glu]Pg and [Lys]Pg analogs bind SK with ϳ5-fold lower affinity than the native proteins (26,48,58). We compared the binding kinetics of labeled Pg and Pm that have their active sites similarly locked in a substrate-binding conformation by the tripeptide chloromethyl ketone.
Here we explore for the first time the steps on the pathway of SK binding to Pg and identify critical differences with Pm binding (59) that are the basis for the ϳ4,000-fold lower affinity of [5F]FFR-[Lys]Pg for SK (27). Stopped-flow kinetics of SK binding to labeled [Lys]Pg and [Glu]Pg defined the forward reactions of complex stabilization. Reverse reactions were studied by competitive displacement of labeled [Lys]Pg by active site-blocked FFR-Pm in the complex with SK. Forward and reverse reactions were biexponential, and overall off-rates were fast, requiring stopped-flow monitoring. Parameters from numerical integration of full forward and reverse time traces were consistent with those from the SK dependences of the forward rates of these conformational changes. This approach allowed a comparison of the elementary reaction steps in the sequence of SK⅐Pg* and SK⅐Pm formation. The data support a three-step mechanism of encounter complex formation followed by two tightening conformational steps as shown previously for SK⅐Pm (59) but with dramatic decreases in affinity of the encounter complex and ϳ10 -40-fold increases in off-rates for both conformational steps. Based on selective blocking of LBSs on Pg and Pm and binding experiments with SK lacking Lys 414 , we propose that the SK⅐Pg* complex is stabilized by SK Lys 414 binding to LBSs on Pg and non-LBS interactions of SK within the Pg catalytic domain. These interactions are also present in the SK⅐Pm complex; however, additional contributions of SK lysines other than Lys 414 may be partially responsible for the substantially tighter SK-Pm interaction.
Active Site Labeling of Pg- [Glu]Pg and [Lys]Pg were labeled at the active site as described previously (27,58,63). The SK⌬(R253-L260)⌬K414-His 6 mutant activates [Lys]Pg conformationally, but the complex does not readily cleave Pg to Pm, and the use of this SK construct for Pg labeling significantly increased the yield of labeled Pg and reduced the preparation time (63). Labeled Pg concentration and probe incorporation (ϳ90%) were determined from the probe and protein absorbances in 6 M guanidine as described (48,54,55). Proteins were homogeneous by SDS gel electrophoresis.
Stopped-flow Kinetics of nSK, WT-SK, and SK⌬K414 Binding to [5F]FFR-Pg-Complete progress curves of SK binding to labeled Pg were captured with an Applied Photophysics SX-18MV stopped-flow spectrofluorometer in single mixing mode with excitation at 500 nm and an emission cut-on filter (Melles-Griot) with 50% transmission at 515 nm. Changes in fluorescence intensity were measured for all the reactions, and for the interaction of native SK with [5F]FFR-[Lys]Pg in the absence of lysine analogs, changes in fluorescence anisotropy were also monitored. The reaction volume was 200 l, the path length was 2 mm, and experiments were performed at 25°C. Subtracting background scattering was critical as the signal-tonoise ratio for reactions with [5F]FFR-Pg (ϳ25% quench and ϳ7.5% maximal scattering) was up to 5-fold smaller than that for reactions with [5F]FFR-Pm (ϳ50% quench and ϳ6% maximal scattering) (59). Experiments were limited to labeled Pg concentrations up to 20 nM due to the lower solubility of the SK⅐Pg complex compared with that of SK⅐Pm, which resulted in a substantial increase in background scattering at Pg concentrations above ϳ40 nM. None of our previously published studies of SK binding and kinetics have used [Lys]Pg and [Glu]Pg concentrations exceeding 20 and 30 nM, respectively. Averaged time traces were analyzed using Equation 1.
