Distinct multisite synergistic interactions determine substrate specificities of human chymase and rat chymase-1 for angiotensin II formation and degradation.

Human chymase and rat chymase-1 are mast cell serine proteases involved in angiotensin II (Ang II) formation and degradation, respectively. Previous studies indicate that both these enzymes have similar P1 and P2 preferences, which are the major determinants of specificity. Surprisingly, despite the occurrence of optimal P2 and P1 residues at the Phe8↓ and Tyr4↓ bonds (where ↓, indicates the scissile bond in peptide substrates) in Ang I (DRVYIHPFHL), human chymase cleaves the Phe8↓ bond with an ∼750-fold higher catalytic efficiency (kcat/Km) than the Tyr4↓ bond in Ang II (DRVYIHPF), whereas rat chymase-1 cleaves the Tyr4↓ bond with an ∼20-fold higher catalytic efficiency than the Phe8↓ bond. Differences in the acyl groups IHPF and DRVY at the Phe8↓ and Tyr4↓ bonds, respectively, are chiefly responsible for the preference of human chymase for the Phe8↓ bond. We show that the IHPF sequence forms an optimal acyl group, primarily through synergistic interactions between neighboring acyl group residues. In contrast to human chymase, rat chymase-1 shows a preference for the Tyr4↓ bond, mainly because of a catalytically productive interaction between the enzyme and the P′1 Ile5. The overall effect of this P′1 Ile interaction on catalytic efficiency, however, is influenced by the structure of the acyl group and that of the other leaving group residues. For human chymase, the P′1 Ile interaction is not productive. Thus, specificity for Ang II formation versus Ang II degradation by these chymases is produced through synergistic interactions between acyl or leaving group residues as well as between the acyl and leaving groups. These observations indicate that nonadditive interactions between the extended substrate binding site of human chymase or rat chymase-1 and the substrate are best explained if the entire binding site is taken as an entity rather than as a collection of distinct subsites.

Human chymase and rat chymase-1 are mast cell serine proteases involved in angiotensin II (Ang II) formation and degradation, respectively. Previous studies indicate that both these enzymes have similar P 1 and P 2 preferences, which are the major determinants of specificity. Surprisingly, despite the occurrence of optimal P 2 and P 1 residues at the Phe 82 and Tyr 42 bonds (where 2 , indicates the scissile bond in peptide substrates) in Ang I (DRVYIHPFHL), human chymase cleaves the Phe 82 bond with an ϳ750-fold higher catalytic efficiency (k cat /K m ) than the Tyr 42 bond in Ang II (DRVYI-HPF), whereas rat chymase-1 cleaves the Tyr 42 bond with an ϳ20-fold higher catalytic efficiency than the Phe 82 bond. Differences in the acyl groups IHPF and DRVY at the Phe 82 and Tyr 42 bonds, respectively, are chiefly responsible for the preference of human chymase for the Phe 82 bond. We show that the IHPF sequence forms an optimal acyl group, primarily through synergistic interactions between neighboring acyl group residues. In contrast to human chymase, rat chymase-1 shows a preference for the Tyr 42 bond, mainly because of a catalytically productive interaction between the enzyme and the P 1 Ile 5 . The overall effect of this P 1 Ile interaction on catalytic efficiency, however, is influenced by the structure of the acyl group and that of the other leaving group residues. For human chymase, the P 1 Ile interaction is not productive. Thus, specificity for Ang II formation versus Ang II degradation by these chymases is produced through synergistic interactions between acyl or leaving group residues as well as between the acyl and leaving groups. These observations indicate that nonadditive interactions between the extended substrate binding site of human chymase or rat chymase-1 and the substrate are best explained if the entire binding site is taken as an entity rather than as a collection of distinct subsites.
