Targeting a Surface Cavity of α1-Antitrypsin to Prevent Conformational Disease*

Conformational diseases are caused by a structural rearrangement within a protein that results in aberrant intermolecular linkage and tissue deposition. This is typified by the polymers that form with the Z deficiency variant of α1-antitrypsin (Glu-342 → Lys). These polymers are retained within hepatocytes to form inclusions that are associated with hepatitis, cirrhosis, and hepatocellular carcinoma. We have assessed a surface hydrophobic cavity in α1-antitrypsin as a potential target for rational drug design in order to prevent polymer formation and the associated liver disease. The introduction of either Thr-114 → Phe or Gly-117 → Phe on strand 2 of β-sheet A within this cavity significantly raised the melting temperature and retarded polymer formation. Conversely, Leu-100 → Phe on helix D accelerated polymer formation, but this effect was abrogated by the addition of Thr-114 → Phe. None of these mutations affected the inhibitory activity of α1-antitrypsin. The importance of these observations was underscored by the finding that the Thr-114 → Phe mutation reduced polymer formation and increased the secretion of Z α1-antitrypsin from a Xenopus oocyte expression system. Moreover cysteine mutants within the hydrophobic pocket were able to bind a range of fluorophores illustrating the accessibility of the cavity to external agents. These results demonstrate the importance of this cavity as a site for drug design to ameliorate polymerization and prevent the associated conformational disease.

Conformational diseases arise when a protein undergoes a change in size or a fluctuation in shape, which results in self-association and tissue deposition (1). This process is now recognized to underlie a whole range of diseases including the amyloidoses, prion encephalopathies, glutamine repeat diseases, and Alzheimer's and Parkinson's disease (2). The paradigm for the conformational diseases was provided by the serpinopathies, which result from mutations in members of the serine proteinase inhibitor or serpin superfamily. The most well characterized of these is the severe plasma deficiency that is associated with the Z allele of ␣ 1 -antitrypsin (3).
␣ 1 -Antitrypsin is the most abundant circulating proteinase inhibitor and the archetypal member of the serpin superfamily (4,5). Most individuals have two M ␣ 1 -antitrypsin alleles, but ϳ1 in 2000 are homozygous for the Z variant. The Z mutation results from a Glu 3 Lys substitution at amino acid 342 (6) and leads to retention of ␣ 1 -antitrypsin as inclusion bodies within the hepatocyte. These inclusions predispose to neonatal hepatitis, juvenile cirrhosis, and adult hepatocellular carcinoma (7)(8)(9). The resulting secretory defect accounts for the low circulating plasma level of ␣ 1 -antitrypsin, which is only 15% of normal in the Z homozygote. This plasma deficiency exposes the lungs to uncontrolled proteolytic attack that in turn causes early onset panacinar emphysema particularly in Z ␣ 1 -antitrypsin homozygotes who smoke (10).
The structure of ␣ 1 -antitrypsin is based on a five-stranded ␤-sheet A and a mobile reactive center loop (11)(12)(13). Our previous studies have shown that the Z mutation promotes opening of ␤-sheet A to facilitate a sequential interaction between the reactive center loop of one molecule and ␤-sheet A of a second, resulting in polymer formation (3, 14 -16). These polymers tangle within the rough endoplasmic reticulum of hepatocytes to form the periodic acid-Schiff-positive inclusions that are associated with liver disease (3,17). The significance of the reactive loop-␤-sheet linkage was underscored by two other ␣ 1 -antitrypsin variants, Siiyama (Ser-53 3 Phe) and Mmalton (⌬52 3 Phe) that also resulted in hepatic inclusions and severe plasma deficiency of ␣ 1 -antitrypsin. Both of these mutants spontaneously formed polymers in vivo (18,19). Moreover, this linkage accounts for the mild plasma deficiency observed with both S (Glu-264 3 Val) and I (Arg-39 3 Cys) ␣ 1 -antitrypsin (20,21).