where F o is the starting fluorescence, F M is the final fluorescence, A 1 is the fractional amplitude of the fast exponential component, (1 Ϫ A 1 ) is the fractional amplitude of the slow exponential component, and k obs 1 and k obs 2 are the observed first-order rate constants for the fast and the slow conformational changes. The rate constants were analyzed as a function of the total SK concentration ([SK] o ) using Equation 2.
where K 1 is the dissociation constant for the SK⅐Pg encounter complex and k lim 1,2 and k off 1,2 are the limiting rates and the reverse rate constants for each conformational step, respectively. Unlike the SK reactions with labeled Pm, the reactions with Pg were not stoichiometric due to weaker SK binding, and a range of SK and FFR-Pm concentrations was used to obtain time traces at various degrees of SK saturation with labeled [Lys]Pg and FFR-Pm. The faster and much larger displacement signal for FFR-Pm compared with FFR-Pg and the vastly lower scattering background of the SK⅐FFR-Pm complex were major reasons for performing displacement experiments with FFR-Pm rather than with FFR-Pg. The time traces were fit by a double exponential function (analogous to Equation 1) to obtain the observed first-order rate constants k disp 1 and k disp 2 for the fast and slow displacement processes.
Equilibrium Binding of [5F]FFR-Pg to SK and SK⌬K414 in the Presence of Benzamidine-[5F]FFR-Pg (10 nM) was titrated with SK or SK⌬K414 at 25°C in 50 mM HEPES, 0.075 M NaCl, 1 mM EDTA, 1 mg/ml PEG 8000 buffer, pH 7.4 containing 50 mM benzamidine, 1 mg/ml BSA, and 1 M FFR-CH 2 Cl. Fluorescence titrations were performed with a Photon Technology International, Inc. fluorometer at excitation and emission wavelengths of 500 and 516 nm, respectively, with 2/8-nm excitation/emission band passes. Fluorescence changes were measured after equilibration for 5-10 min. Measurements were corrected for background (Յ10%) by subtraction of blanks lacking [5F]FFR-Pg. Data were analyzed by the quadratic equation for binding of a single ligand (55). This analysis gave the dissociation constant (K D ) for binding of SK or SK⌬K414 to [5F]FFR-Pg and the maximum fluorescence intensity change (⌬F max /F o ) with a stoichiometric factor (n) of 1 for binding of SK or SK⌬K414 to labeled Pg. Two-exponential time traces of forward and reverse reactions, SK dependences of k obs 1,2 , and equilibrium binding of SK and SK⌬K414 to [5F]FFR-Pg in benzamidine buffer were analyzed by nonlinear least square fitting with SCIENTIST Software (MicroMath). All reported estimates of error represent Ϯ2 S.D.

Numerical Integration Analysis of the Forward and Reverse
Reactions-Arrays of progress curves for SK⅐[5F]FFR-Pg formation and displacement of labeled Pg from the complex were analyzed globally with the numerical integration program Kin-Tek Explorer 3.0 (67-69) for each set of reactants, concentration ranges, and buffer conditions. Five arrays were performed in the absence of lysine analogs: fluorescence amplitude changes of [ SK ϩ Pg L | ; The mechanism included Scheme 1 for three-step SK⅐[5F]FFR-Pg binding and Scheme 2 for competitive three-step SK⅐FFR-Pm binding. The dissociation constants K 1 and K 4 for formation of the SK⅐Pg 1 and SK⅐Pm 1 encounter complexes represent the ratios k Ϫ1 /k 1 and k Ϫ4 /k 4 where k 1 and k 4 are the second-order association rate constants and k Ϫ1 and k Ϫ4 are the first-order rate constants for dissociation of the encounter complex. K 1 , k 2 , k Ϫ2 , k 3 , and k Ϫ3 in this mechanism