Recombinant DNA-derived Forms of Human crl-Proteinase Inhibitor STUDIES ON THE ALANINE 358 AND CYSTEINE 358 SUBSTITUTED MUTANTS*

The specificity and reactivity of human al-protein- ase inhibitor has been investigated by in vitro mutagenesis of the reactive site PI methionine 358 residue to alanine 358 and cysteine 358. A comparison of the second-order association rates of both uncharged mu- tants with 9 serine proteinases indicated that each reacted similarly to either the normal plasma inhibitor or to a mutant containing valine in this position (Travis, J., Owen, M., George, P., Carrell, R., Rosen-berg, S., Hallewell, R. A., and Barr, P. J. (1985) J. Biol. Chem. 260, 4384-4389) when tested against either neutrophil or pancreatic elastase. However, ox- idation, carboxymethylation, or aminoethylation of the cysteine mutant to yield a charged PI residue resulted in a significant decrease in association rates with both elastolytic enzymes, and aminoethylation created an excellent trypsin and plasmin inhibitor. These results indicate that the specificity of al-proteinase inhibitor is determined in a general manner by the class of amino acid residue in the PI position. Substitution within the same category, such as from valine to alanine or cys- teine among the aliphatic hydrophobic residues, has little effect on association rates with the elastolytic enzymes tested. However, alteration from an un- charged to a charged residue may cause considerable by as described above, followed by incubation with excess 2-chloroethylamine (150 pmol) overnight at room temperature, under nitrogen. The alkylated sample was then dialyzed 0.1 M Tris-HC1, pH Determination of Second-order Association Rate Constants-The rates of association of the aI-PI variants with several proteinases were determined as previously described (2, 11). Briefly, human plasma al-PI was standardized against porcine pancreatic trypsin which had been active site-titrated with nitrophenyl p‘-guanidino- benzoate. This inhibitor was then utilized as a secondary standard for determining the activity of all other proteinases utilized in this study. The activities of the yeast alanine and cysteine (ul-PI mutants were measured against human neutrophil elastase while the carbox- ymethyl cysteine and aminoethylcysteine derivatives were titrated against bovine chymotrypsin. For the determination of association rates, equimolar amounts of the al-PI variants and proteinases were incubated at room tempera- ture in 0.1 M sodium phosphate buffer, pH 7.4, 0.15 M NaCl, for appropriate time periods followed by the addition of saturating con- centrations of substrate specific for each proteinase. The residual enzyme activity was then determined and compared to a control sample containing enzyme alone. The final molarity of substrates was 2.0 mM for human neutrophil elastase, human plasmin, S. aureus V8 proteinase, and porcine pancreatic elastase, 1.0 mM for bovine chymotrypsin and human cathepsin G, and 0.9 mM for porcine pancreatic trypsin and papain. All of the association rates reported are average values with standard deviations of 20% or less.

sues is sufficiently high to control this enzyme and thus prevent connective tissue damage, especially in the lung; however, if the level of active al-PI in the blood is reduced as occurs, for example, in individuals synthesizing mutant forms of the inhibitor, lung damage may occur resulting in the development of pulmonary emphysema (4).
In individuals with normal plasma protein levels of al-PI there are probably several pathways by which functional activity is decreased. A major mechanism suggested involves oxidation of the critical reactive site methionine residue at position 358 in the primary structure of the protein to the sulfoxide derivative by chemicals present in cigarette smoke and by enzymatically derived oxidants produced by the neutrophil (5-7). This alteration not only results in a 2000-fold decrease in the second-order association rate with neutrophil elastase (2) but also renders the oxidized inhibitor ineffective in controlling elastin degradation by this enzyme (8).
Confirmation of the importance of methionine 358, referred to as the PI residue, in dictating specificity has been obtained through the isolation of a natural mutant form of the inhibitor, referred to as al-PI (Pittsburgh), from a patient with a severe bleeding disorder (9). This protein was found to contain an arginine residue in the PI position with a resultant altered specificity. Indeed, its reactivity was directly opposite to that of normal plasma al-PI in that it inactivated thrombin rapidly and elastase very slowly. Using recombinant DNA technology it is now possible to produce human al-PI mutants with amino acids other than methionine in the PI or PI' positions (10, ll), as well as substitutions elsewhere in the primary structure of this protein, and examine both the range of specificities and the mechanism by which enzyme-inhibitor complex formation occurs with this protein. Furthermore, since yeasts are able to produce large quantities of such inhibitors, it may also prove practical to utilize such mutant forms as therapeutic agents against a variety of serine proteinase-related diseases, including emphysema (10).
