The kinetic mechanism of serpin-proteinase complex formation. An intermediate between the michaelis complex and the inhibited complex.

Serine proteinase inhibitors (serpins) form enzymatically inactive, 1:1 complexes (denoted E*I*) with their target proteinases that release free enzyme and cleaved inhibitor only very slowly. The mechanism of E*I* formation is incompletely understood and continues to be a source of controversy. Kinetic evidence exists that formation of E*I* proceeds via a Michaelis complex (E·;I) and so involves at least two steps. In this paper, we determine the rate of E*I* formation from α-chymotrypsin and α1-antichymotrypsin using two approaches: first, by stopped-flow spectrofluorometric monitoring of the fluorescent change resulting from reaction of α-chymotrypsin with a fluorescent derivative of α1-antichymotrypsin (derivatized at position P7 of the reactive center loop); and second, by a rapid mixing/quench approach and SDS-polyacrylamide gel electrophoresis analysis. In some cases, serpins are both substrates and inhibitors of the same enzyme. Our results indicate the presence of an intermediate between E·;I and E*I* and suggest that the partitioning step between inhibitor and substrate pathways precedes P1-P1′ cleavage.

␣ 1 -Antichymotrypsin (ACT) 1 is a human serine proteinase inhibitor (serpin), a superfamily of proteins believed to have evolved from a common ancestral gene over ϳ500 million years (1)(2)(3). The involvement of ACT in Alzheimer's disease (4,5) and in the regulation of the inflammatory response (6) 2 as well as of prostate-specific antigen activity (8) makes it a particularly interesting protein for study. As is typical of serpins, ACT (I) forms an enzymatically inactive, 1:1 complex (denoted E*I*) with its target proteinases (for example, chymotrypsin) that releases free enzyme and cleaved ACT (I*) only very slowly (9,10). The complex is designated E*I* to indicate that conformational change has taken place in both the enzyme (10,11) and inhibitor (12,13) moieties. For some serpin-proteinase interactions, full inhibition of proteinase requires Ͼ1 eq of serpin. The ratio of moles of inhibitor required per mole of proteinase for 100% inhibition is defined as the stoichiometry of inhibition (SI). Values of SI Ͼ 1 reflect a partitioning of serpin between inhibitor and substrate pathways, giving rise to "suicide inhib-itor"-type kinetics (9, 14 -16).
A characteristic feature of the serpin-proteinase complex is that it is stable to both heat and SDS treatment, implying covalent bond formation between enzyme and serpin. The nonlability of E*I* may be due either to distortion of the enzyme active site within the complex (10,11) or to inaccessibility of the covalent E-I linkage toward attacking nucleophilic water, or both. The position of cleavage in released I* occurs within a so-called "reactive center loop," which in intact I extends out from the rest of the molecule, contains a segment of modified ␣-helix (17)(18)(19)(20)(21), and is the primary interaction site between the inhibitor and the target proteinase. Following standard nomenclature (22), the position of cleavage takes place between the P1 and P1Ј sites of the inhibitor, which in ACT corresponds to positions 358 and 359. Residues proceeding toward the N and C termini from P1 and P1Ј, respectively, are labeled with higher P and PЈ numbers. The reactive center loop extends from approximately P17 to P9Ј. Cleavage of intact I to form I* is accompanied by a large decrease in free energy and a substantial gain in stability toward denaturation by either heating or denaturing agents (23,24). In I*, residues P1-P14 are inserted into ␤-sheet A, the dominant structural element in ACT, as strand 4A. ␤-Sheet C is also reinforced, with the result that the P1 and P1Ј residues are separated by 70 Å (25).
Kinetic evidence has existed for some time that formation of the nonlabile serpin-proteinase complex involves at least two steps (26 -28): the second-order association of E and I in an "encounter" or Michaelis complex (E⅐I), followed by its conversion, in a first-order process, to E*I* (Scheme 1).
For at least some serpin-proteinase pairs, there is evidence that complex formation is reversible (29 -31), although the equilibrium constant strongly favors E*I*. Recent results that E*I* formation coincides with liberation of a new N terminus at P1Ј (32,33) demonstrate that within E*I*, I has been cleaved at the P1-P1Ј bond. These results, coupled with the observation that E*I* dissociates on treatment with hydroxylamine (34), provide strong evidence that the covalent bond within E*I* is between the proteinase active site serine hydroxyl (Ser-195 in ␣-chymotrypsin) and the liberated P1 carboxyl of the serpin and corresponds either to acyl-enzyme or to the tetrahedral intermediate (35) resulting from water attack on the acylenzyme. Indeed, these species may be in mobile equilibrium with each other.
