Biochemical Journal

Research article

Characterization of domain-selective inhibitor binding in angiotensin-converting enzyme using a novel derivative of lisinopril

Jean M. Watermeyer, Wendy L. Kröger, Hester G. O'Neill, B. Trevor Sewell, Edward D. Sturrock


Human ACE (angiotensin-converting enzyme) (EC is an important drug target because of its role in the regulation of blood pressure via the renin–angiotensin–aldosterone system. Somatic ACE comprises two homologous domains, the differing substrate preferences of which present a new avenue for domain-selective inhibitor design. We have co-crystallized lisW-S, a C-domain-selective derivative of the drug lisinopril, with human testis ACE and determined a structure using X-ray crystallography to a resolution of 2.30 Å (1 Å=0.1 nm). In this structure, lisW-S is seen to have a similar binding mode to its parent compound lisinopril, but the P2′ tryptophan moiety takes a different conformation to that seen in other inhibitors having a tryptophan residue in this position. We have examined further the domain-specific interactions of this inhibitor by mutating C-domain-specific active-site residues to their N domain equivalents, then assessing the effect of the mutation on inhibition by lisW-S using a fluorescence-based assay. Kinetics analysis shows a 258-fold domain-selectivity that is largely due to the co-operative effect of C-domain-specific residues in the S2′ subsite. The high affinity and selectivity of this inhibitor make it a good lead candidate for cardiovascular drug development.

  • angiotensin-converting enzyme (ACE) inhibitor
  • domain-selective inhibition
  • lisinopril
  • testis
  • X-ray crystallography


The RAAS (renin–angiotensin–aldosterone system) plays a critical role in the cardiometabolic continuum. Not surprisingly, drugs that target the RAAS, such as inhibitors of ACE (angiotensin-converting enzyme) (EC, have substantial benefits in a range of cardiovascular conditions, including hypertension and heart failure, despite a number of adverse side effects [1,2]. The somatic form of ACE (sACE) is a type I membrane-anchored dipeptidyl carboxypeptidase consisting of two catalytic extracellular domains, the N domain and C domain, each bearing the zinc-binding motif HEMGH (His-Glu-Met-Gly-His) [3]. A second isoform of this enzyme expressed exclusively in developing sperm cells, testis ACE (tACE), comprises only one extracellular domain identical with the C domain of sACE [4,5].

It is now understood that the presence of two active sites in ACE does not simply amount to a duplication of activity, but rather a dual activity, since the 45% of residues which are non-identical confer a degree of dissimilarity in kinetic properties between the N and C domains. For example, the N domain has been shown to be more thermally stable than the C domain [6], and less strongly activated by Cl ions [7]. Importantly, when interpreting the effects of inhibitor therapy, it has been known for some time that the domains show differences in substrate and inhibitor specificity; differences which were not appreciated at the time of discovery of the currently available ACE inhibitor therapies [7,8]. The best-known activities of ACE in the RAAS are the C-terminal dipeptidase reactions, whereby it converts the inactive peptide AngI (angiotensin I) into the potent vasoconstrictor AngII (angiotensin II) and degrades the vasodilator bradykinin [9,10]. The results of a number of more recent studies have suggested that, whereas both domains contribute equally to bradykinin elimination and conversion of AngI by soluble sACE that has been cleaved from its membrane anchor, the C domain is primarily responsible for the production of AngII by membrane-associated sACE [1114].

The differences in substrate specificity of the ACE domains suggest a new direction for structure-based drug design [15]. Although numerous ACE inhibitors are available for clinical use, these are not without undesirable side effects, which include persistent cough and the potentially life-threatening angio-oedema [16]. Both of these side effects have been attributed to an excess of bradykinin caused by the non-selective inhibition of both domains [1719]. It has been suggested that a C-domain-selective inhibitor might have a different clinical profile from available treatments, possibly eliminating some of the side effects common to the current non-selective ACE inhibitor therapies [20,21]. To this end, a number of highly domain-selective ACE inhibitors have been described [2225].

