Abstract

Angiotensin-converting enzyme (ACE) is a zinc metalloprotease best known for its role in blood pressure regulation. ACE consists of two homologous catalytic domains, the N- and C-domain, that display distinct but overlapping catalytic functions in vivo owing to subtle differences in substrate specificity. While current generation ACE inhibitors target both ACE domains, domain-selective ACE inhibitors may be clinically advantageous, either reducing side effects or having utility in new indications. Here, we used site-directed mutagenesis, an ACE chimera and X-ray crystallography to unveil the molecular basis for C-domain-selective ACE inhibition by the bradykinin-potentiating peptide b (BPPb), naturally present in Brazilian pit viper venom. We present the BPPb N-domain structure in comparison with the previously reported BPPb C-domain structure and highlight key differences in peptide interactions with the S4 to S9 subsites. This suggests the involvement of these subsites in conferring C-domain-selective BPPb binding, in agreement with the mutagenesis results where unique residues governing differences in active site exposure, lid structure and dynamics between the two domains were the major drivers for C-domain-selective BPPb binding. Mere disruption of BPPb interactions with unique S2 and S4 subsite residues, which synergistically assist in BPPb binding, was insufficient to abolish C-domain selectivity. The combination of unique S9–S4 and S2′ subsite C-domain residues was required for the favourable entry, orientation and thus, selective binding of the peptide. This emphasizes the need to consider factors other than direct protein–inhibitor interactions to guide the design of domain-selective ACE inhibitors, especially in the case of larger peptides.

Introduction

Angiotensin-converting enzyme (ACE, EC 3.4.15.1) is a zinc metalloprotease with two catalytically active domains, the N- and C-domain, which is central to the regulation of blood pressure via the renin–angiotensin–aldosterone and kallikrein–kinin systems [1]. Generation of the potent vasoconstrictor angiotensin II (AngII) from the angiotensin I (AngI) decapeptide occurs mainly through the dicarboxypeptidase action of the ACE C-domain [2,3]. Additionally, the two homologous domains (60% sequence identity and 89% active site similarity [4]) function to hydrolyze the vasodilatory peptide bradykinin (BK), with the C-domain being slightly more efficient [5,6]. ACE is thus an attractive target for antihypertensive treatment and, indeed, non-domain-selective ACE inhibitors are commonly used in the clinic. Their use, however, has been associated with the side effect of life-threatening angioedema [7] which is believed to be due to the systemic build-up of BK upon inhibition of both ACE domains. Preliminary evidence suggests that a new generation of C-domain-selective ACE inhibitors could effectively decrease blood pressure with reduced side effects by allowing BK hydrolysis through the active N-domain [1,8,9].

Design of the first ACE inhibitor, captopril, was inspired by the structure of natural bradykinin-potentiating peptides (BPPs) present in venom of the Brazilian pit viper (Bothrops jararaca) [10]. After envenoming, these peptides result in hypotensive shock of the prey by inhibiting ACE, thereby preventing BK hydrolysis, and potentiating the action of BK on its B2 receptor possibly through cross-talk with ACE [1113]. To date, numerous BPPs have been isolated from the venom of snakes [1416], spiders [17], scorpions [18] and secretions of amphibian skin [19,20]. BPPs isolated from snake venom are typically 5–13 amino acids in length and contain an N-terminal pyroglutamate (Pca) residue and C-terminal -Ile-Pro-Pro motif. This motif is required for BK potentiation and only present in venom derived from male snakes [21].

Interestingly, some BPPs display domain-selective inhibition of ACE. Using single ACE active site knockout mutants, Cotton et al. [22] observed that inhibition by BPPb was 360-fold C-selective whereas BPPa (also known as BPP9a or teprotide) was 160-fold C-selective and BPPc non-selective (Figure 1, Table 1). In a previous study by Hayashi and Camargo [11], BPP10c (given as BPP2 in Cotton et al.) displayed 400-fold C-selectivity. Moreover, mice treated with BPP10c showed undesired alterations in spermatogenesis [23,24]. With the current research, we therefore sought to uncover the molecular basis for BPPb C-domain selectivity to identify crucial moieties and guide the future design of domain-selective ACE inhibitors.

Schechter-Berger diagram of BPP peptides binding to the ACE active site.

Figure 1.
Schechter-Berger diagram of BPP peptides binding to the ACE active site.

Residues differing between the C-selective BPPb and non-selective BPPc are given in bold.

Figure 1.
Schechter-Berger diagram of BPP peptides binding to the ACE active site.

Residues differing between the C-selective BPPb and non-selective BPPc are given in bold.

Table 1
Binding affinity (Ki) and domain selectivity values of BPP peptides previously determined using ACE proteins containing a single functional active site
 Ki (nM) C-domain selectivity Ki(N-domain)/Ki(C-domain) Reference 
C-domain N-domain 
BPPb 25 9000 360 [22
BPPa 0.5 80 160 [22
BPPc 70 55 0.79 [22
BPP21 220 110 [22
BPP10c1 0.5 200 400 [11
 Ki (nM) C-domain selectivity Ki(N-domain)/Ki(C-domain) Reference 
C-domain N-domain 
BPPb 25 9000 360 [22
BPPa 0.5 80 160 [22
BPPc 70 55 0.79 [22
BPP21 220 110 [22
BPP10c1 0.5 200 400 [11
1

The peptide sequences for BPP2 and BPP10c are identical.

Materials and methods

Construction of ACE mutants

Truncated wild-type N-domain [25] and C-domain [26], C-domain mutants (S516N [27], E403R [26] and C1–163 Ndom-ACE [28]) and N-domain mutant SEDSTE_YR [29] were previously cloned into pcDNA3.1+ (Invitrogen) for mammalian cell expression. The wild-type C-domain construct in pBlueScript SK(II)+ (pBS SK(II)+, Stratagene) was used to create the single mutant E143S whereas S516N pBS SK(II)+ was used as template for addition of S517V to yield the SS/NV double mutant via PCR-based site-directed mutagenesis. The triple mutants SSE143S and SSE403R were subsequently created using sequential site-directed mutagenesis reactions with SS/NV pBS SK(II)+ as template. For expression in mammalian cells, the mutant constructs were cloned into pcDNA3.1+ using the BamHI and EcoRI restriction sites. The quadruple mutant SSEE was then created by sub-cloning the region containing E143S mutation into SSE403R pcDNA3.1+ using NotI and Kpn2I. Positive incorporation of desired mutations in all mutant constructs was confirmed by nucleotide sequencing.

Heterologous expression of wild-type and mutant ACE

The constructs were transfected into Chinese hamster ovary (CHO-K1) cells using the calcium phosphate Profection® Mammalian Transfection System (Promega Corp.), heterologously expressed and the proteins purified from culture medium via lisinopril–sepharose affinity chromatography, as previously described [30]. Protein purity was assessed by SDS–PAGE and Coomassie staining. Purified protein was quantified by absorbance with the use of a NanoDrop® spectrophotometer and extinction coefficients of 162 070 M−1 cm−1 and 143 620 M−1 cm−1 for the N- and C-domains, respectively.

Kinetic characterization of substrate hydrolysis

ACE hydrolysis of the peptide Z-phenylalanylhistidylleucine (Z-FHL) (Bachem Ltd.) was performed as described by Schwager et al. [31]. Enzyme was diluted to 1.25 nM (or 6.6 nM in the case of SSEE) and a serial dilution of Z-FHL prepared at 10 concentrations using a 100 mM potassium phosphate buffer, pH 8.3, containing 300 mM NaCl and 10 µM ZnSO4. Equal volumes of enzyme and substrate were mixed and incubated for 15 min at 37°C for hydrolysis to occur and the rest of the procedure followed as described [31]. Kinetic constants were calculated by analysis of the initial reaction velocities (at <10% substrate hydrolysis) with the Michaelis–Menten model in GraphPad Prism v.6.0.

Kinetic characterization of inhibitor binding

To evaluate BPPb-binding affinity, a modified Z-FHL assay was used. Lyophilized BPPb (AbbiotecTM) was dissolved in deionized water and a serial dilution prepared in 100 mM potassium phosphate buffer, pH 8.3, containing 300 mM NaCl and 10 µM ZnSO4. Equal volumes of enzyme (4 nM) and BPPb were mixed and 25 µl of the reaction aliquoted to a 96-well plate in triplicate. Twenty-five microliters of Z-FHL at 1 mM was added to each well, incubated for 15 min at 37°C and the remainder of the assay performed as described by Schwager et al. [31]. Initial reaction velocities were analysed using non-linear regression in GraphPad Prism v.6.0 to obtain IC50 values from which Ki values were calculated using the Cheng–Prusoff equation (1) [32] (where [S] is the final substrate concentration of 0.5 mM Z-FHL).

 
formula
(1)

Enzyme for structural characterization

Minimally glycosylated (N389) human ACE N-domain protein was generated by expression in cultured mammalian CHO-K1 cells and purified to homogeneity as described previously [33]. Purity was assessed using SDS–PAGE and shown to be >95% pure.

