Structural insight into the catalytic mechanism and inhibitor binding of aminopeptidase A

Aminopeptidase A (APA) is a membrane-bound monozinc aminopeptidase. In the brain, APA generates angiotensin III which exerts a tonic stimulatory effect on the control of blood pressure in hypertensive animals. The oral administration of RB150 renamed firibastat by WHO, an APA inhibitor prodrug, targeting only the S1 subsite, decreases blood pressure in hypertensive patients from various ethnic origins. To identify new families of potent and selective APA inhibitors, we explored the organization of the APA active site, especially the S 2 ’ subsite. By molecular modeling, docking, molecular dynamics simulations and site-directed mutagenesis, we revealed that Arg368 and Arg386, in the S2’ subsite of human APA established various types of interactions in major part with the P2’residue but also with the P1' residue of APA inhibitors, required for their nanomolar inhibitory potency. We also demonstrated an important role for Arg368 in APA catalysis, in maintaining the structural integrity of the GAMEN motif, a conserved sequence involved in exopeptidase specificity and optimal positioning of the substrate in monozinc aminopeptidases. This arginine together with the GAMEN motif are key players for the catalytic mechanism of these enzymes. e arginine residues, Arg368 and Arg 386 located in the S 2 ’ subsite of hAPA that established either hydrogen bonds and salt bridges with the P1’ and P2’ residues of APA inhibitors. We also showed that Arg368 played a critical role in catalysis by ensuring the optimal orientation of the GAMEN loop, a key motif of monozinc aminopeptidases responsible for their exopeptidase specificity.


INTRODUCTION
Aminopeptidase A (APA; EC 3.4.11.7) is a 160 kDa homodimeric type II membrane-bound monozinc aminopeptidase. The activity of this enzyme is enhanced by Ca 2+ , which increases the hydrolysis, by APA, of N-terminal acidic residues from substrates [1]. Thus, APA hydrolyzes the N-terminal glutamate or aspartate residues of peptide substrates, such as angiotensin II (AngII) or cholecystokinin-8, in vitro [2,3]. This enzyme has been identified in several brain nuclei involved in controlling body fluid homeostasis and cardiovascular functions, together with other components of the brain renin-angiotensin system [4]. Using specific and selective APA inhibitors, such EC33 [(S)-3-amino-4-mercaptobutanesulfonate, Ki = 3 x 10-7 M] (Table S1) [5], we showed that APA cleaved brain AngII to generate angiotensin III (AngIII) in vivo and that AngIII exerts a tonic stimulatory action on the control of blood pressure (BP) in hypertensive animals [6][7][8]. The inhibition of brain APA with EC33 normalizes BP in alert hypertensive rats, suggesting that brain APA constitutes an interesting candidate target for the treatment of hypertension (HTN) [9]. This led us to develop an orally active APA inhibitor prodrug of EC33, RB150 renamed firibastat by World Health Organization, which blocks brain APA activity and AngIII formation and decreases BP in hypertensive rats [8,10,11]. Firibastat was then shown to decrease BP in hypertensive patients of multiple ethnic origins [12,13]. To identify more potent and selective APA inhibitors, we designed efficient and selective APA inhibitors, using a combinatorial approach with (3-amino-2-mercapto-acyl)-dipeptides able to fit into the S 1 , S 1 ' and S 2 ' subsites of APA [14], we previously showed that the S 1 subsite was specific of a glutamate residue, and that the S 1 ' subsite of an hydrophobic residue whereas the S 2 ' subsite preferentially recognized negatively charged residues derived from aspartic acid. This led us to synthesize the  Table S1) into the APA active site [15]. In this 3D model ( Fig.1), the zinc atom was coordinated by the two histidine residues (His385 and His389) of the consensus zinc-binding motif, HEXXH [16], a water molecule and Glu408, which has been shown to be the third zinc ligand. We also showed that Tyr471 and Glu352 residue of the GAMEN motif were responsible for transition state stabilization [17] and APA exopeptidase specificity respectively [18,19]. We then introduced Ca 2+ into the APA 3D model and observed that there is one Ca 2+ atom in the S1 subsite. The crystal structure of human APA (hAPA) (residues 76 to 956) was recently resolved [20] and a comparison of this structure with our 3D homology mAPA model showed a perfect overlap for the APA active site and the same structural organization of the S1 subsite (Fig. S1). However, the residues constituting the S 2 ' subsite and interacting with the electronegative P2' moiety of pseudotripeptide APA inhibitors such as CD497b [14] were unknown. Here, to identify these residues, we performed the molecular docking of pseudotripeptide APA inhibitors in the refined crystal structure of hAPA.
Molecular dynamics simulations and site-directed mutagenesis studies identified two arginine residues, Arg368 and Arg 386 located in the S 2 ' subsite of hAPA that established either hydrogen bonds and salt bridges with the P1' and P2' residues of APA inhibitors. We also showed that Arg368 played a critical role in catalysis by ensuring the optimal orientation of the GAMEN loop, a key motif of monozinc aminopeptidases responsible for their exopeptidase specificity.

