Characterization of a Human a 1 -Antitrypsin Variant That Is as Stable as Ovalbumin*

The metastability of inhibitory serpins (serine proteinase inhibitors) is thought to play a key role in the facile conformational switch and the insertion of the reactive center loop into the central b -sheet, A-sheet, during the formation of a stable complex between a serpin and its target proteinase. We have examined the folding and inhibitory activity of a very stable variant of human a 1 -antitrypsin, a prototype inhibitory serpin. A combination of seven stabilizing single amino acid substitutions of a 1 -antitrypsin, designated Multi-7, in- creased the midpoint of the unfolding transition to al-most that of ovalbumin, a non-inhibitory but more stable serpin. Compared with the wild-type a 1 -antitryp- sin, Multi-7 retarded the opening of A-sheet significantly, as revealed by the retarded unfolding and la-tency conversion of the native state. Surprisingly, Multi-7 a 1 -antitrypsin could form a stable complex with a target elastase with the same kinetic parameters and the stoichiometry of inhibition as the wild type, indicat-ing that enhanced A-sheet closure conferred by Multi-7 does not affect the complex formation. It may be that the stability increase of Multi-7 a 1 -antitrypsin is not suffi- cient to influence the rate of loop insertion during the complex formation. The serpin (serine proteinase inhibitor) -antitrypsin were purchased from Sigma. All other chemicals were reagent grade. Equilibrium Unfolding Transition— folding function of urea was monitored fluorescence spectroscopy and CD spectroscopy, details of which were (12, The buffer used for folding experiments was 10 m M phosphate, unless described otherwise. The native protein was incubated in the folding buffer containing various concentrations of urea at 25 °C. The protein concentration for the unfolding transition was 20 m g/ml for fluorescence spectroscopy and 50 m g/ml for CD spectroscopy. Experimental data of the fluorescence measurement were fitted to a two-state unfolding model, as described previously (12, 15). Kinetics of Unfolding and Refolding— The kinetic study of unfolding and refolding of a 1 AT was performed by measuring the change of fluorescence intensity ( l ex 5 280 nm and l em 5 360 nm) of the native or unfolded proteins upon exposure to various concentrations of urea (12). The final protein concentration was 5 m g/ml, and the experimental temperature was 25 °C. The major kinetic phase was examined, and the data were fitted to a two-state (for unfolding) or three-state (for refolding) model to obtain the relaxation time: F ( t ) 5S F ( i ) e ( 2 t / t i ) 1 F ‘ , where F ( t ) is the fluorescence at time t , F ( i ) is the fluorescence of phase i at zero time, t i is the relaxation time for phase i , and F ‘ is the fluorescence at infinite time (Sigma Plot, Jandel Corp.). Transverse Urea Gradient Gel Electrophoresis— Gels were prepared with a gradient of 0–8 M urea perpendicular to the direction of electrophoresis with an opposing gradient of acrylamide from 15 to 11% (20). Four slab gels (100 3 80 mm) were prepared simultaneously in a multigel caster (Hoeffer) by using a gradient maker and a single chan- nel peristaltic pump. The electrode buffer was 50 m M Tris acetate, 1 m M EDTA, pH 7.5. The native protein (20 m g in 100 m l), the protein unfolded in 8 M urea (10 min at room temperature), or the in vitro translation product was applied across the top of the gel. The gels were run at a constant current of 6 mA fo r 3 h at acontrolled temperature of 25 °C. The protein bands were visualized either by Coomassie Brilliant Blue staining or by autoradiography. Conversion into the Latent Form— Formation of the latent a 1 AT was examined with in vitro translation products, as described (12). Conditions for the formation of latent a 1 AT have been reported previously (21). The translation products labeled with [ 35 S]methionine were incubated at 60 °C for 4 h in 20 m M Tris, pH 7.4, which contained 0.7 M sodium citrate. Various conformations of a 1 AT were analyzed by transverse urea gradient gel electrophoresis. Complex Formation Proteinases— Complex formation of Multi-7 a 1 AT with a proteinase was examined by monitoring the SDS-resistant a 1 AT-proteinase complex (22, 23). Purified a 1 AT was incubated in assay buffer (30 m M phosphate, 160 m M NaCl, 0.1% PEG 6000, 0.1% Triton X-100, pH 7.4) with porcine pancreatic elastase or human leukocyte elastase at designated molar ratios of a 1 AT to proteinase. Samples were incubated at 37 °C for 10 min and were analyzed by 10% SDS-polyacryl- amide gel electrophoresis. The protein bands were visualized by Coomassie Brilliant Blue