Early comparative studies on chymase specificities by Powers et al. (11) using peptide 4-nitroanilide substrates indicated that the S 1 to S 4 subsites 3 of human chymase and rat chymase-1 are similar. In both human chymase and rat chymase-1, the key features for optimal acyl group interactions are a P 1 hydrophobic aromatic residue, a P 2 hydrophobic residue or Pro, and a P 3 hydrophobic residue. S 4 subsite interactions are less restrictive. Thus, these studies could not explain why human chymase is an Ang II-forming enzyme and rat chymase-1 is an angiotensinase (3,5) and suggested to us that enzyme-substrate interactions other than those occurring at the S 1 to S 4 subsites of these enzymes could be important for determining specificity. In a previous paper we explored the S 1 subsite as well as the SЈ 1 to SЈ 2 subsites of human chymase using decapeptide Ang I analogs. We showed that a P 1 hydrophobic aromatic residue was necessary but that several non-* This work was supported by National Institutes of Health Grant HL44201. 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.
‡ To whom correspondence should be addressed: Dept. of Molecular Cardiology, FF30, Research Inst., The Cleveland Clinic Foundation, 9500 Euclid Ave., Cleveland, OH 44195. Tel.: 216-444-2057; Fax: 216-444-9410; E-mail: husaina@cesmtp.ccf.org. 1 Mast cell chymotrypsin-like proteases have in the past been variously referred to as chymases, mast cell chymases, and mast cell proteases. The term mast cell protease has also been used to designate mouse mast cell tryptases and carboxypeptidases; thus, not all mouse enzymes referred to as mast cell proteases are chymases; the number designation associated with this nomenclature is usually 1, 2, etc., but L has also been used. The number assignment has usually been given in order of discovery. Because homologs in different species have been discovered in differing orders, numbers do not necessarily correspond to homologs. In this paper, we have used the term chymase to describe a distinct group of leukocyte serine proteases (7) but have retained the original number designation given at the time of discovery. 2 The abbreviations used are: Ang I, angiotensin I; Ang II, angiotensin II; 2 , scissile bond in peptide substrates; HPLC, high pressure liquid chromatography; ES complex, enzyme-substrate complex; ES ‡ complex, transition state complex; ⌬⌬G 7 ‡ , difference between two substrates in the free energy required for transition state stabilization (i.e. free energy required to reach ES ‡ complex from E ϩ S. 3 The nomenclature used for the individual amino acids (P 1 , PЈ 1 , etc.) of a substrate and the subsites (S 1 , SЈ 1 , etc.) of the enzyme is that of Schechter and Berger (21). Amino acid residues of substrates numbered P 1 , P 2 , etc. are toward the N-terminal direction, and PЈ 1 , PЈ 2 , etc. are toward the C-terminal direction from the scissile bond.
conservative changes in PЈ 1 and PЈ 2 positioned residues produced small effects (i.e. a ϳ3-fold change in the specificity constant k cat /K m ) on the cleavage of the Phe 82 Xaa 9 bond (12). To determine if the Tyr-Ile P 1 -PЈ 1 combination forms a poor cleavage site for human chymase or if the context of the bond within the substrate is important, we synthesized an peptide analog of Ang I that contained two Tyr-Ile bonds, one that naturally occurs at the 4 -5 position and one introduced at the 8 -9 position. The Tyr 4 -Ile 5 bond was resistant to cleavage by human chymase, but the Tyr 8 -Ile 9 bond was readily cleaved; the difference in k cat /K m for the cleavage of the Tyr 8 -Ile 9 bond versus the Tyr 4 -Ile 5 bond was Ͼ500:1 (12). We proposed, therefore, that for human chymase, the structural context of the scissile bond within the polypeptide substrate is likely to be an important determinant of specificity. What is the structural context that makes the Phe 8 -His 9 bond in Ang I, relative to the Tyr 4 -Ile 5 bond in Ang II, highly susceptible to cleavage by human chymase? We show that in human chymase this structural context is generated through synergistic interactions between neighboring P 4 to P 1 residues of the substrate acyl group. Rat chymase-1, on the other hand, has a distinct PЈ 1 preference that distinguishes its specificity from human chymase.