Further support for polymer formation as the mechanism responsible for the retention of mutant ␣ 1 -antitrypsin within hepatocytes came from studies utilizing the Xenopus oocyte expression system. Point mutations that attenuated polymerization of Z ␣ 1 -antitrypsin in vitro (14,22) increased the secretion of Z ␣ 1 -antitrypsin in vivo (23). Our understanding of the mechanism underlying polymerization has allowed the design of strategies to prevent polymer formation (3,16). To date, however, these have been based on peptides that bind to ␤-sheet A and as a consequence inactivate ␣ 1 -antitrypsin as a proteinase inhibitor. A more useful strategy would be to identify cavities in ␣ 1 -antitrypsin that can bind peptides, or their mimetics, and block polymerization without a loss of inhibitory activity.
Our high resolution crystal structure of ␣ 1 -antitrypsin revealed a large hydrophobic cavity bounded by strand 2 of ␤-sheet A and helices D and E (Fig. 1). The cavity is present in monomeric ␣ 1 -antitrypsin but is obliterated during polymerization (24). This cavity could provide an ideal target for drug design to prevent polymer formation and the associated liver disease (11,13,24). We have used site-directed mutagenesis to explore the role of this surface cavity in the conformational transitions of ␣ 1 -antitrypsin in vitro and in vivo.

EXPERIMENTAL PROCEDURES
Mutagenesis, Expression, and Purification of Recombinant ␣ 1 -Antitrypsin-Pittsburgh ␣ 1 -antitrypsin (Met-358 3 Arg) was used as the wild type protein. Replacement of the P 1 methionine to arginine renders the protein a potent inhibitor of thrombin rather than neutrophil elastase. It otherwise has the same biophysical properties and rate of polymerization as Met-358 ␣ 1 -antitrypsin (15,25). The ␣ 1 -antitrypsin mutants (Leu-100 3 Phe, Thr-114 3 Phe, Gly-117 3 Phe, Leu-100 3 Phe/Thr-114 3 Phe, Leu-100 3 Cys/Cys-232 3 Ser, and Thr-114 3 Cys/Cys-232 3 Ser) were prepared by site-directed mutagenesis, and the sequences were confirmed by automated DNA sequence analysis. The ␣ 1 -antitrypsin variants were cloned into the pET16b plasmid and transformed into BL21(DE3) Escherichia coli, and expression was induced with 0.4 mM isopropyl-␤-D-thiogalactopyranoside. Recombinant ␣ 1 -antitrypsin was extracted from the crude E. coli extract by Q-, then zinc chelating-, and finally glutathione-Sepharose chromatography (11). However this method was not efficient for purification of the cysteine cavity mutants. Therefore, the ␣ 1 -antitrypsin cavity mutants were also expressed in the pQE31 vector that contained an aminoterminal MRSHHHHHH tag. Recombinant proteins were then purified from the soluble fraction of E. coli lysate by HiTrap Ni-chelating and Q-Sepharose column chromatography as detailed previously (26). The proteins were dialyzed into 50 mM Tris, 50 mM KCl, pH 7.4, and purity was confirmed by 12% (w/v) SDS-PAGE.
