Design and characterization of a hybrid miniprotein that specifically inhibits porcine pancreatic elastase

Studying protease/peptide inhibitor interactions is a useful tool for understanding molecular recognition in general and is particularly relevant for the rational design of inhibitors with therapeutic potential. An inhibitory peptide (PMTLEYR) derived from the third domain of turkey ovomucoid inhibitor and optimized for specific porcine pancreatic elastase inhibition was introduced into an inhibitor scaffold to increase the proteolytic stability of the peptide. The trypsin-specific squash inhibitor EETI II from Ecballium elaterium was chosen as the scaffold. The resulting hybrid inhibitor HEI-TOE I (hybrid inhibitor from E. elaterium and the optimized binding loop of the third domain of turkey ovomucoid inhibitor) shows a specificity and affinity to porcine pancreatic elastase similar to the free inhibitory peptide but with significantly higher proteolytic stability. Isothermal titration calorimetry revealed that elastase binding of HEI-TOE I occurs with a small unfavorable positive enthalpy contribution, a large favorable positive entropy change, and a large negative heat capacity change. In addition, the inhibitory peptide and the hybrid inhibitor HEI-TOE I protected endothelial cells against degradation following treatment with porcine pancreatic elastase.     

Substrate specificity of human pancreatic elastase 2

The substrate specificity of human pancreatic elastase 2 was investigated by using a series of peptide p-nitroanilides. The kinetic constants, kcat and Km, for the hydrolysis of these peptides revealed that this serine protease preferentially hydrolyzes peptides containing P1 amino acids which have medium to large hydrophobic side chains, except for those which are disubstituted on the first carbon of the side chain. Thus, human pancreatic elastase 2 appears to be similar in peptide bond specificity to the recently described porcine pancreatic elastase 2 [Gertler, A., Weiss, Y., & Burstein, Y. (1977) Biochemistry 16, 2709] but differs significantly in specificity from porcine elastase 1. The best substrates for human pancreatic elastase 2 were glutaryl-Ala-Ala-Pro-Leu-p nitroanilide and succinyl-Ala-Ala-Pro-Met-p-nitroanilide. However, there was little difference among substrates with leucine, methionine, phenylalanine, tyrosine, norvaline, or norleucine in the P1 position. Changes in the hydrolysis rate of peptides with differing P5 residues indicate that this enzyme has an extended binding site which interacts with at least five residues of peptide substrates. The overall catalytic efficiency of human pancreatic elastase 2 is significantly lower than that of porcine elastase 1 or bovine chymotrypsin with the compounds studied.     

Isolation and expression in Escherichia coli of a cDNA clone encoding porcine pancreatic elastase

We have cloned a DNA that is complementary to the messenger RNA that encodes porcine pancreatic elastase 1 from pancreas using rat pancreatic elastase 1 cDNA as a probe. This complementary DNA contains the entire protein coding region of 798 nucleotides which encodes an elastase of 266 amino acids, and 22 and 136 nucleotides of the 5' and 3'-untranslated sequences. When this deduced amino acid sequence was compared with known amino acid sequences, a carboxy-terminal 240 amino acids long peptide was found to be identical with a mature form of porcine pancreatic elastase 1, except for two amino acids. The porcine enzyme contains the same number of amino acid residues as the rat enzyme, and their amino acid sequences are 85% homologous. Taking the above findings together, we conclude that the cloned cDNA encodes a mature enzyme of 240 amino acids including a leader and activation peptide of 26 amino acids. We expressed the cloned porcine pancreatic elastase 1 cDNA in E. coli as a lac-fused protein. The resulting fused protein showed enzymatic activity and immunoreactivity toward anti-elastase serum.     

