Evidence that translocation of the proteinase precedes its acylation in the serpin inhibition pathway.

The inhibition of proteinases by serpins involves cleavage of the serpin, acylation, and translocation of the proteinase. To see whether acylation precedes or follows translocation, we have investigated the pH dependence of the interaction of fluorescein isothiocyanate-elastase with rhodamine alpha(1)-proteinase inhibitor (alpha(1)PI) using two independent methods: (i) kinetics of fluorescence energy transfer which yields k(2,f), the rate constant for the fluorescently detected decay of the Michaelis-type complex (Mellet, P., Boudier, C., Mély, Y., and Bieth, J. G. (1998) J. Biol. Chem. 273, 9119-9123); (ii) kinetics of elastase-catalyzed hydrolysis of a substrate in the presence of alpha(1)PI, which yields k(2,e), the rate constant for the conversion of the Michaelis-type complex into irreversibly inhibited elastase. Both rate constants were found to be pH-independent and close to each other, indicating that acylation, a pH-dependent phenomenon, does not govern the decay of the Michaelis-type complex and, therefore, follows translocation. On the other hand, anhydro-elastase reacts with alpha(1)PI to form a Michaelis-type complex that translocates into a second complex with a rate constant close to that measured with active elastase, confirming that acylation is not a prerequisite for translocation. Moreover, the anhydro-elastase-alpha(1)PI complex was found to be thermodynamically reversible, suggesting that translocation of active elastase might also be reversible. We propose that serpins form a Michaelis-type complex EI(M), which reversibly translocates into EI(tr) whose acylation yields the irreversible complex EI(ac). [see text]


REACTION 1
Serpins are a superfamily of proteins that arouse increasing interest. There are now more than 130 serpin entries in the Swiss-Prot data bank originating mainly from eukaryotic organisms but also from some viruses. In higher organisms they are present extracellularly but also in various intracellular compartments. A large number of these proteins are active serine proteinase inhibitors (1), but inhibition of some cysteine proteinases has also been reported (2). The most striking fea-ture of the inhibition of proteinases by serpins is the formation of an irreversible inhibitory complex. The structure of native serpins includes a long flexible and exposed reactive site loop anchored to a highly conserved scaffold made of nine ␣-helices and three ␤-sheets (3). Some non-inhibited proteinases cleave the exposed loop, which inserts into ␤-sheet A as a new ␤-strand (4). Inhibited proteinases form a SDS-stable complex with serpins, which is probably an acylenzyme (5-7) between the carbonyl group of the P1 site of the loop and the catalytic serine of the proteinase. During the time separating the formation of the first binding complex and the formation of the final complex, a conformational change leading to the translocation of the proteinase was observed (8). The location of the proteinase within the stable complex is still a matter of debate (9 -11). Only the x-ray crystallography of the complex, not available at present, would reveal precisely how the irreversible complex is held together and at which site the proteinase finally binds.
Proteolysis may be regulated by another type of inhibitors, the canonical inhibitors. By contrast with serpins, these smaller proteins have a rigid and short reactive site loop that forms a reversible lock-and-key complex with proteinases. The adaptive advantages by which evolution has selected serpins rather than canonical inhibitors in many regulatory processes are presently not easy to understand. Canonical inhibitors may have loose or strict specificities and low or high affinities for proteinases. Many of them have such low K i values that the inhibition may be pseudo-irreversible. Thus, irreversible inhibition is not a unique feature of serpins. Similarly, the modulation of the activity by an effector like heparin extensively studied with the serpin antithrombin III (12) has also been observed with the canonical mucus proteinase inhibitor (13). Moreover, serpins are large inhibitors that are more sensitive than canonical inhibitors to denaturation and to point mutations even at amino acids distant from the reactive loop (reviewed in Ref. 14). However, serpins have been selected by evolution to regulate delicate proteolytic events like the blood coagulation cascade, the complement cascade, or fibrinolysis. It is perhaps the complicated multistep process leading to the irreversible complex that is the basis of the specificity of serpin inhibitors.