X-ray crystallographic studies

The minimally glycosylated N-domain was pre-incubated with BPPb peptide (20 mM) in a 4:1 v/v ratio of protein (5 mg ml−1 in 50 mM HEPES, pH 7.5, 0.1 mM PMSF) for 1 h at room temperature. Co-crystals were obtained with 1 µl of the protein–inhibitor complex mixed with an equal volume of Molecular Dimensions Morpheus A9 reservoir solution (30% PEG 550 MME/PEG 20000, 0.1 M Tris/Bicine [pH 8.5], and 60 mM divalent cations), which was suspended above the well as a hanging drop.

X-ray diffraction data were collected on station I04 at the Diamond Light Source (Didcot, U.K.), with the crystal kept at constant temperature (100 K) under the liquid nitrogen jet during data collection. Images were collected using a PILATUS 6M-F detector (Dectris, Switzerland). Raw data images were indexed and integrated with DIALS [34] and then scaled using AIMLESS [35] from the CCP4 suite [36]. Initial phases were obtained by molecular replacement with PHASER [37] using PDB ID: 6F9V [38] as the search model. Further refinement was initially carried out using REFMAC5 [39] and then Phenix [40], with COOT [41] used for rounds of manual model building. Ligand and water molecules were added based on electron density in the FoFc Fourier difference map. MolProbity [42] was used to help validate the structures. Crystallographic data statistics are summarized in Table 4. All figures showing the crystal structures were generated using CCP4mg [43], and schematic Dimplot binding interactions are displayed using LigPlot+ [44].

Results

Mutation strategy

Since the two domains of ACE are highly homologous and formed from a single polypeptide chain, studies into the domain selectivity of inhibitor binding are commonly performed using domain-selective substrates or inactivation of one domain by mutation of the zinc-binding histidine residues to lysine [2,22,45]. Data obtained using these approaches, however, could still be confounded by cooperativity between the two domains in binding of either the substrate, inhibitor or both [46]. To enable detailed investigation into the structure–activity relationship of BPPb domain selectivity, we therefore utilized truncated N- and C-domain ACE constructs, previously engineered to consist of a single catalytic domain [25,47].

Examination of the BPP sequences studied by Cotton et al. [22] revealed that the C-selective BPPb (Pca-Gly-Leu-Pro-Pro-Arg-Pro-Lys-Ile-Pro-Pro) and non-selective BPPc (Pca-Gly-Leu-Pro-Pro-Gly-Pro-Pro-Ile-Pro-Pro) differed only at positions 6 and 8 (Figure 1). It thus appears that the Arg6 and Lys8 residues of BPPb, which bind to the S4 and S2 pockets of ACE, respectively, could be the principal drivers for domain selectivity. This is supported by the co-crystal structure of BPPb where these residues reside in pockets containing amino acids unique to the C-domain of ACE [48] (Figure 2b). We therefore created various C-domain ACE mutants where unique C-domain residues were converted to their N-domain counterparts and assessed their effects on BPPb-binding affinity. The single mutants studied were E403R (S2), E143S (S4) and S516N (S4). Possible synergism between these residues and between residues of different subsites was studied using the combination mutants SS/NV (S516N, S517V), SSE143S, SSE403R and SSEE (see Materials and Methods for a detailed description). Previously, it was shown that binding of BPPb induces rearrangement of the first three N-terminal helices which form a lid-like structure over the active site cleft [48]. Since this region is poorly conserved between the two domains, we investigated whether the C-domain lid is more accommodating for BPPb binding using a chimera (C1–163 Ndom-ACE) where the C-domain lid region was replaced by that of the N-domain. This incorporated mutation of Glu143 but not Ser516, Ser517 or Glu403. In addition, it includes residues involved in the S6–9 subsites.

Interaction of BPPb with the C-domain of ACE.

Figure 2.
Interaction of BPPb with the C-domain of ACE.

(a) The C-domain (PDB ID: 4APJ) [48] surface is given in grey and sliced to display BPPb (dark cyan sticks) positioned between the two subdomains. (b) Ionic interactions are possible between Lys8 of BPPb and the S2 residue Glu403 (green sticks), unique to the C-domain. Arg6 of BPPb faces the unique S4 residue Glu143 and the backbone carbonyls of Ser516 and Ser517 (green sticks). Although Glu143 is distal (4.48 Å), Arg6 flexibility could possibly afford interaction.

Figure 2.
Interaction of BPPb with the C-domain of ACE.

(a) The C-domain (PDB ID: 4APJ) [48] surface is given in grey and sliced to display BPPb (dark cyan sticks) positioned between the two subdomains. (b) Ionic interactions are possible between Lys8 of BPPb and the S2 residue Glu403 (green sticks), unique to the C-domain. Arg6 of BPPb faces the unique S4 residue Glu143 and the backbone carbonyls of Ser516 and Ser517 (green sticks). Although Glu143 is distal (4.48 Å), Arg6 flexibility could possibly afford interaction.

Due to the largely occluded nature of the active site of ACE, it has been proposed that hinging between the two subdomains would be required for substrate or inhibitor binding [33,49]. We have previously shown that unique S2 residues and S2′ residues (the latter located at an interface between these subdomains) govern N-selective inhibitor binding and proposed that the C-domain active site might be more exposed [29]. Given the large size of BPPb, we evaluated whether mutation of these residues in the N-domain could alter its binding affinity. The N-domain mutant SEDSTE_YR contains Y369F mutation and the introduction of Glu403 (C-domain residue*.) into the N-domain S2 subsite via R381E. The unique N-domain residues Ser260, Glu262, Asp354, Ser357, Thr358 and Glu431 at the subdomain interface were also converted to their C-domain counterparts (Thr282, Ser284, Glu376, Val379, Val380 and Asp453). Although the latter residues are distal to the BPPb P2′ moiety Pro11, they form part of the large, poorly defined prime subsite cavity and are therefore referred to as S2′ subsite residues based on the original classification of ACE subsites [25,26].

Catalytic efficiency of ACE mutants

The mutants were expressed in CHO-K1 cells and purified to homogeneity by lisinopril affinity chromatography. Since the active site was mutated, their catalytic integrity was verified by assessing hydrolysis of the non-selective substrate Z-FHL. The C-domain displayed slightly greater affinity than N-domain for Z-FHL binding, as previously observed [28], with KM values of 0.10 mM and 0.59 mM, respectively (Table 2). Although mutation introduced some variation in Z-FHL-binding affinity, the mutants all displayed KM values within the wild-type range. Importantly, catalytic integrity was not compromised, as evidenced by their kcat values (Table 2).

Table 2
Kinetic parameters describing binding affinity (KM) and rate (kcat) of Z-FHL hydrolysis1by wild-type and mutant C- and N-domain proteins
 Residues involved Pocket KM (mM) kcat (s-1
C-domain N-domain    
C-domain    0.10 155 
N-domain    0.59 200 
C-domain mutants 
 E143S E143 S119 S4 0.04 68 
 E403R E403 R381 S2 0.08 60 
 S516N S516 N494 S4 0.13 265 
 SS/NV S516, S517 N494, V495 S4 0.29 120 
 SSE143S S516, S517, E143 N494, V495, S119 S4 0.29 405 
 SSE403R S516, S517, E403 N494, V495, R381 S2, S4 0.26 213 
 SSEE S516, S517, E143, E403 N494, V495, S119, R381 S2, S4 0.11 80 
 C1–163 Ndom-ACE S1 – P163 L1 – P141 S4, S6–9 0.19 126 
N-domain mutant 
 SEDSTE_YR T282, S284, E376, V379, V380, D453, F391, E403 S260, E262, D354, S357, T358, E431, Y369, R381 S2′, S2 0.13 120 
 Residues involved Pocket KM (mM) kcat (s-1
C-domain N-domain    
C-domain    0.10 155 
N-domain    0.59 200 
C-domain mutants 
 E143S E143 S119 S4 0.04 68 
 E403R E403 R381 S2 0.08 60 
 S516N S516 N494 S4 0.13 265 
 SS/NV S516, S517 N494, V495 S4 0.29 120 
 SSE143S S516, S517, E143 N494, V495, S119 S4 0.29 405 
 SSE403R S516, S517, E403 N494, V495, R381 S2, S4 0.26 213 
 SSEE S516, S517, E143, E403 N494, V495, S119, R381 S2, S4 0.11 80 
 C1–163 Ndom-ACE S1 – P163 L1 – P141 S4, S6–9 0.19 126 
N-domain mutant 
 SEDSTE_YR T282, S284, E376, V379, V380, D453, F391, E403 S260, E262, D354, S357, T358, E431, Y369, R381 S2′, S2 0.13 120 
1

Initial velocities were measured in triplicate and kinetic parameters determined with the Michaelis–Menten model in GraphPad Prism v.06.

Selectivity for BPPb binding

The 11-mer peptide BPPb displayed 122-fold greater binding affinity for the C-domain than N-domain of ACE with Ki values of 20 nM and 2.43 µM, respectively. To gain insight into the basis for this C-selectivity, the affinity of BPPb binding was assessed across a range of constructs comprising mutations in different active site pockets.