EXPERIMENTAL PROCEDURES
Peptides-Peptides used in this study were synthesized by The Protein Core Facility of The Cleveland Clinic Foundation. Peptides were purified (purity Ͼ99%) on a C 18 reverse phase HPLC column and characterized by amino acid analysis and by analytical C 18 reverse phase HPLC. Peptide concentrations were standardized by amino acid analysis.
Enzymes and Enzyme Kinetics-Human chymase was purified to homogeneity from human left ventricular tissue (5). Rat chymase-1 was purified from rat peritoneal mast cells as described previously by Le Trong et al. (3). Identity of these enzymes was established by N-terminal sequence analysis. To determine K m and V max values for the human chymase and rat chymase-1 reactions, initial velocities (v) were determined as described by us previously (12). Fifteen concentrations of substrate ranging between 0.8 and 1,000 M with human chymase or rat chymase-1 were incubated at 37°C in 20 mM Tris-HCl buffer, pH 8.0, containing 0.5 M KCl, and 0.01% Triton X-100 (final volume, 50 l) for 20 min (in the cleavage of Ang II by human chymase a 120-min incubation period was used). For each peptide substrate, enzyme concentration was adjusted to between 0.02 and 4 nM to ensure that Ͻ15% of the substrate was utilized at the lowest substrate concentration. Under these conditions, product formation was linear with respect to time over the duration of the incubation. Reactions were terminated by the addition of 300 l of ice-cold ethanol, and the resulting solution was evaporated to dryness. The residue was resuspended in 125 l of distilled water, and 100 l was applied to a C 18 reverse phase HPLC column (Vydac, Hesperia, CA). The column was developed with linear acetonitrile gradients containing either 25 mM triethylammonium phosphate buffer, pH 3.0, 0.1% trifluoroacetic acid, or 5 mM hexane sulfonic acid at a flow rate of 2 ml/min. The column effluent was monitored at 214 nm. The elution positions of Ang I and of Ang II and its analogs were also determined using pure synthetic standards. The peak area corresponding to Ang II or its analog was integrated to calculate Ang II or Ang II analog formation. Products were separated by reverse phase HPLC and identified by amino acid analysis. K m and V max values were calculated by nonlinear regression using the equation . Correlation coefficients were routinely Ͼ0.99 but never Ͻ0.97. The concentration of human chymase and rat chymase-1 was determined using bicinchoninic acid protein assay reagent (Pierce) with pure bovine ␣-chymotrypsin as the standard. The overall rate constant k cat was calculated by the formula  (13). ⌬⌬G binding and ⌬⌬G cat represent the difference between two substrates in the free energy required to form the enzyme-substrate complex (ES) from E ϩ S and to convert the ES complex to the ES ‡ complex, respectively.

RESULTS AND DISCUSSION
Why Human Chymase Is an Ang II-forming Enzyme-Human chymase converts Ang I to Ang II by splitting the Ang I Phe 8 -His 9 bond with high catalytic efficiency (k cat /K m ϭ 3.6 M Ϫ1 ⅐s Ϫ1 ). At a human chymase concentration (ϳ20 pM) that allows a rapid and efficient cleavage of this bond in Ang I, the chymotrypsin-sensitive bond in Ang II (i.e. the Tyr 4 -Ile 5 bond) is not appreciably cleaved. Using a high concentration of human chymase (ϳ4 nM) and prolonged incubation times, we determined that the catalytic efficiency and the overall rate constant k cat for the cleavage of the Ang II Tyr 4 -Ile 5 bond are ϳ780-fold and ϳ3,500-fold lower, respectively, than that for the cleavage of the Ang I Phe 8 -His 9 bond (Table I). Because previous studies indicated that the P 1 , P 2 , and PЈ 1 residues at both the Try 4 and Phe 8 bonds are near optimal for human chymase (11,12), we considered whether the acyl group and/or the leaving group at these cleavage sites as a whole could explain these remarkable differences in k cat and catalytic efficiency.
