Distribution of the Native Strain in Human a 1 -Antitrypsin and Its Association with Protease Inhibitor Function*

Serine protease inhibitors (serpins) are metastable in their native state. This strain, which is released upon binding to target proteases, is essential for the inhibitory activity of serpins. To understand the structural basis of the native strain, we previously characterized stabilizing mutations of a 1 -antitrypsin, a prototypical inhibitory serpin, in regions such as the hydrophobic core. The present study evaluates the effects of single point mutations throughout the molecule on stability and protease inhibitory activity. We identified stabilizing mutations in most secondary structures, suggesting that the native strain is distributed throughout the molecule. Examination of the substitution patterns and the structures of the mutation sites revealed surface hydrophobic pockets as a component of the native strain in a 1 -antitrypsin, in addition to the previously identified unusual interactions such as side chain overpacking and cavities. Interestingly, many of the stabilizing substitutions did not affect the inhibitory activity significantly. Those that affected the activity were confined in the regions that are mobilized during the complex formation with a target enzyme. The results of our study should be useful for designing proteins with strain and for regulating the stability and functions of serpins. Human a 1 -antitrypsin ( a 1 AT) serpin protease inhibitor) that a common tertiary structure composed of three -sheets and several -helices

Human ␣ 1 -antitrypsin (␣ 1 AT) 1 is a prototype of the serpin (serine protease inhibitor) superfamily that shares a common tertiary structure composed of three ␤-sheets and several ␣-helices (1). Serpins include protease inhibitors in blood plasma such as ␣ 1 AT, ␣ 1 -antichymotrypsin, antithrombin III, plasminogen activator inhibitor-I, C1 inhibitor, and ␣ 2 -antiplasmin, as well as non-inhibitory members such as ovalbumin and angiotensinogen (1,2). One salient feature of the inhibitory serpin structure is the strain in the native conformation (3)(4)(5)(6)(7), which is necessary for biological functions such as protease inhibition and ligand binding (1,2,8,9). The inhibition process of serpins can be described as a suicide substrate mechanism (10 -12) in which serpins, upon binding proteases, partition between cleaved serpins and stable serpin-enzyme complexes. The stoichiometry of inhibition (SI; the number of moles of inhibitor required to completely inhibit 1 mol of a target protease) is designated as 1 ϩ k substrate /k inhibition , in which k substrate is the rate constant for the substrate pathway toward the cleaved serpin and k inhibition is the rate constant for the inhibitory pathway toward the complex formation. For cognate target proteases, most serpin molecules partition into the complex formation, bringing the SI values close to 1. During the complex formation, the reactive center loop (RCL) of inhibitory serpins is cleaved (13)(14)(15) and inserted into the major ␤-sheet, A sheet, forming a stable complex between the serpin and the protease (11,12,(15)(16)(17). It has been suggested that the rate of loop insertion is critical for inhibitory function; retardation of the loop insertion would alter the partitioning between the inhibitory and substrate pathways in such a way that the SI values would increase (11,12,18,19). Release of the native strain of serpins, upon interaction with a protease, may regulate conformational changes such as loop insertion during the complex formation.
To elucidate the structural basis and functional roles of the native strain, we have previously characterized the effects of single amino acid substitutions of ␣ 1 AT, which increased the stability of the molecule presumably by releasing the native strain at the substitution sites (5,20). Our previous studies focused on the hydrophobic core (strands 4 -6 of B ␤-sheet and helix B) and the region that presumably accepts the inserting RCL during the inhibitory complex formation (strands 3 and 5 of A sheet, helix F, and the loop connecting helix F and strand 3 of A sheet). Various unfavorable interactions such as overpacking of side chains, buried polar groups, and cavities were suggested as the structural basis of the native strain in ␣ 1 AT (5,6,20).
In the present study, the mutational analyses were expanded over the entire molecule, in order to locate additional sites critical to its stability and activity. We excluded the RCL from the target region because mutations at almost every site in the RCL are likely to affect the inhibitory activity (18) without any relevance to the strain, as in the case of many genetic variations (21). Results from the current study revealed that the native strain of inhibitory serpins is distributed throughout the molecule. Interestingly, many of the stabilizing substitutions did not affect the inhibitory activity significantly. The activityregulating strain appears to be highly localized to those regions that are presumably mobilized during the complex formation with a target enzyme.