MATERIALS AND METHODS
Preparation of the target 3D structure. The hAPA crystallographic structure (PDB ID 4KXD) was used as a starting point for our molecular docking and molecular dynamics simulations. This choice was driven by the presence of both the Ca 2+ atom and the glutamate ligand, reflecting the fully activated conformation of the enzyme. We first removed the glutamate ligand, together with the N-acetyl-D-glucosamine and non-conserved water molecules (obtained by comparing water positions within all the available APA PDB files 4KX7, 4KX8, 4KX9, 4KXA, 4KXB, 4KXC, 4KXD) from the X-ray structure. Active site Zn 2+ and Ca 2+ atoms were retained, and a loop missing from all available X-ray structures was added for residues 608 to 611. The ModLoop server was used for this purpose [21].

Molecular docking.
Reference inhibitors were docked into the refined 3D structure obtained with GOLD v5.1, using the ChemPLP scoring function [22]. Several penalties were applied in the scoring function, in the form of protein H-bonds constraints, following the GOLD user guide.
The constraint weights are unitless user defined multiplying factor, and they were established based on the analysis of hAPA crystallographic structures complexes [20], as well as on previous structure activity relationship data, in order to appropriately reproduce experimental poses. In the absence of such constraints, the ligands would bind far from the catalytic site. Crystallographic structures revealed conserved hydrogen bonds (as deduced from angles and distances between possible hydrogen donors and acceptors) between the ligands (either nonselective inhibitors amastatin or bestatin and amino acid ligands arginine, glutamate or aspartate) and residues Downloaded from http://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20200307/893492/bcj-2020-0307.pdf by guest on 22 September 2020 Glu223, Glu360, Glu394, Glu416, Tyr479 and Arg887 of hAPA (Glu215, Glu352, Glu386, Glu408, Tyr471 and Arg878 in mAPA). We first defined and applied a constraint weight of 20 if no interaction occurred with the Glu223, Glu360, Glu394 or Glu416 carboxylate groups. All Xray ligands made at least two hydrogen bonds within this cluster of glutamates, which have previously been shown to be key residues for APA and monozinc aminopeptidase activity [23].
Secondly, a constraint weight of 10 was defined and applied if no interaction occurred with the Arg887 guanidinium group. The position of this residue is displaced by bestatin binding and could be targeted by any acidic group bound close to the Ca 2+ atom, contributing of APA specificity for acidic side chains. Lastly, a constraint weight of 5 was applied if no interaction occurred with the hydroxyl group of Tyr479. This is a secondary interaction observed on all Xray structures. Poses were selected based on the docking score and visual inspection, in order to avoid false positives. We looked for poses with less intra and intermolecular clashes, that respected the planarity of peptide bonds, and that agreed with the well-established SAR.