are equivalent to K 1 , k lim 1 , k off 1 , k lim 2 , and k off 2 , respectively, in Equation 2. The threestep mechanism for SK⅐Pm stabilization was validated in a previous study (59). Time traces of fluorescence quenches were transformed to increases by plotting ⌬F/F o expressed as functions of the formation and stabilization of SK⅐Pg 1, SK⅐Pg 2, and SK⅐Pg 3 complexes using positive amplitude factors as KinTek Explorer does not accept negative parameters. The set of fluorescence anisotropy increases was analyzed without transformation. The fractional change in fluorescence intensity or anisotropy was expressed as Fitting Strategy-The on-rate constant k 1 for formation of the encounter complex was initially constrained at 1 ϫ 10 8 M Ϫ1 s Ϫ1 as determined experimentally for SK binding to unlabeled [Lys]Pg (7), and the assumption was made that similar on-rates would apply for reactions of SK with [5F]FFR-[Lys]Pg and [5F]FFR- [Glu]Pg and of SK⌬K414 with [5F]FFR-[Lys]Pg in all of our experimental buffers. Upon refinement of the other parameters, fitting k 1 yielded values that were close to 1 ϫ 10 8 M Ϫ1 s Ϫ1 under all these conditions, justifying our choice of this value as an initial estimate. The parameters K 1 , k 2 , k Ϫ2 , k 3 , and k Ϫ3 were initially constrained to K 1 , k lim 1 , k off 1 , k lim 2 , and k off 2 obtained from the SK dependences (Table 1, superscript b), and refinement of these initial estimates ultimately provided the final fits (superscripts a and aa).
Analysis of [5F]FFR-[Lys]Pg displacement required known concentrations of free Pg and the intermediates SK⅐Pg 1, SK⅐Pg 2, and SK⅐Pg 3 present at the start of the reaction with FFR-Pm as there was substantial partitioning among these species at equilibration of the SK⅐[5F]FFR-[Lys]Pg complex. They were calculated iteratively using the starting concentrations of SK and [5F]FFR-[Lys]Pg used to form the complex, the forward and reverse rate constants, and the known dissociation constant for the competitive, unlabeled SK⅐FFR-Pm complex. The sum of the calculated free Pg, SK⅐Pg 1, SK⅐Pg 2, and SK⅐Pg 3 concentrations was in agreement with the total Pg concentration, indicating that mass balance was conserved during the fits. Complexes of SK with labeled and unlabeled Pm have indistinguishable affinities in the absence of lysine analogs (26,48) and in 6-AHA (26,27), suggesting that the binding parameters for SK are very similar for labeled and unlabeled FFR-Pm. This allowed fixing K 4 , k 5 , k Ϫ5 , k 6 , and k Ϫ6 to our previously determined values for [5F]FFR-Pm binding in each buffer system (59). Displacement of [5F]FFR-[Lys]Pg binding to SK and SK⌬K414 in benzamidine was analyzed with fitted K D values of 227 Ϯ 11 and 200 Ϯ 20 pM for FFR-Pm binding, respectively, in agreement with the previously determined 130 and 250 pM (59).
The large scattering background introduced variable uncertainty in the amplitude factors f 2 and f 3 for [SK⅐Pg 2] and [SK⅐Pg 3] at increasing SK concentrations, resulting in non-random residuals when imposing global f 2 and f 3 fits on the complete data sets. Initial estimates of the rate constants obtained by global fitting of f 2 and f 3 were fixed, and individual f 2 and f 3 amplitude factors were assigned as fitted parameters for time traces at each SK concentration. This largely eliminated the non-random deviations. Subsequent fixing of all the individual amplitude parameters provided further refinement of the fitted rate constants with only subtle differences from the original estimates.