To determine if the side chain interactions provided by the Ang I leaving group His-Leu influence catalysis, we compared the cleavage by human chymase of DRVYIHPF 2 HL (Ang I) with that of DRVYIHPF 2 GG (peptide hc1). Table I shows that these peptides are cleaved with similar catalytic efficiency, indicating that His-Leu side chain interactions are not important for catalysis. The effect of backbone leaving group interactions on catalytic efficiency is summarized in Table I. Fewer than two backbone interactions produce a decrease in catalytic efficiency, e.g. cleavage of the Phe 8 -Gly 9 bond in DRVYIHPF 2 G (peptide hc2, also see hc3) occurs with a ϳ55fold lower catalytic efficiency than of the Phe 8 -Gly 9 bond in DRVYIHPF 2 GG (peptide hc1), and an increase in leaving group backbone interactions beyond those provided by two residues was without effect (peptides hc4, hc5, and hc6). Thus, backbone interactions provided by the His-Leu leaving group are necessary for the efficient catalysis of the Phe 8 -His 9 bond but side chain interactions are not important.
Ang II degradation by human chymase (DRVY 2 IHPF; k cat /K m ϭ 0.0046 M Ϫ1 ⅐s Ϫ1 ) occurs with a ϳ780-fold lower catalytic efficiency than Ang II formation. In Ang II degradation, IHPF forms the leaving group. Cleavage of DRVY 2 GG (peptide hc7), an Ang II analog in which the leaving group side chains and the nonessential part of the peptide backbone are deleted, occurs with an ϳ11-fold higher catalytic efficiency than the cleavage of Ang II (Table I). Effect of individual leaving group side chain interactions on Ang II catalysis by human chymase are summarized in Table I. Deletion of PЈ 1 (peptide hc8), PЈ 2 (peptide hc9), PЈ 3 (peptide hc10), or PЈ 4 (peptide hc11) side chains in Ang II leads to a 2-14-fold increase in K m , a 7-170fold increase in k cat , and a 2-15-fold increase in k cat /K m . One way to interpret these data 4 is that the binding energy im- 4 Our interpretation of the these and other kinetic data described in this paper is based on the arguments presented by Fersht (15) in his discussions of enzyme-substrate complementarity and the use of binding energy in catalysis. Also germane to these discussions is the commonly held view that for serine protease-dependent amide bond cleavage, the rate-determining step is the acylation step (k 2 is the acylation rate constant), whereas for ester bond cleavage, the deacylation step is rate-limiting (k 3 is the deacylation rate constant). However, recent studies suggest that some amide bonds in substrates with highly efficient acyl groups are hydrolyzed by leukocyte serine proteases with rate-limiting deacylation (11,22). We found that the k cat for human chymase-dependent Ang II-ethyl ester (peptide hc23) and Ang I cleavage is 1,740 Ϯ 5.0 s Ϫ1 and 146 Ϯ 3.3 s Ϫ1 (in each case n ϭ 3), respectively. Because k 2 Ͼ Ͼ k 3 for ester bond cleavage, the rate constant k 3 for the deacylation of the human chymase-Ang II complex will be equal to k cat . Thus, using the formula 1/k cat ϭ 1/k 2 ϩ 1/k 3 , k 2 is ϳ160 s Ϫ1 for human chymase-dependent conversion of Ang I to Ang II. We speculate that in these highly efficient human chymase reactions, k cat adequately parted by these PЈ interactions leads to a more stable ES complex, but the stability of the ES ‡ complex is decreased. The activation energy of the k cat therefore increases substantially, so the reaction rate and catalytic efficiency both decrease. Thus, the IHPF leaving group at the Ang II Tyr 42 bond is detrimental for catalysis, and all leaving group side chains contribute to this effect.