Characterization of Recombinant Wild Type and Mutant ␣ 1 -Antitrypsin-The recombinant proteins were characterized by non-denaturing and 0 -8 M transverse urea gradient PAGE. Inhibitory activity was calculated by incubating bovine ␣-chymotrypsin (5 pmol) of known active site (27) with increasing concentrations of ␣ 1 -antitrypsin (estimated active site concentration of 0.1 M) in a total volume of 100 l with 0.03 M sodium phosphate, 0.16 M NaCl, 0.1% (w/v) PEG 4000, pH 7.4, reaction buffer. The reaction proceeded for 10 min at room temperature, and the residual proteolytic activity was determined by the addition of the substrate succinyl-L-alanyl-L-alanyl-propyl-L-phenylanalyl-p-nitroanilide to a final concentration of 0.16 mM (18). The change in the A 405 over 3 min was observed. Active site values were obtained by plotting residual proteolytic activity against the volume of ␣ 1 -antitrypsin and extrapolating to the x intercept (28). Binary complexes were formed by incubating 50 -100-fold molar excess of the antithrombin 12-mer peptide (P 14 -3 ; Ac-Ser-Glu-Ala-Ala-Ala-Ser-Thr-Ala-Val-Val-Ile-Ala-OH) or ␣ 1 -antitrypsin 6-mer peptide (P 7-2Ј ; Ac-Phe-Leu-Glu-Ala-Ile-Gly-OH) with each ␣ 1 -antitrypsin variant at 0.5 mg/ml in 50 mM Tris, 50 mM KCl, pH 7.4, at 37°C for up to 48 h. Samples at different time points were assessed on a 7.5% (w/v) non-denaturing gel containing 8 M urea (16). All proteins were visualized by Coomassie Blue or silver staining. The melting temperature and far ultraviolet (250 -195 nm) CD spectrum were obtained for each ␣ 1 -antitrypsin mutant as described previously (14) Assessment of Polymerization of Recombinant Wild Type and Mutant ␣ 1 -Antitrypsin-Polymer formation was assessed by incubating each of the recombinant ␣ 1 -antitrypsin variants at 0.1 mg/ml in 50 mM Tris, 50 mM KCl, pH 7.4, at 52°C. The samples were then separated by 7.5% (w/v) non-denaturing PAGE, and the protein was visualized by silver staining. Loss of intensity of the monomeric protein band was determined by densitometry (Quantity One, Bio-Rad). The half-life for polymer formation was calculated from the semi-log plot of the ln fractional loss against time in seconds.
Assessment of ␣ 1 -Antitrypsin Secretion from the Xenopus Oocyte-The cavity mutants Leu-100 3 Phe, Thr-114 3 Phe, and Gly-117 3 Phe were inserted into the sp64T plasmid containing either M or Z ␣ 1 -antitrypsin by site-directed mutagenesis, and the sequences were confirmed as before. The in vitro transcription and assessment of ␣ 1antitrypsin secretion from the Xenopus oocyte were undertaken as described previously (23).
Fluorophore Labeling of Recombinant Wild Type and Mutant ␣ 1 -Antitrypsin-The cysteine cavity mutants were labeled with a 20-fold molar excess of either tetramethylrhodamine-5-iodoacetamide (5-TM-RIA), 1 at both 20 and 37°C, for up to 72 h at pH 7.4 according to the manufacturer's instructions (Molecular Probes Inc., Eugene, OR). The reaction was terminated by the addition of 1 l of 14.3 M ␤-mercaptoethanol, and the labeled protein was separated from excess label on a NAP-10 gel filtration column equilibrated in 50 mM Tris, 50 mM KCl, pH 7.4.

RESULTS
Three residues of ␣ 1 -antitrypsin were selected for site-directed mutagenesis in order to explore the role of the cavity in polymer formation. Leu-100 on hD and Thr-114 on s2A have side chains that point into the hydrophobic pocket, whereas Gly-117 is located at the base of the cavity on s2A (Fig. 1B). Introducing large phenylalanine residues, or cysteine residues that could be labeled with bulky fluorophores, at these sites was predicted to fill the cavity and mimic the effect of binding a small molecule inhibitor. The one naturally occurring cysteine at position 232 in ␣ 1 -antitrypsin was replaced by a serine residue to ensure that only the newly introduced cavity cysteine was available for labeling. The Cys-232 3 Ser mutation has no effect on the inhibitory activity or polymerization of recombinant ␣ 1 -antitrypsin (15,29,30).