Formation of the E*I* complex requires completion of a minimum of three processes: 1) establishment of P1 interaction with the S1-binding site of the enzyme and of perhaps subsite interactions as well, necessitating at least partial unwinding of the helical portion of the reactive center loop; 2) partial or full insertion of strand 4A into ␤-sheet A (18, 36); and 3) P1-P1Ј cleavage and formation of the covalent acyl-enzyme (or tetrahedral intermediate). However, the relative timing of these processes is unknown.
In this paper, we determine the rate of E*I* formation from E⅐I using two approaches. In the first, we derivatize the A352C-rACT variant (at position P7) with the fluorescent reagent 4-bromomethyl-7-methoxycoumarin (BMMC), demonstrate that the resulting (7-methoxycoumaryl-4)-methyl derivative of A352C-rACT (MCM-A352C-rACT) inhibits ␣-chymotrypsin in the normal manner, and measure its rate of complex formation with ␣-chymotrypsin by stopped-flow spectrofluorometry. In the second, we apply a rapid mixing/quench approach and SDS-PAGE analysis to determine the rate of E*I* formation on reaction of either MCM-A352C-rACT or rACT with ␣-chymotrypsin. Our results indicate the presence of an intermediate between E⅐I and E*I* and suggest that the partitioning step between inhibitor and substrate pathways precedes P1-P1Ј cleavage.

EXPERIMENTAL PROCEDURES
Materials-Bovine N ␣ -p-tosyl-L-leucine chloromethyl ketone-treated chymotrypsin and human neutrophil elastase were obtained from Calbiochem or Sigma. The concentrations of these enzymes and of rACT were determined as described earlier (16). All chromophoric proteinase substrates, dithiothreitol, and phenylmethylsulfonyl fluoride were obtained from Sigma. BMMC was acquired from Molecular Probes, Inc. (Eugene, OR). SDS-PAGE analysis was performed according to Laemmli (37). Standard proteins were from Bio-Rad.
Construction, Expression, and Purification of rACTs-A352C-rACT was constructed using sequence overlap expression polymerase chain reaction (38,39) and the ACT expression vector described previously (16,40). The internal primers coding for the Ala 3 Cys mutation are as follows: 5Ј-GCCACATGCGTCAAAATCACCCTCC-3Ј and 5Ј-GATTTT-GACGCATGTGGCAGCAGATGCT-3Ј (the mutation sites are in boldface). The polymerase chain reaction product, representing the entire coding region, was cut with BstXI, gel-purified, and inserted in the correct reading orientation in pZMS. Full gene sequencing confirmed a single codon change. A352C-rACT and rACT were purified to homogeneity as described earlier (9,16,41).
Preparation of Derivatized A352C-rACT-A352C-rACT (2 M in 50 mM Tris and 50 mM KCl, pH 8.3) was reacted with a 100-fold excess of BMMC for 12 h at 0°C in an amber-colored reaction vessel. The protein was then concentrated using an Amicon concentrator; diluted 10-fold in 2 M (NH 4 ) 2 SO 4 and 100 mM sodium phosphate, pH 7.0; and purified by hydrophobic interaction chromatography using a procedure similar to that of Kvassman and Shore (42) for the separation of active from latent plasminogen activator inhibitor-1. Application of the diluted sample onto a Synchropak propyl column (250 ϫ 7.8 mm; SynChrom, Inc.) was followed by washing with 1 void volume of 1.8 M (NH 4 ) 2 SO 4 and 100 mM sodium phosphate, pH 7.0 (Buffer A). Protein was then eluted with a two-step linear gradient of decreasing antichaotropic salt: 100 to 28% Buffer A in 15 min, with a flow rate 2 ml/min; and 28 to 0% Buffer A in 60 min, with a flow rate of 0.7 ml/min. Protein elution was monitored by A 215 . MCM-labeled protein elution was monitored by A 330 as well. This procedure separates modified from unmodified protein. Solutions of pure inhibitor were concentrated and buffer-exchanged using Centriprep-30 and Centricon-30 concentrators (Amicon, Inc.).
Characterization of MCM-A352C-rACT-The stoichiometry of MCM labeling was determined by absorbance measurements. Total bound MCM concentration was determined using ⑀ 330 ϭ 13,000 cm Ϫ1 (43). Total protein concentration was determined using ⑀ 280 ϭ 39,000 cm Ϫ1 , calculated as the sum of the coefficients of the unlabeled protein (36,000 cm Ϫ1 , calculated from an ACT solution standardized by Bradford analysis (44)) and the MCM group (3000 cm Ϫ1 , as determined by the A 280 / A 330 ratio for BMMC). Purified protein had a calculated stoichiometry of 1.1 MCM/protein. Titration reactions of ␣-chymotrypsin activity were performed at pH 7.5 as described (16).
