Antichymotrypsin Interaction with Chymotrypsin

Serpins form enzymatically inactive covalent complexes (designated E*I*) with their target proteinases, corresponding most likely to the acyl enzyme that resembles the normal intermediate in substrate turnover. Formation of E*I* involves large changes in the conformation of the reactive center loop (residues P17 to P9′) and of the serpin molecule in general. The “hinge” region of the reactive center loop, including residues P10–P14, shows facile movement in and out of β-sheet A, and this movement appears to be crucial in determining whether E*I* is formed (the inhibitor pathway) or whether I is rapidly hydrolyzed to I* (the substrate pathway). Here, we report stopped-flow and rapid quench studies investigating the pH dependence of the conversion of the α1-antichymotrypsin·α-chymotrypsin encounter complex, E·I, to E*I*. These studies utilize fluorescent derivatives of cysteine variants of α1-antichymotrypsin at the P11 and P13 residues. Our results demonstrate three identifiable intermediates, EIa, EIb, and EIc, between E·I and E*I* and permit informed speculation regarding the nature of these intermediates. Partitioning between inhibitor and substrate pathways occurs late in the process of E*I* formation, most likely from a species occurring between EIc and E*I*.

Here, in order to explore further the mechanism of E*I* formation, we extend our studies by investigating the interaction with Chtr of two derivatives of rACT in which a fluorescent group, 7-[4-(aminosulfonyl)-2,1,3,benzoxadiazole)] (ABD) has been placed at residues P11 and P13 within the RCL. These residues are of special interest because they fall within the so-called "hinge" region of the RCL (residues P10 -P14) that plays a key role in the partitioning between the substrate and inhibitor pathways. Thus, as recently reviewed (7,29), mutations at residues P14 and P12, particularly those in which Arg or Pro replaces a smaller residue, such as Ala or Ser, convert serpin inhibitors into substrates, presumably because they affect the rate at which the RCL inserts into ␤-sheet A during E*I* formation. In addition, annealing the peptide corresponding to residues P8 -P14 to intact ␣1-protease inhibitor converts this serpin from an inhibitor of trypsin to a trypsin substrate by preventing RCL insertion (30).

EXPERIMENTAL PROCEDURES
Materials-Bovine Chtr was obtained from Calbiochem. Enzyme concentration was estimated using A 280 nm equal to 2.1 ml/mg/cm, giving values consistent with those determined by the hydrolysis rate against a standard substrate (1). The chromophoric proteinase substrate SucAAPF-p-nitroanilide, bovine serum albumin, dithiothreitol, and phenylmethylsulfonyl fluoride were obtained from Sigma. Standard proteins for SDS-PAGE calibration were from Bio-Rad. HPLCgrade acetonitrile was from Fisher, and HPLC-grade trifluoroacetic acid was from Pierce (sequanal grade). DNA primers for PCR reactions were synthesized at the Nucleic Acid Facility, University of Pennsylvania Cancer Center.
Construction, Expression, and Purification of T345C-rACT, E346C-rACT, and S348C-rACT-The three variants were constructed using sequence overlap expression polymerase chain reaction (31,32) and the ACT expression vector described previously (1,33). Polymerase chain reaction products, representing the entire coding region, were cut with BstXI, gel-purified, and inserted in the correct reading orientation in pZMS. Mutation sites were verified by DNA sequencing. Variant proteins were purified to homogeneity as described earlier (1,34) except that in the second chromatography step, SE-Sepharose replaced doublestranded DNA cellulose. This chromatography was carried out at pH 6.5 (10 mM MES) using a NaCl gradient. Variant proteins eluted at 0.4 M NaCl. Their concentrations were estimated by A 280 nm , using a value of 0.86 mg/ml/cm (35), giving values consistent with those determined by Bradford analysis (30).
Preparation of ABD Derivatives-A typical labeling reaction mixture contained 20 M of variant protein, 200 M dithiothreitol, and 1 mM ABD-F (Wako Bioproducts) (4-(aminosulfonyl)-7-fluoro-2,1,3, benzoxadiazole, added from a stock solution of 20 mM ABD-F in DMF) in 50 mM Tris-HCl, 50 mM KCl, pH 8.3. The reaction was allowed to proceed in the dark at 4°C for 16 h and stopped by adding 20 mM dithiothreitol. Unreacted probe was removed by exhaustive exchange with 20 mM NaPi, pH 7.0, buffer in a Centriprep-30 (Amicon).