Molecular dynamics simulations.
For all the ligands considered, the best poses obtained in the docking process described above were used for MD simulations with NAMD [24]. For all the chemicals, the necessary nonstandard parameters to be added to the Charmm36 [25] force field were obtained with BIOVIA Discovery Studio Visualizer 4.5 (Dassault Systèmes BIOVIA, San Diego), by assigning atom types manually and generating a custom residue. Charge distributions were calculated using the PM7 method in MOPAC 2016 (J.J.P. Stewart, Stewart Computational Chemistry, Colorado Springs, CO, USA), and added to the previously obtained topology files. Ligand-protein complexes were generated with the VMD [26]AUTOPSF plugin, using the topology files provided Downloaded from http://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20200307/893492/bcj-2020-0307.pdf by guest on 22 September 2020 by BIOVIA. The VMD MUTATOR plugin was used to build systems with mutated APA (R368A and R386A). The obtained systems were then included in a water box using the VMD SOLVATE plugin, followed by charge neutralization through the IONIZE plugin. The final systems, comprising approximately 100,000 atoms, were then subjected to three stages of conjugate gradient energy minimization of 6400 steps each. In the first stage, the ligand-protein complex was fixed, and water was allowed to move freely. The second stage allowed ligand motions, and the last stage consisted of energy minimization for the entire system. The resulting systems were then subjected to a short MD simulation of 500 ps, to equilibrate pressure and temperature before the production runs. MD production simulations were carried out in triplicate, using the particle mesh Ewald approach for treating electrostatic interactions, in the NPT ensemble using Langevin dynamics and piston methods to fix temperature (298 K) and pressure (1 atm), with periodic boundary conditions. Simulations were performed with 1 fs per time step, during 55,000,000 steps, generating three trajectories of 55 ns for each system. The first 5 ns of each trajectory were considered as equilibration time and were not included in further analysis. Trajectory analyses such as h-bonding, secondary structure and RMSD were performed with VMD [26], pairwise interaction energy calculations were performed with NAMD and local structural fluctuations were obtained with MDLovoFit [24].
Site-directed mutagenesis. The Xpress-polyHis-tagged mAPA (His-mAPA) [27] cDNA was the template for the PCR-based site-directed mutagenesis. The arginine residues 360, 378 and 882 in mAPA were replaced with alanine residues using mutagenic primers A (forward) and B (reverse) were: ). Then site-directed mutagenesis was performed as previously described [28]. AlexaFluor 488-conjugated polyclonal anti-rabbit secondary antibody (Life Technologies) as previously described [29]. Leica TCS SP V (Leica Microsystem, Heidelberg, Germany) microscope was used to analyze the cells as previously described [29].
Purification and Western-blot analysis recombinant His-mAPAs. Stable CHO cell lines (3x10 9 cells) were harvested and a crude membrane preparations were prepared as previously described [27]. Crude membrane preparations were solubilized in 20 mM Tris-HCL buffer (pH 8) containing 100 mM NaCl and 1% nonidet P-40 (loading buffer, 1 mL/10 6 cells ratio) for 16 hours at 4°C. The solubilized membrane preparation was centrifuged at 100,000 x g for 90 min and the supernatant was subjected to metal affinity chromatography with a metal chelate resin column (Talon Co 2+ , Clontech) [27].

RESULTS
Molecular docking of APA inhibitors targeting the S1, S1' and S2' subsites in the refined crystal structure of human APA We performed the molecular docking of different inhibitors in the X-ray structure of hAPA ( Fig.   2A). We applied a series of soft constraints, in the form of docking penalties (for details, see Appendix), to maintain the docking area near the catalytic site. These penalties were formulated following close inspection of the ligand-bound X-ray structures of hAPA with either amastatin or bestatin (nonselective aminopeptidase inhibitors) or with the amino acids arginine, glutamate or aspartate (PDB IDs 4KX7, 4KX8, 4KX9, 4KXA, 4KXB, 4KXC, 4KXD, [20]). Together with our previous structure-activity relationship data [23], these informations allowed to calibrate soft constraints during the docking protocol. After molecular docking with these penalties applied, all pseudotripeptide APA inhibitors (CD409b-11, CD497b and SC1021, Table S2) interacted with the Zn 2+ atom in the S 1 subsite of APA via their thiol groups, and with Glu223, Glu360, and Glu416 via their free alpha amino groups ( Fig. 2B-D). The sulfonate group of these inhibitors was embedded at the bottom of the S 1 subsite, interacting simultaneously with the Ca 2+ atom (bound to residues Asp221 and Asp226) and with Arg887. Furthermore, in the S 1 ' subsite, the phenolic moiety of CD409b-11 and the isoleucine side-chain of CD497b or SC1021 established hydrophobic interactions with the side chains of the Val390 or Tyr479 residues located in this region. Finally, in the P 2 ' position of the inhibitor, the carboxyproline group of CD497b, and the aspartate of CD409b-11 or SC1021 interacted with the guanidinium group of the Arg386 residue ( Fig. 2B-D). Remarkably, the CD409b-11 inhibitor displayed potential interactions with both the Arg368 and Arg386 residues of hAPA (Fig. 2B).