In Ang II formation, the sequence DRVYIHPF (Ang II) forms the acyl group. Table I shows that catalytic efficiency does not change when this Ang I acyl group is reduced in length from DRVYIHPF to IHPF (peptide hc12). IHPF 2 HL (peptide hc12) and IHPF 2 GG (peptide hc13) are cleaved with similar catalytic efficiencies (Table I), again indicating that side chain interactions of the His-Leu leaving group are not consequential in the cleavage of the Phe 8 -His 9 bond. Therefore, of the acyl group interactions, those provided by the Ang I residues IHPF are sufficient for the high catalytic efficiency with which human chymase cleaves the Phe 8 -His 9 bond in Ang I. Because the acyl group involved in Ang II degradation, i.e. DRVY, is also four residues long, in the next series of experiments we directly compared the influence of the IHPF and DRVY acyl groups on catalytic efficiency.
The catalytic efficiency for human chymase-dependent cleavage of IHPF 2 GG (peptide hc13) was ϳ45-fold higher than for DRVY 2 GG (peptide hc7) cleavage (Table I). This difference in catalytic efficiency was almost entirely due to an ϳ44-fold difference in k cat , indicating that the IHPF acyl group, relative to the DRVY acyl group, does not affect the stability of the ES, but instead its binding energy is realized only in the ES ‡ , i.e. the structure of the acyl group binding site is much more complementary to the transition state structure of the IHPF acyl group than to that of the DRVY acyl group.
The difference in DRVY and IHPF acyl groups reactivities was surprising because in both these acyl groups the P 1 and P 2 residues were previously predicted to be optimal (11). This observation prompted us to examine which residue(s) was the chief determinant of the reactivity difference observed between these acyl groups. Table II shows that replacement of the P 4 Asp with Ile (peptide hc15), P 3 Arg with His (peptide hc16), or P 2 Val with Pro (peptide hc17) in the DRVY acyl group of Ang II produced decreases in the free energy required to reach the ES ‡ during hydrolysis (⌬⌬G T ‡ ϭ Ϫ6.37 kJ⅐mol Ϫ1 for P 4 Asp 3 Ile, Ϫ2.01 kJ⅐mol Ϫ1 for P 3 Arg 3 His, and Ϫ0.67 kJ⅐mol Ϫ1 for P 2 Val 3 Pro). Replacement of the P 1 Tyr with Phe (peptide hc18) in the DRVY acyl group of Ang II produced a small increase (0.71 kJ⅐mol Ϫ1 ) in ⌬⌬G T ‡ . Therefore, with respect to individual component differences between the DRVY and IHPF acyl groups, the P 4 Asp 3 Ile change has the most favorable effect on reactivity. Additivity analysis indicates a good agreement between the observed [⌬⌬G T ‡ (multiple)] and calculated [⌺⌬⌬G T ‡ (components)] decreases in transition state stabilization energy for two-or three-component transitions in the Ang II acyl group, e.g. DRVY 3 IHVY, DRVY 3 DRPF, or DRVY 3 IHVF (peptides hc19 -21) (Table II). Remarkably, however, when we compared the observed decrease in transition state stabilization energy associated with the DRVY 3 IHPF change in the Ang II acyl group [⌬⌬G T ‡ (multiple) ϭ Ϫ14.3 kJ⅐mol Ϫ1 ] with that calculated on the basis of individual P 4 to P 1 changes [⌺⌬⌬G T ‡ (components) ϭ Ϫ8.34 kJ⅐mol Ϫ1 ], it is apparent that a significant component (Ͼ40%) of the high reactivity of the IHPF acyl group for human chymase is due to synergistic behavior between all four these acyl group residues. Additional additivity analyses based on calculations of ⌬⌬G binding (from K m values in Table I  residues in producing a highly reactive human chymase substrate is seen in the energetics to reach ES ‡ complex but is not observed in the initial energetics of substrate binding that leads to the formation of the ES complex. We speculate that the extended substrate binding site of human chymase, particularly the region that binds the P 4 to P 1 acyl group residues, has specialized to allow the Phe 82 His 9 bond in Ang I to bind in a highly productive mode. This optimal acyl group is generated through synergistic interactions between neighboring acyl group residues; these interactions, we believe, form the basis of the "structural context" that has allowed human chymase to become an efficient Ang II-forming enzyme. The critical nature of these synergistic interactions in determining specificity is illustrated by the fact that the Ang II Tyr 4 -Ile 5 bond and the Ang I Phe 8 -His 9 bond, which seemingly contain optimal P 1 and P 2 residues and near optimal PЈ 1 residues (11,12), are cleaved by human chymase with an ϳ3,500fold difference in k cat and an ϳ780-fold difference in k cat /K m .