Labeling stoichiometry was determined spectrophotometrically by absorption measurements at 280 nm and 384 nm, using the following extinction coefficients: for ACT, ⑀ 280 36,000 (26) and ⑀ 384 1,100 M Ϫ1 cm Ϫ1 (this work); for bound ABD (25), ⑀ 280 3,000 and ⑀ 384 7,800 M Ϫ1 cm Ϫ1 . To correct for light scattering, observed absorbances at 500 nm were subtracted from values at other wavelengths. This correction was always Յ8% of measured values at 384 nm. Control experiments showed negligible reaction with the unique Cys residue in WT-rACT, Cys-237, which is buried in the hydrophobic core of the protein.
Preparation of HNE-cleaved ABD-E346C-rACT-Cleaved ABD-E346C-rACT was formed by incubation of human neutrophil elastase and intact ABD-E346C-rACT (18 g) at a molar ratio of 1:40 in a final volume of 0.7 ml containing 20 mM NaPi, pH 7.5, at room temperature for 6 h. Reaction was stopped by addition of N-methoxysuccinyl-AAPVchloromethylketone (Bachem) to a final concentration of 15 M. Verification that cleavage was complete was provided by SDS-PAGE analysis.
Determination of SI-SI values for ACT variants and derivatives were determined either by inhibition of Chtr activity, as described (26), using the chromophoric substrate Suc-AAPF-p-nitroanilide, or by SDS-PAGE analysis. In the latter approach, reaction mixtures containing equal volumes (10 l) of Chtr (varying concentration) and of 4 M (final concentration) WT-rACT, variant rACT, or an ABD derivative of an rACT variant in 20 mM NaPi (pH 7.0) were incubated at 25°C for 15 min. Reactions were stopped by adding SDS nonreducing loading buffer. The samples were then subjected to SDS-PAGE analysis (see below), which separates intact ACT from both cleaved ACT and ACT complexed with Chtr. Values of SI were determined from the extrapolated concentration of Chtr just sufficient to completely remove the band due to intact ACT (Fig. 1).
Determination of Second-order Rate Constants for Inhibition of Chtr-Rate constants (k i ) for inhibition of Chtr were measured at 25°C under second-order conditions. The reaction mixture contained Chtr (5 nM) and the inhibitor at a concentration of SI ϫ 5 nM in 100 mM Tris-HCl, pH 8.3, 0.005% Triton X-100 (1). The reaction was quenched at various times by adding 20 l of 10 mM Suc-AAPF-p-nitroanilide in 90% Me 2 SO into 1 ml of reaction mixture, and residual Chtr activity was determined. Plots of (enzyme activity) Ϫ1 versus time were linear, with slopes equal to k i .
Fluorescence Measurements-Stopped-flow fluorescence emission spectra were acquired using an Applied Photophysics stopped-flow spectrofluorometer with excitation at 384 nm and detection at 480 nm.
Rapid Quenched-flow-Rapid quenched-flow kinetic studies were carried out using a KinTek Chemical-Quench-Flow model RQF-3 as previously (11). Reactions were quenched with 0.1 N HCl and analyzed by either SDS-PAGE or reverse phase-HPLC.
SDS-PAGE Analysis-Analyses were performed on quenched samples as previously (1), using 12% polyacrylamide gels. Aliquots containing 2-5 g of total protein were precipitated with an equal volume of 20% trichloroacetic acid for 30 min. The wet pellet was taken up in 20 l of nonreducing sample buffer (0.031 M Tris-HCl (pH 6.8), 1% SDS, 5% glycerol, and 0.001% bromphenol blue), except as otherwise indicated. Small amounts of Tris base were added to neutralize any remaining trichloroacetic acid, and phenylmethylsulfonyl fluoride was added to a final concentration of 2 mM to rapidly inactivate any renatured Chtr. Samples were boiled for 4 min prior to electrophoresis. Gels were stained with Coomassie Blue and destained with 7.5% AcOH and 5% MeOH. Gel scanning was performed on a ScanMaker (9600XL) from Microtek, using Adobe Photoshop, over a concentration range for which band intensity is proportional to the amount of protein present. Gel images were processed using the National Institutes of Health Image program (version 1.61). The reproducibility of band intensities at specific time points was within Ϯ5%.