Using an approach that combines X-ray crystallographic structure determination with site-directed mutagenesis and enzyme kinetics, we have begun to tease apart the relative importance of active-site residues for the binding and selectivity of these novel domain-selective inhibitors [26,27]. Our studies have revealed the importance of residues in the S2′ subsite, together with the P2 Phe391 and P1 Val518 (tACE numbering; tyrosine and threonine respectively in the N domain) for the C domain selectivity of ketomethylene and phosphinic peptide inhibitors kAW {(5S)-5-[(N-benzoyl)amino]-4-oxo-6-phenylhexanoyl-L-tryptophan}, kAF {(5S)-5-[(N-benzoyl)-amino]-4-oxo-6-phenyl-hexanoyl-L-phenylalanine} and RXPA380 respectively. In the present paper, we show the X-ray crystal structure of human tACE in complex with lisW-S, a novel highly C-domain-selective derivative of the non-selective clinical ACE inhibitor lisinopril (Figure 1), together with an investigation by mutagenesis and kinetic analysis into the mechanism of domain selection by this potential lead compound.

Figure 1 Chemical structures of lisW-S and its parent compound, lisinopril

Chemical structures of lisinopril and its derivative, lisW-S, showing residue positions relative to the zinc-binding group (P1, P1′, P2′). The chiral centre determining the R- and S-enantiomers of lisW is indicated with an asterisk.


tACE constructs and preparation of purified material

A minimally glycosylated tACE mutant, tACE-G13 [28], truncated after Ser625 and lacking 36 O-glycosylated N-terminal residues as well as all but two N-glycosylation sites, was used for crystallization. Site-directed mutagenesis was performed as described previously [29] on an SphI/EcoRI fragment of a fully N-glycosylated version of this construct, tACEΔ36NJ [30], in pGEM11zf(+) (Promega). Oligonucleotides were synthesized by Inqaba Biotech, and reagents were supplied by Promega. Colonies were screened by restriction digest, and positive clones were confirmed by nucleotide sequencing. The resulting C domain mutants, having selected active-site residues converted into their N domain counterparts, were cloned into the BamHI and EcoRI restriction sites of pcDNA3.1(+) (Invitrogen) for expression purposes. Heterologous expression and purification were carried out exactly as described previously [26].


Inhibitor lisW was synthesized as described previously [31]. Separation of the enantiomers from a racemic mixture was achieved by HPLC, using a semi-preparative reversed-phase column (Tracer Excel TR-016174 120ODSB, 5 μm, 25 cm×1 cm; from Teknokroma), with a gradient elution of 26–30% (w/v) acetonitrile in 0.1% trifluoroacetic acid over 30 min, at a flow rate of 2.5 ml·min−1. Peaks were detected at the UV wavelengths 215 and 280 nm. The R-enantiomer eluted at tR=27 min, whereas the S counterpart eluted at tR=33 min. The structure and purity of the S-enantiomer of lisW were confirmed by NMR spectroscopy and MS.


Crystals were grown by microseeding as described previously [26]. Diffracting crystals were grown within 2 weeks in drops containing 250 μM lisW-S with 1.45 mg/ml tACE-G13, over precipitant solution comprising 15% (w/v) PEG [poly(ethylene glycol)] 4000, 10 μM ZnSO4 and 10 mM sodium acetate (pH 4.7).

Crystallographic data collection and processing

Data were collected from a single crystal at beamline BM14(UK) of the European Synchrotron Radiation Facility, Grenoble, France, using a wavelength of 1.033 Å (1 Å=0.1 nm) and a temperature of 100 K. HKL2000 [32] was used for data processing, and EPMR 2.5 [33] was used for molecular replacement with protein atoms from the crystal structure of tACE-G13 (PDB code 2IUL) as a model. 2FoFc, FoFc and composite omit maps (simulated annealing with 5% omission) were calculated using CNS [34] and manual model building was carried out using O version 9.0.7 [35]. CNS was also used to carry out maximum likelihood minimization and simulated annealing of selected residues, and to find water positions using water_pick. Ligand molecules were built using PRODRG [36] and ARP/wARP [37] or by hand using O version 9.0.7. Model validation was carried out using PROCHECK and SFCHECK from the CCP4 program suite [37], and Molprobity [38]. Hydrogen bonds and close contacts were identified using HBplus [39]. Figures were generated using PyMOL version 0.99 (DeLano Scientific;