There are at least three steps in the inhibition process by serpins. Each of these steps will decide on the choice of the proteinase and on its fate. The first reversible binding step is governed by the equilibrium dissociation constant for the given proteinase-serpin pair. Several examples covering a wide range of equilibrium dissociation constants have been described in the literature, indicating that this step is effectively used to discriminate the target proteinases. It is important to note that from this step on, the proteinase activity is inhibited (15). The * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  second known step is acylation. The rate of acylation will depend on pH conditions, but, more significantly in physiological conditions, it will be specific of the serpin-proteinase pair. Since acylation is an irreversible step, the lifetime of the reversibly bound complex will depend on the acylation rate. Thus, acylation is another way to subtly regulate the inhibition of proteinases by shortening or expanding the time during which the complex is reversible and may be dissociated by surrounding substrate molecules. The conformational change that leads to the translocation of the proteinase may also be a way of regulating the inhibition, unique to serpins. Recently, we have described a new method combining stopped-flow kinetics and detection of fluorescence resonance energy transfer between FITC 1 -labeled elastase and tetramethylrhodamine-labeled ␣ 1 PI (8). This enabled us to detect the sequential appearance of two complexes, a reversible one, EI M and an irreversible one, EI 2,f , each characterized by a distinct interchromophore distance.
The kinetic constants for the formation and disappearance of EI M were as follows: k 1 ϭ 1.5 ϫ 10 6 M Ϫ1 s Ϫ1 , k Ϫ1 ϭ 0.58 s Ϫ1 , and k 2,f ϭ 0.13 s Ϫ1 , the latter being the rate constant for the translocation of elastase. However, several questions remained unanswered about the translocation step. 1) Is the translocation of the proteinase triggered by the cleavage of the serpin reactive site loop? 2) Is translocation rate-limiting or is it preceded by another step that controls it (i.e. is acylation a prerequisite for translocation)? 3) After which step is the enzyme-inhibitor binding irreversible? This paper attempts to answer these questions.

EXPERIMENTAL PROCEDURES
Elastase was isolated according to the procedure of Shotton (16) with an additional purification step on S-Sepharose (Amersham Pharmacia Biotech) and active site titrated as described previously (17). Recombinant ␣ 1 PI and eglin c expressed in Escherichia coli were obtained from Novartis (Basel) and were titrated with human neutrophil elastase (17,18). Buffer solutions at pH 5.5, 6, 6.5, and 7.1 were made with 50 mM MES and 0.15 M NaCl. Buffer solutions at pH 7, 7.5, 8, and 8.5 were made with 50 mM HEPES and 0.15 M NaCl.
Preparation of Anhydro-elastase-Anhydro-elastase was prepared by alcali treatment of phenylmethanesulfonyl fluoride-treated elastase (20 mg) as described by Ako et al. (19). It was isolated by affinity chromatography as described by Williams et al. (20), except that the column was made with eglin c instead of turkey ovomucoid inhibitor. The affinity gel was prepared by coupling 100 mg of eglin c with 0.5 g of epoxy-activated Sepharose (Amersham Pharmacia Biotech) according to the manufacturer's instructions. The fractions of elastase that retained affinity for eglin c but had negligible catalytic activity were pooled. The pooled sample retained less than 0.09% of the activity expected from a sample of elastase with the same absorbance at 280 nm.
Fluorescent Labeling-Labeling of elastase with FITC (Molecular Probes), of ␣ 1 PI with tetramethylrhodamine-5-maleimide (Molecular Probes), and of eglin c with the succinimidyl derivative of tetramethylrhodamine (Molecular Probes) were done as described previously (8). Anhydro-elastase was labeled with FITC in the same conditions as elastase. The tentative concentration of the stock solution of FITCanhydro-elastase was measured by comparing its absorbance at 280 and 495 nm with that of a titrated solution of active FITC-elastase. The absorbances at 280 and 495 nm gave concentrations of 3.1 ϫ 10 Ϫ5 and 2.6 ϫ 10 Ϫ5 M, respectively. Since a more precise value was not needed, we arbitrarily used a concentration value of 3 ϫ 10 Ϫ5 M for all experiments.