Two yeast-produced al-PI mutants, with methionine or valine in the P1 position, have previously been studied (10,11). The methionine 358 protein appeared to be similar to normal plasma al-PJ except for decreased stability in the absence of reducing agents, while the valine mutant was both stable and oxidation-resistant. In a further effort to define the effects of various P1 residue replacements on both the inhibitory activity and specificity of al-PI we describe in this report the properties of both the structurally related alanine 358 and cysteine 358 mutants. The latter was chosen because it is amenable to chemical modification at the PI-reactive site position.

Materials
Human neutrophil elastase and human neutrophil cathepsin G were prepared as described previously (12,13). Human plasminogen was prepared by the method of Deutsch and Mertz (14) and activated to plasmin by the procedure of Wiman (15). Porcine pancreatic elastase, porcine trypsin, bovine chymotrypsin, and papain were purchased from Sigma, as was the trypsin substrate Bz-L-Arg-OEt, the chymotrypsin substrate Bz-L-Tyr-OEt, and the elastase substrates t-Boc-L-Ala-NP ester and N-Suc-L-Ala-L-Ala-L-Ala-p-NA, and the active site titrant nitrophenyl-p'-guanidinobenzoate. Staphylococcus aureus V8 proteinase was from Miles. The neutrophil elastase substrate, t-Boc-L-Ala-L-Ala-L-Norval-SBzl, the plasmin and trypsin substrate, t-Boc-L-Ala-L-Ala-L-Arg-SBzl, and the V8 proteinase substrate, t-Boc-L-Ala-L-Ala-L-Glu-SBzl, were gifts from Dr. James Powers (Georgia Institute of Technology, Atlanta, GA). Iodoacetic acid and 2-chloroethylamine were from Aldrich.

Methods
Synthesis of Inhibitor Mutants in Yeast"M13 site-specific mutagenesis was performed as described previously using the natural q -PI cDNA (10). For insertion of a unique silent PstI site at the codon for alanine 347 of this inhibitor, the mutagenic primer of sequence 5'-CTAAAAACATGGCGCCTGCAGCTTCAGT-3' was used. The resulting mutant gene was cloned together with the glyceraldehyde-3-phosphate dehydrogenase promoter and transcriptional terminator sequences into the yeast plasmid pC1/1 as described previously (11). This mutant gene construction was shown to synthesize fully active wild-type q -P I when expressed in yeast. For construction of al-PI mutants containing alanine and cysteine residues in place of methionine at position 358, we introduced the appropriate DNA sequence as a synthetic 40/48-mer oligonucleotide duplex between the mutant PstI site and the natural AuaI site in the cDNA (10). Briefly, a SalI/ PstI fragment containing the GAPDH promoter and the 5' region of the a,-PI gene, the kinase-treated synthetic oligonucleotide duplex and an Am-BamHI fragment containing the 3' region of the q -P I gene and the glyceraldehyde-3-phosphate dehydrogenase transcriptional terminator were cloned directly into BarnHI/SalI-digested pC1/1 (11). Recombinant plasmids were amplified in Escherichia coli stain HBlOl and used to transform yeast strain AB110. Oligonucleotides were synthesized by the phosphoramidite method using Applied biosystems 380A DNA synthesizers. Yeast cells containing mutant forms of oI-PI were grown as previously described (11).
Modification of Inhibitor Mutants-The cysteine mutant was activated each day prior to use by dialysis against 0.001 M dithiothreitol for 1 h at 4 "C. Oxidation of each of the mutant forms of &]-PI was performed by incubation with a 20-fold molar excess of N-chlorosuccinimide for 1 h in 0.1 M Tris-HC1, pH 7.0, followed by dialysis overnight in the cold against the same buffer. Carboxymethylation of the cysteine mutant was achieved by reduction of the protein (0.16 pmol) with dithiothreitol (6.0 pmol) for 1 h at room temperature in 0.1 M Tris-HC1, pH 8.0, followed by addition of excess iodoacetic acid (150 pmol). After 1 h of incubation, the carboxymethylated protein was dialyzed overnight in the cold against 0.1 M Tris-HC1, pH 7.0. Aminoethylation of the cysteine mutant (0.16 pmol) was accomplished by reduction as described above, followed by incubation with excess 2-chloroethylamine (150 pmol) overnight at room temperature, under nitrogen. The alkylated sample was then dialyzed against 0.1 M Tris-HC1, pH 7.0.