Purification of Cavity Mutants-Wild type, Leu-100 3 Cys/ Cys-232 3 Ser, Thr-114 3 Cys/Cys-232 3 Ser, and Gly-117 3 Phe ␣ 1 -antitrypsin were purified from the supernatant following lysis of the E. coli (11). However, this approach relies on glutathione affinity chromatography, and the ␣ 1 -antitrypsin cavity mutants Leu-100 3 Cys/Cys-232 3 Ser and Thr-114 3 Cys/Cys-232 3 Ser were unable to bind to the glutathione resin. Wild type, Leu-100 3 Cys/Cys-232 3 Ser, Thr-114 3 Cys/Cys-232 3 Ser, Leu-100 3 Phe, Thr-114 3 Phe, Gly-117 3 Phe, and Leu-100 3 Phe/Thr-114 3 Phe ␣ 1 -antitrypsin were therefore cloned into an expression vector containing the MR-SHHHHHH tag at the amino terminus. They were expressed in E. coli and purified to homogeneity by Ni-chelating-and Q-Sepharose column chromatography. Subsequent experiments were performed with purified recombinant ␣ 1 -antitrypsin variants containing an amino-terminal His tag. All the recombinant proteins migrated as a characteristic doublet on 7.5% (w/v) non-denaturing PAGE and as a single band on 12% (w/v) SDS-PAGE apart from Leu-100 3 Phe/Thr-114 3 Phe ␣ 1antitrypsin which contained a minor contaminant (data not shown). All the mutants had an unfolding profile that was similar to wild type ␣ 1 -antitrypsin on a 0 -8 M transverse urea gradient gel with the exception of Gly-117 3 Phe, which unfolded at a higher urea concentration (ϳ4 M) indicating increased stability (data not shown). Finally the far UV CD spectrum of all the variants was similar to that of wild type ␣ 1 -antitrypsin confirming that the mutations did not cause a significant perturbation in the overall structure of the molecule (data not shown).
Thermal Stability of Recombinant Wild Type and Mutant ␣ 1 -Antitrypsin-The melting temperature (T m ) of wild type and mutant ␣ 1 -antitrypsin was examined by circular dichroic spectroscopy. The amino-terminal histidine tag increased the melting temperature of all the recombinant ␣ 1 -antitrypsin variants by 6°C when compared with the proteins purified from E. coli lysate that lacked the histidine tag (data not shown). All the mutations introduced into strand 2A elevated the T m of the protein. In particular, Gly-117 3 Phe ␣ 1 -antitrypsin had a melting temperature that was more than 8°C higher than the wild type protein. Helix D mutations had differing effects according to their size. Leu-100 3 Phe lowered the melting temperature by 3.5°C, whereas a cysteine residue at the same position did not significantly alter the T m when compared with the wild type control. Furthermore, the addition of a second phenylalanine residue within the cavity at position 114 (Leu-100 3 Phe/Thr-114 3 Phe) resulted in an increase in thermal stability, reversing the effect of Leu-100 3 Phe mutation alone (Table I). Thermal stability was also assessed by incubating recombinant wild type or cavity mutants of ␣ 1 -antitrypsin between 30 and 100°C, at increments of 10°C for 15 min, and assessing the samples by 7.5% (w/v) non-denaturing PAGE. Leu-100 3 Cys/Cys-232 3 Ser and Leu-100 3 Phe ␣ 1 -antitrypsin had thermal stabilities similar to that of wild type protein. However, Gly-117 3 Phe ␣ 1 -antitrypsin was the most thermostable as it remained monomeric following incubation at 60°C for 15 min, which was 10°C higher than wild type ␣ 1antitrypsin. Thr-114 3 Cys/Cys-232 3 Ser, Thr-114 3 Phe, and Leu-100 3 Phe/Thr-114 3 Phe ␣ 1 -antitrypsin all had intermediate thermal stabilities (data not shown). Thus, the differences in melting temperatures were mirrored in the thermal stability of wild type and mutant ␣ 1 -antitrypsin when assessed by heating and separation on non-denaturing PAGE.