MATERIALS AND METHODS
Recombinant ␣ 1 AT Proteins-Plasmids for ␣ 1 AT expression in Escherichia coli and purification of recombinant ␣ 1 AT protein were described previously (22). Protein concentration was determined in 6 M * This work was supported by a National Creative Research Initiatives grant from the Korean Ministry of Science and Technology. 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 U.S.C. Section 1734 solely to indicate this fact.
Screening of Thermostable Mutants-Random mutagenesis at target regions of ␣ 1 AT cDNA was performed as described previously, using degenerative oligonucleotides (5,20,22). The randomly mutated clones were expressed in E. coli, and the recombinant cells were lysed with lysozyme and Triton X-100. The lysates were heated for 90 min at 60°C to inactivate unstable ␣ 1 AT activity. Mutant clones showing thermostable ␣ 1 AT activity were identified after PPE treatment. Representative clones obtained in the first screening were analyzed further by kinetic measurement of heat-induced activity loss of the lysates at 53°C. The residual inhibitory activity was detected by treatment with PPE. The clones that showed increased half-life in heat-induced inactivation were selected for further analyses. The mutation sites were identified by DNA sequencing. Most mutations were single amino acid substitutions, and many such substitutions were obtained more than once, suggesting that the mutations were saturated. Several mutants carried substitutions at more than one residue. To evaluate the mutational effect at each site, these multiple substitutions were separated to single amino acid substitutions. The substitutions at specific sites were generated by oligonucleotide-directed mutagenesis (24).
Urea-induced Equilibrium Unfolding-Equilibrium unfolding as a function of urea concentration was monitored by fluorescence spectroscopy at 25°C in 10 mM potassium phosphate, 50 mM NaCl, 1 mM EDTA, 1 mM ␤-mercaptoethanol, pH 6.5. Experimental data of the fluorescence measurement were fitted to a two-state unfolding model, details of which were described previously (5,20,22).
Determination of the Stoichiometry of Inhibition-The SI was determined as described (25). The active concentration of PPE was determined by measuring the initial rates of hydrolysis of 1 mM N-succinyl-(Ala) 3 -p-nitroanilide. The active concentration of HLE was determined as described previously (26) with trypsin-titrated human plasma ␣ 1 AT and a substrate, N-methoxysuccinyl-Ala-Ala-Pro-Val-p-nitroanilide. Various amounts of purified recombinant ␣ 1 AT proteins were incubated in 50 l of assay buffer (30 mM phosphate, 160 mM NaCl, 0.1% polyethylene glycol 8000, and 0.1% Triton X-100, pH 7.4) with 100 nM PPE or HLE at designated molar ratios of ␣ 1 AT to protease. After incubation with the protease at 37°C for 10 min, the reaction mixture was diluted 10-fold with the assay buffer and residual enzyme activity was determined (27).
Determination of Association Rate Constants-The association rate constant for the interaction of recombinant ␣ 1 AT with PPE was measured in a reaction mixture containing equimolar concentrations (8 nM) of the protease and the inhibitor (28).