Characterization of substrate hydrolysis

The hydrolysis of the fluorigenic peptide Abz-FRK(Dnp)P-OH [o-aminobenzoic acid-Phe-Arg-Lys(2,4-dinitrophenyl)-Pro-hydroxide] (a gift from Professor Adriana Carmona, Department of Biophysics, Universidade Federal de São Paulo, São Paulo, Brazil) was monitored by measuring fluorescence at λex=320 nm and λem=420 nm. Continuous assays were performed at 25 °C under initial rate conditions, with an enzyme concentration of approx. 0.2 nM and a range of substrate concentrations (0–12 μM) in 50 mM Hepes buffer (pH 6.8), containing 200 mM NaCl and 10 μM ZnCl2. Kinetic constants were calculated using the direct linear plot method [40,41].

Inhibition kinetics

An appropriate concentration range of inhibitor was added to the enzyme (~2 nM) in 50 mM Hepes buffer (pH 6.8), containing 200 mM NaCl and 10 μM ZnCl2, and incubated for 60 min at ambient temperature (22 °C). Then, 20 μl of this mixture was added to 280 μl of reaction mixture [50 mM Hepes buffer (pH 6.8), containing 200 mM NaCl, 10 μM ZnCl2 and 4 or 8 μM Abz-FRK(Dnp)P-OH], in triplicate. Fluorescence measurements were taken at λex=320 nm and λem=420 nm, at 25 °C, and inhibition constants were calculated using the Dixon method [42]. The change in Gibbs free energy of inhibitor binding (ΔG °) associated with a mutation was calculated using the following equation: Embedded Image where R is the ideal gas constant [1.98×10−3 kcal·mol−1 (1 kcal=4.184 kJ)], T is temperature (298 K), and Ki,mut/Ki,C is the ratio of mutant to wild-type Ki.


Co-crystal structure of tACE-G13 with lisW-S

We have determined a co-crystal structure of lisW-S in the active site of tACE in order to understand the molecular basis of its domain selectivity. This novel inhibitor was derived from lisinopril by incorporating tryptophan instead of proline into the P2′ position, and has been shown to demonstrate selectivity for the C domain active site of ACE [31]. Two enantiomers of lisW (Figure 1) were prepared by a three-step synthesis, and the S-enantiomer was confirmed to be strongly selective for the C domain (see Table 2). Inhibitor co-crystals were grown with a minimally glycosylated tACE construct, tACE-G13, containing only two intact glycosylation sites [28,43]. An X-ray diffraction dataset was collected from a single co-crystal and phased using molecular replacement. The resulting model was refined to a final crystallographic R-factor of 22.6%. Crystallographic data processing and refinement statistics are presented in Table 1.

View this table:
Table 1 Crystallographic data processing and refinement statistics

Values in parentheses refer to the highest resolution shell. Rmerge=[ΣhΣj|Ij(h)−<I(h)>|]/ΣhΣj|Ij(h)|, where Ii(h) and <I(h)> are the ith and the mean measurements of the intensity of reflection h respectively. Rcrysth|FoFc|/ΣhFo, where Fo and Fc are the observed and calculated structure factor amplitudes of reflection h of the working set respectively. Rfree is equal to Rcryst for h belonging to the test set of reflections. R.m.s.d., root mean square deviation.

The protein component of the co-crystal structure is in a conformation identical with that of the unliganded structure (PDB code 2IUL), as was evident from an all-atom r.m.s.d. (root mean square deviation) of 0.36. This is consistent with previously published tACE co-crystal structures, which show little change in the protein conformation [26,4446]. Overall, this structure is mostly α-helical and ellipsoid in shape, containing two Cl ions, with the catalytic Zn2+ ion buried deep in the active-site cleft (Figure 2).