Polymerization of Recombinant ␣ 1 -Antitrypsin Variants-Polymerization was assessed at 0.1 mg/ml and 52°C for up to 7 days as these conditions led to polymer formation of histidinetagged recombinant ␣ 1 -antitrypsin that could be visualized by non-denaturing PAGE. The rate of polymer formation was calculated from the loss of intensity of the monomeric protein band using densitometry scanning (Table I). Wild type ␣ 1antitrypsin almost completely polymerized within 24 h when heated at 52°C (Fig. 3a). Leu-100 3 Phe ␣ 1 -antitrypsin accelerated polymer formation in keeping with its lower melting temperature (Fig. 3b). However, replacing Leu-100 with a cysteine residue (Fig. 3c) or introducing another bulky phenylalanine residue at position 114 (Fig. 3d) within the cavity reversed this effect as these mutants polymerized at a rate similar to wild type ␣ 1 -antitrypsin (Table I). Interestingly, all the mutations introduced on to s2A, independent of size, slowed polymer formation as would be predicted from their melting temperatures (Fig. 3, e-g) ( Table I). The most thermostable mutant Gly-117 3 Phe ␣ 1 -antitrypsin dramatically impeded polymer formation, as polymers were evident only after incubating at 52°C for 72 h.
Binary Complex Formation between Recombinant Wild Type and Mutant ␣ 1 -Antitrypsin and Exogenous Reactive Loop Peptides-A 12-mer peptide corresponding to the reactive center loop of antithrombin (P 14 -3 ) was used to assess the patency of ␤-sheet A. Binary complexes were formed by incubating recombinant ␣ 1 -antitrypsin (0.5 mg/ml) with a 50 -100-fold molar excess of amino-terminal acetylated 12-mer peptide at 37°C for 48 h. Samples were examined on 7.5% (w/v) non-denaturing PAGE containing 8 M urea. Recombinant wild type ␣ 1 -antitrypsin and the cavity mutants all formed a binary complex with the 12-mer peptide with a 1:1 stoichiometry (Fig. 4, a-g). Gly-117 3 Phe ␣ 1 -antitrypsin formed a binary complex with the peptide at a rate faster than wild type ␣ 1 -antitrypsin (Table II  and Fig. 4b), whereas binary complex formation was significantly retarded by Leu-100 3 Phe ␣ 1 -antitrypsin (Fig. 4c). The other cavity mutations all similarly slowed annealing of the peptide to ␤-sheet A. Neither recombinant wild type nor Leu-100 3 Phe ␣ 1 -antitrypsin was able to form a binary complex with the 6-mer peptide, corresponding to P 7-2 of the reactive loop of ␣ 1 -antitrypsin, under the same conditions at 24 h (data not shown).
Secretion of Recombinant ␣ 1 -Antitrypsin from the Xenopus Oocyte Expression System-The effects of the mutants were then assessed on the polymerization of the Z variant of ␣ 1antitrypsin. This mutant is too unstable to be expressed as a recombinant protein, and the mutants were therefore assessed for their effect on the secretion of Z ␣ 1 -antitrypsin in vivo. 62% (S.E. Ϯ 4%) of the wild type protein was secreted from the FIG. 1. A, 2-Å crystal structure of monomeric ␣ 1 -antitrypsin illustrating the mobile reactive loop (red) and the ␤-sheet A (green). The hydrophobic surface cavity of interest (arrow) is bounded by strand 2 of ␤-sheet A (s2A), helix D (hD) and helix E (hE). This area is obliterated during conformational transitions that involve reactive loop insertion into ␤-sheet A as demonstrated by the cleaved conformation (24). B is a model of the interior of the hydrophobic cavity displaying the position of the amino acid side chains (blue). The residues Leu-100 on hD and Thr-114 and Gly-117 on s2A were chosen as sites to introduce cavityfilling mutations. oocytes compared with 10% (Ϯ2%) of Z ␣ 1 -antitrypsin (p ϭ 0.0001, Student's t test with Welch correction). Gly-117 3 Phe and Leu-100 3 Phe had little effect on the secretion of Z ␣ 1 -antitrypsin (17 Ϯ 5 and 18 Ϯ 5%, respectively). However Thr-114 3 Phe more than doubled the secretion of Z ␣ 1 -antitrypsin to 23 Ϯ 4% (p ϭ 0.0018 compared with Z ␣ 1 -antitrypsin). The results are the mean of 5-9 separate experiments.