Pose refinement and evaluation of docking results by molecular dynamics simulations
We evaluated the predicted interaction models described above, by performing molecular dynamics simulations of the hAPA/CD409b-11 inhibitor complex. Pair interaction energies were evaluated to determine whether the binding pose of this inhibitor was stable. Interaction energies (E int ) between the inhibitor and the solvent (Fig. 3A) were less favorable (-210.8 ± 39 kcal/mol) than interactions between the inhibitor and APA (-487.6 ± 48.03 kcal/mol), indicating a strong interaction between the enzyme and the inhibitor, which was stabilized after 15 ns of simulation.
We then investigated the participation of the two arginine residues detected by our docking protocol (Arg368 and Arg386) in the putative S2' subsite, by evaluating changes in the distance between the carboxylic acid oxygen from the inhibitor P2' side chain and the nitrogen atoms of the guanidinium groups (Fig. 3B). Throughout the simulation, the distances between the inhibitor and Arg368 fluctuated more than the distances between the inhibitor and Arg386. Indeed, such distances suggested that the inhibitor aspartate P2' residue fluctuates between salt bridges (average usual salt bridge distances around 4.5 Å) and hydrogen bonds with Arg368 (average usual hydrogen bonds distance around 2.5 Å), whereas Arg386 may establish hydrogen bonds with the P2' residue. In addition, Arg386 is well positioned to establish cation- interactions with the tyrosine aromatic ring of P1' residue (average usual cation- distances around 2.5 Å).
We further analyzed pairwise interaction energies between individual APA amino-acid residues and the CD409b-11 inhibitor (Fig. 3C), to identify other putative APA residues responsible for Arg386 were well-conserved in molecular dynamics simulations, strongly suggesting that both residues belong to the APA S 2 ' subsite. All these interactions were summarized on  (Table S3 and S4). We first checked by Western blot that Ala substitution of Arg882 did not affect protein expression (Fig. S6). Then we evaluated the kinetic parameters of the mutated enzyme in the presence of of 4 mM Ca 2+ using GluNA as a substrate (Table S3) and showed that the k cat /K m , was reduced by a factor of 38 when compared to the wild-type mAPA.
We then analyzed the effect of the replacement of Arg882 with an alanine residue on the binding of the pseudotripeptide APA inhibitor, CD497b, and showed that this inhibitor displayed the same K i for R882A mutant than for wild-type mAPA (Table S4). This result showed that Arg882 (Arg891 in hAPA) did not play a role in the S2' subsite, demonstrating that this second pose is unlikely the right one. Thus the R882A mutated mAPA was not further characterized and we focused our work on the Arg360 (Arg368 in hAPA) and Arg378 (Arg386 in hAPA) residues.

Site-directed mutagenesis and analysis of kinetic parameters of the purified recombinant wild-type and mutated mouse His-APAs
The Arg360 (Arg368 in hAPA) and Arg378 (Arg386 in hAPA) residues of His-mAPA were substituted by an alanine residue and recombinant His-mAPAs were stably expressed in Chinese hamster ovary (CHO) cells and purified by metal affinity chromatography. After checking that recombinant APAs were produced and processed similarly to wild-type His-mAPA ( Fig. 4A-C) and that none of the mutations modified the EC 50 for Ca 2+ relative to that of wild-type APA (Fig.   4D, E), we analyzed the enzymatic activities of purified wild-type and mutated His-mAPAs by determining kinetic parameters (K m and k cat ) in the presence of 4 mM Ca 2+ , using GluNA as a substrate. The results are summarized in Table 1. The replacement of Arg360 (Arg368 in humans) with an alanine residue strongly decreased k cat /K m , by a factor of 94 relative to the wild type. This decrease resulted from a slight increase in K m , by a factor of 5.2, and a large decrease in hydrolysis velocity (k cat ), by a factor of 18. The replacement of Arg378 (Arg386 in humans) with an alanine residue lead to a decrease in k cat /K m by a factor of 14 relative to the wild type.
This was due to a slight increase in K m , by a factor of 3, and a decrease in hydrolysis velocity by a factor of 4.5.