Figure 2 Overview of tACE co-crystal structure with lisW-S

Cartoon representation showing an overview of the structure of lisW-S in the active site of tACE-G13; view looking into the active-site cleft. LisW-S is shown at the centre in black stick representation, next to the active Zn2+ ion (black sphere). The protein backbone is shown in cartoon representation, with Cl ions as spheres and glycan residues as sticks.

Glycan residues could only be built at Asn109, and not at Asn72 owing to weak density at this site. As in all ACE structures solved to date, surface loops 102–112, 296–299 and 434–439 showed a degree of structural disorder [43]. High B-factors and weak density indicative of disorder were also seen in the N- and C-termini as well as surface loop 154–157, while residues 37–39 (N-terminus) and 617–627 (C-terminus) could not be built. LisW-S was built into clear density present in the active site, where it takes an extended backbone conformation similar to that seen in other ACE–inhibitor co-crystal structures (Figure 3).

Figure 3 LisW-S interactions in the active site: comparison between N and C domain residues

Stick representation showing lisW-S in the first σA-weighted FoFc difference density map (purple mesh) into which the inhibitor model was built, contoured at 2σ. The active-site Zn2+ ion is shown as a green non-bonded sphere, and zinc-binding residues are shown in black lines in the background. Inhibitor-contacting residues that differ between domains are shown in yellow (tACE residues from the co-crystal structure) and purple (N domain residues from the aligned N domain structure; PDB code 2C6F) stick representation. Asp415, which forms a hydrogen bond with the P2′ tryptophan ring N, is also shown in yellow stick representation. Residue labels are given as tACE/N domain, and the P1, P1′ and P2′ inhibitor moieties are labelled.

Variation in the orientation of the P2′ tryptophan residue between inhibitors

Interestingly, the P2′ tryptophan residue of lisW-S takes a different conformation from that observed in the co-crystal structures of tACE with two other C-domain-selective inhibitors, kAW [26] and RXPA380 [44], both of which also have a tryptophan residue in this position. In lisW, the tryptophan residue is rotated through approx. 70 ° and 180 ° about side chain torsions χ1 and χ2 respectively, relative to kAW (Figure 4A). The change in orientation was demonstrated unambiguously by the electron density (Figure 3). In this orientation, the indole nitrogen makes a new hydrogen bond with Asp415 (Figures 3 and 4). This conformation is probably favoured in lisW because of hydrophobic contacts between the P2′ tryptophan ring, the P1′ lysine side chain and Val380, which is sandwiched between the two (Figure 4B). kAW and RXPA380 lack a P1′ lysine residue, so that the conformation observed in the present study is probably less constrained in these structures. The potential for a P2′ tryptophan moiety to take two different conformations highlights the large volume of the S2′ pocket, which allows this side chain to sample different orientations during binding.

Figure 4 Variation in the orientation of the P2′ tryptophan in different inhibitors

(A) Alignment of lisW-S (green sticks) with inhibitor kAW (purple sticks; PDB code 3BKL) in the active site of tACE, showing differences in the orientation of the P2′ tryptophan moiety. The structures were aligned by pairwise alignment of the α-carbons of the zinc-binding residues. The P2′ tryptophan of lisW-S is rotated about χ1 (yellow arrow) and χ2 (orange arrow), relative to kAW. Asp415 is shown in green stick representation, and the hydrogen bond with the lisW-S tryptophan ring is marked as an orange dotted line. The active Zn2+ ion is shown as a grey sphere, and the surrounding residues are shown in green line representation. (B) Hydrophobic stabilization of this new conformation of the P2′ tryptophan of lisW-S (green sticks and transparent spheres) in the active site of tACE (cartoon representation). The side chain of Val380 (orange sticks and transparent spheres) is intercalated between the P1′ lysine and P2′ tryptophan moieties.