The overall K D values for the final, stabilized complexes were calculated from Equation 3 using the rate constants obtained by numerical analysis and compared with K D, overall obtained independently from equilibrium binding.  Fig. 1. Colored lines represent global fits of forward and reverse reactions by numerical analysis. The first-order rate constants k obs 1 and k obs 2 for the fast and slow fluorescence changes obtained from the individual biexponential fits increased hyperbolically with increasing SK concentration. Fig. 2 shows the nSK and WT-SK dependences in the absence of effectors. The hyperbolic dependences of k obs 1 and the much smaller k obs 2 , respectively, indicated saturation of the encounter complex and the subsequent conformational intermediate (Fig. 2, inset). The parameters K 1 , k lim 1 , k lim 2 , k off 1 , and k off 2 obtained by fitting the binding rate constants by Equation 2 are given in Table 1 (superscript b). The reverse rate constants k off 1 and k off 2 for the two conformational steps given by the extrapolated intercepts of k obs 1 and k obs 2 at zero SK were 4.0 Ϯ 1.0 and 0. 25 Fig. 4. Blocking the LBSs on kringles K1, K4, and K5 with 6-AHA decreased the affinity of the encounter complex to a K 1 value of 7 M (Table 1, superscript b), which is comparable with that of SK⌬K414 binding in the absence of 6-AHA. The SK and SK⌬K414 dependences of these weak binding interactions were not saturable, preventing accurate determination of K 1 ; hence values ranging from 7 to 20 M may be considered comparable. A weak encounter complex with a K 1 value of 14 M was also observed for SK⌬K414 binding in 6-AHA (Fig. 5). The limiting rate constants k lim 1 and k lim 2 and the off-rates k off 1 and k off 2 determining the conformational changes following encounter complex formation were similar for SK and SK⌬K414 binding to [5F]FFR-[Lys]Pg in the absence and presence of 6-AHA. Total amplitudes of the time traces fit by Equation 1 reflected overall maximal fluorescence changes (⌬F max / F o ) for SK and SK⌬K414 binding to [5F]FFR-Pg in agreement with equilibrium binding results in the absence and presence of 6-AHA (27,48 were not resolvable into two phases and appeared as single exponential curves, whereas SK⌬K414 binding was clearly biphasic. Progress curves for binding of SK and SK⌬K414 and their concentration dependences of k obs 1 and k obs 2 in benzamidine are shown in Fig. 6. The limiting rate k lim 1 in benzamidine was ϳ5-fold faster for SK binding and ϳ2.3-fold faster for SK⌬K414 binding compared with the values in 6-AHA and in the absence of lysine analogs (Table 1, superscript b). The ϳ30% lower maximal fluorescence changes than those for equilibrium binding in the presence of benzamidine described below (Table 1) may be due to the scattering properties of Pg complexes in benzamidine being differentially affected by the optical cell geometry and path length of the Photon Technology International, Inc. fluorometer and the stopped-flow instrument.  Table 1 as described under "Experimental Procedures."  Table 1 for the reactions with nSK and WT-SK, respectively.

Stopped-flow Kinetics of SK Binding to [5F]FFR-[Glu]Pg-Biexponential binding of SK to [5F]FFR-[Glu]
Pg was not saturable, and the k 2 /K 1 ratios in the absence and presence of 6-AHA were indistinguishable and similar to those for [5F]FFR-[Lys]Pg binding to SK⌬K414 in the absence of kringle ligands and to nSK and SK⌬K414 binding in 6-AHA (Fig. 7). Fitting of these near linear dependences was performed using fixed, lower limit K 1 values that were reasonably resolvable by numerical integration (see below). The limiting rate constants k lim 1 and k lim 2 and the off-rates k off 1 and k off 2 were similar to those for SK and SK⌬K414 binding to [5F]FFR-[Lys]Pg in the absence of lysine analogs and in 6-AHA (Table 1, Table 1. Experiments were performed and analyzed as described under "Experimental Procedures."