staining. Concentrations of a 1 AT were deter- mined by A 280 in 6 M guanidine hydrochloride (17). The active site concentration of human leukocyte elastase was determined as described previously (22) with trypsin-titrated human plasma a 1 AT and a substrate, N -methoxysuccinyl-Ala-Ala-Pro-Val- p -nitroanilide. The active concentration of porcine pancreatic elastase was determined by measuring the initial rates of hydrolysis of 1 m M N -succinyl-(Ala) 3 - p -nitro- anilide by increasing the concentration of elastase, using a previously reported standard titration curve (24). The stoichiometry of inhibition was determined by titration reactions as described (25). The reaction mixture (50 m l volume) in the assay buffer contained 100 n M porcine or human elastase. After incubation with various amounts of recombinant wild-type or Multi-7 a 1 AT for 10 min at 37 °C, the reaction mixture was diluted 10-fold with the assay buffer, and residual enzyme activity was determined. Inhibitory Parameters of a 1 AT— Inhibition kinetic studies for the interaction of a 1 AT with porcine pancreatic elastase were performed by analyzing progress curve kinetic experiments (26). The active concentration of porcine pancreatic elastase was determined as described above. The active site titration of a 1 AT was measured by employing 1 m M N -succinyl-(Ala) 3 - p -nitroanilide and a known activity of porcine pancreatic elastase (18). The assays were performed at 25 °C in reaction buffer containing 50 m M Tris, 50 m M NaCl, 0.1% PEG 4000, and 0.05% v/v Triton X-100, pH 8.0, and started by the addition of elastase, at the final concentration of 1.25 n M . A typical progress curve experiment consisted of 6 assays (1 zero and 5 non-zero concentrations of inhibitor) and the slow development of inhibition was determined by continuously monitoring the appearance of p -nitroaniline at 410 nm. The amount of product formed was calculated by using a molar absorption coefficient of 8,800 M 2 1 cm 2 1 for p -nitroaniline at 410 nm. The inhibition of the enzyme ( E ) by a 1 AT (I) is described in Scheme I, where S is the substrate, and P is p -nitroaniline.

The serpin 1 (serine proteinase inhibitor) superfamily includes inhibitors such as ␣ 1 -antitrypsin (␣ 1 AT), antithrombin, ␣ 1 -antichymotrypsin, C1 inhibitor, and non-inhibitory members such as ovalbumin and angiotensinogen (1). Serpins share a common tertiary structure composed of three ␤-sheets and several ␣-helices (Fig. 1). One of the intriguing aspects of serpin structure is that the native conformation of the inhibitory serpins is strained (2)(3)(4). Proteolytic cleavage of the reactive center loop or the conversion into the more stable latent form (Fig. 1) accompanies a complete insertion of the reactive center loop into the central ␤-sheet, A-sheet, with concomitant release of the strain (5,6). In a non-inhibitory but more stable serpin, ovalbumin, the cleavage of the loop does not induce a drastic conformational switch as in inhibitory serpins (7). The ability of loop insertion of serpins thus appears to be very critical to the conformational switch and inhibitory function. Upon binding a target proteinase, the reactive center loop of a serpin is thought to be inserted into the A-sheet to form a stable complex between the proteinase and the inhibitor (8 -10). It has been suggested that a critical factor for the formation of a stable complex is the rate of loop insertion (8,11). The metastable structure of inhibitory serpins has the advantage of a facile conversion into an alternative stable conformation (12). It is possible that the native strain of inhibitory serpins is utilized for the facile loop insertion during the complex formation with a target proteinase (9,(12)(13)(14). However, the precise mechanism underlying the complex formation is yet to be elucidated.
Previously we have identified several hydrophobic core mutations of ␣ 1 AT that increased the conformational stability, presumably by enhancing the closure of the A-sheet (12). The mutations did not affect proteinase binding, as revealed by the unchanged association rate constants with porcine elastase. It was suspected, however, that these mutations would affect the formation of the stable complex upon proteinase binding by retarding the loop insertion, resulting in substrate-like behavior. In the present study we tested this assumption by analyzing a very stable ␣ 1 AT carrying seven stabilizing mutations ( Fig. 1: F51L, T59A, T68A, A70G, M374I, S381A, and K387R) identified previously (12). The mutant was designated Multi-7. The mutational effects on the inhibitory activity as well as other structural properties such as folding-unfolding transition were also examined.