Preparation of a Binary ␣ 1 PI-Peptide Complex-5.9 mg of the synthetic P 1 -P 14 peptide Ac-Thr-Glu-Ala-Ala-Gly-Ala-Met-Phe-Leu-Glu-Ala-Ile-Val-Met (Neosystems, France) was added to 1.5 ml of a 105 mM solution of TMR-␣ 1 PI in 50 mM HEPES, 150 mM NaCl, pH 7.5. The mixture was incubated for 60 h at 37°C in the presence of sodium azide and dialyzed against the above buffer. Peptide insertion was considered complete since no residual inhibitory activity on elastase could be detected. Absorbance at 555 nm was used to determine the peptideinserted TMR-␣ 1 PI concentration.
Kinetics of Hydrolysis of Suc-Ala 3 -pNA by FITC-elastase as a Function of pH-The initial rates of Suc-Ala 3 -pNA (Bachem) hydrolysis by FITC-elastase were measured spectrophotometrically at 410 nm in the buffered solutions described above. The concentration of Suc-Ala 3 -pNA was varied between 0.216 and 2.16 mM in the presence of a final concentration of 1% N-methylpyrrolidone. The FITC-elastase concentration was 3 ϫ 10 Ϫ8 M throughout. At each pH K m and V m were obtained through non-linear least-square fits of the data to the Michaelis-Menten equation.
Kinetics of Fluorescence Resonance Energy Transfer-Kinetics of interaction of FITC-elastase or FITC-anhydro-elastase with TMR-␣ 1 PI or TMR-eglin c were monitored by fluorescence resonance energy transfer from fluorescein to tetramethylrhodamine using a Bio-Logic SF3 stopped flow apparatus with a dead time of 1.7 ms (Bio-Logic, Claix, France). The concentrations were 8 ϫ 10 Ϫ7 M for FITC-elastase and FITC-anhydro-elastase and 8 ϫ 10 Ϫ6 M for TMR-␣ 1 PI and TMR-eglin c.
pH Dependence of the Rate of Inhibition of Elastase by ␣ 1 PI in the Presence of Suc-Ala 3 -pNA-The rate of inhibition was measured by adding elastase to a mixture of ␣ 1 PI and substrate and recording the release of product as a function of time. Mixing and recording were done with the above stopped-flow apparatus. For comparison purposes FITClabeled elastase was used in these experiments, although its fluorescence properties were not needed. Its concentration was 1/10 or less the concentration of the inhibitor to maintain pseudo-first order conditions. The substrate was 2 mM Suc-Ala 3 -pNA in the presence of a final concentration of 1% N-methylpyrrolidone. At each pH the rate of inhibition was measured with initial concentrations of ␣ 1 PI varying between 1 ϫ 10 Ϫ6 M and 5 ϫ 10 Ϫ5 M.
Differential Scanning Microcalorimetry-Spectra were performed on a DASM-4 microcalorimeter. For analysis and integration, the software ORIGIN (Microcal Inc.) was used. The thermograms were recorded in 20 mM phosphate buffer at pH 7.4 with a protein concentration of 1.5 mg/ml. The heating rate was 1°C/min between 20 and 120°C.

Kinetics of Hydrolysis of Suc-Ala 3 -pNA by FITC-elastase as a
Function of pH-In proteolysis by serine proteinases, the acylation and deacylation steps are controlled by the ionization state of the histidyl residue of the catalytic triad. Thus, both steps are pH-dependent. The extent of the variation of the catalytic constant by pH is specific to the proteinase and to the buffer system used. Hence, as a control experiment K m and k cat for Suc-Ala 3 -pNA hydrolysis by FITC-elastase were determined within the pH range 5.5-8.5. The K m data will be used later in this paper. Fig. 1 shows that k cat increases 18-fold over this pH range.
