Angiotensin I-converting Enzyme Transition State Stabilization by His1089

Angiotensin (Ang) I-converting enzyme (ACE) is a member of the gluzincin family of zinc metalloproteinases that contains two homologous catalytic domains. Both the N- and C-terminal domains are peptidyl-dipeptidases that catalyze Ang II formation and bradykinin degradation. Multiple sequence alignment was used to predict His1089 as the catalytic residue in human ACE C-domain that, by analogy with the prototypical gluzincin, thermolysin, stabilizes the scissile carbonyl bond through a hydrogen bond during transition state binding. Site-directed mutagenesis was used to change His1089 to Ala or Leu. At pH 7.5, with Ang I as substrate,k cat/K m values for these Ala and Leu mutants were 430 and 4,000-fold lower, respectively, compared with wild-type enzyme and were mainly due to a decrease in catalytic rate (k cat) with minor effects on ground state substrate binding (K m ). A 120,000-fold decrease in the binding of lisinopril, a proposed transition state mimic, was also observed with the His1089 → Ala mutation. ACE C-domain-dependent cleavage of AcAFAA showed a pH optimum of 8.2. H1089A has a pH optimum of 5.5 with no pH dependence of its catalytic activity in the range 6.5–10.5, indicating that the His1089 side chain allows ACE to function as an alkaline peptidyl-dipeptidase. Since transition state mutants of other gluzincins show pH optima shifts toward the alkaline, this effect of His1089 on the ACE pH optimum and its ability to influence transition state binding of the sulfhydryl inhibitor captopril indicate that the catalytic mechanism of ACE is distinct from that of other gluzincins.

Angiotensin I (Ang I) 1 -converting enzyme (ACE, EC 3.4.15.1, peptidyl-dipeptidase A) is a chloride-activated peptidase with broad substrate specificity that releases a C-terminal dipeptide from substrates (1). The somatic form of human ACE has two homologous catalytic domains (1). These N-and C-domains most likely are the result of an ancient gene duplication event that occurred during vertebrate evolution (2). Invertebrate ACE has a single catalytic domain (3). The physiological substrates of ACE include Ang I, bradykinin, substance P, and AcSDKP. AcSDKP, the principal substrate of N-domain ACE, may play a role in hemopoietic cell differentiation, whereas both domains are thought to be important for regulating tissue and blood levels of the vasoactive hormones angiotensin II (Ang II) and bradykinin. Inhibition of these ACE activities has proved to be important in the treatment of hypertension and congestive heart failure (4).
ACE belongs to the gluzincin family (clan MA) of metalloproteases, of which thermolysin is the prototypical member (5). With thermolysin, catalysis occurs by a general base-type mechanism (6,7). The proposed mechanism involves the displacement of a zinc-bound water molecule by the approaching substrate followed by the attack of this water molecule on the scissile carbonyl bond to form an oxyanion (Fig. 1). The attack of the water molecule is facilitated by the active site glutamic acid, Glu 143 . The resulting tetrahedral intermediate subsequently decomposes to yield the products. The negative charge on the scissile bond carbonyl oxygen that develops during transition state binding is stabilized by hydrogen bonding interactions with a protonated His, His 231 , and a tyrosine, Tyr 157 , in the active site. His 231 is thought to be maintained in a protonated state through a hydrogen bonding interaction with Asp 226 . However, the functional consequence of the interaction between His 231 and Asp 226 , inferred on the basis of crystallographic evidence, has not been confirmed by site-directed mutagenesis studies. The crucial stabilization of the oxyanion by His 231 occurs after the formation of the Michaelis enzymesubstrate complex and greatly influences catalytic rate.