Determination of Second-order Association Rate Constants-The rates of association of the aI-PI variants with several proteinases were determined as previously described (2, 11). Briefly, human plasma al-PI was standardized against porcine pancreatic trypsin which had been active site-titrated with nitrophenyl p'-guanidinobenzoate. This inhibitor was then utilized as a secondary standard for determining the activity of all other proteinases utilized in this study. The activities of the yeast alanine and cysteine (ul-PI mutants were measured against human neutrophil elastase while the carboxymethyl cysteine and aminoethylcysteine derivatives were titrated against bovine chymotrypsin.
For the determination of association rates, equimolar amounts of the al-PI variants and proteinases were incubated at room temperature in 0.1 M sodium phosphate buffer, pH 7.4, 0.15 M NaCl, for appropriate time periods followed by the addition of saturating concentrations of substrate specific for each proteinase. The residual enzyme activity was then determined and compared to a control sample containing enzyme alone. The final molarity of substrates was 2.0 mM for human neutrophil elastase, human plasmin, S. aureus V8 proteinase, and porcine pancreatic elastase, 1.0 mM for bovine chymotrypsin and human cathepsin G, and 0.9 mM for porcine pancreatic trypsin and papain. All of the association rates reported are average values with standard deviations of 20% or less.

RESULTS
Expression of Inhibitor Mutants in Yeast-The al-PI coding sequence was derivatized by silent third position mutagenesis to form a unique PstI site close to the active center methionine 358 residue (10). By using this restriction site together with the natural AuaI site unique in the cDNA for al-PI, synthetic DNA coding for alanine 358 or cysteine 358 mutants was inserted directly into the yeast expression vector pC1/1. Using identical conditions to those described previously for the production of wild-type and valine 358 al-PI (ll), high level expression of both the alanine and cysteine mutants was achieved in yeast.
Isolation and Properties of Alanine 358 and Cysteine 358 Inhibitor Mutants-Both of the mutants studied were isolated exactly as described previously (11). Each migrated as a single band with a M, of 46,000 after NaDodSO, electrophoresis with reduction (Fig. 1, lane 1: Fig. 2, lane 4). This M, is lower than that of the native plasma inhibitor ( M , = 52,000) because of the lack of glycosylation in the yeast system utilized. Initially, both were found to inhibit porcine pancreatic and human neutrophil elastase. However, the cysteine mutant rapidly lost this activity upon storage, presumably due to dimer formation involving not only cysteine 358 but also possibly the other cysteine residue known to be present at position 232 (18,19) (Fig. 2, lane 1 ) . The dimerized protein did not form complexes with either neutrophil or pancreatic elastases (Fig. 2, lanes 2 and 3), instead being degraded by each of these enzymes. However, if the cysteine mutant was preincubated with 0.001 M dithiothreitol only the monomeric form of the inhibitor was detectable (Fig. 2, lane 4 ) and inhibitory activity was restored.
Complex Demonstration between Alanine 358 or Cysteine 358 Mutants and Proteinases-The interaction of the alanine mutant of aI-PI with human neutrophil elastase yielded NaDodS04-stable complexes as shown in Fig. 1. When the inhibitor and human neutrophil elastase (5:l molar ratio) were incubated together for 1 min an NaDodS04-stable complex of M, about 70,000 was formed, representing a 1:1 complex between the two proteins (Fig. 1, lane 2). As the proportion of neutrophil elastase relative to the alanine mutant was increased more complex formed until equivalence, after which excess elastase could be demonstrated and also slight degradation of the complex (Fig. 1, lane 4 ) .
As described above, the cysteine mutant could only inactivate neutrophil or pancreatic elastase after reduction with 0.001 M dithiothreitol. As shown in Fig. 2, lanes 5 and 6, some complex formation could be detected at a 2:l molar ratio of inhibitor to enzyme. However, degradation of inhibitor and/ or complex was also apparent presumably because of slower inactivation of the enzymes tested.
Although papain is known to readily convert human plasma cu,-PI into a modified form by cleavage of the methionine 3581 serine 359 peptide bond (20), the presence of a cysteine residue in the PI position of this inhibitor could have impaired papain function. However, as shown in Fig. 3, lanes 2 and 3, the cysteine mutant was converted into a lower M , form at low enzyme:inhibitor molar ratios and completely degraded at higher molar ratios. This pattern was also observed with the plasma form of the inhibitor (Fig. 3, lanes 4 and 5).