Fluorescence Measurements-Stopped-flow measurements were made and fluorescence emission spectra were acquired using an Applied Photophysics !SX.18MV stopped-flow spectrofluorometer with excitation at 330 nm. For rate measurements, emission was monitored at 400 nm. First-order rate constants were obtained by fitting a single exponential to unsmoothed traces.
Rapid Quenched Flow-Rapid quenched-flow kinetic studies were carried out using a KinTek Chemical-Quench-Flow Model RQF-3 ma-chine with an estimated dead time of 5 ms (45,46). Quenched-flow experiments were carried out by rapid mixing of the enzyme solution in one sample loop with the inhibitor solution in the other and quenching with 0.1 N HCl. SDS-PAGE analysis was performed on aliquots of the quenched reaction mixture using gels containing 12 or 15% polyacrylamide. Prior to analysis, all samples were treated with 1 M Tris base (2 l/20 l of quenched solution), to neutralize the excess HCl present in the quenched samples, and with phenylmethylsulfonyl fluoride at a final concentration of 2 mM, to rapidly inactivate any renatured ␣-chymotrypsin. Samples were boiled for 4 min prior to application to the gels. Gels were stained for ϳ1 h in Coomassie Blue staining solution (0.1% Coomassie Blue R-250 in 40% MeOH and 10% AcOH) and then destained for 12-15 h with 7.5% AcOH and 5% MeOH.
Image Scanning and Densitometry-The protein bands on the destained gels were quantified using densitometric analysis. Scanning was performed at a protein concentration range for which band intensity was proportional to the concentration of protein present. Gel photographic images were stored as GRAYSCALE pictures in the ⅐TiFF format and were processed using the National Institutes of Health IMAGE program (Version 1.60) on a Macintosh computer using National Institutes of Health soft-FPU. Sample data are presented in Fig. 1.

RESULTS
MCM-A352C-rACT-Virtually complete derivatization of A352C-rACT was achieved under the conditions used for preparing MCM-A352C-rACT. By contrast, reaction of rACT with BMMC under exactly the same conditions gave only minor derivatization, reflecting the low reactivity of the single Cys residue (Cys-237) within wild-type ACT, which is buried within ␤-sheet B of rACT (20), as well as slow nonspecific reaction with carboxylate residues in the protein (43).
MCM-A352C-rACT inhibits ␣-chymotrypsin with a stoichiometry of inhibition of ϳ1 (data not shown). The MCM-A352C-rACT⅐␣-chymotrypsin complex is stable toward SDS denaturation, as shown by SDS-PAGE analysis (see below), in the usual manner of serpin-proteinase complexes and is cleaved by catalytic amounts of human neutrophil elastase, similar to what is found for wild-type rACT.
Rate of MCM-A352C-rACT⅐␣-Chymotrypsin Complex Formation by Stopped-flow Spectrofluorometry-Excitation of a solution of MCM-A352C-rACT at 330 nm gives the emission spectrum shown in Fig. 2, with max ϭ 399 nm. Human neutrophil elastase-induced cleavage leads to an increase in emission intensity and a slight blue shift ( max ϭ 397 nm), consistent with the expected shift to a more hydrophobic environment as the chromophore moves from a solvent-exposed position to one that is at least partly buried within ␤-sheet A. The fluorescence spectrum of the MCM-A352C-rACT⅐␣-chymotrypsin complex is similar to that of cleaved MCM-A352C-rACT.
Mixing of ␣-chymotrypsin and MCM-A352C-rACT led to an increase in fluorescence at 400 nm, which was used to measure the rate constant for complex formation in a stopped-flow spectrofluorometer. All stopped-flow reactions were carried out under pseudo first-order conditions, with ␣-chymotrypsin (denoted E) present in large molar excess over MCM-A352C-rACT (denoted I). Results obtained at pH 7.0 and 40°C are shown in Fig. 3. Rate constants for complex formation as a function of [I], temperature, pH, and ionic strength are presented in Table I Our results are similar to those of Shore et al. (13), who carried out an analogous study measuring the rate of complex formation between a fluorescent derivative of plasminogen activator inhibitor-1 (derivatized at P9) and plasminogen activator. These workers reported a value for k f (the rate constant determined by the change in fluorescence) of 4 s Ϫ1 (at pH 7.4 and 25°C) and found that the large increase in fluorescence intensity and the 13-nm blue shift accompanying complex formation were virtually identical to the changes observed on cleavage of derivatized plasminogen activator inhibitor-1.