The catalytic mechanism in ACE is not known but has been inferred on the basis of that proposed for thermolysin. Some key active site residues in ACE have been determined because of similarity in local primary structure, such as those in the HEXXH and the EXXD 2 motifs. Thus it is known that His 959 , His 963 , and Glu 987 are the zinc-ligating residues in human ACE C-domain, and Glu 960 likely acts as a general base during catalysis (1,8). Although it is realized that ACE is related by divergent evolution to thermolysin and neprilysin, key ACE residues involved in catalysis and substrate binding, other than the zinc-ligating residues and Glu 960 , have not been identified because the estimated overall primary sequence identity between ACE and other gluzincins is less than 15%. To construct structural models of ACE in the absence of a crystal structure, it is necessary to identify key, structurally conserved catalytic and substrate-binding residues so that the primary structures can be aligned with less ambiguity. Here we report the identification of the catalytic His, His 1089 , in human ACE C-domain equivalent to His 231 in thermolysin. These studies indicate for the first time that the structure of ACE, C-terminal to the zinc-ligating residues, shows similarity to thermolysin and neprilysin. We show that the His 1089 side chain allows ACE to function as an alkaline peptidyl-dipeptidase, and its interactions with inhibitors reveal major differences in binding modes when compared with other gluzincins.

EXPERIMENTAL PROCEDURES
Construction of Human ACE C-domain Gene-A human ACE Cdomain gene with 47 unique restriction sites was designed by strategies used previously (9) and was chemically synthesized and cloned into the shuttle expression vector pcDNA3 (Invitrogen). The synthetic gene encodes amino acids 1-29 (signal peptide) and 611-1201 of human somatic ACE with an 8-residue FLAG epitope recognized by a commercially available antibody (M2, Sigma) at the C terminus. This ACE synthetic gene does not include the transmembrane sequence found in somatic ACE. Mutations were constructed in the synthetic gene by cassette mutagenesis or site-directed mutagenesis, and all mutations were confirmed by DNA sequence analysis.
Transfection of COS-7 Cells-COS-7 cells (ATCC) were cultured under an atmosphere of 5% CO 2 at 37°C and transfected with plasmid DNA using the Gene Pulser system (Bio-Rad). 12 h after transfection, cells were washed and further cultured with serum-free Dulbecco's modified Eagle's media (Life Technologies, Inc.) for 48 -72 h. C-domain ACE and its mutants released into conditioned media were collected and used as the starting point of the purification. The following ACE C-domain mutations were made: H1002A, H1089A, H1089L, D1083A, D1083N, and D1083A/H1089A. Fig. 2 shows the effect of these mutations on glycosylation patterns (Western blot) and protein expression levels. A polyclonal antiserum generated against pure human kidney ACE (gift from Dr. R. Ramchandran, Cleveland Clinic Foundation) was used in Western blots. Expression levels of H1089A and H1089L proteins by COS-7 cells was similar to that of wild-type ACE C-domain and that of D1083A, D1083N, and D1083A/H1089A was Ϸ50% lower (Fig. 2B). Glycosylation pattern was not affected by these mutations (Fig. 2A).
Purification of Recombinant Human C-domain ACE and Its Mutants-An anti-FLAG M2 affinity gel was used to purify human Cdomain ACE according to the procedure described by the manufacturer (Sigma). This ACE protein preparation was Ͼ95% pure as determined by Coomassie Blue staining, and the protein amount was quantified by amino acid analysis. This preparation of human ACE C-domain was used as a standard. For kinetic studies, human C-domain ACE and its mutants were partially purified by ion-exchange HPLC using a Bio Scale Q2 column (Bio-Rad). Cell culture media containing ACE Cdomain or its mutant was dialyzed against 20 mM Tris-HCl buffer, pH 7.5, containing 20 mM NaCl. The dialyzed media (5-10 ml) was applied to the ion-exchange HPLC column which was developed using a 45-min linear NaCl gradient (20 -500 mM) at a flow rate of 1 ml/min. Two to three 1-ml fractions with the highest activity were pooled. The purity of recombinant ACE in the pooled peak fractions was Ϸ50%, with bovine serum albumin as the only detectable contaminant. ACE was not further purified to remove the contaminating bovine serum albumin since the stability of pure (Ͼ95%) ACE C-domain at 4°C was markedly lower than that of partially purified ACE containing bovine serum albumin.