Fluorophore Labeling of Recombinant ␣ 1 -Antitrypsin Cysteine Cavity Mutants-The accessibility of the cavity was examined by labeling the cysteine variants Leu-100 3 Cys/Cys-232 3 Ser and Thr-114 3 Cys/Cys-232 3 Ser ␣ 1 -antitrypsin with a number of fluorophores having different length side chains (Table III). The labeling reactions were performed in the dark at either 20 or 37°C for up to 72 h. Incubation of Leu-100 3 Cys/Cys-232 3 Ser ␣ 1 -antitrypsin with 5-TMRIA at 20°C resulted in 14, 24, 20, and 20% labeling when incubated for 12, 24, 48, and 72 h, respectively. Likewise, incubation of Thr-114 3 Cys/Cys-232 3 Ser ␣ 1 -antitrypsin with 5-TMRIA at 20°C resulted in 15, 20, 26, and 24% labeling when incubated for 12, 24, 48, and 72 h, respectively. These results imply that maximal labeling of the cavity cysteine residues with 5-TMRIA was achieved within 24 h. Other fluorescent probes were assessed for their ability to label the cysteine variants in an attempt to improve the labeling efficiency (Table III). The addition of the reducing agent tris-(2-carboxyethyl)phosphine and raising the reaction temperature both increased the amount of protein labeled with 5-IAF but promoted labeling of other susceptible residues (histidines, methionines, and lysines) as evidenced by multiple bands on non-denaturing PAGE (data not shown). This experimental artifact could be overcome by raising the pH to 8.5 and limiting the reaction time to 2 h at 37°C. With this method 35% of Leu-100 3 Cys/Cys-232 3 Ser ␣ 1antitrypsin and 28% of Thr-114 3 Cys/Cys-232 3 Ser ␣ 1antitrypsin were labeled with 5-IAF (Table III). As only approx- a This represents the cysteine variants maximally labeled with 5-IAF in 50 mM Tris, pH 8.5, at 37°C for 2 h. The inhibitory activity was determined against bovine ␣-chymotrypsin, and melting temperature was calculated from circular dichroic spectrum analysis at 222 nm. Polymer formation was assessed by incubating each ␣ 1 -antitrypsin variant at 0.1 mg/ml and 52°C for 24 h. The t1 ⁄2 was calculated from the loss of intensity of the monomeric protein band using densitometry scanning. The results are the mean Ϯ S.D. of three measurements. Aliquots were removed and snap-frozen prior to assessment on 7.5% (w/v) non-denaturing PAGE. All lanes contain 4 g of ␣ 1 -antitrypsin. N represents native ␣ 1 -antitrypsin and P polymers of ␣ 1 -antitrypsin.
imately a third of the protein labeled with fluorophore, it was difficult to interpret the effect of these agents on the rate of polymerization.

DISCUSSION
Our high resolution crystal structure of recombinant M ␣ 1antitrypsin demonstrated a hydrophobic cavity bounded by s2A, hD, and hE that was present in the monomeric structure but predicted to reduce in size by Ͼ70% during polymer formation (24). As such this cavity provides a target for rational drug design to prevent polymerization and ameliorate the associated disease. In order to explore the role of the cavity in polymer formation, three residues, whose side chains border the cavity, were selected for site-directed mutagenesis (Fig. 1B). Introducing large phenylalanine residues at Leu-100 on hD and Thr-114 and Gly-117 on s2A is likely to fill the cavity. A detailed assessment has been undertaken to determine the effect of these mutations on polymer formation in order to determine whether the hydrophobic cavity would be a suitable target for rational drug design.