It is interesting to note that trandolapril, a high-affinity ACE inhibitor drug compound, lacks the domain selectivity seen in lisW-S and RXPA380 despite the fact that all three compounds have bulky hydrophobic P2′ residues [8]. One possible explanation for the lack of selectivity of trandolapril is that its tryptophan-like moiety lacks an indole nitrogen, which would prevent it from forming the hydrogen bond seen here with Asp415. However, one has to take into consideration the fact that the trandolapril tryptophan-like side chain has more degrees of freedom in its six-membered ring due to the absence of conjugated double bonds, and that the five-membered ring is fused with the backbone in a similar manner to the side chain of proline in lisinopril. Thus the trandolapril P2′ side chain probably shares part of the S1′ pocket with the smaller P1′ alanine side chain. Clearly, the combination of these effects renders trandolapril less selective than compounds such as RXPA380 and lisW-S, which have an unmodified tryptophan moiety.

Active-site interactions of lisW-S

LisW-S makes 11 polar contacts (potential hydrogen bonds and ionic interactions) with 11 protein side chains in the active site (see Supplementary Table S1 at Hydrogen bonds are also made with four water molecules, two of which are conserved in the co-crystal structures of tACE with the ketone inhibitors kAW and kAF [26]. Most of these polar interactions are conserved in the structure of tACE with lisinopril (PDB code 1O86), which binds in a conformation nearly identical with that of lisW-S (Figure 5). There are some differences in the orientation of the amine group of the P1′ lysine moiety, which might be due to the influence of the P2′ tryptophan moiety, and in the zinc-co-ordination distances; however, these could equally be attributed to the lower resolution of these data in comparison with those used to determine the lisinopril structure.

Figure 5 LisW-S takes a similar conformation to lisinopril in the tACE active site

Alignment of the co-crystal structures of tACE with lisW-S (grey sticks) and parent molecule lisinopril (PDB code 1O86; black sticks) in the active site. Structures were aligned by pairwise alignment of the α-carbons of the zinc-binding residues. The Zn2+ ion is shown as a grey sphere, and adjacent protein components of both structures are shown in line representation to indicate the degree of alignment.

The non-polar atoms of lisW-S are in close proximity to 14 protein side chains (see Supplementary Table S2 at Most of these contacts (His353, Ala354, Ser355, Val379, Val380, His383, Phe457, Phe512, Ser516, Val518, Tyr523 and Phe527) are probably entropically favourable, although they include some charged moieties (His383, Asp415 and Glu143). In addition to these, nine water molecules are in close proximity to non-polar inhibitor moieties in the S1 and S2′ pockets. Non-polar contacts in the S1′ and S1 pockets are conserved in the structure of tACE with lisinopril (PDB code 1O86).

Among these polar and non-polar contacts are nine residues that differ between the N and C domains, and are thus likely to contribute to the domain-selective binding of lisW-S (Figure 3; Supplementary Tables S1 and S2). Two of these, Glu162 and Asp377 of the C domain, are close enough to the P1′ lysine side chain amine nitrogen to make ionic interactions or weak hydrogen bonds (Supplementary Table S1). Glu162 is replaced by the shorter Asp140 in the N domain, and Asp377 by the polar Gln355, substitutions which have both previously been suggested to play a role in the modest domain selectivity of lisinopril [46].

In the S2′ and S1 pockets, lisW-S contacts the same unique C domain residues that have been shown to contribute to the C domain selectivity of kAW and RXPA380, namely Glu376 and Val380 in the S2′ pocket, and Val518 in the S1 pocket (Figure 3) [26,27]. Owing to the lack of a P2 moiety in lisW, there would be no steric clash with Tyr396 of the N domain, as was seen in both kAW and RXPA380. In addition to these contacts, Glu143 and Ser516, which are replaced by Ser119 and Asp140 in the N domain respectively, come into distant contact with the P1 phenylalanine ring (Figure 3 and Supplementary Table S2). However, since these residues are charged and polar, they are unlikely to contribute favourably to lisW binding. Because of the conformational change seen in the P2′ moiety, the relative importance of the S2′ interactions is likely to be different in comparison with kAW and RXPA380.

Kinetic analysis of wild-type tACE inhibition by lisW-S

Our kinetic analyses using the fluorigenic substrate Abz-FRK(Dnp)P-OH show that lisW-S is 258-fold selective for the C domain, 64-fold more selective than lisinopril which shows only 4-fold domain selectivity (Table 2). Thus the replacement of the P2′ proline moiety with tryptophan makes this inhibitor strongly C-domain-selective. The reduction in affinity for the N domain is similar to that observed with ketone inhibitors kAW and kAF, relative to their parent compound, kAP {keto-ACE: (5S)-5-[(N-benzoyl)-amino]-4-oxo-6-phenyl-hexanoyl-L-proline} [25,26].