Peptides-Peptides used in this study were synthesized by The Protein Core Facility, The Cleveland Clinic Foundation or Auspep (Parkville, Australia). 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 determined by amino acid analysis.
Enzymes and Enzyme Kinetics-To determine K m and V max values for human ACE C-domain reactions, initial velocities (v) were determined as described by us previously (10). Twelve concentrations of Ang I ranging between 5 and 250 M were incubated with wild-type or mutants forms of ACE at 37°C in 50 mM HEPES buffer, pH 7.5, containing 50 mM NaCl and 10 M ZnSO 4 (final volume 50 l) for 30 -60 min. Enzyme concentration was adjusted to between 0.1 and 16 nM to ensure that Ͻ15% of the substrate was consumed at the lowest substrate concentration. Under these conditions, product formation was linear with respect to time over the duration of the incubation. For pH dependence studies, AcAla-Phe-Ala-Ala-COOH was used as substrate at a concentration (40 M) 10-fold lower than its K m value, and thus variations in activity with respect to pH reflect changes in k cat /K m values. Buffers used in pH studies 3 are as follows: 50 mM sodium acetate containing 10 mM ZnSO 4 , 50 mM MES containing 0.1 mM ZnSO 4 , 50 mM HEPES containing 10 M ZnSO 4 , and 50 mM CHES containing 10 M ZnSO 4 for the pH ranges 3.4 -5.6, 5.5-6.9, 6.7-8.8, and 8.4 -10.5, respectively. Reactions were terminated by the addition of 70 l of ice-cold 0.1% trifluoroacetic acid or 2% H 2 PO 4 . The resulting solution (100 l) was applied to a C 18 reverse phase HPLC column (XTerra RP18 3.5 m, 4.6 ϫ 50 mm column, Waters, Milford, MA). The column was developed with linear acetonitrile gradients containing either 25 mM triethylammonium phosphate buffer, pH 3.0 or 7.0, or 0.1% trifluoroacetic acid at a flow rate of 2 ml/min. The column effluent was monitored at 214 nm. The elution positions of Ang I, Ang II, AcAla-Phe-Ala-Ala-COOH, and AcAla-Phe-COOH were determined us-3 ZnSO 4 concentrations were increased progressively as the buffer pH was reduced, because zinc begins to dissociate from the enzyme below pH 7 (32). The concentrations of zinc used were those used by Shapiro et al. (33).
FIG. 1. Proposed mechanism of thermolysin-catalyzed hydrolysis. The reaction proceeds by the attack of the zinc-bound water molecule on the scissile bond. The resulting tetrahedral intermediate, with enzyme residues involved in its stabilization, is shown. This subsequently decomposes to yield the products. Adapted from Matthews (6). The nomenclature used for the individual amino acids (P 1 , P 1 Ј, etc.) of a substrate and the subsites (S 1 , S 1 Ј, etc.) of the enzyme is that of Schechter and Berger (34). Amino acid residues of substrates numbered P 1 , P 2 , etc. are toward the N-terminal direction, and P 1 Ј, P 2 Ј, etc. are toward the C-terminal direction from the scissile bond. ing pure synthetic standards. The peak area corresponding to Ang II or AcAla-Phe-COOH was integrated to calculate product 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 v ϭ V max ϫ [S]/(K m ϩ [S]). Correlation coefficients were routinely Ͼ0.99 but never Ͻ0.97.
Assays with inhibitors were performed in 50 mM HEPES buffer, pH 7.5, containing 50 mM NaCl and 10 M ZnSO 4 at 37°C with 40 M Ang I as substrate in a total volume of 50 l; incubation period was 30 min.