TABLE 1 Kinetic and equilibrium binding parameters for the formation of SK⅐Pg and SK⌬K414⅐Pg complexes
Kinetic constants obtained from simultaneous numerical integration of the forward and reverse reactions a , forward reactions measured as anisotropy changes aa , and SK dependences of the fast and slow phases of the forward reactions b are listed for reaction Scheme 1 in the absence of kringle ligands (no effector) and in the presence of saturating 6-AHA or benzamidine. K D , overall was calculated from the individual kinetic parameters c and measured by fluorescence titration d . For analysis of near-linear SK dependences K 1 was fixed to the value obtained by numerical analysis fixed . Amplitudes of change in fluorescence intensity (⌬F max /F o ) were from numerical analysis (f 2 and f 3 for SK⅐Pg 2 and SK⅐Pg 3, respectively) and equilibrium binding (overall value). Reported errors are 2 ϫ S.D. and were calculated by error propagation for compound parameters. this affinity with K D, overall calculated from the forward and reverse constants obtained by the binding kinetics (Fig. 10). Analysis of the titrations indicated that SK bound with a K D of 200 Ϯ 20 nM and ⌬F max /F o of Ϫ30 Ϯ 1%. The affinity was ϳ5-fold weaker than in the absence of kringle ligands but still ϳ3-fold tighter than in 6-AHA. SK⌬K414 bound labeled [Lys]Pg with a ⌬F max /F o of Ϫ38 Ϯ 1% and K D of 800 Ϯ 100 nM. This affinity was similar to that of SK in 6-AHA, SK⌬K414 with and without 6-AHA (48), and SK binding to [5F]FFR-[Glu]Pg with and without 6-AHA (27,48).
Numerical Integration Analysis of the Forward and Reverse Reactions-Fitted values for K 1 , the rate constants for both conformational steps, and the fluorescence amplitudes were in good agreement with those obtained from two-exponential analysis and equilibrium binding and are given in Table 1 (superscript a, fluorescence intensity, and superscript aa, fluorescence anisotropy). The results indicated that formation of fluorescently silent SK⅐Pg 1 occurs in the dead time of the reaction and that subsequent partitioning occurs between SK⅐Pg 2 and SK⅐Pg 3.
The SK⅐[5F]FFR-[Lys]Pg encounter complex was weakened ϳ10 -20-fold by blocking LBSs on the Pg kringles and by loss of Lys 414 . The rate constant k 2 for the first conformational step ranged from 25 to 45 s Ϫ1 in the absence and presence of 6-AHA  Table 1. Experiments were performed and analyzed as described under "Experimental Procedures."  Table 1. Experiments were performed and analyzed as described under "Experimental Procedures." but increased substantially in benzamidine, suggesting a decrease in conformational restraint.
The rate constants k Ϫ2 and k Ϫ3 for the reverse reactions were equivalent to k off 1 and k off 2 from hyperbolic fitting of the SK dependences of the forward reaction rates and to k disp 1 and k disp 2 for the biexponential appearance of free [5F]FFR-[Lys]Pg in competitive displacement by FFR-Pm. The analytical solution of the overall k off value for a three-step reaction is only straightforward under conditions of single exponential kinetics (70); however, the agreement of k off 2 and k disp 2 with k Ϫ3 from numerical analysis suggests that dissociation is limited by k Ϫ3 . The off-rates were unaffected by lysine analogs.
In the absence of effectors, K D, overall for SK binding to [5F]FFR-[Lys]Pg ranged from 30 Ϯ 18 to 86 Ϯ 18 nM in agreement with the results from equilibrium binding (27,48). Dele-tion of SK Lys 414 or blocking the LBSs with 6-AHA caused an increase of K D, overall to 0.5-0.9 M, which is identical to that for [Glu]Pg binding. In benzamidine, K D, overall for binding of intact SK to labeled [Lys]Pg was 0.2 M, possibly reflecting the contribution of the large forward rate for the first tightening step.
Within global data sets, the errors in the amplitude factors f 2 for the fast conformational step and f 3 for the slow step were ϳ30 and ϳ18%, respectively (2 ϫ S.D.). Numerical integration fits for the forward and reverse reactions are shown as colored lines in the figures.