Although the C-domain BPPb crystal structure showed direct interaction of BPPb with a unique S2 pocket residue and a possible role for unique S4 residues in peptide binding, mutation of C-domain Glu143, Glu403 or Ser516 to their N-domain counterparts led to only modest decreases in binding affinity with Ki values of 90 nM, 70 nM and 60 nM, respectively (Table 3). The modest change with Ser516 mutation is not surprising since the interaction formed with BPPb Arg6 is via the serine backbone carbonyl and thus also possible with the N-domain counterpart Asn494. Similarly, the Ser517 backbone carbonyl interaction would be possible with the N-domain counterpart Val495. To assess whether the unique S4 subsite residues synergistically provide a more favourable environment for binding of the BPPb Arg6, a double mutant having Ser516 and Ser517 mutated was created (SS/NV) as well as a triple mutant containing additional mutation of Glu143 (SSE143S). These mutants displayed slightly greater decreases in BPPb-binding affinity with Ki values of 120 nM and 160 nM, respectively (Table 3), yet the domain selectivity was affected less than 10-fold (Figure 3a). We therefore assessed the possibility of inter-subsite synergism by creating a triple and quadruple mutant where, in addition to Ser516 and Ser517, either Glu403 (SSE403R) or Glu143 and Glu403 (SSEE) of the S4 and S2 subsites were mutated. For the triple mutant SSE403R, a Ki value of 120nM was obtained, similar to that of SS/NV (Table 3). Further addition of E143S, and thus simultaneous disruption of interactions on opposite faces of BPPb, led to a 17-fold change in binding affinity versus wild-type (Ki of 340 nM for SSEE). The effect of SSEE mutation on domain selectivity surpassed that of SSE143S, suggesting that there not only exists synergism between Ser516, Ser517 and Glu143 of the S4 subsite but also between Glu143 and the S2 subsite Glu403 (Figure 3a). This does not, however, fully account for the 122-fold C-selectivity of BPPb.

Relative BPPb-binding affinities of C-domain and N-domain mutants.

Figure 3.
Relative BPPb-binding affinities of C-domain and N-domain mutants.

Increasing absolute values represent a decrease in affinity relative to that of wild-type and thus a change towards a more (a) N-domain-like or (b) C-domain-like Ki.

Figure 3.
Relative BPPb-binding affinities of C-domain and N-domain mutants.

Increasing absolute values represent a decrease in affinity relative to that of wild-type and thus a change towards a more (a) N-domain-like or (b) C-domain-like Ki.

Table 3
Kinetic parameters for inhibition of wild-type and mutant C- and N-domain proteins by BPPb
 Pocket IC501 (nM) Ki2 (nM) Ki(mutant)/Ki(WT C-domain) 
C-domain  130 20  
N-domain  4480 2430 122 
C-domain mutants 
 E143S S4 1400 90 4.5 
 E403R S2 520 70 3.5 
 S516N S4 300 60 
 SS/NV S4 330 120 
 SSE143S S4 450 160 
 SSE403R S2, S4 360 120 
 SSEE S2, S4 1890 340 17 
C1–163 Ndom-ACE S4, S6–9 1350 370 18.5 
N-domain mutant 
 SEDSTE_YR S2′, S2 520 110 5.53 
 Pocket IC501 (nM) Ki2 (nM) Ki(mutant)/Ki(WT C-domain) 
C-domain  130 20  
N-domain  4480 2430 122 
C-domain mutants 
 E143S S4 1400 90 4.5 
 E403R S2 520 70 3.5 
 S516N S4 300 60 
 SS/NV S4 330 120 
 SSE143S S4 450 160 
 SSE403R S2, S4 360 120 
 SSEE S2, S4 1890 340 17 
C1–163 Ndom-ACE S4, S6–9 1350 370 18.5 
N-domain mutant 
 SEDSTE_YR S2′, S2 520 110 5.53 
1

IC50 values were determined from triplicate initial reaction velocities using non-linear regression in GraphPad Prism v.6.0 with R2 values ranging 0.95–0.99.

2

Ki values were calculated using the Cheng–Prusoff equation where Ki = IC50/(1 + [S]/KM) and [S] = 0.5 mM Z-FHL.

3

Relative to C-domain.

Table 4
X-ray data collection and refinement statistics
 N-domain BPPb 
Resolution (Å) [74.05–9.86] (1.83–1.80) 
Space group P
Cell dimensions (a,b,c72.68, 77.10, 82.33 Å 
Angles (α,β,γ88.85, 64.60, 75.01° 
Molecules/asymmetric unit 
Total/unique reflections 1428701 
 141950 
Completeness (%) [91.7] 98.7 (95.4) 
Rmerge [0.058] 0.104 (1.698) 
Rpim [0.019] 0.034 (0.618) 
I/σ(I)〉 [29.5] 10.2 (1.2) 
CC1/2 [0.998] 0.997 (0.622) 
Multiplicity [9.9] 10.1 (8.5) 
Refinement statistics 
Rwork/Rfree 0.202/0.232 
RMSD in bond lengths (Å) 0.011 
RMSD in bond angles (°) 1.119 
Ramachandran statistics (%) 
Favoured 97.5 
Allowed 2.5 
Outliers 0.0 
Average B-factors (Å2) 
Protein 44.9 
BPPb 47.2 
Ligand 61.0 
Water 41.1 
Number of atoms 
Protein 19 627 
BPPb 260 
Ligand 678 
Water 645 
PDB code 6QS1 
 N-domain BPPb 
Resolution (Å) [74.05–9.86] (1.83–1.80) 
Space group P
Cell dimensions (a,b,c72.68, 77.10, 82.33 Å 
Angles (α,β,γ88.85, 64.60, 75.01° 
Molecules/asymmetric unit 
Total/unique reflections 1428701 
 141950 
Completeness (%) [91.7] 98.7 (95.4) 
Rmerge [0.058] 0.104 (1.698) 
Rpim [0.019] 0.034 (0.618) 
I/σ(I)〉 [29.5] 10.2 (1.2) 
CC1/2 [0.998] 0.997 (0.622) 
Multiplicity [9.9] 10.1 (8.5) 
Refinement statistics 
Rwork/Rfree 0.202/0.232 
RMSD in bond lengths (Å) 0.011 
RMSD in bond angles (°) 1.119 
Ramachandran statistics (%) 
Favoured 97.5 
Allowed 2.5 
Outliers 0.0 
Average B-factors (Å2) 
Protein 44.9 
BPPb 47.2 
Ligand 61.0 
Water 41.1 
Number of atoms 
Protein 19 627 
BPPb 260 
Ligand 678 
Water 645 
PDB code 6QS1 

Inner shell, overall and outer shell statistics are given in square brackets, un-bracketed and round brackets, respectively.

We therefore investigated whether unique C-domain residues distal to BPPb indirectly enhanced binding to this large peptide using a chimeric protein (C1–163 Ndom-ACE) where the C-domain lid region was replaced by that of the N-domain. The C-domain chimera incorporated mutation of Glu143 but not Ser516, Ser517 or Glu403 and greatly increased the Ki from 20 nM to 370 nM (Table 3), thus decreasing the C-selectivity 7-fold (Figure 3a). This points to the involvement of C-domain lid residues in addition to Glu143, contained in this chimera, in accommodating BPPb binding, which are likely to be those identified from the C-domain complex to be involved in the S6–9 subsites.

Given the size of the BPPb undecapeptide, we further examined the role of distal unique C-domain S2′ residues previously suggested to increase active site exposure [29] in peptide-binding affinity. Interestingly, the SEDSTE_YR N-domain mutant displayed a C-domain-like affinity for BPPb (Figure 3b) and decreased the Ki 22-fold from 2.43 µM to 110 nM (Table 3). Although this effect could, in part, be explained by introduction of a salt-bridge to Lys8 of BPPb through R381E mutation (contained in this construct), it is unlikely to be the only driver of BPPb binding since affinity was only slightly decreased upon loss of this interaction in the inverse E403R C-domain mutant.

Overall structure of the N-domain BPPb complex

The structure of BPPb co-crystallized with the N-domain was determined to gain further insight into the molecular basis for C-selective binding of BPPb by comparison to the C-domain BPPb structure [48]. The high-resolution 1.8 Å structure of the N-domain BPPb complex crystallized in the P1 space group with two molecules in the asymmetric unit (Table 4). The overall structure of both molecules, which is typical for ACE domains, shows an ellipsoid comprised mainly of α-helices with a buried two-lobed cavity that forms the S′ and S subsites, and the active site, including a zinc ion, located at the junction of these lobes (Figure 4a). This active site zinc ion is observed with the typical coordination to His361, His365 and Glu389, although unusually for N-domain it has been refined with an occupancy of 70% and 64% for chain A and B, respectively. A flexible lid-like structure comprised of sections from the first 100 residues is likely involved in controlling peptide access to the active site by opening and closing [50]. Consistent with this region being flexible, these residues show high-temperature factors (B-factors) in both molecules of the asymmetric unit for the N-domain BPPb structure described here, though more pronounced in chain B. In addition, it is often observed that the electron density for these residues is poor in one chain and shows evidence of a second conformation. In the N-domain BPPb complex structure, while the electron density for this region in chain B is not as clear as in chain A, it is better than what is usually observed, albeit there is still some evidence of a second conformation.