Effect of pH on the Rate of Formation and Translocation of the Elastase-␣ 1 PI Complex-This was done by measuring the time-dependent change in fluorescence energy transfer upon mixing FITC-elastase and TMR-␣ 1 PI with a stopped-flow apparatus. As shown previously (8), the reaction of FITC-elastase with TMR-␣ 1 PI quenches the fluorescence of FITC and enhances that of TMR, indicating donor-acceptor energy transfer. Also, binding of unlabeled elastase or ␣ 1 PI with labeled partners only marginally affected the fluorescence of the latter, indicating that the fluorescence changes are associated with nonradiative energy transfer. We observed two consecutive first-order reactions: the appearance of EI M governed by k obs ϭ k 1 [I] 0 ϩ k Ϫ1 and the conversion of EI M into EI 2,f governed by k 2,f (see Scheme 1). All experiments were done using [I] 0 ϭ 10 [E] 0 ϭ 8 ϫ 10 Ϫ6 M, which yields nearly saturation of the enzyme as an EI M complex and hence the highest amplitude at pH 7.4 (8). Under these concentration conditions, we were able to follow the reaction down to pH 5.5 despite an important loss in the quantum yield of fluorescein. At all pH values we observed the same biphasic trace as described previously. Fig. 2 illustrates this observation with two examples, one at pH 8.5, the other one at pH 5.5. The difference in the amplitudes of the signal is mainly due to changes in the quantum yield of fluorescein with pH but also to the variations of the settings of the detector. k obs did not significantly vary with pH (data not shown), suggesting that formation of EI M does not involve ionic interactions. Interestingly, k 2,f was also pH-independent from pH 5.5 to 8.5 (Fig. 3).
Interaction of FITC-elastase with the Binary TMR-␣ 1 PI-Peptide Complex-Insertion of a 14-mer peptide corresponding to the P 1 to P 14 sequence of the reactive site loop into ␤-sheet A of serpins is known to change them from inhibitors to substrates by preventing loop insertion (21). As a control experiment and to investigate the nature of the rearrangement monitored by our fluorescence transfer method, we observed the reaction of FITC-elastase with 14-mer peptide-inserted TMR-␣ 1 PI at pH 7.5 and at concentrations identical to those used with free TMR-␣ 1 PI. In Fig. 4, three steps are visible: (i) a fast exponential decrease in fluorescence corresponding to the formation of the initial Michaelis-type complex, (ii) a steady-state phase corresponding to the expected turnover of peptide-bound TMR-␣ 1 PI whose concentration is 10-fold higher than that of elastase, and (iii) a very slow (ϳ100 s) non-exponential return of the fluorescence intensity to its initial value. Such a return to the initial fluorescence intensity indicates that elastase and ␣ 1 PI are again separated from each other following substratelike cleavage of the reactive site loop. Thus, the exponential translocation (t 1/2 ϳ 5 s) observed with free TMR-␣ 1 PI does not take place with peptide-inserted TMR-␣ 1 PI. It may therefore be concluded that the interchromophore distance change observed with active ␣ 1 PI is related to the insertion of the reactive site loop into ␤-sheet A.
Effect of pH on the Rate of Inhibition of Elastase by ␣ 1 PI in the Presence of Suc-Ala 3 -pNA-The rate of inhibition, i.e. the rate of disappearance of free elastase, was measured by recording the release of product in mixtures of FITC-elastase, ␣ 1 PI, and Suc-Ala 3 -pNA as described under "Experimental Procedures." As expected for irreversible inhibition, all progress curves were simple exponentials from which the pseudo-firstorder rate constants k were calculated by nonlinear regression analysis. At each pH value, k was determined using a series of ␣ 1 PI concentrations. All k versus [␣ 1 PI] 0 plots were hyperbolic (see two examples in Fig. 5), strongly suggesting two-step irreversible inhibition (22) as illustrated in Scheme 2.  Fig. 6 shows that k 2,e does not significantly vary with pH except at pH 5.5, where it is about twice as high as at pH ϭ 6.
Kinetics of Reaction of FITC-anhydro-elastase with TMR-␣ 1 PI as Followed by Fluorescence Transfer-FITC-labeled anhydro-elastase was mixed with TMR-␣ 1 PI in the stopped-flow apparatus at pH 7.5, and the reaction was followed by fluorescence. The trace shown in Fig. 6A is biphasic as for active elastase (Fig. 2) suggesting that anhydro-elastase reacts with ␣ 1 PI via a two-step mechanism similar to that previously demonstrated with active elastase and illustrated in Scheme 1. Moreover, the rate constant for the translocation of anhydroelastase (0.11 Ϯ 0.02 s Ϫ1 ) is very close to that for the translocation of active elastase (0.13 Ϯ 0.03 s Ϫ1 ).