Investigation of the active center of rat pancreatic elastase

We have isolated rat pancreatic elastase I (EC 3.4.21.36) using a fast two-step procedure and we have investigated its active center with p-nitroanilide substrates and trifluoroacetylated inhibitors. These ligands were also used to probe porcine pancreatic elastase I whose amino acid sequence is 84% homologous to rat pancreatic elastase I as reported by MacDonald, et al. (Biochemistry 21, (1982) 1453-1463). Both proteinases exhibited non-Michaelian kinetics for substrates composed of three or four residues: substrate inhibition was observed for most enzyme substrate pairs, but with Ala3-p-nitroanilide, rat elastase showed substrate inhibition, whereas porcine elastase exhibited substrate activation. With most of the longer substrates, Michaelian kinetics were observed. The kcat/Km ratio was used to compare the catalytic efficiency of the two elastases on the different substrates. For both elastases, occupancy of subsite S4 was a prerequisite for efficient catalysis, occupancy of subsite S5 further increased the catalytic efficiency, P2 proline favored catalysis and P1 valine had an unfavorable effect. Rat elastase has probably one more subsite (S6) than its porcine counterpart. The rate-limiting step for the hydrolysis of N-succinyl-Ala3-p-nitroanilide by rat elastase was essentially acylation, whereas both acylation and deacylation rate constants participated in the turnover of this substrate by porcine elastase. For both enzymes, trifluoroacetylated peptides were much better inhibitors than acetylated peptides and trifluoroacetyldipeptide anilides were more potent than trifluoroacetyltripeptide anilides. A number of quantitative differences were found, however, and with one exception, trifluoroacetylated inhibitors were less efficient with rat elastase than with the porcine enzyme.     

Primary structure of two distinct rat pancreatic preproelastases determined by sequence analysis of the complete cloned messenger ribonucleic acid sequences

The mRNA sequences for two rat pancreatic elastolytic enzymes have been cloned by recombinant DNA technology and their nucleotide sequences determined. Rat elastase I mRNA is 1113 nucleotides in length, plus a poly(A) tail, and encodes a preproelastase of 266 amino acids. The amino acid sequence of the predicted active form of rat elastase I is 84% homologous to porcine elastase 1. Key amino acid residues involved in determining substrate specificity of porcine elastase 1 are retained in the rat enzyme. The activation peptide of the zymogen does not appear related to that of other mammalian pancreatic serine proteases. The mRNA for elastase I is localized in the rough endoplasmic reticulum of acinar cells, as expected for the site of synthesis of an exocrine secretory enzyme. Rat elastase II mRNA is 910 nucleotides in length, plus a poly(A) tail, and encodes a preproenzyme of 271 amino acids. The amino acid sequence is more closely related to porcine elastase 1 (58% sequence identity) than to the other pancreatic serine proteases (33-39% sequence identity). Predictions of substrate preference based upon key amino acid residues that define the substrate binding cleft are consistent with the broad specificity observed for mammalian pancreatic elastase 2. The activation peptide is similar to that of the chymotrypsinogens and retains an N-terminal cysteine available to form a disulfide link to an internal conserved cysteine residue.     

Elafin: an elastase-specific inhibitor of human skin. Purification, characterization, and complete amino acid sequence

A potent inhibitor of human leukocyte elastase (EC 3.4.21.37) and porcine pancreatic elastase (EC 3.4.21.36) was purified to homogeneity from human horny layers. It inhibits human leukocyte elastase and porcine pancreatic elastase in a 1:1 molar ratio and shows equilibrium dissociation constants of 6 x 10(-10) M and 1 x 10(-9) M, respectively. Inhibition of plasmin, trypsin, alpha-chymotrypsin, and cathepsin G was not observed. This inhibitor proved to be an acid stable basic peptide with an isoelectric point of 9.7. The complete amino acid sequence appears to be unique with 38% homology to the C-terminal half of antileukoprotease. The sequence shows that the inhibitor is composed of 57 amino acids and predicts a Mr of 7017. The high affinity as well as the apparent specificity for elastases suggests a functional role in preventing elastase-mediated tissue proteolysis. It is suggested that the term "elafin" be used to designate this inhibitor.     

Turkey ovomucoid third domain inhibits eight different serine proteinases of varied specificity on the same ...Leu18-Glu19 ... reactive site

We show that eight different serine proteinases--bovine chymotrypsins A and B, porcine pancreatic elastase I, proteinase K, Streptomyces griseus proteinases A and B, and subtilisins BPN' and Carlsberg--interact with turkey ovomucoid third domain at the same Leu18-Glu19 peptide bond, the reactive site of the inhibitor. Turkey ovomucoid third domain was converted to modified (the reactive site peptide bond hydrolyzed) form as documented by sequencing. Complexes of all eight enzymes both with virgin and with modified inhibitor were prepared. All 16 complexes were subjected to kinetically controlled dissociation, and all 16 produced predominantly virgin (greater than 90%) inhibitor, thus proving our point. During this investigation, we found that both alpha-chymotrypsin and especially S. griseus proteinase B convert virgin to modified turkey ovomucoid third domain, even in the pH range 1-2, a much lower pH than we expected. We have also measured rate constants kon and kon* for the association of virgin and modified turkey ovomucoid third domain with several serine proteinases. The kon/kon* ratio is 4.8 X 10(6) for chymotrypsin, but it is only 1.5 for subtilisin Carlsberg. A number of generalizations concerning reactive sites of protein proteinase inhibitor are proposed and discussed.     