Prior to the addition of substrate (5 l), enzyme was preincubated with inhibitors for 1 h. K i values were determined from plots of enzyme activity versus inhibitor concentration where inhibitor concentration was corrected by using the K m and concentration of the substrate used for the competition ( Pure human wild-type ACE C-domain of known concentration was used in a Western blot protocol to determine enzyme concentration of partially purified ACE and mutant ACE enzyme preparations. The overall rate constant k cat was calculated by the formula , ⌬⌬G binding , and ⌬⌬G cat , were calculated as described by Wells (11). ⌬⌬G T ‡ , ⌬⌬G binding , and ⌬⌬G cat represent the difference between two enzymes in the free energy required for transition state stabilization (i.e. difference in free energy required to reach E⅐S ‡ from E ϩ S), to form the enzyme-substrate complex (E⅐S) from E ϩ S, and to convert the E⅐S complex to the transition state complex (E⅐S ‡ ), respectively.

RESULTS AND DISCUSSION
Prediction of His 1089 as a Transition State Stabilizing Residue in Human ACE C-domain-In thermolysin, thermolysinlike enzymes, and neprilysin, in which the identity of the catalytic His has been confirmed by site-directed mutagenesis (12,13) and x-ray crystallography (14,15), this residue is found C-terminal to the EXXD zinc-binding motif and is preceded by a conserved Asp with 1 or 4 intervening residues. The distance between the EXXD motif and the transition state His is 49 -60 residues in vertebrate neprilysin-like and bacterial thermolysin-like sequences.
Human ACE C-domain possesses two conserved histidines C-terminal to the EXXD zinc-binding motif (Fig. 3). The first of these, His 1002 , is 10 residues downstream of the EXXD motif and does not possess a nearby upstream Asp. The second, His 1089 , is 97 residues downstream of the EXXD motif and is preceded by an Asp with 5 intervening residues. A conserved Pro two residues downstream of the catalytic His in neprilysinlike sequences is also found two residues downstream of His 1089 in vertebrate ACEs. On the basis of these features, we predicted and tested if His 1089 is the transition state stabilizing His.
Design and Properties of Human ACE C-domain-Human ACE has two fully functional catalytic domains, N and C. To assign function to individual residues in human ACE through kinetic analyses, it is necessary to express each domain separately or to inactivate the domain not under study. Both approaches have been used previously (16). We constructed a synthetic human ACE C-domain gene consisting of exons 14 -25 (coding for residues 611-1201) of the human ACE gene. This region is homologous to invertebrate ACE and to exons 1 (excluding the signal peptide ϩ 6 residues) to 12 of the human ACE gene, that is the ACE N-domain. The human ACE Cdomain construct contains the catalytic residues and sequence up to the region that is reported to be cleaved during ACE shedding from the plasma membrane but does not contain the unique C-terminal transmembrane spanning and cytosolic domains encoded by exon 26. K m , k cat , and k cat /K m values for Ang I conversion to Ang II by human ACE C-domain were 21 Ϯ 2.5 M, 18.8 Ϯ 1.8 s -1 , and 0.9 M⅐s -1 , respectively (n ϭ 3). These values are similar to those obtained for Ang I to Ang II conversion by full-length human ACE where the N-domain is inactivated through mutation of the zinc-ligating residues (17).
Role of His 1089 in Ground State and Transition State Substrate Binding-Substrate binding to ground and transition states of wild-type ACE C-domain and its mutants was studied using Ang I. The results are summarized in Table I. The His 1089 to Ala or Leu mutations in human ACE C-domain produced a small increase in K m and a large decrease in k cat values. A loss in transition state binding energy of 3.74 kcal⅐mol Ϫ1 is associated with the His 1089 to Ala mutation and 5.13 kcal⅐mol Ϫ1 with the Leu mutation. This loss of transition state binding energy is consistent with the loss of a strong hydrogen bond in the tetrahedral intermediate as would be expected between a strong hydrogen bond donor and acceptor. Equivalent mutations in a thermolysin-like enzyme produced a change in ⌬G T ‡ of 3.8 kcal⅐mol Ϫ1 in transition state binding of substrate (12), and in neprilysin the change in ⌬G T ‡ was 2.2 kcal⅐mol Ϫ1 (13). No change in phenotype was observed with the His 1002 3 Ala mutation (data not shown), and thus further studies with this His mutant were not performed.