Data Analysis-Stopped-flow spectrophotometric data were fit using Pro-Kineticist global analysis and simulation software, as described by the manufacturer (Applied Photophysics). Typically, five traces were averaged with data smoothing. Quenched-flow data were fit using Hopkinsim (version 1.7) software from Johns Hopkins University. In fitting results to Scheme III, stopped-flow data gave values for k 1 , k 2 , and k 3 (assuming that the fluorescence of EI c is essentially identical to that of E*I*). Values of k 4 were obtained by fits of quenched-flow data, using values for k 1 , k 2 , and k 3 derived from stopped-flow analysis. Attempts to fit data allowing reactions in Schemes II or III to be reversible gave values of reverse rate constants that were highly variable, although always very low compared with the forward rate constants. Hence, the reactions were treated as irreversible. Fits of rate constant data to Scheme IV were performed with Igor Pro 3.13.

RESULTS
SIs and Second-order Rate Constants-All three variants (T345C-rACT (P14), E346C-rACT (P13), and S348C-rACT (P11)), as well as the two fluorescent derivatives (ABD-E346C-rACT and ABD-S348C-rACT), form SDS-stable complexes with Chtr. T345C-rACT has an SI value of 3.8 as measured by titration of Chtr activity (Table I). The other four variants or derivatives have SIs just over 1.0, as measured either by titration of Chtr activity or by SDS-PAGE analysis, measuring the decrease in intensity of the band due to intact inhibitor as a function of the Chtr/inhibitor ratio ( Fig. 1; see also Fig. 4). The second-order rate constants (k i ) for inhibition of Chtr by ABD-E346C-rACT and ABD-S348C-rACT are virtually identical to the rate constant for inhibition by WT-rACT (Table I), paralleling the lack of effect of amino acid substitution at residues P11 (38) and P13 (39) on inhibitory activity of other serpins. These results are in accord with the expectation that mutations of "P-odd" residues in the P10 -P14 hinge region should not affect inhibitory activity, because, on insertion into ␤-sheet A, P-odd side chains point into solution (40).
Fluorescence Emission Spectra-Excitation of a solution of ABD-E346C-rACT at 384 nm gives the emission spectrum shown in Fig. 2, with max equal to 502 nm. Addition of a small excess of ABD-E346C-rACT to Chtr leads to formation of the E*I* complex and a large increase in fluorescence intensity accompanied by a slight blue shift (1-2 nm), consistent with a shift to a more hydrophobic environment on complex formation. Excess ABD-E346C-rACT was employed to avoid proteolysis of the complex by Chtr (14). The estimated increase of fluorescence intensity at 502 nm on full conversion of I to the E*I* complex is 46%. The emission spectrum of ABD-S348C-rACT shows a somewhat larger increase (ϳ2-fold) in intensity and a small blue shift (ϳ3 nm) on complexation with Chtr, measured at pH 7.0 (data not shown). In contrast with complex formation, HNE-cleavage of ABD-E346C-rACT produces a much smaller increase in emission intensity (ϳ13%) and a slight red shift (ϳ3 nm) (Fig. 2).
The fluorescence emission spectra shown in Fig. 2 were obtained at pH 7.5. Qualitatively similar results were obtained at pH values of 5.5 and 6.5.