Effect of Modification of Cysteine Residues on Complex For- mation-The reason for modifying the cysteine mutant was to determine whether changing the charge on the PI residue would affect inhibitor specificity. Carboxymethylation of cysteine with iodoacetic acid introduced an acidic side chain reminiscent of glutamic acid while aminoethylation with 2chloroethylamine yielded a basic side chain with a structure analogous to lysine. The carboxymethylated derivative was used to examine interactions with S. aureus V8 proteinase, an enzyme which accommodates a glutamic acid residue in its SI pocket (21).
After reductive carboxymethylation, the protein migrated as a single band (Fig. 4, lane 1 ) in a nonreducing system and could inactivate a number of proteinases tested. However, complex formation with the V8 proteinase was not detected (Fig. 4, lane 2) and the inhibitor was, instead, converted to a modified form of lower Mr. Since a glutamic acid residue is known to occur in the Ps position it is possible that the enzyme cleaved the modified mutant inhibitor either at this position or at the modified Pl position, thus inactivating the inhibitor.
While reduction and/or carboxymethylation of the airoxidized cysteine inhibitor caused restoration of inhibitory activity, it should be noted that stronger oxidizing agents, such as N-chlorosuccinimide, caused the conversion of the protein into a non-dimerizable and inactive form (Fig. 4, lane 3) which was degraded by neutrophil elastase in a manner similar to that of the air-oxidized protein (Fig. 4, lanes 4 and   6).
The aminoethylcysteine 358 derivative was used to examine the effect of placing a basic residue analogous to lysine (22) in the P1 position on interactions with trypsin and plasmin.
NaDodS04-gel electrophoresis patterns showed that most of the modified protein migrated as a monomer in a nonreducing system (Fig. 5, lane 4 ) , indicating nearly complete conversion to the aminoethylcysteine derivative. However, traces of a dimer could also be detected which may represent unreacted protein or, possibly, products of the long reaction time with 2-chloroethylamine. The former is not likely, however, since treatment of this protein with reducing agents had no effect on the migration pattern. After incubation with either porcine trypsin or plasmin (latter not shown), complex formation was readily detected with either the reduced cysteine or the aminoethylcysteine mutant (Fig. 5, lanes 3 and 5 ) .
Association Rate Constants for Alanine, Cysteine, and Modified Cysteine Mutants with Serine Proteinases-The secondorder association rates determined for the interaction of several proteinases with mutant forms of al-PI utilized in the current study are given in Table I, together with those previously determined with the plasma form of this protein (2). Significantly, the alanine and cysteine mutants reacted rapdily with both elastases tested. However, neither was a good inhibitor of neutrophil cathepsin G. Differences did occur, however, when bovine chymotrypsin or porcine trypsin were tested with the cysteine mutant which acted as a somewhat better inhibitor of these two enzymes than the alanine mutant.
As would be expected, N-chlorosuccinimide oxidation of the alanine variant had little effect on its interaction with either human neutrophil elastase or porcine pancreatic elastase while oxidation of the cysteine variant destroyed its inhibitory activity against all proteinases tested except human neutrophil elastase. Some slight inhibitory activity towards this enzyme could be measured, with the rate of interaction being approximately 3,000-fold less than the reduced cysteine variant and 34,000-fold less than that with human al-PI.
Surprisingly, the rates of interaction of the carboxymethylcysteine 358 mutant with bovine chymotrypsin and porcine trypsin were not significantly altered by this modification. However, the modified inhibitor did react a t a rate nearly 300-fold more slowly with human neutrophil elastase than the reduced cysteine 358 mutant and 3000-fold slower than human al-PI. It also lost all of its activity towards cathepsin G, and only very slowly inactivated pancreatic elastase. Finally, it showed no effect as an inhibitor of the S. aureus V8 proteinase, as previously indicated by the gel electrophoresis profiles in Fig. 4.
Association rates of the aminoethylcysteine mutant with either human neutrophil elastase or porcine pancreatic elastase were considerably slower than those found with either the cysteine or alanine mutants, but chymotrypsin inhibition " 1 2 3 4' 5" 3"-  rates were nearly equal. Significantly, the rates of inhibition of trypsin and plasmin were much higher than with any of the other mutants tested and far better than those obtained previously with plasma a,-PI. In this case the prediction that a change to a lysine-like amino acid residue in the PI position of al-PI should yield a good inhibitor of trypsin-like proteinases was fulfilled.