View this table:
Table 2 Kinetic parameters for the inhibition of ACE by lisinopril and lisW-S

Inhibition constants for wild-type tACE and N domain, as well as tACE active-site mutants containing corresponding N domain residue substitutions, using the fluorigenic peptide Abz-FRK(Dnp)P-OH, were determined using the Dixon method [42]. n.d., not determined. ED/DQ, double mutation of Glu162 and Asp377 to their N domain counterparts. VV/ST, double mutation of Val379 and Val380 to their N domain counterparts. TEVD, multiple mutation of Thr282, Glu376, Val380 and Asp453 to their N domain counterparts. TEVVD, as for TEVD but with the additional mutation of Val379.

The potency of lisW-S for the C domain determined in the present study is significantly higher than that determined previously, and indicates that this inhibitor has an affinity of the same order as lisinopril (Table 2) [31]. This observation is probably the result of the improved separation of enantiomers achieved using the present preparation protocol. Interestingly, the R-enantiomer displayed a low-micromolar Ki for the C domain (results not shown), which is approx. 1000-fold higher than that of lisW-S. This decreased affinity is probably due to steric interference between the β-carbon of the P1 side chain and the hydroxy moiety of the conserved catalytic residue Tyr523, which are only 3.2 Å apart if the R-enantiomer is built into the active site instead of lisW-S (results not shown). Such a steric effect is in keeping with the activity of ACE on biological peptides, since the P1 pseudo-phenylalanine residue of lisW-S has the chirality of an L-amino acid, whereas that of lisW-R has D-amino acid chirality.

Kinetic analysis of tACE active-site mutants

In order to test the relative importance for domain-selective binding of inhibitor-contacting residues that differ between domains, selected residues in tACE were converted into their N domain counterparts (Table 2 and Figure 6). The effect of each mutation on inhibitor binding was measured by determining the Ki, as for the wild-type enzymes.

Figure 6 Relative binding affinities of tACE mutants for lisW-S and lisinopril compared with wild-type tACE

Comparison of the relative binding affinities of tACE active-site mutants for lisinopril (black bars) and lisW-S (grey bars) with that of wild-type tACE (C domain). Values greater than zero represent a decrease in affinity relative to that of tACE, towards a more N-domain-like Ki. Ki values were not determined for the F391Y, T282S, E376D, V379S, D453E, TEVD and TEVVD mutations with lisinopril, as these residues were not expected to play a role in lisinopril binding. C dom, C domain; tACE, testis ACE (equivalent to C domain); N dom, N domain; ED/DQ, double mutation of Glu162 and Asp377 to their N domain counterparts; VV/ST, double mutation of Val379 and Val380 to their N domain counterparts; TEVD, multiple mutation of Thr282, Glu376, Val380 and Asp453 to their N domain counterparts. The active-site pockets (S2, S1, S1′, S2′) in which these residues are located are indicated.

LisW-S affinity was not affected by F391Y and V518T (in this and all subsequent instances, numbering is given for tACE, with the equivalent N domain residue following), the two single mutations found to have the biggest effect on the C domain selectivity of kAW and RXPA380 (Table 2 and Figure 6) [26,27]. In the case of Phe391, this was expected since lisW has no P2 moiety and hence does not interact with this side chain. The absence of a P2 moiety also explains the lack of change for the V518T mutation, since the P1 side chain is free to shift into the S2 subsite to avoid clashing with a threonine residue in this position. This leaves the S1′ and S2′ subsites as the major contributors to domain selectivity for lisW. Since lisinopril displays only a modest 4-fold selectivity for the C domain (Table 2), the majority of the domain selectivity displayed by lisW is likely to be contributed by the P2′ tryptophan residue.