RESULTS
Characterization of Thermostable Mutations of ␣ 1 AT-Thermostable mutations of ␣ 1 AT were obtained at 50 sites, which were distributed over most secondary structures of ␣ 1 AT ( Fig.  1) except strand 6 of A sheet (s6A), strands 1 and 3 of B sheet (s1B, s3B), and strands 3 of C sheet (s3C). These mutations were selected based on the enhanced kinetic stability against heat-induced inactivation (see "Materials and Methods"). The conformational stability of each mutant ␣ 1 AT was measured through equilibrium unfolding in urea by monitoring changes in intrinsic tryptophan fluorescence intensity (Fig. 2). The midpoint of the unfolding transition, C m , of the wild-type ␣ 1 AT in urea at 25°C was 1.8 M. Most of the thermostable mutations shifted the C m values to a higher urea concentration, which indicates that conformational stability was also increased. Changes in the free energy of stabilization by these substitutions are summarized in Table I. Results from the previous studies on the mutations at the hydrophobic core region and the RCL insertion site are also included. Half of the 50 substitutions documented in Table I increased the conformational stability more than 1 kcal mol Ϫ1 , while 11 of them increased the conformational stability less than 0.3 kcal mol Ϫ1 . To investigate the effect of the mutations on the inhibitory activity, the SI values toward PPE were measured ( Fig. 3; Table I). Most of the mutations did not affect the inhibitory activity signifi-FIG. 2. Urea-induced equilibrium unfolding of thermostable ␣ 1 AT mutants. The unfolding transition was measured by the increase in fluorescence emission intensity ( ex ϭ 280 nm, ex ϭ 360 nm). Samples were equilibrated in a urea solution containing 10 mM potassium phosphate, 50 mM NaCl, 1 mM EDTA, 1 mM ␤-mercaptoethanol, pH 6.5 at 25°C for 8 h. The protein concentration was 10 g/ml. q, wild-type; E, T114I; Ⅺ, A142V; ‚, R196S; ƒ, K335G; छ, P362L. cantly, causing the changes of less than 10% in SI (Fig. 1, blue  beads). However, several mutations increased the SI values significantly (Fig. 1, red beads; Table I,

bold numbers).
Characterization of the Activity-affecting Mutation Sites-Among the newly identified mutations that affected the inhibitory activity more than 10% (Table I, bold numbers), G117V increased the conformational stability substantially. This substitution mapped to the region that presumably is mobilized during the complex formation (Fig. 1). However, the other four mutations, R196S, Q212E, L245I, and P362L, increased the stability marginally. To understand the relationship between the stability increase and the activity loss at these sites, various other substitutions were introduced at each site. The effects of individual substitutions on the stability and the inhibitory activity toward both PPE and HLE are summarized in Table II. Except for the substitutions at Gln-212, which did not affect the inhibitory activity toward HLE, most of the substitutions caused the loss of activity toward both elastases. The activity losses by the mutations at Leu-245 and Pro-362 were substantial, but were not correlated with the stability increase at all. In order to examine if any of these substitutions affect protease binding, the second order association rate constant b One-letter codes for amino acids are used. Residues in front of numbers are the wild-type residues, and those following the numbers are the mutant residues. c From ⌬C m ϫ 3.1 (average m-value) in kcal mol Ϫ1 . The experimental errors are Ϯ0.16 kcal mol Ϫ1 . ⌬C m is the difference between the transition mid point of the mutant and that of the wild-type ␣ 1 AT in urea-induced equilibrium unfolding. d Relative activity against PPE is determined as the ratio of the wild-type SI value over the mutant value. The wild-type SI value was 1.67. Activity loss greater than 10% is indicated in bold numbers. The experimental errors are Ϯ0.07. e Stabilization modes: SP, filling surface pockets; S, size reduction at buried residues; (S), size reduction at exposed residues; C, cavity-filling; Po, removal of buried polar groups; I, ionic interactions.
was determined. As shown in Table II, the association rate constant (k a ) of the variant ␣ 1 AT with PPE did not differ significantly from the wild-type value, 5.5 ϫ 10 5 M Ϫ1 s Ϫ1 . The association rate with HLE was over 10 7 M Ϫ1 s Ϫ1 in all cases.