MATERIALS AND METHODS
Recombinant ␣ 1 AT Proteins and Plasmids-The plasmid for ␣ 1 AT expression in Escherichia coli (15), the purification of recombinant ␣ 1 AT protein (15), and the detailed method for in vitro translation products of ␣ 1 AT (16) were described previously. The plasmid for the in vitro translation of ovalbumin was constructed by substituting ␣ 1 AT cDNA in the in vitro translation vector, pF(BLG)AT (16), with the cDNA of ovalbumin from pYOV5 (gift from Dr. H. J. Kim). Concentrations of ␣ 1 AT were determined in 6 M guanidine hydrochloride using a value of A 1 cm 1 % ϭ 4.3 at 280 nm, calculated from the tyrosine and tryptophan content of the ␣ 1 AT protein (17) and based upon M r ϭ 44,250. ␣ 1 AT activity was measured as residual porcine pancreatic elastase activity employing 1 mM N-succinyl-(Ala) 3 -p-nitroanilide as a chromogenic substrate (18). Individual thermostable mutations of ␣ 1 AT were previously reported (12). Combination of the mutations of ␣ 1 AT and site-specific mutations of ovalbumin were made by oligonucleotide-directed mutagenesis (19).
Urea-induced Equilibrium Unfolding Transition-Equilibrium unfolding as a function of urea was monitored by fluorescence spectroscopy and CD spectroscopy, details of which were described previously (12,15 1. Comparison of the native, cleaved, and the latent structures of inhibitory serpins. The Native structure is a ribbon diagram of the crystal structure of Multi-7 ␣ 1 AT (32). The side chain atoms for the seven stabilizing substitutions in Multi-7 are represented. The Cleaved ␣ 1 -Antitrypsin Variant That Is as Stable as Ovalbumin 2510 50 mM NaCl, 1 mM EDTA, 1 mM ␤-mercaptoethanol, pH 6.5, unless described otherwise. The native protein was incubated in the folding buffer containing various concentrations of urea at 25°C. The protein concentration for the unfolding transition was 20 g/ml for fluorescence spectroscopy and 50 g/ml for CD spectroscopy. Experimental data of the fluorescence measurement were fitted to a two-state unfolding model, as described previously (12,15).
Kinetics of Unfolding and Refolding-The kinetic study of unfolding and refolding of ␣ 1 AT was performed by measuring the change of fluorescence intensity ( ex ϭ 280 nm and em ϭ 360 nm) of the native or unfolded proteins upon exposure to various concentrations of urea (12). The final protein concentration was 5 g/ml, and the experimental temperature was 25°C. The major kinetic phase was examined, and the data were fitted to a two-state (for unfolding) or three-state (for refolding) model to obtain the relaxation time: is the fluorescence at time t, F(i) is the fluorescence of phase i at zero time, i is the relaxation time for phase i, and F ϱ is the fluorescence at infinite time (Sigma Plot, Jandel Corp.).
Transverse Urea Gradient Gel Electrophoresis-Gels were prepared with a gradient of 0 -8 M urea perpendicular to the direction of electrophoresis with an opposing gradient of acrylamide from 15 to 11% (20). Four slab gels (100 ϫ 80 mm) were prepared simultaneously in a multigel caster (Hoeffer) by using a gradient maker and a single channel peristaltic pump. The electrode buffer was 50 mM Tris acetate, 1 mM EDTA, pH 7.5. The native protein (20 g in 100 l), the protein unfolded in 8 M urea (10 min at room temperature), or the in vitro translation product was applied across the top of the gel. The gels were run at a constant current of 6 mA for 3 h at a controlled temperature of 25°C. The protein bands were visualized either by Coomassie Brilliant Blue staining or by autoradiography.
Conversion into the Latent Form-Formation of the latent ␣ 1 AT was examined with in vitro translation products, as described (12). Conditions for the formation of latent ␣ 1 AT have been reported previously (21). The translation products labeled with [ 35 S]methionine were incubated at 60°C for 4 h in 20 mM Tris, pH 7.4, which contained 0.7 M sodium citrate. Various conformations of ␣ 1 AT were analyzed by transverse urea gradient gel electrophoresis.