Rates of Complex Formation with Chtr by Stopped-flow Spectrofluorometry at pH 7, 25°C and 40°C-Mixing of Chtr and ABD-E346C-rACT at pH 7, 25°C leads to a biphasic response, with the fluorescence intensity at 480 nm first decreasing and then increasing (Fig. 3). Rate constants for both phases were calculated by fitting the data to Scheme II (Table II). Shown in Fig. 3 are results obtained for initial concentrations of Chtr and ABD-E346C-rACT of 125 and 25 M, respectively. Essentially the same results were obtained when the Chtr and ABD-E346C-rACT concentrations were varied, maintaining a 5:1 ratio, over a [Chtr] range of 25-125 M (Table II) or at higher E:I ratios (data not shown), demonstrating that both rate constants measure first-order processes, i.e. processes taking place after formation of the encounter complex (E⅐I) between ABD-E346C-rACT and Chtr. A similar biphasic response was seen at 40°C, with correspondingly higher values for k I and k II . Gel scanning was performed as described under "Experimental Procedures." The apparent x-intercept value of 1.11 at pH 7.0, corresponding to an SI of 0.90, was adjusted to give an SI of 1.1 in Table I, based on the minor formation of cleaved I (see under "Results"), also detected by PAGE.
a Determined as described under "Experimental Procedures." See also Fig. 1. b ND, not determined.
As mentioned above, the SI for ABD-E346C-rACT interaction with Chtr is slightly greater than 1. Formation of cleaved ABD-E346C-rACT was ignored in our calculations, a procedure that is justified not only because the amount of such formation was small (ϳ6%), as judged by SDS-PAGE analysis (see below) but also because the fluorescence change at 480 nm accompanying ABD-E346C-rACT cleavage (Fig. 2, sample iii) was Ͻ15% that accompanying complex formation with Chtr.
As contrasted with ABD-E346C-rACT, mixing ABD-S348C-rACT with Chtr leads to a monophasic increase in fluorescence intensity at 480 nm that is well fit to a single exponential (data not shown). In the concentration range studied, using [Chtr] in excess over [ABD-S348C-rACT], the observed rate constant at 25°C, 4.6 s Ϫ1 , is independent of [Chtr] and corresponds to a first order process (Table II). The similarity in this value to k II above (4.2 s Ϫ1 ) leads us to believe that these constants measure the rates of similar processes for the two rACT derivatives. The failure of ABD-S348C-rACT to display a fast first-order process on reaction with Chtr, corresponding to k I above, may either be due to the absence of such a process (i.e, direct conversion of E⅐I to E*I* in Scheme II) or to its being undetectable by this approach (e.g. if the k I /k II ratio was greater than or equal to that for ABD-E346C-rACT but little change in fluorescence intensity accompanied E⅐I to EI ␣ conversion). We believe the latter possibility more likely (see under "Discussion"). Further kinetic studies, described below, were conducted with ABD-346C-rACT, because this derivative offered the greater promise for detecting intermediates on conversion of E⅐I to E*I*.
Rate of ABD-E346C-rACT-Chymotrypsin Complex Formation by Quenched-flow-In an experiment exactly paralleling the stopped-flow experiment described above, rapid quench kinetics was employed to determine the rates of E*I* formation between Chtr and ABD-E346C-rACT, by SDS-PAGE analysis (Fig. 4), and of postcomplex fragment formation, by reverse phase-HPLC analysis. Both analyses gave virtually identical results, consistent with our earlier demonstration for WT-ACT (2) that E*I* and postcomplex fragment are formed at the same rate. Both data sets were fit to Scheme II (Fig. 3), using the Hopkinsim program, yielding rate constants indistinguishable from those obtained by fitting of the stopped-flow data (Table  II). Thus, at pH 7 and 25°C, stopped-flow and quenched stopped-flow kinetics are consistent with formation of E*I* via Scheme II, in which fluorescence intensities vary in the order E*I*Ͼ E⅐IϾ EI ␣ and neither the encounter complex nor the intermediate complex is SDS-stable, thus accounting for the lag in SDS-stable complex formation.
It was also possible to quantify the minor amount (ϳ6%) of cleaved ABD-E346C-rACT formed as a function of time by SDS-PAGE analysis (Fig. 3), although with less precision than for E*I* formation. Nevertheless, it is clear that E*I* and cleaved I are formed at approximately the same rate.
Rates of ABD-E346C-rACT-Chtr and WT-rACT-Chtr Complex Formation at 10°C-More extensive studies of rates of ABD-E346C-ACT interaction with Chtr as a function of pH in the range 5-8 were conducted at 10°C to take advantage of slower rate constants that might permit observation of more than one intermediate. Typical stopped-flow spectrophotometric and quenched-flow data are displayed in Fig. 5. Taken together, they provide evidence for a pathway for E*I* formation (Scheme III) involving three intermediates between the encounter complex E⅐I and the final E*I* complex.