DISCUSSION
The present mutational studies addressed several specific questions about native strain in ␣ 1 AT. What parts of the molecule are strained? What is the structural basis of the native strain in ␣ 1 AT? Is the native strain at each identified site related to the functional regulation? Results from the current studies revealed that the native form of ␣ 1 AT is suboptimally folded and the strain is distributed throughout the molecule. However, the strain directly regulating the protease inhibitory activity appears to be highly localized.

Modes of Destabilization in the Native Form of ␣ 1 AT
Surface Pockets-Examination of the wild-type ␣ 1 AT crystal structure (29) revealed that many of the newly identified substitutions map at surface hydrophobic pockets. For instance, the side chain of Ser-36 on helix A is close to the surface pocket formed by Leu-306 of helix I and Asn-81 of the loop connecting helix C and helix D (Fig. 4). Substitution to bulky arginine at the 36 site (S36R) may provide better packing interactions in the pocket. Another mutation site in helix A, Asn-29, is very close to Phe-82 of the loop connecting helix C and helix D (Fig.  4A). The substitution to tyrosine (N29Y) may provide good aromatic-aromatic interactions with Phe-82. Several other substitutions appear to permit better packing interactions at surface pockets, which are indicated in Table I (denoted as SP). A surface hydrophobic pocket near T114I formed between helix D and strand 2 of A sheet was also identified previously by structural examination (29). The T114I substitution increased the conformational stability by 2 kcal mol Ϫ1 , apparently due to filling of the pocket by the mutation. Our results suggest strongly that the surface pockets are part of the structural design for the metastable native structure of ␣ 1 AT.
Overpacking and Cavities-Previous sets of thermostable mutations of ␣ 1 AT (marked in Table I) showed that size reduction or cavity filling (denoted as S and C in Table I, respectively) inside the molecule increased the conformational stability (5,20). This was surprising because size reduction inside the proteins usually makes the protein unstable (30,31), and cavities other than active-site pockets are rarely observed (32). The substitution patterns observed in the new set also showed size reduction at several sites (Table I). Size reduction at the buried sites (for example T27S, A34G, L241I, etc.) may provide better packing by allowing rearrangement of the interlocked side chains. The mutants carrying the K163T, K168I, K174T, E199D, and L245Q substitutions also show the size reduction. These sites are highly exposed, and many of the substitutions show a marginal stabilizing effect. The stabilizing mechanism at these sites is not clear, and these substitutions are accordingly denoted as "(S)" in Table I. Several substitutions in the new set, such as A31V, G117V, A183V, A142V, A248V, and V364I, showed size increase rather than size reduction ( Table I). Examination of these sites in the crystal structure (29) of ␣ 1 AT revealed the existence of cavities nearby. Cavities are a likely source of energetic cost in conformational stability (33), and the stabilizing substitutions appear to provide better packing by filling the cavities.
Other Destabilizing Mechanisms-Previous studies of the hydrophobic core mutations (5, 6) suggested that certain buried polar groups provide another structural feature for the native strain in ␣ 1 AT (designated as Po in Table I) because they do not participate in hydrogen bonding. It has been shown that unpaired polar groups cause energetic cost (34). Creation of a polar network may be another structural mechanism of the native strain in ␣ 1 AT, as suggested by the R196S mutation. Arg-196, located beneath the RCL, forms a positively charged network with Arg-223 and Arg-281 (7,29). The R196S substitution might relieve the charge repulsion at the site. Several substitutions appear to create salt bridge interactions (Table I, denoted as I). These surface substitutions marginally increased the conformational stability (Ͻ0.9 kcal mol Ϫ1 ). It is not clear,  however, if any of these sites in the wild-type protein contribute to the native strain.