The crystallographic data of the present study highlight the close connection between the S1′ and S2′ subsites, with Val380 making close contact with both the P1′ and P2′ side chains (Figure 4B). In the S1′ subsite, which binds the inhibitor lysine side chain, lisinopril binding was seen to remain roughly unchanged following mutation of Glu162 and Asp377 to their N domain counterparts; however, the D377Q and ED/DQ mutations did cause increases in Ki for lisW-S (2.7- and 4-fold respectively; Table 2 and Figure 6). This runs counter to the previous suggestion that the selectivity of lisinopril is the result of interactions of the lysine side chain with Glu162 in the C domain [47]; at least for lisW-S, it seems that Asp377 is the more important residue. The reason for the difference between lisW-S and lisinopril binding in this subsite is unclear, but may be due to a small conformational change in the inhibitor lysine moiety, caused by the presence of the bulky P2′ tryptophan residue in the neighbouring S2′ subsite.

In the S2′ subsite, the single mutations T282S, E376D and D453E showed only small (1.5–3.7-fold) decreases in affinity for lisW-S, relative to the wild-type C domain, whereas the mutations V379S and V380T both showed modest (3- and 6-fold respectively) increases in affinity for lisW-S and the double mutation of Val379 and Val380 to serine and threonine respectively (mutant VV/ST), showed little or no change in affinity (Table 2 and Figure 6). These results are surprising in the light of the strong domain selectivity caused by the introduction of the tryptophan residue, when compared with lisinopril (Table 2), and taken together with the contrasting result seen for the double mutant, point to co-operativity in inhibitor binding between these residues, which will be discussed further below.

In previous studies, the V379S mutation led to an increase in affinity for inhibitors kAW and RXPA380 [26,27], and the similar result seen for lisW-S in the present study suggests that this is a common feature of inhibitors having a P2′ tryptophan residue. The suggested reason for this effect is the introduction of a new water-mediated hydrogen bond with the tryptophan moiety, which would have to take on a new conformation in the mutated active site [26]. In contrast, increased affinity of V380T has not been observed previously for any C-domain-selective ACE inhibitor; rather, this residue was found to be favourable for the binding of RXPA380 and kAW [26,27]. However, when the V380T mutant was tested with the parent compound lisinopril, a similar increase in affinity was also seen, indicating that this effect is common to these two inhibitors (Table 2 and Figure 6). Whereas lisinopril and lisW differ in the P2′ position, both have a P1′ lysine side chain that also contacts Val380 (Figure 4B). These results thus suggest that the N domain threonine residue at this position is more favourable for the binding of the P1′ lysine side chain than the C domain valine residue, possibly due to the formation of a water-mediated hydrogen bond with the lysine amine.

The decrease in affinity following mutation of Glu376 and Asp453 is difficult to account for from the crystal structure, since the conformation of the tryptophan residue seen in the present study precludes the formation of any sort of water-mediated hydrogen bonds between these side chains and the tryptophan ring nitrogen. The close proximity of Glu376 to the S1′ pocket might mean that changes in the hydrogen-bonding network in this subsite due to replacement of Glu376 with aspartate could alter the affinity for the inhibitor lysine side chain, while the D453E effect may simply be due to the larger size of the glutamate side chain.

The importance of the S2′ subsite as a whole was confirmed by the simultaneous mutation of all of the S2′ subsite residues to their N domain counterparts, which resulted in a 16-fold decrease in affinity of lisW-S (mutant TEVVD; Table 2 and Figure 6). In contrast with what was seen for RXPA380, the TEVD multiple mutant, which lacks the V379S mutation, had lower affinity than that seen for TEVVD, again suggesting that the behaviour of the P2′ tryptophan moiety in this inhibitor is somewhat different from RXPA380 (Table 2 and Figure 6) [27]. The change in Gibbs free energy for both multiple mutants was markedly more positive than the summed changes of the individual mutations (Table 3), indicating that the side chains in this subsite have a co-operative rather than an additive contribution to inhibitor binding. This co-operativity probably involves mobility of the tryptophan side chain as well as of the active-site residues. From the crystal structures, it is clear that the S2′ subsite is large enough to allow for a variety of tryptophan conformations, which might conceivably vary from one mutant to the next. It also seems possible that under conditions of different pH or ionic strength, such as in the buffer used for crystallization, or in vitro, a different tryptophan conformation may be favoured to those explored under assay conditions. For the purposes of inhibitor design, these results again emphasize the large size and complexity of the S2′ subsite, and suggest that a bulkier P2′ moiety might take better advantage of the possible domain-specific interactions.