Considering first the results obtained at pH 6.5 (Fig. 5B), the observed fluorescence changes provide clear evidence of a three-phase reaction, in which fluorescence first increases rapidly on EI a formation, then slightly decreases due to EI b formation, and finally increases again on EI c formation. The clear lag in E*I* formation as compared with the third phase of fluorescence change necessitates invocation of a fourth phase of reaction-i.e. there are three intermediates (EI a , EI b , and EI c ) between the encounter complex and E*I*-and implies that the fluorescence of E*I* is similar to that of EI c . The same pattern is seen at pH 7.0 (Fig. 5C), although here the lag between the final phase of fluorescence change and E*I* formation is much less pronounced. As the pH is raised toward 8.0 (Fig. 5D), the lag between the final phase of fluorescence change and E*I* formation disappears, reflecting an increase in k 4 relative to k 3 . Finally, as the pH is lowered toward 5.0 (Fig. 5A), the phase corresponding to decreased fluorescence intensity disappears, presumably because the fluorescence of EI b becomes greater than or equal to that of EI a . On the other hand, the lag between the final phase of fluorescence change and E*I* formation becomes even more pronounced.
Also shown in Fig. 5A are quenched-flow results for E*I* formation from WT-rACT and Chtr at pH 5. As these data make clear, not only is the lag phase seen with ABD-346C-rACT also seen with WT-rACT, but the rates of E*I* formation in both cases are virtually identical.
At each pH, values for rate constants k 1 -k 3 for ABD-346C-rACT reaction were obtained by fitting the stopped-flow fluorescence data to the first three steps of Scheme III. Each of these constants is first-order, describing processes following encounter complex formation, as shown by their insensitivity to changes in Chtr and/or ABD-E346C-ACT concentration over the ranges employed (Table III). As above, the formation of cleaved ABD-E346C-rACT was ignored. Values for k 4 were determined by fitting quenched-flow data to Scheme III, using values for k 1 -k 3 determined by fitting the stopped-flow fluorescence data.
The observed pH dependences of rate constants k 1 -k 4 are displayed in Fig. 6  implies that conversion of the encounter complex to the first intermediate complex proceeds most rapidly via a neutral form, HE⅐I, and that either protonation of this complex to form H 2 E⅐I or deprotonation to form E⅐I leads to a large decrease in rate. Due to the narrowness of the pH optimum (i.e. the closeness of the two relevant pK a values, 7.0 and 7.1), the maximum fraction of the encounter complex in the most reactive form does not exceed one-third, accounting for the observed maximal rate constant at pH 7 being well below the calculated rate constant for HE⅐I.
The rate constant for conversion of the first intermediate to the second intermediate depends to only a minor extent on pH. Interpreted according to Scheme IV, HEI a conversion to HEI b proceeds about 1.5 times as fast as H 2 EI a conversion to H 2 EI b , but deprotonation to EI a has no effect on rate. Similarly, both EI b and HEI b are converted to the third intermediate with the same rate constant, but both of these reactions proceed much more rapidly (10-fold) than the corresponding conversion of H 2 EI b . Finally, conversion of the third intermediate to E*I* proceeds most rapidly at high pH, with EI c being Ͼ5-fold more reactive than HEI c , which is in turn Ͼ3-fold more reactive than H 2 EI c .