Localization of the Strain That Regulates Inhibitory Function
Most of the activity-affecting mutations reside either near the RCL or at the loop insertion site on A sheet (Fig. 1, colored  green). If the strain is relevant to inhibitory function, increasing the stability by releasing the strain should decrease the serpin's inhibitory activity. This appears to be the case for the previously identified mutations at Lys-335 and Lys-331, which showed increased conformational stability with a concomitant decrease in the inhibitory activity (20). In the crystal structure of the native ␣ 1 AT (Fig. 1), these residues are located in strand 5 of A sheet (s5A) and interact with the residues in strand 3 of A sheet (s3A), helix F (hF), and the adjacent loop connecting strand 3 of A sheet and helix F. Among the newly identified sites where stabilizing substitutions reduced the activity, Gly-117 maps in this region. Gly-117 interacts with Tyr-160 of hF and Val-185 in strand 3 of A sheet. It is very likely that the residue interactions in this region must be mobilized to accept the RCL during the complex formation with a target protease. Another site where the native strain may promote the inhibitory activity is Arg-196. Arg-196 has been suggested to form a positively charged pocket near the RCL (7,29). Such an unfavorable interaction can drive the conformational change during the complex formation with a target protease. Indeed, there is a modest correlation between activity loss and stability increase among the substitutions at Arg-196 (Table II). The strain at the identified sites may facilitate the conformational change.
None of the stabilizing mutations in the other regions of ␣ 1 AT appeared to decrease the inhibitory activity (Table I). Filling the surface hydrophobic pocket between strand 2 of A sheet and helix D was predicted to inhibit the transition to a more stable conformation (29). However, T114I, which increased the stability substantially (2 kcal mol Ϫ1 ), presumably by filling the predicted pocket, failed to decrease the inhibitory activity significantly (Table I). These results suggest that only the strain in critical regions of ␣ 1 AT, such as the regions that are directly involved in the complex formation with a target enzyme, regulates the serpin's inhibitory function.

Mutations Affecting Inhibitory Activity without Stability Increase
Thermostable mutations at some sites profoundly affected the inhibitory activity without any associated stability increase. Two such sites (Lys-168 and Ile-169) were previously identified (20), and three sites (Gln-212, Leu-245, and Pro-362) were identified in the present study. Various other substitu-tions at the three sites, Gln-212, Leu-245, and Pro-362, also reduced the inhibitory activity without any associated stability increase (Table II). The mutations do not seem to affect the binding of a target protease, because they have only marginal effects on the rate constant for association (k a ) with PPE (Table  II). It was not surprising that the substitutions at Pro-362 affected the inhibitory activity, because Pro-362 is located at the end of RCL ( Fig. 1; P4Ј position, 4 residues away from the active site residue, Met-358), at the site where RCL kinks and joins C ␤-sheet (Fig. 1). This proline may help maintain the rigidity of the RCL needed to maximize the strain loaded on the RCL, which may be directly utilized for the distortion of the protease active site (35). Leu-245 is exposed at the end of strand 2 of B sheet (Fig. 1, s2B), and appears to fix the proximal hinge region of RCL near Thr-345 (6,29). The importance of this hinge region of the RCL has been implied by many genetic mutations in this region of inhibitory serpins (21). Gln-212 is exposed at the end of strand 4 of C sheet and is far away from the RCL. It is not clear how the substitutions at Gln-212 affect the activity. Further studies must define the precise roles of these residues in the inhibitory mechanism.

Conclusion
Our examination of thermostable mutations of ␣ 1 AT showed that the native strain is scattered throughout the molecule. The structural instabilities in the native form may well be mobilized during the complex formation with a target protease. Interestingly, however, the destabilization component for regulating the inhibitory function of ␣ 1 AT is highly localized ( Fig.  1 and Table I). Our approach has some potential limitations. Since the mutagenesis method adopted in this study is highly likely to yield only the residue substitutions induced by single nucleotide change (36), we might have missed some important sites where critical substitutions require more than a single nucleotide change. Despite these limitations, the current mutational analyses clearly provide us perspectives on the stability-function relationship of ␣ 1 AT by revealing the distribution and structural mechanism of the native strain in ␣ 1 AT. Information obtained from this study will be a valuable guide in regulating the stability and function of serpins, including developing stable therapeutic serpins without activity compensation.