Complex Formation with Proteinases-Complex formation of Multi-7 ␣ 1 AT with a proteinase was examined by monitoring the SDS-resistant ␣ 1 AT-proteinase complex (22,23). Purified ␣ 1 AT was incubated in assay buffer (30 mM phosphate, 160 mM NaCl, 0.1% PEG 6000, 0.1% Triton X-100, pH 7.4) with porcine pancreatic elastase or human leukocyte elastase at designated molar ratios of ␣ 1 AT to proteinase. Samples were incubated at 37°C for 10 min and were analyzed by 10% SDS-polyacrylamide gel electrophoresis. The protein bands were visualized by Coomassie Brilliant Blue staining. Concentrations of ␣ 1 AT were determined by A 280 in 6 M guanidine hydrochloride (17). The active site concentration of human leukocyte elastase was determined as described previously (22) with trypsin-titrated human plasma ␣ 1 AT and a substrate, N-methoxysuccinyl-Ala-Ala-Pro-Val-p-nitroanilide. The active concentration of porcine pancreatic elastase was determined by measuring the initial rates of hydrolysis of 1 mM N-succinyl-(Ala) 3 -p-nitroanilide by increasing the concentration of elastase, using a previously reported standard titration curve (24). The stoichiometry of inhibition was determined by titration reactions as described (25). The reaction mixture (50 l volume) in the assay buffer contained 100 nM porcine or human elastase. After incubation with various amounts of recombinant wild-type or Multi-7 ␣ 1 AT for 10 min at 37°C, the reaction mixture was diluted 10-fold with the assay buffer, and residual enzyme activity was determined.
Inhibitory Parameters of ␣ 1 AT-Inhibition kinetic studies for the interaction of ␣ 1 AT with porcine pancreatic elastase were performed by analyzing progress curve kinetic experiments (26). The active concentration of porcine pancreatic elastase was determined as described above. The active site titration of ␣ 1 AT was measured by employing 1 mM N-succinyl-(Ala) 3 -p-nitroanilide and a known activity of porcine pancreatic elastase (18). The assays were performed at 25°C in reaction buffer containing 50 mM Tris, 50 mM NaCl, 0.1% PEG 4000, and 0.05% v/v Triton X-100, pH 8.0, and started by the addition of elastase, at the final concentration of 1.25 nM. A typical progress curve experiment consisted of 6 assays (1 zero and 5 non-zero concentrations of inhibitor) and the slow development of inhibition was determined by continuously monitoring the appearance of p-nitroaniline at 410 nm. The amount of product formed was calculated by using a molar absorption coefficient of 8,800 M Ϫ1 cm Ϫ1 for p-nitroaniline at 410 nm. The inhibition of the enzyme (E) by ␣ 1 AT (I) is described in Scheme I, where S is the substrate, and P is p-nitroaniline.
The progress data were fitted by nonlinear regression to Equation 1 describing this mechanism of slow, tight-binding mechanism as below (26).
where P is the amount of product formed at time, t, d is a function of E t , I t , and K i Ј , and kЈ is a function of these parameters and the observed second-order rate constant (k i Ј ) for the interaction between the inhibitor and enzyme. The analyses yield values for apparent association rate constant (k i Ј ) and the apparent dissociation constant (K i Ј ). The value of the Michaelis constants (K m ) required in these calculations was 1.15 mM for Suc-(Ala) 3 -p-nitroanilide (24).

RESULTS
Equilibrium Stability of Multi-7 ␣ 1 AT-Equilibrium unfolding transition of Multi-7 ␣ 1 AT was examined as a function of increasing urea concentration at 25°C. When the transition was monitored by intrinsic fluorescence intensity ( Fig. 2A), the midpoints of transition of the wild-type and the mutant protein were 1.8 and 4.8 M urea, respectively, yielding ⌬⌬G of 8 kcal mol Ϫ1 . A similar shift in the transition midpoint was observed by far UV CD signal (Fig. 2B) or by transverse urea gradient gel electrophoresis (Fig. 2C). The unfolding transition of the wild-type ␣ 1 AT monitored by far UV CD signal or urea gradient gel electrophoresis exhibited at least two phases, indicative of at least one equilibrium folding intermediate. Because of the midpoint shift of the first unfolding phase, the equilibrium unfolding intermediate was not detected with the Multi-7 mutant protein. The unfolding transition was reversible for both the wild-type and the mutant proteins, as revealed by refolding on urea gradient gel electrophoresis (Fig. 2C, Refolding). The transition of Multi-7 was rather abrupt at the transition midpoint.