To confirm that the second exponential represents translocation of EI M into EI 2,f and not an artifact, we reacted the same sample of FITC-anhydro-elastase with TMR-eglin c, a canonical reversible inhibitor that reacts with active elastase in only one step (8). Fig. 7B confirms that eglin c also reacts in one step with anhydro-elastase. Thus, translocation of elastase from the first binding site to another site does not require the enzyme's catalytic machinery.
Recovery of ␣ 1 PI following Dissociation of the Anhydro-elastase-␣ 1 PI Complex-A mixture of 30 M FITC-anhydro-elastase and 30 M ␣ 1 PI was chromatographed at pH 7.5 on a Sepharose-eglin c column. FITC-anhydro-elastase was retained at the top of the column, as evidenced by its color while ␣ 1 PI was collected in the flow-through volume. Titration of the dissociated ␣ 1 PI with neutrophil elastase confirmed that all of the inhibitor could be recovered. This implies that eglin c was able to fully dissociate the complex. The rate of inhibition of active elastase by the recovered ␣ 1 PI was measured in the presence of substrate as described under "Experimental Procedures." Using two different concentrations of ␣ 1 PI, we got a second-order association rate constant of 5 ϫ 10 5 Ϯ 0.8 M Ϫ1 s Ϫ1 at pH 7.5. This figure agrees with the rate constant usually found in the buffer conditions used in our laboratory for unreacted ␣ 1 PI. The ␣ 1 PI sample recovered from the complex was also analyzed by differential scanning calorimetry. The thermogram revealed a thermal transition at 60.5°C for both the control ␣ 1 PI and the sample recovered from the complex with anhydro-elastase. Thus, the translocated ␣ 1 PI-anhydro-elastase complex is in true thermodynamic equilibrium with free anhydro-elastase and ␣ 1 PI, since its dissociation is quantitative and yields ␣ 1 PI molecules whose inhibitory capacity and thermal stability are identical to those of the native inhibitor.

DISCUSSION
Translocation Precedes Acylation-In recent years many kinetic, chemical, and crystallographic data have been collected in view of understanding the mechanism by which serpins inhibit serine proteinases. It is commonly accepted that the enzyme and the inhibitor first form a Michaelis-type complex within which the serpin is subsequently cleaved to yield an acylenzyme linking the serine residue of the proteinase with the P 1 residue of the serpin (5-7). Formation of the acylenzyme is believed to relax the strained conformation of the serpin, to incorporate the N-terminal part of the reactive site loop into ␤-sheet A and hence to cause a translocation of the proteinase that stabilizes the acylenzyme bond. This model is summarized in Scheme 3.
EI M , EI ac , and EI tr are the Michaelis-type complex, the acylenzyme, and the translocated acylenzyme, respectively; K i is the equilibrium dissociation constant of the Michaelis-type complex; k ac and k tr are the acylation and the translocation rate constants, respectively. As will be demonstrated below, our kinetic data on the interaction of ␣ 1 PI with elastase and anhydro-elastase invalidate Scheme 3 and provide strong evidence that translocation precedes acylation as shown in Scheme 4.