Fluorescence Intensities of Intermediates-The fitting procedure for the stopped-flow fluorescence data allows calculation    (14). Numbers at bottom are times of quenching (seconds). St, molecular mass standards. Lanes E and I are Chtr and ABD-E346C-rACT, respectively. of relative fluorescence intensity differences at 480 nm for each intermediate relative to E⅐I as a function of pH (Fig. 7). A large positive difference is seen for EI c over the whole pH range, with the largest value observed at pH 6.5. EI a shows a positive difference at pH 5, which falls to virtually nothing at pH 8. A more dramatic change, however, is seen for EI b , in which a positive difference is replaced by a negative difference as the pH is raised from 5 to 8. The relative fluorescence intensities of both EI a and EI b are modulated by a group (or groups) of apparent pK a 6.5-7.0 (see under "Discussion"). DISCUSSION Earlier, we showed that at pH 7.0 and 40°C, the spectral change resulting from MCM-A352C-rACT interaction with Chtr proceeded more rapidly than E*I* formation, allowing the inference that an intermediate formed between the encounter complex E⅐I and the acyl enzyme E*I* (26). The major result of the current study is the direct demonstration that there are at least three identifiable intermediates, EI a , EI b , and EI c , between the encounter complex formed between ABD-E346C-rACT and Chtr and E*I*. It might be argued that ABD-E346C-rACT interaction with Chtr reflects a peculiarity due to the presence of a bulky group in the hinge region and is not a good model for WT-rACT interaction. Strong evidence that this is not so comes from the similarities of the two molecules as Chtr inhibitors, as judged by their k i and SI values (Table I) and their values for k 3 and k 4 at pH 5 (Table III and Fig. 5A). Furthermore, even though the complexity of Scheme III introduces ambiguity into comparisons of overall first-order rate constants for E*I* formation, it is worth noting the similar magnitudes of the single rate constants observed for WT at pH 7.5 (7 Ϯ 1 and 17 Ϯ 2 s Ϫ1 at 25 and 40°C, respectively) and k II for ABD-E346C-rACT at pH 7.0 (4.6 Ϯ 0.5 and 22 Ϯ 2 s Ϫ1 at 25 and 40°C, respectively) (Table II). In addition, at pH 7 and 40°C, values of k I and k II for ABD-E346C-rACT (53 Ϯ 10 and 22 Ϯ 2 s Ϫ1 ) are similar to those estimated for MCM-A352C-rACT (30 -45 and 20 -25 s Ϫ1 ) (Table II), in which the bulky fluorescent group is at P7, far from the hinge region.
Below, using the results of this study, especially Figs. 6 and 7, we propose a model relating the structures of intermediates EI a , EI b , and EI c to conformational changes within the RCL that are known to be crucial for the inhibitory activity of a serpin and consider the question of from which of these intermediates, if any, does partitioning occur between the inhibitor and substrate pathways (Scheme I)?
A Proposed Model for Conformational Changes in the RCL on Conversion of E . I to E*I* via EI a , EI b , and EI c -In formation of E*I* from E and I we know, or can safely assume on the basis of prior work (7,8), that the following must occur: (a) the RCL assumes the canonical conformation in fitting into the active site of Chtr; (b) there is a major insertion of the RCL into ␤-sheet A (s4A), although how much insertion occurs is unclear; (c) the structure of Chtr within the complex changes considerably vis-à -vis the native structure, with concomitant inactivation of its catalytic apparatus (14 -16); and (d)  in Chtr is acylated with ACT at residue P1 (11,12), with formation of the postcomplex fragment, a reaction that proceeds via a tetrahedral intermediate.
We propose that in Scheme III, E⅐I conversion to EI a (step 1) principally involves rearrangement of the RCL to the required canonical conformation, which is followed by a conformational change within Chtr to give EI b (step 2). In both EI a and EI b , the P13 residue moves in and out of an inserted position within sheet A, in a process modulated by the ionization state of His-57. Formation of EI c (step 3) involves a closer fitting of the RCL to the catalytic machinery of Chtr (EI c ), as well as insertion of a major portion of the RCL within ␤-sheet A. Finally, conversion of EI c to E*I* (step 4) involves covalent reaction between Chtr residue Ser-195 and the P1 residue of ACT to give the tetrahedral intermediate and ultimately acyl enzyme. Evidence for this proposal is discussed further below.
Relevance of Chtr Hydrolysis of a Simple Substrate to Steps 2-4 in Scheme III-Earlier, Fink (41) identified three intermediates intervening between the encounter complex E⅐S and the acylated enzyme on reaction of N-acetyl-L-Phe-p-nitroanilide FIG. 7. pH effects on normalized relative changes in fluorescence intensities. EI a , OE; EI b , ࡗ; EI c , f. Values shown were obtained by subtracting the fluorescence intensity of the encounter complex, E⅐I, from that of the species indicated, and setting the value for the largest change (EI c at pH 6.5) equal to 1.00. e For reaction of WT-rACT with Chtr. In fitting the quenched flow data to Scheme III, k 1 and k 2 were assumed to have the same values as for ABD-346C-rACT, and k 3 and k 4 were fit. These latter values were quite insensitive to 3-fold increases or decreases in the assumed values of k 1 and k 2 .