A key feature of the catalytic His in gluzincins is that it greatly impacts transition state binding with minimal effects on ground state binding. If K m is equal to the true dissociation constant (K s ) for the enzyme and substrate, the 4.5-fold increase in K m associated with the His 1089 to Ala mutation would correspond to a 0.93 kcal⅐mol Ϫ1 loss in binding energy of the substrate to the ground state structure. The effect of the His 1089 to Ala mutation on ground state structure of the enzyme was also examined using [Phe 8 --His 9 ]Ang I, an Ang I analog where the scissile amide group (-CONH-) in Ang I is replaced by a aminomethylene (-CH 2 NH-) isostere, so that  essential contacts required for transition state binding are absent (Fig. 4, A and B). Loss of [Phe 8 --His 9 ]Ang I binding energy associated with the His 1089 to Ala substitution was 0.05 kcal⅐mol Ϫ1 (Fig. 5A and Table II). These findings indicate that ground state structure is not altered in the His 1089 to Ala mutant. His 1089 and Glu 960 in ACE C-domain play an important role in catalysis. The pK a values of the His imidazole group and the Glu side chain carboxylate group are expected to be in the range 5 to 8 and 2 to 5.5, respectively (18). To study the role of these residues in determining pH dependence of activity, we used the artificial ACE substrate AcAla-Phe-Ala-Ala-COOH. This substrate is preferable to Ang I in that it has no ionizable groups in the pH range 6 -9. The pH optimum of ACE C-domaindependent cleavage of AcAla-Phe-Ala-Ala-COOH is Ϸ8 (Fig.  6A). It is therefore expected that pH dependence of ACE Cdomain activity at around neutral pH will be greatly influenced by His 1089 . Indeed, pH dependence of activity in the pH range 6.5-10.5 was not evident in H1089A, and the mutant was Ϸ100-fold less active than wild type (Fig. 6B). Thus, His 1089 is a major determinant in phenotype in human ACE C-domain that is responsible for the alkaline pH optimum of this enzyme. In H1089A, a sharp pH optimum was observed at 5.5 (Fig. 6B).
The maximal activity at pH 5.5 with H1089A was similar to that observed with wild-type enzyme at this pH. The H1089A mutation likely unmasks the pK a of the remaining ionizable group, and the sharp increase in activity between pH 5 and 5.5 is likely due to the deprotonation of the carboxylate side chain of Glu 960 that facilitates the attack of the zinc-bound water on the peptide carbonyl bond (Fig. 4A). The rapid quenching of  activity between pH 5.5 and 6 was unexpected and suggests that deprotonation of a carboxylate other than that of Glu 960 hinders catalysis. This observation differs markedly from that seen with a thermolysin-like enzyme where the pH optimum shifts from 6 to 8 when a similar mutation is introduced in place of the catalytic His (12).

Role of His 1089 in Captopril and Lisinopril
Binding-Several classes of metalloprotease inhibitors have been described, and their mode of binding with thermolysin and thermolysin-like enzymes has been studied by crystallography. The different classes of inhibitors studied include the mercaptans (19,20), hydroxymates (21), carboxylates (22,23), and phosphoramidates (24). All these inhibitors are coordinated to the catalytic zinc, but interactions with the transition state stabilizing residues His 231 and Tyr 157 differ. Carboxylates and phosphoramidates are hydrogen-bonded to His 231 and Tyr 157 . Hydroxymates form a hydrogen bond with His 231 but do not interact with Tyr 157 , whereas mercaptans do not display hydrogen bonding interactions with either of these residues. Mutation studies with thermolysin have confirmed the presence or absence of these interactions with His 231 . To examine the similarities and differences between ACE and other gluzincins, we studied the His 1089 interaction with two distinct inhibitors of human ACE as follows: lisinopril, a carboxylate inhibitor, and captopril, a mercaptan inhibitor.