All the cavity mutants had a normal far UV circular dichroic spectrum, were active as proteinase inhibitors, and formed SDS-stable complexes. They varied in specific activity when assessed against bovine ␣-chymotrypsin indicating differing stoichiometry of inhibition, but overall the data show that the point mutations did not lead to a significant change in the structure of ␣ 1 -antitrypsin. All of the mutations introduced onto s2A elevated melting temperature and significantly slowed the rate at which ␣ 1 -antitrypsin formed loop-sheet polymers, particularly the introduction of bulky phenylalanine residues at either position 114 or 117. These observations are in keeping with our previous studies (14). Moreover, the addition of a phenylalanine at position 114 restored the rate of Leu-100 3 Phe ␣ 1 -antitrypsin polymerization to that of the wild type protein. Furthermore, the s2A mutations all increased thermal stability of ␣ 1 -antitrypsin which is in accordance with their effect on polymerization. These data provide strong evidence that filling the cavity with mutants on s2A stabilizes ␣ 1antitrypsin and retards polymer formation without compromising inhibitory function in vitro.
Polymerization results from the sequential insertion of the reactive center loop of one molecule into ␤-sheet A of another (15). Point mutations that favor polymerization are predicted to open ␤-sheet A to facilitate incorporation of an exogenous reactive center loop (31). The accessibility of ␤-sheet A was assessed in the mutants by measuring the rate at which they annealed an exogenous reactive loop peptide to form a binary complex (32). Gly-117 3 Phe ␣ 1 -antitrypsin formed a binary complex with the peptide at a rate faster than wild type protein. This was unexpected as the mutant significantly retarded polymer formation. Moreover, the most polymerogenic mutant, Leu-100 3 Phe ␣ 1 -antitrypsin, formed a binary complex with

TABLE III
The labeling efficiency of cysteine cavity mutants of ␣ 1 -antitrypsin with different fluorophores The fluorophores were incubated with recombinant ␣ 1 -antitrypsin at 20 or 37°C between 2 and 16 h in the dark. Labeled protein was separated from excess fluorophore by gel filtration, and the labeling efficiency was calculated spectrophotometrically. ND, not done. The abbreviations used are as follows: FL IA, 4,4-difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-sindacene-3-propionyl)-NЈ-iodoacetylethylenediamine; FL C 1 -IA, N-(4,4difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-s-iodoacetylethylenediamine; TCEP, tris-(2-carboxyethyl)phosphine. the peptide at one of the slowest rates (Table II and Fig. 4). Such dichotomy between polymer formation and speed at which a mutant of ␣ 1 -antitrypsin accepts exogenous reactive loop peptides has also been observed for the naturally occurring Z and Mmalton deficiency variants (19). Indeed the explanation for Z ␣ 1 -antitrypsin has become apparent recently (16). The Z mutation lies at that head of strand 5 of ␤-sheet A and the base of the reactive center loop, where it opens ␤-sheet A to allow partial incorporation of its own reactive loop. This fills the upper portion of ␤-sheet A and consequently retards admission of an exogenous reactive loop 12-mer peptide. However, the lower part of ␤-sheet A remains patent, and this permits the Z variant to accept either an exogenous loop to form polymers or a 6-mer peptide that is homologous to the P 7-2 residues of the reactive loop (16). The effect of the Leu-100 3 Phe mutation is likely to be similar to that of Z ␣ 1 -antitrypsin. Leu-100 3 Phe retards insertion of the exogenous reactive loop peptide by restricting the opening of the top of ␤-sheet A. The lower part of the A sheet must remain accessible to account for the accelerated rate of polymerization. Nevertheless, unlike Z ␣ 1 -antitrypsin, it