BPPb binding to ACE domains.

Figure 4.
BPPb binding to ACE domains.

Zinc ions are shown as lilac spheres, with α-helices and β-strands shown in rose and dark cyan, respectively. (a) Schematic representation of BPPb-ACE N-domain complex crystal structure showing the BPPb molecule (dark orange sticks) bound to the active site zinc ion. Loop regions have been smoothed for clarity. (b) Close-up view of the BPPb molecule bound to N-domain overlaid with the final 2mFo–DFc (blue, contoured at 1σ level) electron density map. The polder map for BPPb residues 1–6 is shown in green. (c) Close-up view of the previously published BPPb molecule bound to C-domain (PDB ID: 4APJ) overlaid with the final 2mFo–DFc (blue, contoured at 1σ level) electron density map. (d) Overlay of the BPPb molecules from the N-domain and C-domain complex structures shown as dark orange and green sticks, respectively. (e, f) Close-up view of the additional density close to zinc ion and Ile3 of BPPb in the N-domain complex structure for molecules 1 and 2, respectively of the asymmetric unit. The final 2mFo–DFc (contoured at 1σ level) electron density map is shown in blue, with the additional density highlighted by the mFo-DFc electron density difference map (contoured at 3σ level) shown in green.

Figure 4.
BPPb binding to ACE domains.

Zinc ions are shown as lilac spheres, with α-helices and β-strands shown in rose and dark cyan, respectively. (a) Schematic representation of BPPb-ACE N-domain complex crystal structure showing the BPPb molecule (dark orange sticks) bound to the active site zinc ion. Loop regions have been smoothed for clarity. (b) Close-up view of the BPPb molecule bound to N-domain overlaid with the final 2mFo–DFc (blue, contoured at 1σ level) electron density map. The polder map for BPPb residues 1–6 is shown in green. (c) Close-up view of the previously published BPPb molecule bound to C-domain (PDB ID: 4APJ) overlaid with the final 2mFo–DFc (blue, contoured at 1σ level) electron density map. (d) Overlay of the BPPb molecules from the N-domain and C-domain complex structures shown as dark orange and green sticks, respectively. (e, f) Close-up view of the additional density close to zinc ion and Ile3 of BPPb in the N-domain complex structure for molecules 1 and 2, respectively of the asymmetric unit. The final 2mFo–DFc (contoured at 1σ level) electron density map is shown in blue, with the additional density highlighted by the mFo-DFc electron density difference map (contoured at 3σ level) shown in green.

In both molecules of the asymmetric unit, the final 2mFo–DFc electron density map shows clear density for the C-terminal Pro7 to Pro11 residues of the 11-mer BPPb peptide, and these are bound in the S3 to S2′ subsites, respectively (Figure 4a,b). The density for the N-terminal residues pyroglutamate (Pca1) to Arg6 is less clear, indicating increased flexibility of these residues, although a combination of the 2mFo–DFc and polder electron density maps allowed the most likely positions to be modelled for molecule A (Figure 4b). The electron density in molecule B was not sufficient to model the N-terminal residues of BPPb, and this may be due to the increased flexibility of the adjacent lid-like region observed in chain B. During the structure refinement process using Phenix, the occupancy of the BPPb chains was also refined with the resulting occupancies being 79% and 85% in molecules A and B of the asymmetric unit, respectively.

In both molecules of the asymmetric unit, between the active site zinc ion and BPPb Ile9 residue, there is a patch of positive density in the final mFo-DFc electron density difference map that is additional to the modelled BPPb ligand (Figure 4e,f). The position of this density is a little different between the two molecules where it is closer to Ile9 in molecule A but closer to the zinc ion in molecule B. There are a variety of possible explanations for this density. Firstly, with the occupancies of the BPPb molecules being ∼80%, the density could be water molecules bound to the active site zinc ion when there is no BPPb bound. A second possibility, at least for molecule A of the asymmetric unit where the patch of density is closer to BPPb Ile9, would be a small amount of the tripeptide Ile-Pro-Pro being cleaved from the BPPb peptide, with the then free N-terminal backbone nitrogen occupying the additional density. Another possible explanation for the density is that a water molecule could be bound to the zinc ion while the BPPb is also present. This explanation is likely to be more applicable to molecule B where the patch of density is closer to N-domain Glu362 and the active site zinc ion and would be consistent with a step in the reaction mechanism. Based on electron density observation alone, it is not possible to distinguish between these possibilities or to rule out other explanations. Therefore, these patches of density have been left un-modelled.

BPPb binding to N-domain

Residues Pro7 to Pro11 of BPPb bind to the two chains of N-domain in the asymmetric unit in an almost identical manner. The following text and related figures describe the binding interactions observed in molecule A where the complete BPPb ligand is visible. As stated above, the C-terminal Ile9 to Pro11 residues of BPPb are bound in the S1, S1′ and S2′ subsites of N-domain close to the active site zinc ion, with the remaining residues occupying the full extent of the non-primed lobe of the binding cavity. The interactions involved in the binding of each residue of BPPb to the N-domain are shown in Figures 57. Figure 5 shows a Dimplot representation of all the interactions involved in BPPb binding to N-domain and C-domain, whereas Figure 6 is a close-up of the interactions formed between each BPPb moiety and the N-domain. Figure 7 shows a schematic structural view of each C-domain subsite binding the BPPb residues, overlaid with the equivalent N-domain and BPPb residues, as well as any proximal residues which are not conserved between N-domain and C-domain.

Dimplot representation of the binding site interactions of BPPb bound to N-domain and C-domain.

Figure 5.
Dimplot representation of the binding site interactions of BPPb bound to N-domain and C-domain.

(a) Residues 1–6 of BPPb bound to N-domain. (b) Residues 1–6 of BPPb bound to C-domain. (c) Residues 7–11 of BPPb bound to N-domain. (d) Residues 7–11 of BPPb bound to C-domain. BPPb C-domain complex was previously published (PDB ID: 4APJ). BPPb residues are depicted below the dotted line in each plot with the interacting ACE domain residues shown above. H-bond/electrostatic interactions are shown in green, hydrophobic interactions in red and water molecules as red spheres. Residues solely involved in hydrophobic interactions are represented by red, semi-circular symbols.

Figure 5.
Dimplot representation of the binding site interactions of BPPb bound to N-domain and C-domain.

(a) Residues 1–6 of BPPb bound to N-domain. (b) Residues 1–6 of BPPb bound to C-domain. (c) Residues 7–11 of BPPb bound to N-domain. (d) Residues 7–11 of BPPb bound to C-domain. BPPb C-domain complex was previously published (PDB ID: 4APJ). BPPb residues are depicted below the dotted line in each plot with the interacting ACE domain residues shown above. H-bond/electrostatic interactions are shown in green, hydrophobic interactions in red and water molecules as red spheres. Residues solely involved in hydrophobic interactions are represented by red, semi-circular symbols.

BPPb interactions with the N-domain.

Figure 6.
BPPb interactions with the N-domain.

The panels show individual BPPb residues (thicker, darker orange sticks) and their interactions. H-bond/electrostatic interactions are shown in green, hydrophobic interactions in red and water molecules as red spheres.

Figure 6.
BPPb interactions with the N-domain.

The panels show individual BPPb residues (thicker, darker orange sticks) and their interactions. H-bond/electrostatic interactions are shown in green, hydrophobic interactions in red and water molecules as red spheres.

Comparison of the N- and C-domain interactions with BPPb.

Figure 7.
Comparison of the N- and C-domain interactions with BPPb.

The panels show individual BPPb residues from the C-domain and N-domain complex structures and the interactions with C-domain. The C-domain and N-domain structures are shown as green and orange sticks, respectively, with the BPPb residues depicted as darker and thicker sticks. H-bond/electrostatic interactions are shown in green, hydrophobic interactions in red and water molecules as red spheres. In addition, residues in the vicinity that are not conserved between C-domain and N-domain are included.

Figure 7.
Comparison of the N- and C-domain interactions with BPPb.

The panels show individual BPPb residues from the C-domain and N-domain complex structures and the interactions with C-domain. The C-domain and N-domain structures are shown as green and orange sticks, respectively, with the BPPb residues depicted as darker and thicker sticks. H-bond/electrostatic interactions are shown in green, hydrophobic interactions in red and water molecules as red spheres. In addition, residues in the vicinity that are not conserved between C-domain and N-domain are included.