An Arrhenius plot of k f at pH 7 (15-40°C) gives a good straight line and an apparent activation energy of 19 kcal/mol, indicating a substantial energy barrier for the process leading to fluorescence change. The rate constant increases 2-fold in the pH range 6 -8, but the addition of 1 M NaCl has little effect.

Rate of MCM-A352C-rACT⅐␣-Chymotrypsin and rACT⅐␣-Chymotrypsin Complex Formation by Quenched Stopped
Flow-Monitoring the buildup of the SDS-stable complex E*I* provides another way of measuring complex formation. E*I* is stable toward 0.1 N HCl (data not shown). As a consequence, rates of E*I* formation could be measured by 0.1 N HCl quenching at various times following rapid mixing of E and I and quantitative scanning of stained SDS-PAGE analyses of the quenched reaction mixtures (Fig. 1).
The results obtained are summarized in Table I. In terms of Scheme 1, our expectation was that E⅐I would not survive the quench and would be measured as free E and intact I. This expectation was confirmed by the result that the observed first-order rate constant for E*I* formation (k g ) saturates as a function of E concentration (Experiments 20 -22). The slight downturn in rate constant at the highest enzyme concentration used (0.2 mM, Experiment 23) might be due to a nonspecific protein effect. All rate constants were determined at [E] o in excess over [I] o and could be measured by either E*I* buildup or I disappearance with similar results. E*I* buildup was chosen for calculation of rate constants because it could be measured with greater precision.
Although the saturated rates of complex formation for MCM-A352C-rACT reaction with ␣-chymotrypsin are not significantly different when measured by stopped-flow spectrofluorometry or by quenched stopped flow at 25°C (Table I, compare Experiments 1 and 2 with Experiment 3), at 40°C, the stopped-flow rate is clearly faster (ϳ2.5-fold) (Fig. 3 and Table  I, Experiments 4 and 6). This difference demonstrates that Scheme 1 is inadequate to explain the kinetics of E*I* formation, as discussed below. We also note that the rate constants for E*I* formation from both rACT and MCM-A352C-rACT are similar in magnitude and that the rate constant for rACT reaction increases 3-6-fold over the pH range 6.0 -9.0 (Table I,  with ␣-chymotrypsin. Measurement of these band intensities allowed us to estimate the fractions of cleaved I formed during the course of E*I* formation, which are 0.20 and 0.17 for MCM-A352C-rACT (at pH 7.0 and 40°C) and rACT (at pH 7.5 and 40°C), respectively.

DISCUSSION
Kinetics of E*I* Formation-At 40°C, the saturated rate of fluorescence change on mixing of MCM-A352C-rACT with ␣-chymotrypsin exceeds the saturated rate of overall E*I* formation (Fig. 3). This result provides clear kinetic evidence for an intermediate between E⅐I and E*I*, which, unlike E*I*, does not survive the quench/SDS-PAGE analysis procedure. Earlier, we (9) and others (14,15,47,48) proposed the existence of an intermediate between E⅐I and E*I* from which partitioning could occur between the inhibitor pathway, leading to E*I* formation, and the substrate pathway, leading to enzyme and cleaved inhibitor (I s ) release, to account for the finding that some serpin-proteinase pairs have SI values Ͼ 1 (Scheme 2). For reasons of parsimony, we equate the intermediate we demonstrate kinetically in this work with EIЈ in Scheme 2.
As noted above, the time course for the increase in fluorescence (Fig. 3) could be well fit with a single exponential, implying that the emission spectrum of EIЈ is similar to that of E*I*. Analysis of the time dependence of the fluorescence spectrum over the wavelength range 350 -450 nm was consistent with this conclusion. Thus, in applying the simulation program HopKINSIM (49) to estimate microscopic rate constants in Scheme 2 from the data in Fig. 3, the rate of fluorescence change was equated with the rate of disappearance of E⅐I. As the rate of complex formation measured by quenched stopped flow reflects E*I* formation, the difference between the two curves in Fig. 3 measures the buildup and decay of EIЈ. Satisfactory simulations were obtained with the following rate constant values: k 2 ϭ 30 -45 s Ϫ1 ; k Ϫ2 Ͻ 10 s Ϫ1 ; and k 3 ϭ 20 -25 s Ϫ1 . Both k 4 and k Ϫ3 were assumed to be negligible. It is clear that k 3 Ͼ Ͼ k 4 for both rACT and MCM-A352C-rACT since, in both cases, SI is little different from 1. k Ϫ3 may be ignored given that the overall equilibrium between E*I* and E⅐I very much favors E*I*.