The Multi-7 ␣ 1 AT Is as Stable as Ovalbumin-The stability of Multi-7 ␣ 1 AT was compared with that of ovalbumin with in vitro translation products on urea gradient gel electrophoresis. Fig. 3 shows that ovalbumin is much more stable than the wild-type ␣ 1 AT, and its unfolding transition follows a two-state process. The overall transition of ovalbumin was very similar to that of Multi-7 ␣ 1 AT. The in vitro translation product of the wild-type ovalbumin yielded two species of the native form that exhibited a slight difference in unfolding transition on urea gradient gel electrophoresis (Fig. 3, Ovalbumin, Wt). Ovalbumin contains one disulfide bond between Cys-73 and Cys-120, which is not present in ␣ 1 AT. To compare the stability of Multi-7 ␣ 1 AT with that of ovalbumin without disulfide contribution, we constructed recombinant versions of ovalbumin that cannot form the disulfide bond by substituting cysteines into alanines. The mutant ovalbumin yielded only a single species of the native form on urea gradient gel with the same unfolding midpoint as Multi-7 ␣ 1 AT (Fig. 3, Ovalbumin, C73A). The structure is that of ␣ 1 AT (2), where the wild-type residues of the seven substitutions in Multi-7 are shown. The Latent form is adopted from the crystal structure of dimeric antithrombin-III (36), where the absence of electron density is depicted as broken lines. In all three structures, the reactive center loop is located by shading, and the active site residue (P1) is indicated. amino acid substitutions themselves do not appear to contribute significantly to the stability because all three different forms of mutant ovalbumin (C73A, C120A, and C73A/C120A) showed the unfolding transition at a urea concentration similar to that of the less stable wild-type species (data not shown). The stability of Multi-7 was similar to that of the mutant ovalbumin which could not form the disulfide. The unfolding transition of both Multi-7 ␣ 1 AT and ovalbumin were abrupt.
Kinetic Analysis of Unfolding and Refolding of Multi-7 ␣ 1 AT-Kinetic unfolding and refolding of Multi-7 ␣ 1 AT were performed by monitoring the intrinsic fluorescence change as a probe. Fig. 4 shows that the unfolding of ␣ 1 AT occurs in an all-or-none process at all urea concentrations, and the unfold-ing of Multi-7 ␣ 1 AT was retarded significantly. Refolding of ␣ 1 AT exhibited multiple kinetic states. There were two major refolding phases in which the intensity of fluorescence decreased: a fast decrease ( ϭ 200 -500 s), and a further slow decrease ( ϭ 1,000 -3,000 s). The amplitudes of the two phases were about same at near zero urea concentration, but the amplitude of slow phase increased as a function of urea. Refolding of Multi-7 ␣ 1 AT was facilitated slightly in both phases, but urea dependence of the refolding rate at lower urea concentrations disappeared in such a way that the difference in refolding rate was minimized when extrapolated to 0 M urea (the relaxation time of 339 and 106 s at 0 M urea for the fast phase of wild-type and Multi-7, respectively). In particular, the rate of the slow refolding phase, which intersects with the unfolding kinetics at the same urea concentration (1.9 and 4.8 M for the wild-type and Multi-7, respectively) as the equilibrium midpoint ( Fig. 2A), was not significantly altered in the Multi-7 mutant (1072 s at 0 M urea for both). The amplitude of both phases was not changed by the mutation. These results indicate that the major effect of the Multi-7 mutation in the unfolding-refolding transition is the retardation of the unfolding rate.
Conversion of Multi-7 ␣ 1 AT into the Latent Form-The native structure of inhibitory serpins is considered as a kinetically trapped folding intermediate because the intact native form can convert into a more stable latent form. It was expected that the mutations that stabilize the native state of serpin will retard the conversion into the latent state. This was the case with Multi-7 ␣ 1 AT, as shown in Fig. 5. Unlike the native form of ␣ 1 AT that undergoes unfolding transition in urea (Fig. 2C), the latent form does not unfold even in the presence of 8 M urea. When heat denaturation was performed under conditions where the production of the latent ␣ 1 AT was favored (about 50% of the wild-type ␣ 1 AT converted into the latent form), the Multi-7 mutant converted into the latent form much less readily than the wild-type protein. The results showed that the Multi-7 mutation retarded the insertion of the reactive center loop into the A-sheet, although the accessibility of the loop insertion was not affected.