ACE domain structures in complex with ligands solved to date have predominantly shown binding to the S1, S1′ and S2′ subsites with many conserved or similar interactions. Residues Ile9, Pro10 and Pro11 of BPPb have these typical interactions and are strongly bound by N-domain (Figures 5c and 6). The BPPb terminal carboxylic acid group of Pro11 forms direct hydrogen bonds with residues Gln259, Lys489 and Tyr498, and interacts via a water molecule with Lys489. The backbone carbonyl of BPPb Pro10 directly interacts with His331 and His491, while for Ile9 the backbone carbonyl interacts with Tyr501 and the active site zinc ion. In addition, the peptide N-atom of Ile9 hydrogen bonds to a water molecule which itself interacts with Glu389, Arg500 and Tyr501. These three BPPb residues also form extensive hydrophobic interactions with N-domain residues His331, Ala332, Ser333, His361, Glu362, Phe435, Phe490, His491, Thr496 and Tyr501.

A strong patch of electron density was observed adjacent to the NZ atom of BPPb Lys8 residue, which could not be explained by modelling in a water molecule but is consistent with a chloride ion that also interacts with the backbone nitrogen of N-domain Gly382 and a water molecule in a typical trigonal planar conformation. In addition to this chloride ion, BPPb Lys8 has two direct interactions from its peptide backbone with the backbone of N-domain Ala334, and hydrogen bonding with Arg381, Pro385 and Glu389 via a single water molecule (Figures 5c and 6). There are also hydrophobic interactions with N-domain residues His385, Arg381 and His388. In contrast with the many N-domain residues interacting with Lys8 of BPPb, only Trp335 of N-domain interacts with BPPb Pro7. However, this is an edge to face stacking orientation with multiple hydrophobic interactions, which is consistent with the electron density for BPPb Pro7 being clear.

As described above, the electron density for residues 1–6 of BPPb is much weaker than that observed for residues 7–11. This is consistent with only a few interactions between N-domain and BPPb residues 2–6 (Figures 5a and 6) and is indicative of this section of BPPb being flexible and not strongly bound to N-domain. In detail, the side chain of BPPb Arg6 is orientated such that it can interact with the backbone carbonyl of N-domain Asn94 but, as stated, the electron density is weak indicating that these interactions are not strong. There is also a water-mediated interaction from the backbone carbonyl of BPPb Arg6 with N-domain Arg381. Residues 3–5 of BPPb form hydrophobic interactions with N-domain involving BPPb Pro5 with N-domain Arg381, BPPb Pro4 with N-domain Leu32 and Ser35, and BPPb Leu3 and N-domain Ala38. BPPb Gly2 has a direct hydrogen bond with N-domain Ser61 (alternate conformation B) and a water-mediated interaction with N-domain Thr97. The BPPb N-terminal Pca1 residue has a direct hydrogen bond with N-domain Tyr197 and a water-mediated interaction with N-domain Ser100. In addition, there are hydrophobic interactions with N-domain residues Leu98, Gly99, Tyr186 and Tyr197. Although there are multiple interactions, the side chain carbon atoms of this BPPb residue are orientated such that they are in close proximity (∼3 Å) with the hydrophilic OH atoms of N-domain Tyr186 and Tyr197.

Discussion

Toxins provide a rich source of molecules with therapeutic actions and toxinologists often state the potential for such molecules in drug development [51]. Design of the widely used antihypertensive agent captopril based on the structure of naturally occurring BPPs is a popular example where a toxin-derived molecule received FDA approval [52]. Since the use of non-domain-selective ACE inhibitors such as captopril has been associated with side effects, current efforts are aimed at developing C-domain-selective inhibitors and dual inhibitors that block the C-domain and targets from other pathways that regulate BP. To guide the design of non-peptidic C-selective inhibitors, we studied the molecular basis whereby the undecapeptide BPPb binds with 122-fold greater affinity to the C-domain of ACE. The residues important for BPPb binding were identified by evaluating the effect of substituting unique C-domain residues by their N-domain counterparts on binding affinity. Further insight into the difference in BPPb affinity between the domains can be gained by comparison of the BPPb N-domain complex presented here with that of the previously published BPPb C-domain complex [48] as described below.

As described above, in the N-domain complex, there is strong electron density only for BPPb residues 7–11, whereas there is clear, complete electron density observed for the whole BPPb bound to C-domain (Figure 4c). This suggests that the binding to C-domain is strong for the full length of BPPb, with less flexibility than observed for residues 1–6 of BPPb bound to N-domain. When the N-domain and C-domain BPPb complex structures are overlaid, it can be seen that even with the lack of a zinc ion and some structural differences in the C-domain complex, the orientations and interactions for BPPb residues 9–11 are very similar (Figures 5d and 7). In contrast, the interactions begin to diverge with BPPb residue Lys8, except for BPPb Pro7 which has similar edge to face stacking with a conserved tryptophan residue in both structures (Trp335 and Trp357 in N-domain and C-domain, respectively). There are clear differences with residues 1–6 and, in particular, residues 1–3 where Cα atom positions differ between 3.3 Å and 5.3 Å, and this suggests that the affinity difference measured for the two proteins is likely caused by variations in the binding of the N-terminal BPPb residues. This apparent importance of the N-terminal residues of BPPb for C-domain selectivity correlates with the non-selective BPPc that differs only from BPPb in its P4 and P2 moieties. Therefore, we initially mutated unique C-domain residues to their N-domain counterparts in the S4 and/or S2 subsites. It was found that disruption of interactions to either the Arg6 (through E143S, S516N, SS/NV or SSE143S mutation in S4) or Lys8 (through E403R mutation in S2) had a modest effect on BPPb-binding affinity. The same was observed upon combined mutation of SSE403R. Further addition of Glu143Ser (SSEE), however, led to a more substantial decrease in binding affinity. This suggests that these residues synergistically contribute to BPPb binding and C-selectivity through formation of Lys8-Glu403, and possibly Arg6-Glu143 interactions, which could stabilize opposite faces of the peptide and orientate its N-terminus. The mutagenesis data demonstrate a greater significance for Glu143 in BPPb binding than indicated by the C-domain BPPb complex structure where Arg6 was rather distal (4.48 Å) to Glu143. Though distal, Glu143 possibly creates a more favourable environment for the proper orientation of Arg6 to interact with the backbone carbonyls of Ser516 and Ser517.

The shift in BPPb-binding orientation at the N-terminus could have partly been caused by the loss of a stabilizing Glu403-Lys8 salt-bridge (despite the conserved nature of the BPPb Lys8 backbone interactions) and the proximity of the N-domain Arg381 side chain (which is larger than the non-conserved Glu403 in C-domain) to the BPPb Pro5 ring (Figure 7). Due to this shift, the interactions of BPPb Pro5 seen in C-domain would not form in N-domain. The N-domain complex structure also showed an altered BPPb Arg6 side chain conformation to that observed in the C-domain, in support of the environmental role for unique S4 subsite residues Glu143, Ser516 and Ser517 in BPPb binding suggested by the mutagenesis data. Despite the loss of Arg6 and Lys8 interactions in SSEE, this shift in BPPb would have been unlikely in this C-domain mutant due to further steric hindrance in the S1–S4 subsites, as described below. This could, in part, explain why the SSEE mutant's affinity for BPPb did not equal that of the N-domain.

Another factor which could have contributed to the shift in orientation of the BPPb residues 1–6 between the two domains is the presence of unique residues at the N-terminal lid region. To investigate the possible involvement of the N-terminal lid region residues on BPPb-binding affinity, a chimeric protein was kinetically characterized. We demonstrated that replacement of the first three N-terminal helices of the C-domain by that of the N-domain (C1–163 Ndom-ACE) drastically decreased BPPb-binding affinity. This chimera included residues from the S4 and S6–9 subsites and thus substitution of K118A, D121T, E123G, R124S and E143S. Since the effect of C1–163 Ndom-ACE surpassed that of the single mutant E143S, it suggests that the smaller side chains allowed a shift in the N-terminal BPPb residues, similar to that observed in the N-domain BPPb structure, upon loss of Arg6 stabilization by Glu143 and thereby decreased the affinity further through loss of interactions with BPPb residues 1–5. BPPb Pro4 in the C-domain complex structure forms a hydrophobic face-to-face stacking interaction with Trp59 but is also surrounded by a hydrophobic environment from C-domain Tyr62, Ile88 and Tyr360 residues (although a little above the 3.9 Å cut-off distance used by LigPlot to identify hydrophobic interactions) (Figure 7). In the N-domain BPPb structure and C-domain chimera, Trp59 is replaced by Leu32 which still interacts with the BPPb Pro4 but contributes to the shift in BPPb orientation due to its bulky shape. In addition, the overall environment for Pro4 is more hydrophilic in N-domain with equivalent residues being Ser35, Ser61 and Tyr338. The backbone interaction between Glu123 and BPPb Leu3 was also abolished and the BPPb peptide shifted into the equivalent space occupied by the bulky Glu123 and Arg124 side chains of C-domain (Figure 7). Due to this change in BPPb orientation, the Gly2 backbone interactions with conserved residues in the C-domain were absent from the N-domain. Additionally, this resulted in the peptide extending a little further into the non-prime lobe of the N-domain binding cavity. BPPb Pca1 is positioned furthest from the active site in both structures, but the equivalent location of this residue in the C-domain structure is sterically hindered by Tyr197 in the N-domain complex (the C-domain counterpart Ser219 is significantly smaller) (Figure 7). Furthermore, in the N-domain complex, BPPb Pca1 adopts a position occupied by the non-conserved Glu123 side chain (Gly99 in N-domain or C-domain chimera). While the BPPb Pca1 residue does partake in a series of hydrophilic and hydrophobic interactions, its ring structure looks strained and puckered, likely caused by the BPPb orientation and surrounding environment.