The catalytic-triad of serine proteinases comprises Asp 102 , His 57 , and Ser 195 . Catalysis involves acylation and deacylation of Ser 195 , both of which are sensitive to pH because they depend upon the state of ionization of His 57 (pK a ϳ6.7). As a control, we have shown that k cat for the elastase-catalyzed hydrolysis of the model substrate Suc-Ala 3 -pNA increases 18-fold between pH 5.5 and 8.5. Therefore, if EI ac would follow immediately EI M (Scheme 3), its rate of appearance should strongly increase with pH. We did not measure the kinetics of EI ac formation but we used two ways to measure the kinetics of EI M decay: fluorescence, which yielded k 2,f (Scheme 1), and substrate hydrolysis, which gave k 2,e (Scheme 2). Both k 2,f and k 2,e were found to be essentially pH-independent. k 2,f is the first-order rate constant for the fluorescently detected decay of the Michaelis-type complex EI M (Scheme 1). In our experiment fluorescence detects only those species that significantly differ in energy transfer efficiency. It is thus not unlikely that fluorescently silent events such as acylation take place during the fluorescence change that characterizes the translocation of EI M . As a consequence, EI 2,f is not necessarily a single molecular species and k 2,f is possibly an apparent rate constant that comprises both the pH-dependent constant, k ac , and the pH-independent one, k tr . Thus, the pH independence of k 2,f is, at first sight, compatible with both of the afore-mentioned reaction schemes but imposes some restrictions. Either Scheme 4 is valid and k 2,f ϭ k tr whatever the magnitude of k ac , or Scheme 3 is valid, in which case k tr must necessarily be rate-limiting for the formation of the final complex EI tr , which implies that k ac Ͼ Ͼ k tr . Therefore, k ac should be much larger than k 2,f , the translocation rate constant measured by fluorescence resonance energy transfer, i.e. 0.13 s Ϫ1 . In summary, Scheme 3 can only be compatible with the observed pH inde- pendence of k 2,f if k ac Ͼ Ͼ 0.13 s Ϫ1 over the whole pH range studied, so that even at pH 5.5, where acylation is greatly reduced, it is still not rate-limiting. k 2,e is derived from the rate of elastase inhibition by ␣ 1 PI in the presence of a competing substrate. It represents the firstorder rate constant for the irreversible or pseudo-irreversible conversion of Michaelis-type complex EI M into EI 2,e , the complex that immediately follows EI M in the reaction pathway (Scheme 2). The calculation and the interpretation of k 2,e do not depend upon whether EI 2,e is the final complex or the first of several consecutive complexes. If Scheme 3 were valid, k 2,e ϭ k ac . Being the rate constant of an acylation step, k 2,e should, therefore, be pH-dependent. Furthermore, the restriction deduced from the study of k 2,f as a function of pH predicts that k ac Ͼ Ͼ k 2,f for Scheme 3 to be valid (see above). This leads to k 2,e Ͼ Ͼ k 2,f . The pH independence of k 2,e and the fact that k 2,e (0.41 s Ϫ1 ) is of the same order of magnitude as k 2,f (0.13 s Ϫ1 ) are thus complementary evidence invalidating Scheme 3 and favoring Scheme 4. Besides, the numerical value of k 2,e may be considered as equal to that of k 2,f if allowance is made for the errors on each constant, one being measured directly by fluorescence, the other being derived enzymatically by a much more indirect method. In conclusion, the kinetic data indicate that the irreversible or pseudo-irreversible step that immediately follows the formation of the Michaelis-type complex EI M is the translocation step that occurs with a rate constant of 0.13-0.41 s Ϫ1 . In our system, translocation is also the rate-limiting step for enzyme inhibition.
On the other hand, the complex formed between ␣ 1 PI and anhydro-elastase is able to translocate with a rate constant (0.11 s Ϫ1 ) close to that measured for the translocation of active elastase (0.13-0.41 s Ϫ1 ).The observed translocation of catalytically inactive elastase thus confirms that acylation of the enzyme is not required for its translocation but that translocation is rather a prerequisite for acylation as indicated in Scheme 4. This model suggests that k ac may be higher or lower than k tr , depending upon the pH conditions and/or the nature of the enzyme-inhibitor pair. In addition, an implicit assumption of this model is that part of the binding energy of the Michaelis-type complex EI M is used to trigger the translocation process via the flexible reactive site loop or a yet unknown exosite. This is reminiscent of the observation by Verhamme et al. (23) that a monoclonal antibody directed against an unidentified epitope of plasminogen activator inhibitor 1 is able to accelerate the conversion of this serpin into a latent inhibitor in which the reactive center loop is inserted into ␤-sheet A. It may be hypothesized that the interaction between the antibody and the serpin partly mimics the interaction of the proteinase with the serpin and triggers the conformational change that results in loop-sheet interaction. The elucidation of the location of the epitope would be of great significance.