Binding of amino acid side chains to preformed cavities: interaction of serine proteinases with turkey ovomucoid third domains with coded and noncoded P1 residues.

In the association of serine proteinases with their cognate substrates and inhibitors an important interaction is the fitting of the P1 side chain of the substrate or inhibitor into a preformed cavity of the enzyme called the S1 pocket. In turkey ovomucoid third domain, which is a canonical protein proteinase inhibitor, the P1 residue is Leu18. Here we report the values of equilibrium constants, Ka, for turkey ovomucoid third domain and 13 additional Leu18X variants with six serine proteinases: bovine alpha chymotrypsin A, porcine pancreatic elastase, subtilisin Carlsberg, Streptomyces griseus proteinases A and B, and human leukocyte elastase. Eight of the Xs are coded amino acids: Ala, Ser, Val, Met, Gln, Glu, Lys, and Phe, and five are noncoded: Abu, Ape, Ahx, Ahp, and Hse. They were chosen to simplify the interamino acid comparisons. In the homologous series of straight-chain side chains Ala, Abu, Ape, Ahx, Ahp, free energy of binding decreases monotonically with the side-chain length for chymotrypsin with large binding pocket, but even for this enzyme shows curvature. For the two S. griseus enzymes a minimum appears to be reached at Ahp. A minimum is clearly evident for the two elastases, where increasing the side-chain length from Ahx to Ahp greatly weakens binding, but much more so for the apparently more rigid pancreatic enzyme than for the more flexible leukocyte enzyme. beta-Branching (Ape/Val) is very deleterious for five of the six enzymes; it is only slightly deleterious for the more flexible human leukocyte elastase. The effect of gamma-branching (Ahx/Leu), of introduction of heteroatoms (Abu/Ser), (Ape/Hse), and (Ahx/Met), and of introduction of charge (Gln/Glu) and (Ahp/Lys) are tabulated and discussed. An important component of the free energy of interaction is the distortion of the binding pocket by bulky or branched side chains. Most of the variants studied were obtained by enzymatic semisynthesis. X18 variants of the 6-18 peptide GlyNH2 were synthesized and combined with natural reduced peptide 19-56. Disulfide bridges were formed. The GlyNH2 was removed and the reactive-site peptide bond X18-Glu19 was synthesized by complex formation with proteinase K. The resultant complexes were dissociated by sudden pH drop. This kinetically controlled dissociation afforded virgin, reactive-site-intact inhibitor variants.     

NMR and enzymatic investigation of the interaction between elastase and sodium trifluoroacetate.

At pH 5.5, sodium trifluoroacetate is a potent competitive inhibitor of porcine elastase (Ki = 2.6 mM) and human leukocyte elastase (Ki = 9.3 mM). For both enzymes the Ki increases strongly with pH. Sodium fluoride is inactive on pancreatic elastase and sodium acetate is a weak inhibitor of this enzyme. Trifluoroethanol inhibits both enzymes but is less active than trifluoroacetate in acidic pH conditions. Bovine trypsin and alpha-chymotrypsin are resistant to the action of sodium trifluoroacetate and trifluoroethanol. The interaction between sodium trifluoroacetate and pancreatic elastase is also demonstrated by 19F NMR spectroscopy. Trifluoroacetyltrialanine is able to displace trifluoroacetate from its complex with pancreatic elastase. In addition, a method using turkey ovomucoid for the active site titration of leukocyte and pancreatic elastase is described.     