Considering the unique nature of interactions on the surface of the N-terminal lid helices, it seems likely that alterations in the dynamics of the C-domain chimera's lid region could have also contributed to its decreased BPPb-binding affinity. The C-domain cleft surface is lined by largely hydrophobic or small residues of helices 2 and 3 (Figure 8a). In the N-domain, however, these residues are replaced by polar, basic or acidic residues which form hydrogen bonds over the active site cleft (Figure 8b). This would likely alter the necessary BPPb-induced rearrangement observed in the C-domain crystal structure [48] (Figure 8c) and result in steric hindrance of the peptide, thereby lowering binding affinity.

Interactions between helices of the lid region in the N- and C-domain.

Figure 8.
Interactions between helices of the lid region in the N- and C-domain.

The C-domain region replaced by the N-domain counterpart in the C1–163 Ndom-ACE chimera is given in dark pink, whereas C-domain helices (PDB ID: 4APJ) are coloured light green, N-domain helices (PDB ID: 6QS1) rose and BPPb (complexed to C-domain) given as sea green sticks. Residues from helices 2 and 3 of the N-terminal lid line the active site cleft surface and are (a) small and hydrophobic in the C-domain but (b) longer and polar in the N-domain. (c) These polar interactions over the N-domain cleft could prevent the necessary rearrangement of helices 1–3 (α1–3) upon BPPb binding.

Figure 8.
Interactions between helices of the lid region in the N- and C-domain.

The C-domain region replaced by the N-domain counterpart in the C1–163 Ndom-ACE chimera is given in dark pink, whereas C-domain helices (PDB ID: 4APJ) are coloured light green, N-domain helices (PDB ID: 6QS1) rose and BPPb (complexed to C-domain) given as sea green sticks. Residues from helices 2 and 3 of the N-terminal lid line the active site cleft surface and are (a) small and hydrophobic in the C-domain but (b) longer and polar in the N-domain. (c) These polar interactions over the N-domain cleft could prevent the necessary rearrangement of helices 1–3 (α1–3) upon BPPb binding.

Given that BPPb only differs from BPPc at Arg6 and Lys8, and that C-selectivity of BPPb is not abolished by loss of interactions to unique residues in these subsites alone, the non-selective nature of BPPc might seem surprising. The lack of selectivity, however, is the result of a dramatically higher affinity of the N-domain for BPPc (Ki value of 55 nM) over BPPb (Ki value of 9000 nM) [22], with the C-domain having similar affinities for both peptides. The presence of Pro8 after Pro7, and the extra flexibility from Gly6 in BPPc (Pca-Gly-Leu-Pro-Pro-Gly-Pro-Pro-Ile-Pro-Pro) could allow for a change in the orientation of the peptide by altering the backbone torsion angles. In the N-domain, the BPPc position towards the lid is likely slightly different from BPPb so that steric hindrance in the S4–S9 subsites is reduced, thereby allowing interaction with and potent inhibition of the N-domain. Due to the nature of the non-prime subsite residues, BPPc likely binds to the C-domain in a similar position to BPPb. Potent C-domain inhibition (Ki value of 70 nM [22]), albeit with a slight reduction in affinity upon loss of the S2 and S4 subsite side chain interactions, and thereby a lack of domain selectivity thus occurs.

Since BPPb is a large peptide and its binding to the C-domain induced a slight conformational change, we studied the effect of protein dynamics on BPPb-binding affinity using an N-domain mutant (SEDSTE_YR) where unique residues of the S2 and S2′ subsites were converted to their C-domain counterparts. This has previously been described to affect binding of an N-selective ACE inhibitor by increasing active site exposure and altering the lid region's dynamics [29]. Interestingly, the affinity of this N-domain mutant for BPPb was only 5-fold lower than that of the C-domain, despite the multiple differences in the S4–S9 subsites of the two domains described above. Considering the aforementioned C-domain mutant data, a change of this magnitude seems unlikely to be the result of mere introduction of a Glu403-Lys8 salt-bridge or Phe391-Lys8 hydrophobic interactions into an N-domain protein (Figure 9). Previously, it was shown that replacement of the polar N-domain S2′ residues by their non-polar C-domain counterparts led to repulsion between the subdomains and opening of the active site cleft [29]. These distal C-domain S2′ residues thus likely contribute to BPPb-binding affinity by allowing greater ease of BPPb entry to the C-domain active site for interaction with unique residues through lid rearrangement.

Positioning of BPPb relative to the interface between the two subdomains.

Figure 9.
Positioning of BPPb relative to the interface between the two subdomains.

BPPb complexed to the C-domain is given as sea green sticks, unique C-domain residues (PDB ID: 4APJ) of the S2 and S2′ subsites as green sticks and their N-domain counterparts (PDB ID: 6QS1) as orange sticks. Conserved residues interacting with Pro11 of BPPb are given as grey sticks. (a) Interaction of Lys8 with Glu403 and Phe391 and (b) close-up view of the location of BPPb relative to unique residues at the S2′ subdomain interface.

Figure 9.
Positioning of BPPb relative to the interface between the two subdomains.

BPPb complexed to the C-domain is given as sea green sticks, unique C-domain residues (PDB ID: 4APJ) of the S2 and S2′ subsites as green sticks and their N-domain counterparts (PDB ID: 6QS1) as orange sticks. Conserved residues interacting with Pro11 of BPPb are given as grey sticks. (a) Interaction of Lys8 with Glu403 and Phe391 and (b) close-up view of the location of BPPb relative to unique residues at the S2′ subdomain interface.

In summary, residues 7–11 of BPPb are strongly bound, with many conserved interactions, in both the N-domain and C-domain complexes. In contrast, BPPb residues 1–6 are similarly well bound in C-domain with multiple interactions, but in N-domain, there is a different orientation of binding involving less favourable interactions. This results in greater flexibility highlighted by the poor electron density for this region. The altered binding orientation of BPPb can be attributed to (a) potential steric clashes in the S5, S6 and S9 subsites with N-domain residues Arg381, Leu32 and Tyr197 which are not conserved in C-domain (Glu403, Trp59 and Ser219, respectively) and (b) extra space in the N-domain S7 subsite caused by Gly99 and Ser100 replacing the C-domain Glu123 and Arg124 residues. These complex structures are consistent with and contribute to the explanation of the measured 122-fold C-domain selectivity of BPPb. We propose that the presence of unique residues in the lid region, S2, S4 and distal S2′ subsites synergistically contribute to this C-domain selectivity by directing the protein's dynamics to increase active site access and optimal orientation of BPPb for interaction with C-domain residues. Since multiple subsites are implicated in the mechanism of BPPb selectivity, designing small, non-peptidic C-selective inhibitors based on the structure of BPPb would be challenging. Furthermore, the lack of oral bioavailability is a major drawback for BPPs and would require implementation and optimization of drug-delivery systems for clinical use [53]. These results demonstrate the potential confounding role of altered protein dynamics, active site entry and inhibitor positioning on domain selectivity and emphasize the need for considering these factors, in addition to direct protein–inhibitor interactions, particularly in the case of larger peptides, when interpreting inhibition data to guide the design of domain-selective ACE inhibitors.

Database Depositions

The atomic co-ordinates and structure factors for N-domain ACE in complex with BPPb have been deposited in the RCSB Protein Data Bank with the code PDB ID: 6QS1. The atomic co-ordinates and experimental data will be released upon article publication.

Abbreviations

     
  • ACE

    angiotensin-converting enzyme

  •  
  • AngI

    angiotensin I

  •  
  • AngII

    angiotensin II

  •  
  • BK

    bradykinin

  •  
  • BPPs

    bradykinin-potentiating peptides

  •  
  • CHO-K1

    Chinese hamster ovary

  •  
  • Pca

    pyroglutamate

  •  
  • Z-FHL

    Z-phenylalanylhistidylleucine

Author Contribution

E.D.S. supervised the enzyme kinetics study and edited the manuscript. L.L. analysed the kinetics data, wrote the first draft and edited the manuscript. G.E.C. performed the crystallography experiments, analysed the data, wrote the structural sections and edited the manuscript. S.L.U.S. and E.B. performed transfections, protein purifications and enzyme assays. E.B. performed the mutagenesis and cloning. L.B.A. performed cloning and edited the manuscript. A.T.A. and E.D.S. conceptualized the enzyme kinetics study. K.R.A. supervised the structural study, analysed the data and edited the manuscript. All authors reviewed the manuscript.

Funding

This work was supported by the Medical Research Council (U.K.) Project Grant MR/M026647/1 (to K.R.A.) and based on the research supported in part by the National Research Foundation of South Africa (Grant Numbers 111545 (to L.L.) and 111798 (to E.D.S.)).