DISCUSSION
The present study demonstrates a minimal three-step sequential mechanism for binding of SK to [5F]FFR-Pg, consisting of an encounter complex with affinity in the low micromolar range followed by at least two resolvable conformational  Table 1. Experiments were performed and analyzed as described under "Experimental Procedures."  Table 1. Experiments were performed and analyzed as described under "Experimental Procedures." steps ( Fig. 11), which increase the affinity of the stabilized complex to ϳ30 -86 nM. The first conformational step is the main tightening event, whereas the second step does not confer additional tightening. However, deletion of this second step in the mechanism resulted in single exponential curves that did not fit the data.
We previously discovered that a three-step mechanism also governs SK binding to labeled plasmin but with ϳ4,000-fold tighter K D, overall values of 7-12 pM (59). The conformational changes caused a ϳ9,000-fold tightening of the Pm encounter complex but only a ϳ50-fold increase in affinity for Pg. We show here that substantial decreases in affinity of the encounter complex and the second conformational event are mainly responsible for the weaker SK binding to Pg in the stabilized complex.
The results suggest that the SK interactions with LBSs on [Lys]Pg are mainly limited to SK Lys 414 binding to K4, whereas  Table  1. Fluorescence titrations were performed and analyzed as described under "Experimental Procedures." plasmin binding involves another SK lysine interacting with K5 in addition to Lys 414 binding to K4. It is noteworthy that non-LBS interactions with the protease domain are significant sources of binding energy in both plasminogen and plasmin binding (71). Until now, SK binding to Pg had only been studied by equilibrium binding, and although the published K D values report the affinities of the final complexes, they do not provide information on the intermediates in this multistep mechanism.
The results support the following sequential steps on the pathway to a stabilized complex with labeled Pg: SK Lys 414 binding to a Pg kringle during formation of a weak, fluorescently silent encounter complex and two conformational steps of SK reorganization from a flexible to a more organized form during binding to the Pg protease domain accompanied by expression of a pro-exosite for binding of a second Pg molecule in the substrate mode. This reorganization is reported by biphasic fluorescence changes of the probe in the active site on the protease domain of Pg. Two striking differences between SK binding to labeled Pm and [Lys]Pg were immediately obvious: a ϳ40-fold weaker binding of SK in the encounter complex illustrated by higher SK concentrations required for saturation of the rates of fluorescence change and the requirement of stopped flow to study Pg displacement from the complex by FFR-Pm evidenced by the large increase in the off-rate constants k Ϫ2 and k Ϫ3 . Whereas displacement from the SK⅐Pm complex required several hours of incubation with excess FFR-Pm, the complexes with [Lys]Pg were easily reversed in a matter of seconds.
Binding to Pg also involves insertion of the NH 2 terminus of SK in the activation pocket of the Pg catalytic domain; however, adding a conformational step to the mechanism did not improve the fits. Stopped-flow fluorescence of SK binding to labeled Pg may not allow identifying the timing of the NH 2terminal insertion or whether NH 2 -terminal insertion contributes to the affinity of the Pg complex, and further studies are required to resolve this complex event.