Inhibitory potency of various classes of compounds against purified wild-type and mutated mouse His-APAs
We then characterized the roles of R360 (R368 in hAPA) and R378 (R386 in hAPA), by evaluating the inhibitory potency (K i value) of various classes of compounds against GluNA (0.  respectively. GluSH also had a lower potency to inhibit R360A and R378A than wild-type APA, by factors of 10 and 2, respectively. Finally, the inhibitory potency of (L,D)-GluPO 3 H 2 was lower for R360A and R378A than for wild-type APA, by factors of 420 and 2, respectively.
We then evaluated the effect of the substitutions on the K i values of the inhibitors CD497b, SC1021 and CD409b-11, which target the S 1 , S 1 ' and S 2 ' subsites of APA, on both wild-type and mutated His-mAPAs (Table 3). CD497b had a lower potency to inhibit R360A and R378A than wild-type APA, by a factor of 8 for both mutants. SC1021 was a less potent inhibitor of R360A and R378A than of wild-type APA, by factors of 7 and 10, respectively. Finally, CD409b-11 was a less potent inhibitor towards R360A and R378A than towards wild-type APA, by factors of 48 and 63, respectively. in the R368A mutant. In both wild-type and R368A APAs, the R368 backbone can form a hydrogen bond with Gly357 (Fig. 6A,B). However, the R368A mutant lacks several of the interactions present in wild-type hAPA (Fig. 6A), and this disturbs the pre-GAMEN loop, as seen in Fig. 5. In the ligand-free (apo) conformation of wild-type hAPA, these interactions included hydrogen bonds between the Arg368 side chain and either the Phe354 backbone from the pre-

Role of the Arg360 on the stabilization of the GAMEN in mAPA
To confirm that Arg368 of the hAPA stabilizes the GAMEN motif and contributes to the recognition of the free α-amino group of the inhibitors, we compared the capacity of L-and D-GluPO 3 H 2 to inhibit wild-type and R360A mAPAs (Table 4)