Acknowledgements

We thank the scientists at stations IO4 (Proposal Number mx17212) of Diamond Light Source, Didcot, Oxfordshire (U.K.), for their support during X-ray diffraction data collection. The authors are grateful for the financial support from the Medical Research Council (U.K.) and the National Research Foundation of South Africa. K.R.A. and E.D.S. also thank the University of Cape Town (South Africa) and the University of Bath (U.K.) for the respective Visiting Professorships.

Competing Interests

The Authors declare that there are no competing interests associated with the manuscript.

References

References
1
Acharya
,
K.R.
,
Sturrock
,
E.D.
,
Riordan
,
J.F.
and
Ehlers
,
M.R.
(
2003
)
ACE revisited: a new target for structure-based drug design
.
Nat. Rev. Discov.
2
,
891
902
2
Wei
,
L.
(
1991
)
The two homologous domains of human angiotensin I-converting enzyme are both catalytically active
.
J. Biol. Chem.
266
,
9002
9008
3
Fuchs
,
S.
,
Xiao
,
H.D.
,
Hubert
,
C.
,
Michaud
,
A.
,
Campbell
,
D.J.
,
Adams
,
J.W.
et al.  (
2008
)
Angiotensin-converting enzyme C-terminal catalytic domain is the main site of angiotensin I cleavage in vivo
.
Hypertension
51
,
267
274
4
Soubrier
,
F.
,
Alhenc-Gelas
,
F.
,
Hubert
,
C.
,
Allegrini
,
J.
,
John
,
M.
,
Tregear
,
G.
et al.  (
1988
)
Two putative active centers in human angiotensin I-converting enzyme revealed by molecular cloning
.
Proc. Natl Acad. Sci. U.S.A.
85
,
9386
9390
5
Jaspard
,
E.
,
Wei
,
L.
and
Alhenc-Gelas
,
F.
(
1993
)
Differences in the properties and enzymatic specificities of the two active sites of angiotensin I-converting enzyme (kininase II). Studies with bradykinin and other natural peptides
.
J. Biol. Chem.
268
,
9496
9503
PMID:
[PubMed]
6
Bernstein
,
K.E.
,
Shen
,
X.Z.
,
Gonzalez-Villalobos
,
R.A.
,
Billet
,
S.
,
Okwan-Duodu
,
D.
,
Ong
,
F.S.
et al.  (
2011
)
Different in vivo functions of the two catalytic domains of angiotensin-converting enzyme (ACE)
.
Curr. Opin. Pharmacol.
11
,
105
111
7
Bicket
,
D.P.
(
2002
)
Using ACE inhibitors appropriately
.
Am. Fam. Physician
66
,
461
468
PMID:
[PubMed]
8
Burger
,
D.
,
Reudelhuber
,
T.L.
,
Mahajan
,
A.
,
Chibale
,
K.
,
Sturrock
,
E.D.
and
Touyz
,
R.M.
(
2014
)
Effects of a domain-selective ACE inhibitor in a mouse model of chronic angiotensin II-dependent hypertension
.
Clin. Sci.
127
,
57
63
9
Sharp
,
S.
,
Poglitsch
,
M.
,
Zilla
,
P.
,
Davies
,
N.H.
and
Sturrock
,
E.D.
(
2015
)
Pharmacodynamic effects of C-domain-specific ACE inhibitors on the renin-angiotensin system in myocardial infarcted rats
.
J. Renin Angiotensin Aldosterone Syst.
16
,
1149
1158
10
Ondetti
,
M.A.
and
Rubin
,
B.
and
Cushman
,
D.W.
(
1977
)
Design of specific inhibitors of angiotensin-converting enzyme: new class of orally active antihypertensive agents
.
Science
196
,
441
444
11
Hayashi
,
M.A.
and
Camargo
,
A.C.
(
2005
)
The bradykinin-potentiating peptides from venom gland and brain of Bothrops jararaca contain highly site specific inhibitors of the somatic angiotensin-converting enzyme
.
Toxicon
45
,
1163
1170
12
Erdös
,
E.G.
Deddish
,
P.A.
and
Marcic
,
B.M.
(
1999
)
Potentiation of bradykinin actions by ACE inhibitors
.
Trends Endocrinol. Metab.
10
,
223
229
13
Marcic
,
B.
,
Deddish
,
P.A.
,
Jackman
,
H.L.
and
Erdös
,
E.G.
(
1999
)
Enhancement of bradykinin and resensitization of its B2 receptor
.
Hypertension
33
,
835
843
14
Ferreira
,
S.H.
,
Bartelt
,
D.C.
and
Greene
,
L.J.
(
1970
)
Isolation of bradykinin-potentiating peptides from Bothrops jararaca venom
.
Biochemistry
9
,
2583
2593
15
Ianzer
,
D.
,
Konno
,
K.
,
Marques-Porto
,
R.
,
Vieira Portaro
,
F.C.
,
Stöcklin
,
R.
,
de Camargo AC
,
M.
et al.  (
2004
)
Identification of five new bradykinin potentiating peptides (BPPs) from Bothrops jararaca crude venom by using electrospray ionization tandem mass spectrometry after a two-step liquid chromatography
.
Peptides
25
,
1085
1092
16
Kodama
,
R.T.
,
Cajado-Carvalho
,
D.
,
Kuniyoshi
,
A.K.
,
Kitano
,
E.S.
,
Tashima
,
A.K.
,
Barna
,
B.F.
et al.  (
2015
)
New proline-rich oligopeptides from the venom of African adders: insights into the hypotensive effect of the venoms
.
Biochim. Biophys. Acta
1850
,
1180
1187
17
Ferreira
,
L.A.
,
Alves
,
W.E.
,
Lucas
,
M.S.
and
Habermehl
,
G.G.
(
1996
)
Isolation and characterization of a bradykinin potentiating peptide (BPP-S) isolated from Scaptocosa raptoria venom
.
Toxicon
34
,
599
603
18
Verano-Braga
,
T.
,
Rocha-Resende
,
C.
,
Silva
,
D.M.
,
Ianzer
,
D.
,
Martin-Eauclaire
,
M.F.
,
Bougis
,
P.E.
et al.  (
2008
)
Tityus serrulatus hypotensins: a new family of peptides from scorpion venom
.
Biochem. Biophys. Res. Commun.
371
,
515
520
19
Arcanjo
,
D.D.R.
,
Vasconcelos
,
A.G.
,
Comerma-Steffensen
,
S.G.
,
Jesus
,
J.R.
,
Silva
,
L.P.
,
Pires
,
O.R.
et al.  (
2015
)
A novel vasoactive proline-rich oligopeptide from the skin secretion of the frog Brachycephalus ephippium
.
PLoS ONE
10
,
e0145071
20
Conceição
,
K.
,
Konno
,
K.
,
de Melo
,
R.L.
,
Antoniazzi
,
M.M.
,
Jared
,
C.
,
Sciani
,
J.M.
et al.  (
2007
)
Isolation and characterization of a novel bradykinin potentiating peptide (BPP) from the skin secretion of Phyllomedusa hypochondrialis
.
Peptides
28
,
515
523
21
Pimenta
,
D.C.
,
Prezoto
,
B.C.
,
Konno
,
K.
,
Melo
,
R.L.
,
Furtado
,
M.F.
,
Camargo
,
A.C.
et al.  (
2007
)
Mass spectrometric analysis of the individual variability of Bothrops jararaca venom peptide fraction. Evidence for sex-based variation among the bradykinin-potentiating peptides
.
Rapid Commun. Mass Spectrom.
21
,
1034
1042
22
Cotton
,
J.
,
Hayashi
,
M.A.
,
Cuniasse
,
P.
,
Vazeux
,
G.
,
Ianzer
,
D.
,
De Camargo
,
A.C.
et al.  (
2002
)
Selective inhibition of the C-domain of angiotensin I converting enzyme by bradykinin potentiating peptides
.
Biochemistry
41
,
6065
6071
23
Gilio
,
J.M.
,
Portaro
,
F.C.
,
Borella
,
M.I.
,
Lameu
,
C.
,
Camargo
,
A.C.
and
Alberto-Silva
,
C.
(
2013
)
A bradykinin-potentiating peptide (BPP-10c) from Bothrops jararaca induces changes in seminiferous tubules
.
J. Venom Anim. Toxins Incl. Trop. Dis.
19
,
28
24
Alberto-Silva
,
C.
,
Gilio
,
J.M.
,
Portaro
,
F.C.
,
Querobino
,
S.M.
and
Camargo
,
A.C.
(
2015
)
Angiotensin-converting enzyme inhibitors of Bothrops jararaca snake venom affect the structure of mice seminiferous epithelium
.
J. Venom Anim. Toxins Incl. Trop. Dis.
21
,
27
25
Corradi
,
H.R.
,
Schwager
,
S.L.
,
Nchinda
,
A.T.
,
Sturrock
,
E.D.
and
Acharya
,
K.R.
(
2006
)