Why Rat Chymase-1 Is an Angiotensinase-In rat chymase-1-dependent Ang I degradation, the product DRVY accumulates ϳ20-fold faster than Ang II (7), suggesting that rat chymase-1 splits the Ang I Tyr 4 -Ile 5 bond ϳ20-fold faster than the Phe 8 -His 9 bond. Table III shows that rat chymase-1 cleaves the Tyr 4 -Ile 5 bond in Ang II with a ϳ20-fold higher catalytic efficiency than the Phe 8 -His 9 bond in IHPF 2 HL (peptide rc1). Therefore, we used DRVY 2 IHPF and IHPF 2 HL as model peptides to study the essential components of acyl and leaving group interactions that are involved in the cleavage of Ang I Tyr 4 -Ile 5 and Phe 8 -His 9 bonds.
No difference in the catalytic efficiency was observed between IHPF 2 HL (peptide rc1) and IHPF 2 GG (peptide rc2) cleavage by rat chymase-1 (Table III). Catalytic efficiency was also not affected when backbone leaving group interactions were increased from those provided by two residues to those provided by five residues; an example of this using the IHPF acyl group is shown in Table III (peptides rc2, rc3, rc4, and rc5). DRVY 2 GG (peptide rc6) cleavage by rat chymase-1 occurs with a ϳ22-fold lower catalytic efficiency than DRVY 2 IHPF cleavage. This decrease in catalytic efficiency is due to a 3.2-fold increase in K m and an 6.7-fold decrease in k cat . These results indicate that leaving group side chain interactions overall facilitate rat chymase-1-dependent catalysis of the Tyr 4 -Ile 5 bond in Ang II but not the Phe 8 -His 9 bond in Ang I. Table III summarizes the effect of leaving group side chain interactions on Ang II catalysis by rat chymase-1. Deletion of the PЈ 1 Ile side chain in Ang II (peptide rc7) produced a 6.5-fold decrease in catalytic efficiency, but deletion of the PЈ 2 His (peptide rc8), PЈ 3 Pro (peptide rc9), or PЈ 4 Phe (peptide rc10) side chains produced a 3.8 -6.6-fold increase in catalytic efficiency. These individual PЈ 1 to PЈ 4 side chain deletions produced small (Ͻ2.2-fold) effects on K m . Additivity analysis predicts that the cumulative detrimental effects of the Ang II PЈ 2 -, PЈ 3 -, and PЈ 4 -side chains on transition state stabilization en-ergy [⌺⌬⌬G T ‡ (PЈ 2 , PЈ 3 , PЈ 4 ) ϭ 12.28 kJ⅐mol Ϫ1 ] should overcome the decrease in transition state stabilization energy produced by the PЈ 1 Ile side chain [⌬⌬G T ‡ (PЈ 1 ) ϭ Ϫ4.84 kJ⅐mol Ϫ1 ] to generate an ϳ7.5 kJ⅐mol Ϫ1 increase in the free energy required to reach the ES ‡ complex during hydrolysis (Table IV). However, the overall observed effect of Ang II leaving group interactions on catalytic efficiency is favorable (⌬⌬G T ‡ ϭ Ϫ7.94 kJ⅐mol Ϫ1 ) (Table IV); this is evident when Ang II cleavage (k cat /K m ϭ 0.085 M Ϫ1 ⅐s Ϫ1 ) is compared to DRVY 2 GG cleavage (peptide rc6; k cat /K m ϭ 0.0039 M Ϫ1 ⅐s Ϫ1 ). These findings suggest that interaction of the Ang II leaving group with the rat chymase-1 SЈ subsite is dependent on secondary structure of the substrate leaving group or perhaps that of the entire substrate and that the PЈ 1 interaction is dominant over other PЈ interactions. These speculations led us 1) to consider whether the effect of the Ang II leaving group would be different if this leaving group was attached to a different acyl group and 2) to examine the effect of a single PЈ 1 Ile side chain on catalysis.