Inhibitory Activity of Multi-7 ␣ 1 AT-To investigate the effect of the Multi-7 mutation in ␣ 1 AT on the inhibitory activity, complex formation of wild-type and Multi-7 ␣ 1 AT with various proteinases was examined. Both wild-type and the mutant inhibitor formed a tight SDS-resistant proteinase-inhibitor complex with porcine and human elastases (Fig. 6A). In addition, the mutation did not alter the partitioning between the inhibitory and substrate pathways, as revealed by the unchanged values of stoichiometry of inhibition in the titration of elastase by ␣ 1 AT (Fig. 6B: 1.1 and 1.7 for human and porcine elastase, respectively). This was also confirmed by densitometric scanning of the SDS-resistant complex formation shown in Fig. 6A. Inhibition kinetic studies showed that the association rate constant (k a ) of the wild type and the Multi-7 mutant ␣ 1 AT for porcine pancreatic elastase was 5.11 Ϯ 0.7 ϫ 10 5 M Ϫ1 s Ϫ1 and 5.03 Ϯ 1.0 ϫ 10 5 M Ϫ1 s Ϫ1 , respectively. The dissociation constant (K i ) was 95 Ϯ 24 pM and 117 Ϯ 29 pM for the wild type and the Multi-7, respectively. The results clearly showed that the ability of ␣ 1 AT to form a complex with a target proteinase was not affected by the Multi-7 mutation. DISCUSSION It has been suggested that rapid insertion of the reactive center loop into the A-sheet during the complex formation with a target proteinase is critical for the inhibitory activity of serpins (8,11). In the present study, we examined the effect of Multi-7 mutation on the inhibitory activity of human ␣ 1 AT, which stabilized the native state of the molecule and retarded opening of the A-sheet. The mutant ␣ 1 AT could form a stable complex with various target proteinases, and the stoichiometry of inhibition was not affected. These results indicate that the A-sheet closure conferred by the stability increase of Multi-7 ␣ 1 -antitrypsin does not affect the complex formation with target proteinases. The results support the notion that, in addition to intramolecular interactions, other interactions such as contact between the inhibitor and its target proteinase contribute significantly to the conformational change needed for the complex formation.
Conformational Properties of Multi-7 ␣ 1 AT-The equilibrium unfolding of the wild-type ␣ 1 AT exhibited at least two phases with one equilibrium intermediate, which is compact (Fig. 2C) and retains approximately 70% of the native CD signal (Fig. 2B) but most of the native fluorescence is dequenched ( Fig. 2A). Unfolding of the Multi-7 ␣ 1 AT molecule did not exhibit the equilibrium unfolding intermediate (Fig. 2, B  and C). Moreover, the unfolding pattern of Multi-7 ␣ 1 AT on urea gradient gel electrophoresis was very similar to that of ovalbumin (Fig. 3). Ovalbumin, although sharing a common tertiary fold with inhibitory serpins, is more stable and is not active as a proteinase inhibitor. Molecular properties of ovalbumin are quite different from those of inhibitory serpins. For instance, the equilibrium unfolding transition of inhibitory serpins including ␣ 1 AT (Fig. 2) is very complex (5,(27)(28)(29), whereas the unfolding transition of ovalbumin is more or less fitted to a two-state model when monitored either by far UV CD signal (5), intrinsic fluorescence (30), or by UV absorbency and intrinsic viscosity (31). The unfolding midpoint of C73A mutant ovalbumin, which could not form disulfide bonds, was very close to that of Multi-7 ␣ 1 AT (Fig. 3). Previous studies on the reversible unfolding of disulfide-reduced authentic ovalbumin also yielded the same value (4.8 M) of transition midpoint (C m ) in urea as Multi-7 ␣ 1 AT at 25°C (30). The difference in C m values of the wild-type and Multi-7 ␣ 1 AT yields ⌬⌬G of 8 kcal/mol, using the m (measure of dependence of ⌬G on denaturant concentration) value of 2.6 determined previously (12). It is interesting to note that neither the Multi-7 mutant ␣ 1 AT nor ovalbumin exhibit a smooth unfolding transition seen in regular globular proteins (Fig. 3). It appears that the equilibrium unfolding of Multi-7 ␣ 1 AT resembles that of ovalbumin.