does not accept the 6-mer peptide, which is likely be due to the more distal location of the Leu-100 3 Phe mutation in ␤-sheet A. Conversely the Gly-117 3 Phe mutation almost certainly closes the lower portion of ␤-sheet A to explain its slowing effect on polymer formation, which leaves the top of the sheet readily available for insertion of an exogenous reactive loop peptide. Surface cavities contribute to the metastability of ␣ 1 -antitrypsin that is essential for its inhibitory function (33). Although filling of these cavities increases thermal stability, it is often associated with a loss of inhibitory activity (34 -37). Our results show that filling specific surface cavities can stabilize the molecule, attenuate polymerization, and yet still retain inhibitory activity. We have shown that the Z variant of ␣ 1antitrypsin adopts a different conformation from the wild type protein (16). Thus, the Z mutation itself may distort the structure of the hydrophobic pocket that has been selected as a target for drug design. This cannot be assessed in vitro, as Z ␣ 1 -antitrypsin is too unstable to be prepared as a recombinant protein in E. coli. To overcome this problem the effect of the mutants on Z ␣ 1 -antitrypsin was characterized in vivo using the Xenopus oocyte expression system. This system reproduces the way in which hepatocytes handle mutants of ␣ 1 -antitrypsin (19,23,38). The importance of these studies is highlighted by the Gly-117 3 Phe mutation, which markedly slowed the polymerization of wild type M ␣ 1 -antitrypsin in vitro without affecting the secretion of Z ␣ 1 -antitrypsin in vivo. However, the Thr-114 3 Phe mutant, which also slowed the polymerization of wild type M ␣ 1 -antitrypsin in vitro, significantly increased the secretion of Z ␣ 1 -antitrypsin from the Xenopus oocyte expression system. Thus the cavity is likely to have a different conformation in the Z ␣ 1 -antitrypsin than in the wild type protein, or has a different structure when glycosylated.
Neither Leu-100 3 Cys/Cys-232 3 Ser nor Thr-114 3 Cys/ Cys-232 3 Ser ␣ 1 -antitrypsin bound to the glutathione-Sepharose column, implying that the cysteine residues were buried within the cavity. The accessibility of the cavity to small mimetics was assessed by fluorophore binding to Leu-100 3 Cys/ Cys-232 3 Ser and Thr-114 3 Cys/Cys-232 3 Ser ␣ 1 -antitrypsin. Both of these mutants bound a range of fluorophores indicating that the cavity was accessible to external agents. The fluorophores had no effect on polymer formation, which implies that more than 30% of molecules must have their cavities filled if a mimetic is to impede polymerization.
Taken together our data show that the conformational change in the hydrophobic cavity bounded by s2A, hD, and hE is important in the polymerization of ␣ 1 -antitrypsin. Inhibiting polymer formation is an important therapeutic goal, and several approaches have been explored including chemical chaperones that stabilize protein folding (39,40) and the use of small peptides that bind to ␤-sheet A (16). Chaperone and ␤-strand blockers have also been used in other conformational pathologies such as Alzheimer's and Huntington's disease (41)(42)(43). Although the peptide approach is promising, it is problematic as blocking ␣ 1 -antitrypsin polymerization, through binding to ␤-sheet A, invariably results in the inactivation of ␣ 1 -antitrypsin. The surface cavity bounded by s2A, hD, and hE is ideal for rational drug design as it is accessible to external agents that can block polymerization without an accompanying loss of inhibitory activity. Thus inhibition of ␣ 1 -antitrypsin polymerization within hepatocytes will prevent the liver disease associated with Z ␣ 1 -antitrypsin. Moreover, an increase in the amount of circulating active ␣ 1 -antitrypsin may offer a treatment for the associated emphysema.