FIG. 6. pH effects on first order rate constants. A, k 1 ; B, k 2 ; C, k 3 ; D, k 4 . All rate constants are from Table III and are in s Ϫ1 . Solid lines represent best fits to Scheme IV. and Chtr at pH 7.6 in 65% aqueous dimethyl sulfoxide (Scheme V).
At Ϫ56°C, k 2* Ͼ k 3* Ͼ k 4* Ͼ Ͼ k 5* , but on extrapolation to 25°C, this order changed to k 2* ϳ k 3* ϳ k 4* Ͼ Ͼ k 5* . Moreover, over the pH range 4.2-7.8, k 2* and k 3* were insensitive to pH, whereas both k 4* and k 5* increased with pH, with kinetically determined pK a values of 5.9 and 7.6, respectively (both measured at 0°C). When Chtr was replaced with its N-Me His-57 derivative, which has very low enzymatic activity, k 2* and k 3* were unaffected, but reaction 4* could not be detected. Fink (41) concluded that reactions 2* and 3* represented conformational changes within Chtr following substrate binding, but that both reactions 4* and 5* involved the Chtr catalytic apparatus. He assigned reaction 4* to the formation of a pre-tetrahedral intermediate (In) that requires His-57 to be in the basic form, and reaction 5* to the formation of acyl enzyme via a tetrahedral intermediate that does not accumulate, arguing that In could not itself correspond to a tetrahedral intermediate, because its rate of breakdown increases as pH is increased. (In contrast, tetrahedral intermediate does accumulate at high pH (9.4) during acyl-elastase formation from elastase and pnitroanilide substrates (42).) There is a striking degree of similarity in the properties of several of the steps seen by Fink (41) and ourselves. The conversion of EI a to EI b (step 2) in Scheme III parallels steps 2* and 3* in Scheme V in taking place comparatively rapidly and in showing little pH dependence. Similarly, k 3 and k 4 , the rate constants for EI b conversion to EI c and EI c conversion to E*I*, respectively, depend on basic groups of apparent pK a values of 6.1 and Ն7.8 (Fig. 6), closely paralleling what Fink (41) reports for k 4* and k 5* in Scheme V. These similarities lead us to speculate that reactions 2, 3, and 4 in Scheme III parallel reactions 2* and 3*, 4*, and 5* in Scheme V, respectively, such that the intermediates that accumulate between E⅐I and E*I* do not involve covalent bond formation between ACT and Chtr. Such a model agrees with conclusions reached by Stone and Le Bonniec (43) for the reaction of the heparin complex of antithrombin with thrombin, but run counter to the model of Olson et al. (44) for the reaction of plasminogen activator inhibitor-1 with trypsin. This apparent disagreement may simply reflect real differences in the relative free energies of intermediates formed in the reactions of different proteinases, as noted above in connection with tetrahedral intermediate accumulation in Chtr versus elastase.
Step 1 in Scheme III-In contrast to steps 2-4, step 1, the formation of EI a , which is both rapid and has strong pH dependence, has no clear analogue in Scheme V, allowing the inference that step 1 represents a transformation unique to the interaction of serpins with Chtr, the conversion of the RCL to the canonical conformation. The necessity of such a conformational change is suggested by the structure of an intact variant of rACT retaining inhibitory activity, in which the RCL is present as a distorted ␣-helix (18). It would be reasonable for the rate of formation of the canonical conformation to depend on the charge relay system being in the catalytically active form (neutral His-57), accounting for the ascending limb of the pH rate profile for k 1 with apparent pK a 7.0 (Fig. 6A). Accounting for the descending limb (pK a 7.1) is less obvious. A requirement for the protonated form of one of the eight remaining His residues in the E⅐I complex (His-40 in Chtr and seven His residues in rACT) is a possibility. Based on proximity within a model of the docked rACT⅐Chtr complex (45) to either His-57 or Ser-195 in Chtr, or to P13, the most likely candidates are His-40 in Chtr and His-204, -224, or -225 in ACT.