Retarded A-sheet Opening of Multi-7 ␣ 1 AT-The x-ray crystal structure of Multi-7 ␣ 1 AT (Fig. 1) was determined with 2.7-Å resolution recently (32). The uncleaved wild-type structure has yet to be determined. Mutation sites of Multi-7 ␣ 1 AT appear to be well packed as in ovalbumin (33) and the presumed native state of the wild-type around the mutation sites would increase steric hindrance. The reactive center loop of Multi-7 ␣ 1 AT is not inserted into the A-sheet at all, as in the crystal structure of intact ␣ 1 -antichymotrypsin (34). A peculiar feature of Multi-7 ␣ 1 AT is that strands 3 and 5 of the A-sheet are hydrogen-bonded all the way up to the top of the sheet, which is not the case in other native structures of inhibitory serpins (34 -36) and ovalbumin (33).
The following experimental results support that the A-sheet of the wild-type ␣ 1 AT is more open than that of Multi-7, although the structure of the intact wild-type ␣ 1 AT is not known. First, kinetic analysis revealed that the unfolding of the Multi-7 mutant ␣ 1 AT was retarded significantly, and refolding was facilitated only slightly (Fig. 4). The refolding rates of Multi-7 in both phases were urea-independent at lower urea concentrations and approach the refolding rates of the wild type. One interpretation of such urea independence is that the observed change in the refolding rates is due to the stabilization of a kinetic folding intermediate, as with the case of ubiquitin mutants (37,38). If this is the case with Multi-7 ␣ 1 AT, the intrinsic refolding rates are not altered significantly by the Multi-7 mutation. The results are consistent that the stability increase by the Multi-7 mutation is mainly due to the stabilization of the native state as opposed to the destabilization of the unfolded state. Another line of experimental evidence is that Multi-7 ␣ 1 AT is not as readily converted into the latent form as the wild type (Fig. 5). The native form of ␣ 1 AT can be converted into the latent form upon heat treatment in the presence of 0.7 M citrate (21), which did not show unfolding transition in urea gradient gel electrophoresis (Fig. 5). It was shown previously that the reactive center loop of the latent form produced under this condition was not accessible for the proteolytic attack by V8 proteinase (12). Since the conversion into the latent form requires the insertion of the reactive center loop into the A-sheet (3), it is very likely that the A-sheet is more closed in Multi-7 than in the wild-type ␣ 1 AT.
Inhibitory Activity of Multi-7 ␣ 1 AT-Various biochemical and structural studies with the mutant forms of inhibitory serpins suggest that the loop insertion is necessary for the formation of a stable complex although not sufficient to confer inhibitory activity (8 -10). Alteration of inhibitory function by oversized or charged amino acid substitutions at the P-even positions (P14, P12, and P10) in the reactive center loop region was attributed to a retarded loop insertion (39 -43). It was FIG. 5. Effect of mutations on the formation of latent ␣ 1 AT. Conversion of the native ␣ 1 AT into the latent form was analyzed by urea gradient gel electrophoresis. The in vitro translation products of ␣ 1 AT labeled with [ 35 S]methionine were heated at 60°C for 4 h in 20 mM Tris, pH 7.4, and 0.7 M sodium citrate. After insoluble aggregates were removed by centrifugation, the sample buffer was exchanged using Centricon concentrator. The samples were analyzed by transverse urea gradient gel electrophoresis. The protein bands were visualized by autoradiography.
expected therefore that mutations that interfere with the loop insertion, like Multi-7 in this study, would affect the formation of the stable complex by retarding the loop insertion, thus inducing substrate-like behavior, even if binding proteinase was not affected. The Multi-7 ␣ 1 AT did not affect the inhibitory activity measured by inhibition kinetic experiments, nor did the Multi-7 mutation significantly influence the partitioning between the inhibitory pathway and the substrate pathway (Fig. 6). In addition, the stability of the complexes was not influenced by the Multi-7 mutation up to 72 h examined (data not shown). It was reported that the partitioning of C1 inhibitor between the inhibitory and substrate pathways are temperature-dependent; the substrate pathway is more favorable at low temperatures (44). This suggests that a kinetic step is involved in complex formation. We examined the complex formation at a low temperature (20°C) but could not detect any mutational effects, although more of the cleaved forms are produced at 20 than at 37°C in all cases (data not shown). These results suggest that enhanced A-sheet closure conferred by the stability increase of the Multi-7 molecule is not sufficient to influence the complex formation with target proteinases.