Crystal structure of the N domain of human somatic angiotensin I-converting enzyme provides a structural basis for domain-specific inhibitor design
.
J. Mol. Biol.
357
,
964
974
26
Kröger
,
W.L.
,
Douglas
,
R.G.
,
O'Neill
,
H.G.
,
Dive
,
V.
and
Sturrock
,
E.D.
(
2009
)
Investigating the domain specificity of phosphinic inhibitors RXPA380 and RXP407 in angiotensin-converting enzyme
.
Biochemistry
48
,
8405
8412
27
Watermeyer
,
J.M.
,
Kröger
,
W.L.
,
O'Neill
,
H.G.
,
Sewell
,
B.T.
and
Sturrock
,
E.D.
(
2008
)
Probing the basis of domain-dependent inhibition using novel ketone inhibitors of angiotensin-converting enzyme
.
Biochemistry
47
,
5942
5950
28
Woodman
,
Z.L.
,
Schwager
,
S.L.
,
Redelinghuys
,
P.
,
Chubb
,
A.J.
,
van der Merwe
,
E.L.
,
Ehlers
,
M.R.
et al.  (
2006
)
Homologous substitution of ACE C-domain regions with N-domain sequences: effect on processing, shedding, and catalytic properties
.
Biol. Chem.
387
,
1043
1051
29
Lubbe
,
L.
,
Sewell
,
B.T.
and
Sturrock
,
E.D.
(
2016
)
The influence of angiotensin converting enzyme mutations on the kinetics and dynamics of N-domain selective inhibition
.
FEBS J.
283
,
3941
3961
30
Bull
,
H.G.
,
Thornberry
,
N.A.
and
Cordes
,
E.H.
(
1985
)
Purification of angiotensin-converting enzyme from rabbit lung and human plasma by affinity chromatography
.
J. Biol. Chem.
260
,
2963
2972
PMID:
[PubMed]
31
Schwager
,
S.L.
,
Carmona
,
A.K.
and
Sturrock
,
E.D.
(
2006
)
A high-throughput fluorimetric assay for angiotensin I-converting enzyme
.
Nat. Protoc.
1
,
1961
1964
32
Cheng
,
Y.
and
Prusoff
,
W.H.
(
1973
)
Relationship between the inhibition constant (KI) and the concentration of inhibitor which causes 50 per cent inhibition (I50) of an enzymatic reaction
.
Biochem. Pharmacol.
22
,
3099
3108
33
Anthony
,
C.S.
,
Corradi
,
H.R.
,
Schwager
,
S.L.
,
Redelinghuys
,
P.
,
Georgiadis
,
D.
,
Dive
,
V.
et al.  (
2010
)
The N domain of human angiotensin-I-converting enzyme: the role of N-glycosylation and the crystal structure in complex with an N domain-specific phosphinic inhibitor, RXP407
.
J. Biol. Chem.
285
,
35685
35693
34
Waterman
,
D.G.
,
Winter
,
G.
,
Gildea
,
R.J.
,
Parkhurst
,
J.M.
,
Brewster
,
A.S.
,
Sauter
,
N.K.
et al.  (
2016
)
Diffraction-geometry refinement in the DIALS framework
.
Acta Crystallogr. D Struct. Biol.
72
,
558
575
35
Evans
,
P.R.
and
Murshudov
,
G.N.
(
2013
)
How good are my data and what is the resolution?
Acta Crystallogr. D Biol. Crystallogr.
69
,
1204
1214
36
Collaborative Computational Project, Number 4
. (
1994
)
The CCP4 suite: programs for protein crystallography
.
Acta Crystallogr. D Biol. Crystallogr.
50
,
760
763
37
McCoy
,
A.J.
,
Grosse-Kunstleve
,
R.W.
,
Adams
,
P.D.
,
Winn
,
M.D.
,
Storoni
,
L.C.
and
Read
,
R.J.
(
2007
)
Phaser crystallographic software
.
J. Appl. Crystallogr.
40
,
658
674
38
Cozier
,
G.E.
,
Schwager
,
S.L.
,
Sharma
,
R.K.
,
Chibale
,
K.
,
Sturrock
,
E.D.
and
Acharya
,
K.R.
(
2018
)
Crystal structures of sampatrilat and sampatrilat-Asp in complex with human ACE – a molecular basis for domain selectivity
.
FEBS J.
285
,
1477
1490
39
Murshudov
,
G.N.
,
Vagin
,
A.A.
and
Dodson
,
E.J.
(
1997
)
Refinement of macromolecular structures by the maximum-likelihood method
.
Acta Crystallogr. D Biol. Crystallogr.
53
,
240
255
40
Adams
,
P.D.
,
Afonine P
,
V.
,
Bunkóczi
,
G.
,
Chen
,
V.B.
,
Davis
,
I.W.
,
Echols
,
N.
et al.  (
2010
)
PHENIX: a comprehensive Python-based system for macromolecular structure solution
.
Acta Crystallogr. D Biol. Crystallogr.
66
,
213
221
41
Emsley
,
P.
and
Cowtan
,
K.
(
2004
)
Coot: model-building tools for molecular graphics
.
Acta Crystallogr. D Biol. Crystallogr.
60
,
2126
2132
42
Chen
,
V.B.
,
Arendall
, 3rd,
W.B.
,
Headd
,
J.J.
,
Keedy
,
D.A.
,
Immormino
,
R.M.
,
Kapral
,
G.J.
et al.  (
2010
)
MolProbity: all-atom structure validation for macromolecular crystallography
.
Acta Crystallogr. D Biol. Crystallogr.
66
,
12
21
43
McNicholas
,
S.
,
Potterton
,
E.
,
Wilson
,
K.S.
and
Noble
,
M.E.
(
2011
)
Presenting your structures: the CCP4mg molecular-graphics software
.
Acta Crystallogr. D Biol. Crystallogr.
67
,
386
394
44
Laskowski
,
R.A.
and
Swindells
,
M.B.
(
2011
)
LigPlot+: multiple ligand–protein interaction diagrams for drug discovery
.
J. Chem. Inf. Model.
51
,
2778
2786
45
Carmona
,
A.K.
,
Schwager
,
S.L.
,
Juliano
,
M.A.
,
Juliano
,
L.
and
Sturrock
,
E.D.
(
2006
)
A continuous fluorescence resonance energy transfer angiotensin I-converting enzyme assay
.
Nat. Protoc.
1
,
1971
1976
46
Skirgello
,
O.E.
,
Binevski P
,
V.
,
Pozdnev
,
V.F.
and
Kost
,
O.A.
(
2005
)
Kinetic probes for inter-domain co-operation in human somatic angiotensin-converting enzyme
.
Biochem. J.
391
,
641
647
47
Yu
,
X.C.
,
Sturrock
,
E.D.
,
Wu
,
Z.
,
Biemann
,
K.
,
Ehlers
,
M.R.
and
Riordan
,
J.F.
(
1997
)
Identification of N-linked glycosylation sites in human testis angiotensin-converting enzyme and expression of an active deglycosylated form
.
J. Biol. Chem.
272
,
3511
3519
48
Masuyer
,
G.
,
Schwager
,
S.L.U.
,
Sturrock
,
E.D.
,
Isaac
,
R.E.
and
Acharya
,
K.R.
(
2012
)
Molecular recognition and regulation of human angiotensin-I converting enzyme (ACE) activity by natural inhibitory peptides
.
Sci. Rep.
2
,
717
49
Watermeyer
,
J.M.
,
Sewell
,
B.T.
,
Schwager
,
S.L.
,
Natesh
,
R.
,
Corradi
,
H.R.
,
Acharya
,
K.R.
et al.  (
2006
)
Structure of testis ACE glycosylation mutants and evidence for conserved domain movement
.
Biochemistry
45
,
12654
12663
50
Natesh
,
R.
,
Schwager
,
S.L.
,
Sturrock
,
E.D.
and
Acharya
,
K.R.
(
2003
)
Crystal structure of the human angiotensin-converting enzyme-lisinopril complex
.
Nature
421
,
551
554
51
Harvey
,
A.L.
(
2014
)
Toxins and drug discovery
.
Toxicon
92
,
193
200
52
McCleary
,
R.J.
and
Kini
,
R.M.
(
2013
)
Non-enzymatic proteins from snake venoms: a gold mine of pharmacological tools and drug leads
.
Toxicon
62
,
56
74
53
Gavras
,
H.
,
Brunner
,
H.R.
,
Laragh
,
J.H.
,
Sealey
,
J.E.
,
Gavras
,
I.
and
Vukovich
,
R.A.
(
1974
)
An angiotensin converting-enzyme inhibitor to identify and treat vasoconstrictor and volume factors in hypertensive patients
.
N. Engl. J. Med.
291
,
817
821
54
Ehlers
,
M.R.
,
Fox
,
E.A.
,
Strydom
,
D.J.
and
Riordan
,
J.F.
(
1989
)
Molecular cloning of human testicular angiotensin-converting enzyme: the testis isozyme is identical to the C-terminal half of endothelial angiotensin-converting enzyme
.
Proc. Natl Acad. Sci. U.S.A.
86
,
7741
7745

Author notes

*

C-domain numbering throughout this paper according to the testis ACE isoform nomenclature described by Ehlers et al. [54].