The Magnitudes and Directions of Fluorescence Intensity Changes (Fig. 7) Support the Proposed Conformational Changes-According to our model, motions within the active site would be coupled to the hinge region of ACT on formation of EI a , so that the changes in fluorescence intensity on conversion of E⅐I to E*I*, although directly reflecting changes in the local environment of the ABD group attached to Cys-346, could also detect changes occurring within the active site of Chtr. We interpret increases and decreases in fluorescence intensity in Fig. 7 as reflecting either further insertion of the ABD group into ␤-sheet A or movement toward a more solvent-exposed environment, respectively, based on the increase in fluorescence we observe on E*I* formation (Fig. 2), which parallels increases seen for fluorescent probes placed at residues P7 (26) and P9 (46). According to this interpretation, the fluorescence changes seen in EI a and EI b as a function of pH reflect small movements of the RCL that are modulated by the protonation state of His-57, with the inactive, protonated form corresponding to more insertion within the A-sheet and the neutral, active form corresponding to less. The movement is more pronounced for EI b , indicating a stronger coupling of the hinge region to the active site following the presumed conformational change in Chtr in step 2. Full exposure of the RCL when bound to the catalytically active form of Chtr is consistent with the hypothesis that a fully exposed RCL is ideal for binding residues P3-P3Ј in the canonical conformation required for productive interaction with a serine proteinase active site (47)(48)(49).
Facile movement of the hinge region of rACT in and out of ␤-sheet A is also suggested by the contrast between the Wei et al. (18) structure of intact rACT, in which the hinge region is fully exposed, and the difficulty we find in derivatizing T345C-rACT, an indication that, in solution, the P14 residue is at least partially inserted within ␤-sheet A. In addition, structural and fluorescence spectroscopy studies show that the binding of heparin to antithrombin induces a conformational change that results in the expulsion of an inserted P14 residue into a solvent-exposed position (48,50), and model building studies of several intact serpin-proteinase complexes show that it is possible to insert the RCL only as far as P12 while maintaining residues P3-P3Ј in the canonical conformation (40). This latter result offers a rationale for the failure of ABD-S348C-rACT, the P11 derivative, to show a change in fluorescence on a time scale appropriate for EI b formation on interaction with Chtr at pH 7, because an ABD group at P11 would be expected to remain exposed on conversion of I to EI b .
In contrast to the pH-dependent changes seen with EI a and EI b , the ABD-group in EI c shows a large increase in fluorescence intensity versus E⅐I over the entire pH range investigated. The magnitude of this change, its relative insensitivity to pH, and the fact that no further fluorescence change is detectable on conversion of EI c to E*I* suggest that it reflects the major insertion of the RCL within ␤-sheet A that accompanies overall E*I* formation. Thus, contrary to an earlier suggestion (29), RCL insertion appears not to be directly coupled to the inactivation of the Chtr catalytic apparatus, resulting in E*I* formation, because such inactivation would be expected to block the conversion of EI c to tetrahedral intermediate and acyl enzyme.
From Which Intermediate Does Partitioning Occur?-Earlier (11) we presented evidence that in the interaction of rACT with Chtr, partitioning between the inhibitor and substrate pathways, leading to formation of the inhibited complex E*I* or the cleaved inhibitor I s , respectively, proceeds from a common in-termediate formed in or following a step that is largely ratedetermining. The current result showing that the rate of formation of the minor product, cleaved ABD-E346C-rACT, is essentially the same as the rate of formation of the major product, E*I* (Fig. 3) supports this notion. As step 3 in Scheme III is largely or completely rate-determining for E*I* formation over the whole pH range investigated (Table III), the possible candidates for the partitioning intermediate are EI c and an as yet unobserved intermediate that occurs between EI c and E*I*: e.g. the tetrahedral intermediate, or a putative "active" form of acyl enzyme that either isomerizes to E*I* or hydrolyzes to E and I*. Although we consider the latter possibility most likely, because it preserves the catalytic apparatus necessary for the relatively rapid covalent reactions in E*I* formation (which occur at essentially the same rate as in substrate turnover; see Ref. 11), a clear challenge for future work will be to convincingly distinguish among these possibilities.