View this table:
Table 3 Effect of mutations in the S2′ subsite on Gibbs free energy of lisW-S binding

TEVD, multiple mutation of Thr282, Glu376, Val380 and Asp453 to their N domain counterparts. TEVVD, as for TEVD but with the additional mutation of Val379.


The crystal structure of lisW-S in the active site of tACE shows a similar binding mode to that observed for the parent compound, lisinopril; however, the close interaction between the P1′ and P2′ moieties of lisW-S reveals co-operation between side chains binding to the S1′ and S2′ pockets. Our mutational kinetic analysis confirms that the P2′ tryptophan residue is largely responsible for the domain selectivity of lisW-S, despite the potential mobility of the tryptophan residue in the large S2′ subsite. A small contribution is also made by the P1′ lysine moiety of lisW-S; however, this is not the case for lisinopril, indicating that the lysine moiety only plays a role in C domain selectivity in the presence of a P2′ tryptophan residue.

This combined structural and kinetic approach provides a detailed understanding of the extent to which particular residues confer selectivity, and highlights the fact that this phenomenon cannot be attributed to isolated amino acids, but rather is the result of the contribution by a number of residues. The present results implicate interactions between lisW-S and the side chains of Glu376, Asp377, Thr282, Val379, Val380 and Asp453 in conferring the potent C domain selectivity of this inhibitor. Furthermore, lisW-S has been shown to display a similar C domain potency to that of the commercially available antihypertensive drug lisinopril, with a C-selectivity comparable with that of both RXPA380 and kAW. This compound therefore shows a lot of promise for future clinical relevance.


The present study is the result of the doctoral research of Jean Watermeyer and Wendy Kröger, which was supervised by Edward Sturrock and co-supervised by Trevor Sewell. Jean Watermeyer carried out the X-ray crystallography component and structure analysis, and Wendy Kröger did the mutagenesis and enzyme kinetics work. Hester O'Neill was involved in the design of, and carried out mutagenesis on, the single tACE mutants outside the S2′ subsite, as well as the multiple mutations. The bulk of the paper was written by Jean Watermeyer, with contributions from Wendy Kröger and editorial supervision by Edward Sturrock and Trevor Sewell.


This work was supported by the Wellcome Trust (Senior International Research Fellowship to E.D.S. [grant number 070060]), the National Research Foundation of South Africa, the Deutscher Akademischer Austauschdienst and the University of Cape Town.


We thank Sylva Schwager for preparing the purified lisW-S enantiomer, Hassan Belrhali and Itai Chitapi for assisting with the data collection and Adriana Carmona for kindly donating the fluorigenic substrate Abz-FRK(Dnp)P-OH.


  • The X-ray crystallographic data and co-ordinates reported for testis angiotensin-converting enzyme in complex with the lisinopril derivative lisW-S have been deposited in the Protein Data Bank under accession code 3L3N.

Abbreviations: Abz-FRK(Dnp)P-OH, o-aminobenzoic acid-Phe-Arg-Lys(2,4-dinitrophenyl)-Pro-hydroxide; ACE, angiotensin-converting enzyme; AngI, angiotensin I; AngII, angiotensin II; kAF, (5S)-5-[(N-benzoyl)-amino]-4-oxo-6-phenyl-hexanoyl-L-phenylalanine; kAP, keto-ACE, (5S)-5-[(N-benzoyl)-amino]-4-oxo-6-phenyl-hexanoyl-L-proline; kAW, (5S)-5-[(N-benzoyl)amino]-4-oxo-6-phenylhexanoyl-L-tryptophan; RAAS, renin–angiotensin–aldosterone system; sACE, somatic ACE; tACE, testis ACE


View Abstract