These simulations predict a lag in the formation of E*I*, which we have been unable to demonstrate unequivocally, for two reasons. First, the predicted lag is most obvious over a time scale (0 -10 ms) that brackets the dead time (ϳ5 ms) of our instrument. Making the lag more obvious by lowering the temperature is not an option because the buildup of EIЈ depends on the higher temperature. Second, background in the gel lowers the precision with which low levels of protein can be quantified.
At 40°C, k 3 (the rate constant for acylation of ␣-chymotrypsin within the complex) is similar in magnitude to k 2 . By contrast, k 3 must be considerably larger than k 2 at 25°C, given the similarity of k f and k q (ϳ7 Ϯ 2 s Ϫ1 ) at this temperature. We infer that k 3 has a small activation energy between 25 and 40°C. This is reminiscent of the temperature dependence of the rate constants for formation of acyl-chymotrypsin from the Michaelis complex of the enzyme with either p-nitrophenyl acetate or p-nitrophenyltrimethyl acetate (50). These rate constants display a dramatic drop in activation energy from 21 kcal/mol (measured from 6 to 20°C) to 1 kcal/mol (measured from 23 to 36°C), reflecting a thermally induced transition (at ϳ25°C) between two forms of enzyme with different catalytic activities.
Acyl-chymotrypsin formation on reaction with model amide substrates depends on His-57 being in the neutral form. The relevant pK a within the enzyme-substrate complex varies between 6 and 7, depending on the substrate (7). The pH dependences we observe for both k q and k f are consistent with the need for a neutral His-57, although other ionizable group(s) within the E⅐I complex may also modulate these processes. The decrease in rates observed at lower pH suggests that clear demonstration of a lag in E*I* formation might be possible for rates measured at 40°C below pH 6.
Structure of EIЈ-In addition to providing evidence for the  kinetic intermediacy of EIЈ between E⅐I and E*I*, our results also permit the interesting inference that the P1-P1Ј linkage remains intact in EIЈ. Only a minor amount of cleaved I is formed during the course of reaction of ␣-chymotrypsin and either rACT or MCM-A352C-rACT (Fig. 1). Furthermore, at 40°C, the rate of cleaved I formation was clearly much slower than the rate of fluorescence change. These observations require that when EIЈ dissociates during the quench/analysis procedure, it forms E and intact I, whereas cleaved I would have been expected if the P1-P1Ј linkage were cleaved in EIЈ.
Here it should be noted that EIЈ rises to a level of ϳ35-40% of total I after 40 ms, so cleaved I arising from EIЈ dissociation would have been easily detectable. Interpreted according to Scheme 2, maintenance of the P1-P1Ј linkage in EIЈ requires that partitioning into substrate and inhibitor pathways occurs prior to acyl-enzyme formation.
The small accumulation of cleaved I may reflect formation of cleaved I in solution (via reaction 4 in Scheme 2), due to partitioning of the E⅐I complex in solution between inhibitor and substrate pathways, or may rather arise from partitioning of E*I* in solution into an SDS-stable complex (the major pathway) and an SDS-unstable complex as a consequence of the quench/analysis procedure. Given the evidence that E*I* may correspond to acyl-enzyme or to the final tetrahedral intermediate (35) resulting from water attack on the acyl-enzyme, an interesting possibility is that the observed cleaved I may arise from the breakdown of the final tetrahedral intermediate during quench/analysis.
A final speculation, based on the pH dependence of k f , is that EIЈ corresponds to a covalent adduct between E and I, as suggested by Olson et al. (48). More specifically, we propose the initial tetrahedral adduct formed by serine 195 attack on the P1 carbonyl. Such an adduct could reasonably be expected to break down to E and intact I on quench/analysis. Experiments underway to directly measure the rate of P1-P1Ј cleavage within the complex should permit direct testing of whether the P1-P1Ј linkage is intact within EIЈ. a Buffers used were as follows: pH 6 and 8, method F, 20 mM phosphate; pH 6.0, 7.5, and 9.0, method Q, 50 mM Tris and 50 mM KCl; and pH 7, methods F and Q, 20 mM phosphate.
b Method F, stopped-flow fluorescence; method Q, quenched stopped flow. c For the rate constants, the error ranges shown are the larger of average deviation or average precision. Numbers of independent observations are in parentheses. d MCM, MCM-A352C-rACT; WT, wild-type rACT.