DISCUSSION
Since it seems well established that: (a) the best inhibitors of serine proteinases are actually excellent substrates, the difference being in the fact that dissociation rates are extremely slow (23); (b) human cyI-PI reacts most rapidly with neutrophil and pancreatic elastase (2); and (e) both enzymes cleave peptide bonds after methionine, valine, and alanine residues (24, 25) but not after methionine sulfoxide residues (26), the introduction of valine (Ilf, alanine, or reduced cysteine into the P1 position is not expected to have any significant effect on elastase inhibition, and this was indeed what was found. Similarly, oxidation of the valine or alanine mutants should not have affected inhibitory activity as is known to occur with either the recombinant methionine mutant or the natural plasma inhibitor (ll), and this was also the case. Oxidative inactivation of the cysteine mutant could also have been predicted on the basis of dimer formation (air oxidation) or probable conversion to a cysteic acid mutant (N-chlorosuccinimide oxidation), and a marked reduction in association rates with neutrophil elastase in both cases was observed. The carboxymethylcysteine and aminoethylcysteine mutants, both of which contained charged groups, were equally ineffective as neutrophil elastase inhibitors, again in agreement with the defined substrate specificity of this enzyme. Porcine pancreatic elastase appeared to be more readily affected by a change in the P, position than the neutrophil enzyme. Although any of these mutants in their native configuration were effective pancreatic elastase inhibitors, the effect of oxidation, carboxymethylation, or aminoethylation of the cysteine variant was to substantially reduce or eliminate a,-PI inhibitory activity towards this enzyme. Bovine chymotrypsin and porcine trypsin, on the other hand, were quite nonspecific in their requirements since all the mutants were reasonably effective as inhibitors of these enzymes with the exception of the oxidized cysteine mutant which was completely inactive against either enzyme. Perhaps the most unexpected finding of this study involved the reaction between trypsin and the carboxymethylcysteine 358 mutant. Trypsin is regarded as selective for substrates and inhibitors with basic side chains in the PI site. Although there are exceptions to this rule, most notably with methionine in the PI position, we believe this is the first report of a substrate or inhibitor with an acidic side chain in PI reacting at a significant rate with trypsin. Presumably, the interaction between the carboxyl group of carboxyrnethylcysteine 358 in the mutant inhibitor and the carboxyl group of aspartyl 189 in trypsin which comprises the bottom of the SI pocket of this enzyme, although unfavorable, is not so great as to overcome other interactions occurring between the reactive site of the modified inhibitor and the substrate binding region of trypsin. Thus, although the PI residue usually dictates the specificity of a proteinase inhibitor other interactions can compensate for an unfavorable interaction.
None of the mutants tested were good inhibit.ors of cathepsin G. This enzyme appears to be controlled by the plasma inhibitor cu,-antichymotrypsin which contains a leucine residue in the PI position (27), and although inhibition by plasma @,-PI is reasonable it is still far slower than that of al-antichymotr~sin, indicating that methionyl residues are not well tolerated by this enzyme. Furthermore, when valine, alanine, cysteine, carboxymethylcysteine, or aminoethylcysteine are present in the PI position, association rates are decreased even further. While it is certainly likely that other amino acids in aI-PI make important contributions to its function, the results presented here would clearly support the premise that the P, residue dictates the primary specificity of this inhibitor. The most significant changes in inhibitor specificity were obtained when the aminoethylcysteine variant was prepared. This mutant became an excellent inhibitor of both trypsin and plasmin, relative to all of the other mutants examined and to the natural inhibitor and behaved in a similar manner to the Pittsburgh mutant (9), further supporting the importance of the P, residue in a,-PI. Unfortunately, the use of the carboxymethylcysteine mutant as an inhibitor of the S. aureus V8 proteinase was unsuccessful. This is probably because the carboxymethylcysteine residue was too large to fit into the binding pocket of the enzyme but also could be due to the fact that the V8 proteinase is of bacterial orgin and only very distantly related to members of the chymotrypsin family (28). Clearly, it would be important to test a mutant with glutamic acid in the PI position against this enzyme not only for the further understanding of the mechanism of action of this inhibitor but also for potential therapeutic use against S. aureus infection. Should such an alteration in specificity be successful in controlling this bacterial proteinase it may be possible to develop other mutant inhibitors directed specifically against serine proteinases elicited from pathogenic organisms, some of which may play a direct role in the development of disease states, such as septicemia.