Molecular cloning and expression in Escherichia coli of a cDNA encoding human pancreatic elastase 2

We have cloned a DNA that is complementary to the messenger RNA that encodes human pancreatic elastase 2 from a human pancreatic cDNA library using a cloned cDNA for rat pancreatic elastase 2 messenger RNA. This complementary DNA contains the entire protein coding region of 807 nucleotides which encodes preproelastase of 269 amino acids, and 4 and 82 nucleotides of the 5'- and 3'-untranslated sequences, respectively. When this deduced amino acid sequence was compared with known amino acid sequences it showed 82% homology with rat pancreatic elastase 2. This deduced sequence also contains a 16-amino-acid peptide identical with the N-terminal sequence determined for native human pancreatic proelastase 2. Taking the above findings together, we conclude that the cloned cDNA encodes a mature enzyme of 241 amino acids including 16 and 12 amino acids for a signal peptide and an activation peptide, respectively. Moreover, the predicted key amino acid residues involved in determining the substrate specificity of mammalian pancreatic elastase 2 are retained in the human enzyme. Cloned human pancreatic elastase 2 cDNA was expressed in E. coli as a mature and pro-form protein. Both resulting proteins showed immunoreactivity toward anti-elastase serum and enzymatic activity. We have also cloned and sequenced a porcine pancreatic elastase 2 cDNA.     

Design, synthesis and evaluation of biomimetic affinity ligands for elastases

A low-molecular-weight biomimetic affinity ligand selective for binding elastase has been designed and synthesized. The ligand was based on mimicking part of the interaction between a natural inhibitor, turkey ovomucoid inhibitor and elastase, and modelled from the X-ray crystallographic structure of the enzyme-inhibitor complex. Limited solid-phase combinatorial chemistry was used to synthesize 12 variants of the lead ligand using the triazine moiety as the scaffold for assembly. The ligand library was screened for its ability to bind elastase and trypsin, and two ligands were studied further. Ligand C4/6 [2-alanyl-alanyl-4-tryptamino-6-(alpha-lysyl)-s-triazine] was found to bind porcine pancreatic elastase, but not trypsin, with a dissociation constant of 6 x 10(-5) M and a binding capacity of 21 mg elastase per ml gel. The adsorbent was used to purify elastase from a crude extract of porcine pancreas. Immobilized ligand C4/5 6 [2-alanyl-alanyl-4-tyramino-6-(alpha-lysyl)-s-triazine] was similarly chosen for optimal binding of elastase from cod and used to purify the enzyme from a crude extract of cod pyloric caeca. Ligand C4/6 was subsequently synthesized in solution and its structure verified by 1H-NMR.     

Synthetic inhibitors of human leukocyte elastase, Part 3. Peptides with alkyl groups at the N- or C-terminus. Non-toxic competitive inhibitors of human leukocyte elastase

A series of carboxy-alkylamidated and N-acetylated amino acids and peptides were synthesized and examined for their ability to inhibit human leukocyte elastase. The Boc-amino acid alkylamides were found to be potent specific and competitive inhibitors of this enzyme. They were found not to or only poorly inhibit several other serine proteinases such as bovine trypsin, alpha-chymotrypsin, porcine pancreatic elastase and human leukocyte cathepsin G at concentrations less than 10(-4) M. Specificity and maximum inhibition of human leukocyte elastase were achieved when the N-terminus of the amino acid was protected by a t-butyloxy-carbonyl (Boc) group, the oligopeptide fragment consisted of valine residues and when the alkyl chain was between 10 and 12 carbon atoms in length and attached to the C-terminus of the peptide fragment. Highest inhibition was obtained with the compound Boc-[Val]3-NH[CH2]11--CH3 (Ki = 0.21 microM). These specific inhibitors were also found to be non-toxic after oral administration to mice and rats (LD50 greater than 3.0 g/kg body weight).     

Preparation of chemically 'mutated' aprotinin homologues by semisynthesis. P1 substitutions change inhibitory specificity

The semisynthesis of homologues of aprotinin (BPTI) is described. The P1 amino acid residue of these homologues was substituted by other amino acids using peptide synthetic methods. The reactive-site-modified inhibitor (with the Lys15-Ala16 peptide bond hydrolyzed) was used as starting material. All carboxyl groups of the modified inhibitor were esterified with methanol, then the Lys15 methyl ester group was hydrolyzed selectively. Afterwards, Lys15 itself was split off. A new amino acid residue was incorporated by using water-soluble carbodiimide combined with an acylation catalyst. tert-Butyl-ester-protected amino acids were used for reinsertion. The method was tested by re-insertion of Lys15 to reconstitute the original inhibitor. Thirteen BPTI homologues with coded (Lys, Glu, Gly, Ala, Val, Ile, Leu) or uncoded amino acids (Abu, Ape, aIle, Ahx, tLeu, Neo) in position 15 were synthesized and the specificity of the inhibitors investigated. Amongst these, [Val15]BPTI was shown to be an excellent inhibitor for human polymorphonuclear leukocyte elastase having a complex dissociation constant of 0.11 nM. This inhibitor showed no detectable affinity to bovine pancreatic trypsin.     