Direct comparisons were made between the Ang II acyl group DRVY and the Ang I acyl group IHPF. Specificity constants for the cleavage of IHPF 2 GG (peptide rc2) and DRVY 2 GG (peptide rc6) were identical (Table III), indicating that within the context the Gly-Gly leaving group rat chymase-1 does not differentiate between these acyl groups. To determine if interactions between the acyl and the leaving group can influence transition state stabilization, we compared the effects of IHPF and Gly-Gly leaving groups on DRVY and IHPF acyl groups (peptides rc2, rc6, rc11, and Ang II). Table IV shows that the IHPF leaving group, relative to the Gly-Gly leaving group, causes a Ϫ7.94 kJ⅐mol Ϫ1 decrease in ⌬⌬G T ‡ when DRVY is the acyl group and a Ϫ13.8 kJ⅐mol Ϫ1 decrease in ⌬⌬G T ‡ when IHPF is the acyl group. Thus, acyl-leaving group interactions can greatly influence the overall effect of the leaving group on transition state stabilization.
To show if the favorable effect of the IHPF leaving group in IHPF 2 IHPF catalysis can be mimicked by the introduction of an Ile side chain at the PЈ 1 position, we examined IHPF 2 IHPF (peptide rc11), IHPF 2 GGGG (peptide rc4), IHPF 2 IGGGG (peptide rc12), and IHPF 2 GGGGG (peptide rc5) catalysis by rat chymase-1. IHPF 2 IHPF was cleaved by rat chymase-1 with a 138-fold higher catalytic efficiency than IHPF 2 GGGG. IHPF 2 IGGGG was cleaved with a 66-fold higher catalytic efficiency than IHPF 2 GGGGG; peptides in which Ile side chains were introduced at additional leaving group sites, e.g. IHPF 2 IGIGG and IHPF 2 IGIGI (peptides rc13 and rc14), were catalyzed with an efficiency similar to that of IHPF 2 IGGGG (Table III). These findings are consistent with the view that the PЈ 1 Ile side chain of the IHPF leaving group provides the dominant favorable effect. This Ile effect on catalytic efficiency is produced by a 12.5-fold decrease in K m and a 5.3-fold increase in k cat (peptides rc5 and rc12) and is position-dependent. In contrast, in the case of human chymase, introduction of an Ile in the PЈ 1 position of the pentaglycyl leaving group (peptides hc6 and hc22) generates a nonproductive binding mode, because this interaction produces an ϳ3-fold decrease in K m , but k cat /K m decreases by a small extent (ϳ2.7-fold) ( Table I).