Binding of SK Lys 414 to a kringle facilitates formation of the encounter complexes with both [Lys]Pg and Pm as a similar 6 -8-fold reduction in their affinity was observed when Lys 414 was deleted. K 1 of the encounter complex with labeled [Lys]Pg increased from ϳ3 to ϳ19 M upon deleting SK Lys 414 . Saturation of the LBSs did not decrease the affinity of SK and SK⌬K414 any further, indicating no other SK lysine-LBS interactions, and the 10 -19 M affinity range of LBS-blocked SK and SK⌬K414 complexes likely represents the contribution of non-LBS binding to the Pg catalytic domain (Table 1). K 1 of the encounter complex with Pm increases from ϳ0.08 to ϳ0.67 M upon SK Lys 414 deletion (59); however, the SK⌬K414⅐Pm encounter complex still exhibits substantial affinity, reflecting the sum of the LBS interactions with other lysine residues and non-LBS interactions with the protease domain. Kringle K5 harbors an LBS that preferentially interacts with ligands not carrying a free carboxylate function, such as alkylamines (51,72,73), and K5 on Pm may bind a non-COOH-terminal SK lysine. Saturation of Pm with 6-AHA disengages Lys 414 and other lysines, and as expected, this affinity is not weakened further by SK Lys 414 deletion. The remaining encounter affinity of ϳ5-8 M likely represents the non-LBS interactions with the Pm catalytic domain (59). Multiple LBS interactions in the tighter encounter complex with Pm may be made possible by an increased flexibility of two-chain Pm compared with single chain Pg. This flexibility might also allow more intimate contacts during stabilization of the SK⅐Pm complex. 6 Benzamidine blocks kringles K1, K2, and K5 and leaves kringle K4 available for lysine binding. The affinity of the SK⅐Pg encounter complex in 6-AHA and benzamidine was similar, suggesting that Lys 414 binding to K4 in Pg does not increase the affinity when K5 is blocked. However, the SK⅐Pm encounter complex was ϳ2-fold tighter in benzamidine than that in 6-AHA and was further weakened by deletion of Lys 414 (59). Further studies are required to clarify these different effects on Pg and Pm binding.
The 42-residue COOH-terminal sequence is not resolved in the SK⅐Pm crystal structure (19). Lys 414 at the end of this disordered, mobile sequence may guide the pathway by initial interaction with the LBS on K4; however, this does not contribute much to the free energy of binding of the encounter complex. Calculating changes in free energy of association for SK FIGURE 11. The three-step mechanism of SK⅐[5F]FFR-[Lys]Pg catalytic complex formation.
[5F]FFR-[Lys]Pg is shown as blue circles in a hypothetical partially extended ␤-conformation. The five kringle domains are small circles with the LBSs of K1, K4, and K5 as tiny black dimples. The zymogen catalytic domain is the larger blue circle with the activated catalytic site in white locked into its conformation by the fluorescein probe (ocher triangle) covalently attached to the peptide chloromethyl ketone that has alkylated His 57 (black stem). SK is shown by three green ovals representing the three ␤-grasp domains marked ␣, ␤, and ␥. The NH 2 terminus of SK is indicated by I, 1 and the COOH-terminal Lys 414 is at the end of a long disordered segment (squiggle) of the ␥-domain. During formation of the initial SK⅐Pg encounter complex (governed by K 1 ), Lys 414 engages the LBS of K4, whereas the domains of SK are thought to not be fully engaged, and this does not produce a change in fluorescence. The first, tightening, conformational change governed by k 2 and k Ϫ2 with the largest decrease in fluorescence (red triangle) is shown hypothetically to involve insertion of SK Ile 1 into the NH 2 -terminal binding cleft forming the Asp 194 salt bridge and settling of the SK domains into a more ordered arrangement. The last conformational step controlled by k 3 and k Ϫ3 completes the arrangement of SK domains accompanied by a smaller fluorescence decrease (maroon triangle). and SK⌬K414 binding to Pg and Pm from ⌬G 0 ϭ RT ln(K D ) using averaged K 1 values from Table 1 and our previous study (59) shows that the non-LBS interactions contribute ϳ83 and ϳ73% of the binding energy in Pg and Pm encounter complex formation, respectively. SK Lys 414 contributes ϳ17 and ϳ13%. Binding of (an)other SK lysine residue(s) to Pm contributes ϳ14%. Although LBS interactions are important for efficient docking of SK, it appears that non-LBS interactions are the major source of the binding energy both for Pg and for Pm encounter. Previous equilibrium binding studies with ␣-domain-truncated SK showed that the LBS-independent interactions with the Pg/Pm protease domain largely reside in the SK ␣-domain, whereas the ␤and ␥-domains participate in LBSdependent interactions with Pg/Pm kringles (71). It is likely that Pm-binding lysines other than the C-terminal Lys 414 reside in the SK ␤and ␥-domains.