Human serum alpha 1-antichymotrypsin is an inhibitor of pancreatic elastases

Incubation of human serum alpha 1-antichymotrypsin with human pancreatic elastase 2 or porcine pancreatic elastase results in the complete inhibition of each enzyme as determined by spectrophotometric assays. alpha 1-Antichymotrypsin reacts much more rapidly with the human than with the porcine enzyme. The inhibitor: enzyme molar ratio, required to obtain full inhibition of enzymatic activity, is equal to 1.25/1 when alpha 1-antichymotrypsin reacts with human pancreatic elastase 2 while it is markedly higher with porcine pancreatic elastase (5.5/1). Patterns obtained by SDS/polyacrylamide gel electrophoresis of the reaction products show the formation with both enzymes of an equimolar complex (Mr near 77 000) and the release of a fragment migrating as a peptide of Mr near 5000. Moreover a free proteolytically modified form of alpha 1-antichymotrypsin, electrophoretically identical with that obtained in the reaction with cathepsin G or bovine chymotrypsin, is produced in the reaction with each elastase but in a much greater amount when alpha 1-antichymotrypsin reacts with porcine elastase than with human elastase. As a consequence of our findings, the specificity of alpha 1-antichymotrypsin, so far limited to the inhibition of chymotrypsin-like enzymes from pancreas and leukocyte origin, has to be extended to the two pancreatic elastases investigated in this work. A contribution of alpha 1-antichymotrypsin to the regulatory balance between plasma inhibitors and human pancreatic elastase 2 in pancreatic diseases is suggested.     

The energetics of phosphate binding to a protein complex

The heat of binding the serine protease, porcine pancreatic elastase, by the inhibitor, turkey ovomucoid third domain, is dependent on the presence of inorganic phosphate. This dependence is saturable and can be accurately modeled as the phosphate binding to a single site on the protease-inhibitor complex; thus, the elastase-ovomucoid system provides a unique opportunity to study phosphate-protein interactions. We have used isothermal titration calorimetry to investigate this binding, thereby providing one of the few complete thermodynamic characterizations of phosphate interacting with proteins. The binding is characterized by a small favorable deltaG degrees, a large unfavorable deltaH degrees, and a positive deltaCp, thermodynamics consistent with the release of water being linked to phosphate binding. These measurements provide insight into the binding of phosphotyrosine containing peptides to SH2 domains by suggesting the energetic consequences of binding phosphate free from other interactions.     

Primary structure of human pancreatic elastase 2 determined by sequence analysis of the cloned mRNA

A cDNA encoding elastase 2 has been cloned from a human pancreatic cDNA library. The cDNA contains a translation initiation site and a poly(A) recognition site and encodes a protein of 269 amino acids, including a proposed 16-residue signal peptide. The amino acid sequence of the deduced mature protein contains a 12-residue activation peptide containing a cysteine at residue 1 similar to that of chymotrypsin. The proposed active enzyme contains all of the characteristic active-site amino acids, including His-57, Asp-102, and Ser-195. The S1 binding pocket is bounded by Gly-216 and Ser-226, making this pocket intermediate in size between chymotrypsins and elastase 1 or protease E, consistent with the substrate specificity of elastase 2 for long-chain aliphatic or aromatic amino acids. Computer modeling studies using the amino acid sequence of elastase 2 superimposed on the X-ray structure of porcine elastase 1 suggest that a change of Gln-192 in elastase 1 to Asn-192 in elastase 2 may account for the lower catalytic efficiency of the latter enzyme. In addition, modeling studies have been conducted to attempt to identify basic amino acids in elastases which are absent in chymotrypsins, and which could account for the specific property of elastolysis. Several basic residues appear to be near the ends of the extended binding pocket of elastases which might serve to anchor the enzyme to the elastin substrate. These studies indicate that elastases 2 and elastase 1 both contain an Arg-65A as well as a basic dipeptide at 223/224 which is not present in chymotrypsins.(ABSTRACT TRUNCATED AT 250 WORDS)     

Kinetic and crystallographic analysis of complexes formed between elastase and peptides from beta-casein.