The predicted mode of binding of lisinopril to the ACE Cdomain is shown in Fig. 4D and is based on crystallographic studies with thermolysin. The His 1089 to Ala mutation produced a 7.2 kcal⅐mol Ϫ1 loss in the binding energy of lisinopril ( Fig. 5C and Table II) suggesting that the bidentate mode of binding of the carboxylate group resembles the presumed geometry of the tetrahedral transition state, with a hydrogen bond being formed between one oxygen of the carbonyl group and His 1089 . Unexpectedly, the His 1089 to Ala mutation produced a 5.3 kcal⅐mol Ϫ1 loss in the binding energy of captopril ( Fig. 4B and Table II) suggesting a direct hydrogen bonding interaction between the imidazole N-H and the captopril sulfur, presumably in the anionic form (23). Fig. 4C illustrates the presumed binding scheme. N-H⅐⅐⅐⅐S hydrogen bonds are described in protein data bases (25,26). Because S is a weak hydrogen bond acceptor, N-H⅐⅐⅐⅐ hydrogen bonds are generally weak (Ͻ-1.2 kcal⅐mol) (27); however, strong hydrogen bonds with S as an acceptor have also been described (28) and could operate in ACE C-domain-captopril complexes.
Previous studies have shown differences between inhibitor binding to thermolysin and neprilysin, as compared with ACE. For example, retro-inverso modification of the amide bond (from -CONH-to -NHCO-) has little effect on inhibitory potencies of mercaptan inhibitors of thermolysin and neprilysin, but this change produces a marked decrease (Ͼ1,000-fold) in ACE inhibitor potency (12,29). These findings illustrate differences in the substrate binding pocket of these gluzincins distal to the site that interacts with the scissile carbonyl bond. Our findings with captopril show for the first time that atoms arranged around the tetrahedral carbon designed to mimic the transition state of the substrate interact in a markedly different manner in ACE compared with thermolysin-like enzyme and neprilysin.
Does Asp 1083 Interact with His 1089 ?-In several gluzincin crystal structures, including those of thermolysin and neprilysin, the transition state His is located in close proximity to an Asp side chain, such that a hydrogen bonding network between these residues has been proposed (15,30). It is believed that this interaction appropriately positions the imidazole side chain and maintains it in a protonated state. In gluzincins with known structures, this Asp is located 2-5 residues N-terminal to the transition state His in the primary structure, but the carboxyl head group superimposes in the tertiary structure. In ACE C-domain, a conserved Asp, Asp 1083 , is located 6 residues N-terminal to His 1089 . To study if Asp 1083 interacts functionally with His 1089 , the effects of single His and Asp mutations to Ala were compared with the double mutant D1083A/H1089A. The results are summarized in Table I. The main effect of the Asp 1083 to Ala mutation was on k cat resulting in a 1.8 kcal⅐mol -1 loss in transition state binding energy (⌬⌬G T ‡ ). The combined effect of the two individual mutations on ⌬⌬G T ‡ was additive to within 7% of that observed in the double mutant indicating that these residues stabilize the transition state by independent mechanisms. A potential hydrogen bonding interaction between Asp 1083 and His 1089 should also influence the pK a of the catalytic imidazole group. If there is such an interaction in ACE C-domain, analogous to that in thermolysin and neprilysin, the loss of this Asp interaction should acid-shift the pH optimum of ACE C-domain. The pH optimum of D1083A, however, is not lower than that of the wild-type enzyme (Fig. 6C). This indicates again that Asp 1083 does not influence the orientation or protonation state of the His 1083 side chain, as is evident from ⌬⌬G T ‡ additivity analyses. The chief effect of the Asp 1083 to Ala or Asn mutation is on transition state binding, since these mutants display a 1.8 or 3.1 kcal⅐mol -1 loss in Ang I binding energy, respectively (Table  I). However, these mutations minimally affect either captopril binding or lisinopril binding (0.91 kcal⅐mol Ϫ1 ) which mimics the tetrahedral intermediate insofar as the scissile bond is concerned ( Fig. 7 and Table II). These findings indicate that in the transition state, Asp 1083 is directly or indirectly involved in making contacts between Ang I and ACE C-domain but is not involved in those interactions that are directed toward the scissile bond. In producing this effect, Asp 1083 appears to make important ionic interactions since the Asp 1083 to Asn mutation, which retains hydrogen bonding potential but is unable to make ion pair interactions, was unable to mimic the requirement of Asp 1083 in transition state substrate binding (Table I).  Table I. In these respects, ACE differs significantly from thermolysin and neprilysin where Asp 226 -His 231 and Asp 709 -His 711 interactions are observed, respectively, in the crystal structure (15,30). Moreover, ACE differs from other gluzincins such as Vibrio anguillarum vibriolysin, where the Asp equivalent to Asp 226 of thermolysin is naturally replaced by Asn.