In these studies with rat chymase-1, we show that the introduction of a single PЈ 1 Ile side chain on a pentaglycyl leaving group markedly increases catalytic efficiency when IHPF is the acyl group; also, we show using the DRVY acyl group that deletion of the PЈ 1 Ile side chain in the IHPF leaving group decreases catalytic efficiency. We believe that the active site of rat chymase-1 does not specifically favor the ␤-branched aliphatic side chain of PЈ 1 Ile but that hydrophobic side chains are generally preferred here. In this regard, Le Trong et al. (3) have reported that rat chymase-1-dependent cleavage of peptides and proteins is more likely to occur between pairs of hydrophobic residues, where the PЈ 1 residue is Ile, Leu, Phe, Trp, or Tyr. Because interactions between acyl and leaving groups and between leaving group residues can influence substrate catalysis by rat chymase-1, the contribution of this PЈ 1 interaction to overall catalytic efficiency is likely to depend on the structure of the rest of the substrate. Despite its slight preference for the IHPF acyl group over the DRVY acyl group, it is likely that rat chymase-1 shows an ϳ20-fold preference for the Tyr 4 -Ile 5 bond over the Phe 8 -His 9 bond because a PЈ 1 Ile exists at the Tyr 4 -Ile 5 bond of Ang I.
General Considerations about the S and the SЈ Subsites of Human Chymase and Rat Chymase-1-Early studies on chymotrypsin and trypsin show that the main feature of serine protease specificity in these digestive enzymes is the interaction of the P 1 substrate residue with the S 1 subsite on the enzyme (14,15). The S 1 subsite is a deep pocket whose structure has been defined through crystallography and mutagenesis, and amino acid side chains that fit best in this pocket have been defined by examining the effect of substrate variations in the P 1 position on catalytic efficiency. More recent studies show that several regulatory serine proteases have a S 1 subsite preference similar to that of chymotrypsin or trypsin but show considerable preference between two or more optimal P 1 residues in polypeptide substrates. It is now clear that trypsin and chymotrypsin can also distinguish between two or more optimal P 1 residues in polypeptide substrates. To understand the basis for this preference, additional S and SЈ subsites have been examined for these enzymes using a linear approach, where the influence of several side chains at each P or PЈ position on catalysis is systematically analyzed; that is, in a polypeptide consisting of n residues, n Ϫ 1 residues are kept constant, and one residue is varied at a time. Implicit in this widely used approach is the view that a distinct S x or SЈ x subsite exists for each P x or PЈ x residue, respectively. This view is supported by the finding that observed changes in transition state stabilization energies due to multiple P or PЈ substitutions in a substrate can be predicted from the sum of transition state stabilization energies calculated from individual changes, i.e. the S and SЈ subsites function independently. For example, simple additive behavior is seen with substrate hydrolysis by trypsin (16), chymotrypsin (17), tissue-type plasminogen activator (16), and subtilisin (18). Nonadditive interactions, i.e. subsite interdependence in transition state stabilization, has also been observed with subtilisin and chymotrypsin in the hydrolysis of some substrates (18,19). In most examples of subsite interdependence, however, it is apparent that the function of one residue is compromised by mutation of another (13,18,19). In contrast, using Ang II formation versus Ang II degradation as examples, we show that human and rat chymase-1 specificities are achieved through synergistic interactions between neighboring residues of the acyl group or the leaving group as well as interactions between these groups. These observations indicate  that interactions between the binding site of human chymase or rat chymase-1 and the substrate are best explained if the entire substrate binding site is taken as an entity rather than as a collection of distinct S x and SЈ x subsites. Thus, these studies suggest that the identification of highly reactive novel substrates for human chymase as well as the design of substrate-derived inhibitors cannot be predicted from simple subsite mapping; on the other hand, combinatorial approaches are likely to be effective. This speculation is strengthened by the recent studies of Bastos et al. (20) that show highly synergistic behavior between certain P 3 -P 2 combinations in combinatorial human chymase inhibitor libraries that could not be predicted from simple subsite maps of human chymase (11). Combinatorial approaches could also prove be useful in defining the specificity of related leukocyte serine proteases such as cathepsin G where linear approaches have failed in identifying highly efficient natural substrates.