The pH independence of the rate-limiting step for the inhibition of chymase by antichymotrypsin was suggested by Schechter et al. (24). However, this study measured the overall inhibition rate constant with no information on the nature of the reaction steps. Serpin-proteinase reaction mechanisms with two consecutive reversible steps have been recently proposed by Kvassman et al. (25) for the inhibition of plasminogen activator by plasminogen activator inhibitor 1, by Nair et al. (26) for the inhibition of chymotrypsin by antichymotrypsin, and by Stone et al. (27) for the inhibition of thrombin by several serpins. However, the methods used in these works were not adapted to determine whether the proteinase was translocated and whether acylation occurred before or after translocation.
Although the fluorescence change that accompanies translocation is described by a simple exponential, it is possible that a further conformational change takes place after acylation since formation of the acylenzyme is accompanied by liberation of the new N terminus PЈ 1 residue. Such a conformational change is not detected by fluorescence transfer. This either means that each translocated molecule is rapidly acylated so that the accumulated fluorescence changes of the two consecutive movements are included in the fluorescence change accompanying the rate-limiting exponential or that the hypothetical conformational change induced by acylation does not significantly modify the interchromophore distance. In the first assumption, the translocation occurs in two steps: (i) one involving the uncleaved reactive site loop that is brought to a position where acylation can safely occur in a solvent-free environment to prevent deacylation, and (ii) the other resulting from the rearrangement consecutive to loop cleavage. The second assumption implies that the proteinase has reached its final position after the translocation step.
Is Translocation Reversible?-Chromatography of the anhydro-elastase-␣ 1 PI complex on a Sepharose-eglin c column was able to dissociate the complex and to recover functionally active and structurally intact ␣ 1 PI, indicating that the reactants are in true thermodynamic equilibrium as outlined in Scheme 5. ahE ϩ I º ahEI M L | ; k tr k Ϫtr ahEI tr SCHEME 5 ahE is anhydro-elastase. Since significant translocation of the initial Michaelis-type complex was observed, k tr must be larger than k Ϫtr . For instance, if k tr ϭ 10 k Ϫtr , the translocation is virtually complete.
It is likely that translocation of active elastase is also reversible or pseudo-irreversible. Active and anhydro-elastase differ only in one residue. Both form a reversible complex with eglin c, and a Michaelis-type complex with ␣ 1 PI, which further undergoes translocation. It is difficult to conceive that the serine residue of the active site of elastase prevents reversibility of translocation. The data of Whisstock et al. (28) favor this view. These authors modeled the interaction of ␣ 1 PI with neutrophil elastase and came to the conclusion that insertion of the uncleaved binding loop up to residue P12 into the ␤-sheet A allowed a possible docking of the proteinase to the translocated site. They reported that the required conformational changes occur with the correct stereochemistry and without steric clashes. Furthermore, the proteinase and the serpin were held together by a large number of van der Waals and hydrogen bonds. There is thus an energetically favorable docking at the translocated site which does not involve the serine residue of the enzyme's active center and does not require reactive site loop cleavage of the serpin. Further experiments are needed to firmly demonstrate that the translocated ␣ 1 PI-active elastase complex is in equilibrium with the Michaelis-type complex.
Implications of the New Model-In our study we show that in the second reversible step the anhydro-proteinase has been translocated. Accordingly, the Michaelis-type complex between an anhydro-proteinase and a serpin, which is rapidly converted into the translocated complex, should no longer be considered as a model for the first encounter complex (10,11,29,30) in which the serpin is probably still close to its native conformation.
Each step described in the minimal Scheme 4 is determinant for the selection of the target proteinase by the serpin and for its fate. The fact that each step may be governed by a special property of the proteinase leads necessarily to a variety of situations. Thus, the ability of the proteinase to form a tight or a loose EI M complex is probably independent of its ability to trigger translocation. Similarly, the features that are determinant in these two first steps are certainly not related to the ability of the proteinase to catalyze the formation of a covalent complex with the serpin. This provides a flexibility in the regulation of proteinase inhibition at each step that may be the cause of the selection of serpins in potentially dangerous processes such as the cascade of the coagulation, the complement cascade, or fibrinolysis. The example of the inhibition of proteinase by ␣ 2 -antiplasmin, which leads to an SDS-stable complex (31) but appears to be reversible for some time (32,33), illustrates what can be achieved with the flexible mechanism proposed above.