SK⅐Pg and SK⅐Pm differ in their rate constants for the two conformational steps. The k 2 values for the SK⅐Pm complex were ϳ10 s Ϫ1 for intact and Lys 414 -deleted SK and ϳ37 s Ϫ1 in 6-AHA (59); the latter is comparable with k 2 for all the SK and SK⌬K414 interactions with Pg in this study except those in benzamidine. For SK⅐Pm, the conformational restraint reflected by a low k 2 may be due to binding of a non-COOHterminal SK lysine to K5, which also makes the SK⅐Pm encounter complex tighter. This restraint is absent in [Lys]Pg, suggesting no binding contribution from SK lysines other than Lys 414 . Benzamidine enhanced k 2 by 5-and 6-fold in SK binding to Pg and Pm, respectively (59). This enhancement was weaker with SK⌬K414, suggesting that Lys 414 binding to K4 may release conformational restraints. The large k 2 value may contribute to a ϳ4-fold tighter K D, overall for SK⅐Pg in benzamidine compared with that for SK⅐Pg in 6-AHA and SK⌬K414⅐Pg in all buffer systems. 6-AHA causes transition of [Glu]Pg from the compact ␣-form to the fully extended ␥-form and of [Lys]Pg from the partially extended ␤to the ␥-form, whereas benzamidine keeps [Lys]Pg in the ␤-form (43). This suggests that the release of constraints on k 2 does not depend on ␣ 3 ␤ 3 ␥ conformational changes in Pg. The k Ϫ2 , k 3 , and k Ϫ3 values were similar in all data sets, indicating that these steps are LBS-independent.
In summary, we demonstrate here for the first time that the three-step kinetic model for the pathway of Pg binding to SK is substantially different from that of Pm binding with the main differences being a weaker encounter complex and increased off-rates for the conformational steps. Whereas cooperative Lys 414 and other lysine interactions with K4 and K5 and non-LBS interactions with the protease domain contribute to formation of the SK⅐Pm complex, the SK⅐Pg* complex assembly appears to be driven by non-LBS interactions and by SK Lys 414 binding to Pg kringle K4. Consistent with the experimentally determined K m of Ն2 M for substrate Pg binding to the SK⅐Pg* complex and of 270 nM for Pg binding to the SK⅐Pm complex (7) and in the absence of a crystal structure of the SK⅐Pg* complex, we hypothesize that the weaker interaction with Pg in the catalytic complex results in expression of a pro-exosite that binds substrate Pg with lower affinity than the corresponding exosite on the SK⅐Pm complex.
Differences in affinity of the SK⅐Pg* and SK⅐Pm complexes may be important in their partitioning on bacterial surface pro-teins and in binding to host fibrin(ogen). Group A streptococcal M-like surface proteins bind Pg with high affinity, and the M1 subset lacking Pm-binding motifs binds fibrinogen (Fbg). This allows indirect activation by way of SK⅐Pg*⅐Fbg ternary complex formation (7,77), which proceeds by Fbg binding to the SK⅐Pg* complex rather than SK recruitment on the Pg⅐Fbg complex (7). Group A Streptococcus SK exhibits significant polymorphism (78 -80), and considerable differences exist among SK allelic variants in their efficiency of activating Pg and their recruitment in complexes with Fbg, Fbg fragment D, fibrin, and the plasminogen-binding group A streptococcal M protein (81).
[Glu]Pg binds fibrin(ogen) through a K1 interaction, whereas K1 and K4 of [Lys]Pg and Pm are involved in fibrin(ogen) binding (74), and these differences are expected to influence localization of SK⅐Pg* and SK⅐Pm complexes. Future stopped-flow studies will identify how fibrin(ogen) and streptococcal surface proteins affect the pathways of SK⅐Pg* and SK⅐Pm formation and will be instrumental in characterizing these pathways in complexes with allelic SK variants.