Implication on the Mechanism of the Complex Formation-It is very likely that the metastability of inhibitory serpins is utilized for the facile loop insertion during the complex formation with a target proteinase (9,(12)(13)(14). It was suggested that the drive toward a more stable state due to metastability results in trapping the proteinase-inhibitor complex as an acylenzyme inhibitor (13,14), possibly accompanying conformational change in proteinases also (45,46), including a distortion of the active site (47). We have constructed a variant ␣ 1 AT that is as stable as ovalbumin (⌬⌬G Ͼ 8 kcal/mol), but the mutational effect on the inhibitory activity is not comparable with the dramatic shift in stability. How can one reconcile such a paradox? It is possible that the drive stemmed from the hydrophobic core of the serpin molecule can indeed influence the conformational switch during the complex formation, but such a drive is only slightly diminished for Multi-7 ␣ 1 AT. The increase in stability by Multi-7 may be only a small fraction of much greater binding energy of the complex between a serpin inhibitor and a target proteinase, and consequently the stability increase of Multi-7 is not sufficient to influence the rate of loop insertion during the complex formation. Since the stoichiometry of inhibition is defined by (1 ϩ k substrate /k inhibitory ), the value is not sensitive to the changes of partitioning when one of the pathways is dominant (e.g. either when the value is close to 1 as with the interactions of many inhibitory serpins and their cognate target proteinases or when the value is very large as with ovalbumin). For instance, increasing the rate of the inhibitory pathway of ovalbumin significantly by hinge region mutations does not lead to detectable complex formation (48). There is, however, another class of mutations that particularly affects the stoichiometry of inhibition significantly. Many of the mutations in the reactive loop region such as G349P (22) or T345R (40) of ␣ 1 AT induce the substrate pathway more effectively, even if they may not increase the conformational stability as much as Multi-7. They may exert the mutational effect by interfering with the loop insertion more directly (e.g. blocking accessibility). None of the seven mutations in Multi-7 is located in the region directly involved in the loop insertion in the proposed model (10). It is worth examining if further increases in the stability of ␣ 1 AT by additional substitutions at the residues not directly involved in the loop insertion would affect the inhibitory activity.
Finally, our results suggest that target enzymes contribute to the complex formation with serpins. The mutational effect of Multi-7 on the inhibitory function was not manifested with the  2 and 6), 0.1 (lanes 3 and 7), 0.2 (lanes 4 and 8), and 0.4 (lanes 5 and 9). Conditions for the incubations are given under "Materials and Methods." The formation of the SDS-resistant ␣ 1 AT-elastase complex was analyzed on 10% SDS-PAGE. The protein bands were visualized by Coomassie Brilliant Blue staining. Lane 1 represents molecular mass standards (from the top, 200 kDa, 97.4 kDa, 68 kDa, 43 kDa, 29 kDa, and 18.4 kDa). The C and I indicate the migration position of the complex and ␣ 1 AT, respectively. B, stoichiometries of inhibition of proteinases by ␣1AT. Human leukocyte elastase (filled symbols) or porcine pancreatic elastase (open symbols) was incubated with increasing amounts of wild-type (q and E) and Multi-7 (f and Ⅺ) ␣1AT at 37°C for 10 min in the assay buffer (30 mM phosphate, 160 mM NaCl, 0.1% PEG6000, 0.1% Triton X-100, pH 7.4). The residual proteinase activity was measured with N-methoxysuccinyl-Ala-Ala-Pro-Val-p-nitroanilide as a substrate for human leukocyte elastase and with N-succinyl-Ala-Ala-Ala-p-nitroanilide for porcine pancreatic elastase. same target proteinase, but the same ␣ 1 AT exhibited different values of the stoichiometry of inhibition on binding human and porcine elastases (Fig. 6B: 1.1 versus 1.7). It was suggested that enzymes also play an important role in inducing conformational changes in inhibitory serpins (49). Many experimental results support that contact between serpins and proteinases is critical. For instance, the stoichiometry of inhibition of T345R (P14 mutant) ␣ 1 AT was quite different for human neutrophil elastase, porcine pancreatic elastase, and trypsin (40). Also G349P (P10 mutant) ␣ 1 AT conferred different effects on human leukocyte elastase and trypsin (22). The S380W variant antithrombin showed a similar result upon binding factor Xa, thrombin, and trypsin (50). These and our results clearly show that the same serpin molecule can confer a different effect depending on target proteinases, supporting the idea that part of the drive comes from the contact. In this context, it is worth noting a recent observation that all sequences from P5 to P16 including surface-exposed residues may be critical for the inhibitory activity of plasminogen activator inhibitor-1 (51). It may be that contact between a serpin and its target proteinase contributes significantly to the conformational change needed for the loop insertion.