Human beta-casomorphin-7 (NH2-Tyr-Pro-Phe-Val-Glu-Pro-Ile-CO2H) is a naturally occurring peptide inhibitor of elastase that has been shown to form an acyl-enzyme complex stable enough for X-ray crystallographic analysis at pH 5. To investigate the importance of the N-terminal residues of the beta-casomorphin-7 peptide for the inhibition of elastase, kinetic and crystallographic analyses were undertaken to identify the minimum number of residues required for effective formation of a stable complex between truncated beta-casomorphin-7 peptides and porcine pancreatic elastase (PPE). The results clearly demonstrate that significant inhibition of PPE can be effected by simple tri-, tetra-and pentapeptides terminating in a carboxylic acid. These results also suggest that in vivo regulation of protease activity could be mediated via short peptides as well as by proteins. Crystallographic analysis of the complex formed between N-acetyl-Val-Glu-Pro-Ile-CO2H and PPE at pH 5 (to 1.67 A resolution) revealed an active site water molecule in an analogous position to that observed in the PPE/beta-casomorphin-7 structure supportive of its assignment as the 'hydrolytic water' in the deacylation step of serine protease catalysis.     

Identification and fine mapping of IgG and IgE epitopes in ovomucoid

Ovomucoid is a major allergen in hen egg white which causes a serious IgE-mediated food allergy reaction. This study determined eight IgG epitopes, 5-11 amino acids in length, and nine IgE epitopes, 5-16 amino acids in length, within the primary sequence in ovomucoid using arrays of overlapping peptides synthesized on cellulose membranes. Pooled sera from eight egg-allergic patients were used to probe the membrane. We also analyzed the amino acids that are critical for antibody binding by substituting a single amino acid within each epitope. Mutational analysis of the epitopes indicated that charged amino acids (aspartic acid, glutamic acid, and lysine) and some hydrophobic (leucine, phenylalanine, and glycine) and polar (serine, threonine, tyrosine, and cystein) amino acids were important for antibody binding. These results provide useful information for the molecular design necessary to reduce the allergenicity of ovomucoid, and a better understanding of structure-function relationships of allergic epitopes in food proteins.     

Conformation and inhibitory properties of peptides based on the tissue kallikrein-aprotinin complex.

A series of peptides encompassing the primary binding segment (residues 12-19) of aprotinin has been synthesized and tested for their ability to inhibit porcine pancreatic kallikrein. A minimum sequence of five amino acids spanning residues 12-16 of aprotinin is necessary for inhibition of porcine pancreatic kallikrein. An octapeptide homologous with the binding segment of aprotinin has a Ki-value of 1.2 x 10(-4) M. The solution structure of the octapeptide was studied by one- and two-dimensional NMR methods for comparison with the known structure of the segment of aprotinin that contacts tissue kallikrein. NMR experiments suggest that the peptide is either a random coil or that it samples several conformations on the NMR time scale. Analysis of the molecular dynamics trajectory of the octapeptide also suggests that the peptide is highly flexible. Thus, inhibition by the octapeptide occurs because of its homology with residues 12-19 of aprotinin. Moreover, the absence of a stable solution conformation similar to that of the binding segment of aprotinin is consistent with the 150,000-fold increase in Ki of the octapeptide compared to intact aprotinin.     

Crystallographic refinement of Japanese quail ovomucoid, a Kazal-type inhibitor, and model building studies of complexes with serine proteases

Japanese quail ovomucoid third domain (OMJPQ3), a Kazal-type inhibitor, was crystallographically refined with energy constraints. The final R-value is 0.20 at 1.9 Å resolution. The four molecules in the asymmetric unit are very similar, with deviations of main-chain atoms between 0.2 and 0.3 Å. An analysis of the side-chain hydrogen-bonding pattern and amino acid variability in the Kazal family shows a high correlation between hydrogen-bonding and conservation.The conformation of the reactive site loop (P2-P2′) of OMJPQ3 is similar to those of basic pancreatic trypsin inhibitor, Streptomyces subtilisin inhibitor, and soybean trypsin inhibitor. This suggests a common binding mode and justifies model-building studies of complexes.Complexes of OMJPQ3 with trypsin, chymotrypsin and elastase were modelled on the basis of the trypsin-basic pancreatic trypsin inhibitor complex structure and inspected by use of a computer graphics system. Stereochemically satisfying models were constructed in each case and detailed interactions are proposed. The complex with elastase is of particular interest, showing that leucine and methionine are good P1 residues. A good correlation is observed between functional properties of ovomucoid variants and the position of the exchanged residues with respect to the modelled inhibitor-protease contact.