Summary and Conclusions-We show that His 1089 in human ACE C-domain stabilizes the rate-limiting transition state by 3.7 kcal⅐mol Ϫ1 with minor effect on the ground state structure. The catalytic His is known to stabilize the transition state by making a hydrogen bond with the oxyanion in the tetrahedral intermediate. In human ACE C-domain, this interaction, schematically illustrated in Fig. 4A, has a binding energy equivalent to that of a strong hydrogen bond. A similar loss of binding energy of lisinopril, an inhibitor of ACE that closely mimics the transition state configuration of the scissile bond, is observed with the His 1089 to Ala mutation. The presence of an Asp 1083 -(Xaa) 5 -His 1089 motif in human ACE C-domain equivalent to that in other gluzincins and the selective effect of the His 1089 Ala mutation on transition state binding imply that His 1089 in human ACE C-domain is the catalytic His equivalent to His 231 in thermolysin and His 711 in neprilysin. Because of the high homology between the N-and C-domains of ACE, it is likely that the equivalent residue in human ACE N-domain, His 491 , is also a catalytic residue.
The overall structure of thermolysin consists of two roughly spherical lobes with a deep cleft between the two that is used to coordinate the catalytic zinc and to bind the substrate (31). Helical structure predominates in the C-terminal lobe, whereas ␤-structure predominates in the N-terminal lobe. Although neprilysin is a much larger protein, the crystal structure shows it to be structurally similar to thermolysin in the C-terminal half of the N-terminal lobe and most of the C-terminal lobe (15). These regions surround the active site and include the zincligating residues and transition state His. Based on the presence of near-identical primary structural motifs for the zincligating residues in ACE N-and C-domains and thermolysin, it is expected that tertiary structure of ACE around the zincbinding sites will be conserved. The prediction that His 1089 is a catalytic residue in ACE was initially based on multiple sequence alignment. This alignment is likely to be largely correct on the basis of our experimental findings and suggests for the first time a more extended conservation between the C-terminal lobe tertiary structure of thermolysin, neprilysin, and ACE.
The pH optimum of human ACE (Ϸ8.0) is higher than that reported for other gluzincins (5.5 to 7.5). pH dependence of human ACE C-domain activity in the pH range 6.5 to 9 is entirely due to His 1089 , such that H1089A is an acid peptidyldipeptidase with a pH optimum Ϸ5.5. The marked effect of the catalytic His on pH dependence is an unusual design feature of this gluzincin. The mode of binding of inhibitors to thermolysin has been well documented (19 -24) and has served as a model of how the different classes of inhibitors bind to gluzincins. Here we provide evidence for a marked difference in the binding mode of a mercaptan to ACE. These variations in the binding pocket suggest that it is possible to design new types of ACE inhibitors, not previously anticipated from studies with other gluzincins. These differences between ACE and thermolysin in the role of the catalytic His in influencing the pH optimum and in the mode of binding of transition state inhibitors question the general use of thermolysin-based mechanistic models for predicting structure-based function in gluzincins.