DISCUSSION
The key biological function of brain APA in modulating arterial BP has generated interest in developing pharmacological tools that can regulate its activity. We investigated the mode of binding of previously developed potent APA inhibitors targeting the S1, S1' and S2' subsites, by docking three pseudotripeptide APA inhibitors, CD497b, CD409b-11 and SC1021, on the X-ray structure of hAPA [20]. The molecular docking of these inhibitors highlighted the putative contributions of two arginine residues, Arg368 and Arg386, to the S2' subsite of the hAPA active site. The side chains of Arg368 and Arg386 interact with the carboxylate side chains of carboxyproline or aspartate in these inhibitors, consistent with the preference of the S2' subsite for negatively charged moieties [14]. These interactions were further supported by molecular dynamics simulations for the hAPA/CD409b-11 system and site-directed mutagenesis studies.
The functional role of Arg360 and Arg378 was then investigated by replacing the Arg360 and Arg378 residues in mAPA (corresponding to Arg368 and Arg386 in hAPA, respectively) with alanine residues by site-directed mutagenesis. First, we showed that recombinant mutated mAPA were expressed similarly to the wild-type mAPA and that substitutions did not modify the capacity of the mutated enzymes to be activated by Ca 2+ . However, these substitutions led to a decrease in mAPA activity, suggesting that Arg360 or Arg378 is involved in mAPA enzymatic activity. Analysis of kinetic parameters showed that mutated mAPAs had a slightly lower affinity and a significant decreased hydrolysis velocity when compared to the wild-type enzyme. These results suggest that Arg360 and Arg378 affect the formation of the Michaelis complex and Given the location of the Arg residues, structurally apart from the S1 subsite, and the impact of their mutation on GluNA hydrolysis, a substrate targeting only the S1 subsite, we further investigated the role of these Arg residues by analyzing the effects of their mutation on the inhibitory potency (Ki values) of APA inhibitors targeting either he S1 or the S1, S1' and S2' subsites of APA. The Ki values of EC33, and GluSH, which target the S1 subsite and mimic the Michaelis complex, were significantly more strongly affected for the R360A-mutated mAPA than for the R378A mutant when compared to wild-type mAPA. This finding suggests that the Arg360 residue plays an indirect role in the S1 subsite during formation of the Michaelis complex. Moreover, a role for Arg360 in catalysis was also supported by the 415-fold decrease in inhibitory potency observed for the analog of the transition state (L,D)-GluPO 3 H 2 , relative to the value obtained for wild-type mAPA. Otherwise, the very small effect of the Arg378 mutation on the Ki of inhibitors targeting the S1 subsite shows that this residue does not have an important role in this subsite. To investigate whether these arginine residues contribute to the specificity of the S2' subsite for electronegative groups, we evaluated the effect of the substitutions on the inhibitory potencies of inhibitors having a carboxyproline or an aspartate in P2' position.
Substitutions strongly and significantly decreased the inhibitory potency of CD497b, SC1021 and CD409b-11, relative to that of wild-type mAPA. Our data support the conclusion that Arg360 and Arg378 interact with the carboxylate side chain of the residue in the P2' position in the inhibitor and explain why the addition of an acidic residue in the P2' position increases the affinity of these inhibitors for APA. Our data also show that Arg378 is involved exclusively in Downloaded from http://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20200307/893492/bcj-2020-0307.pdf by guest on 22 September 2020 the S2' subsite. By contrast, Arg360 seemed to play a role not only in the S2' subsite but also in catalysis and in the binding of smaller inhibitors exclusively targeting the S1 subsite.
We then investigated the role of Arg360 and Arg378 in catalysis, by performing longer molecular dynamics simulations on wild-type and mutated hAPAs, in both the apo and holo configurations.
We found that Arg386 (Arg378 in mAPA), which was thought to contribute to the S2' subsite, was present at the entry of the cavity constituting the APA active site. The replacement of this Arg residue by an Ala residue might therefore affect the electronic environment around the site of APA substrate entry. This might make it harder for the substrate to enter the active site of APA, as illustrated by the increase in K m , and the products might find it harder to leave the cavity, accounting for the lower k cat value obtained for this mutant. Interestingly, molecular dynamics analysis showed that local changes triggered by R368A substitution (R360 in mAPA) modified the flexibility of the region adjacent to this residue.
Simulations with or without the transition state analog inhibitor L-GluPO 3 H 2 revealed that the Arg368 residue stabilized the GAMEN motif by interacting directly with residues flanking this region (the pre-GAMEN loop). Several groups have shown the GAMEN motif to be crucial for gluzincin aminopeptidase activity, because this conserved motif is responsible of recognition of the free N-terminal part of substrates or inhibitors through the Glu residue of the motif [18,[32][33][34]. This interaction, which is initiated on formation of the Michaelis complex, affects catalysis by allowing optimal positioning of the substrate within the active site to trigger its hydrolysis [18,[32][33][34]. We hypothesize that, in wild-type hAPA, Arg368 acts as a molecular pillar in addition to performing its role in the S2' subsite, supporting the structural arch between β-sheets partly shaping the catalytic domain. Interestingly, the plasticity of the pre-GAMEN loop of the recently solved X-ray structure of the human insulin-regulated aminopeptidase (IRAP) in its integrity of the S1 subsite by stabilizing the GAMEN motif. This conclusion may also apply to other monozinc aminopeptidases, because this Arg residue and the GAMEN motif are conserved among a large panel of monozinc aminopeptidases (Fig. S7).
We demonstrated that Arg368 and Arg386 participate in the S 2 ' subsite of APA through their guanidinium groups. From our MD simulation Arg368 fluctuates between electrostatic salt with the carboxylate of the residue in the P2'. From our docking experiment with CD409b-11, the aromatic ring of the P1' residue may also establish cation-π interactions with Arg386. Such cation- interactions implying Arg were already described [37] as playing important role in protein-ligand recognition and binding [38,39]. Our data regarding the binding of CD409b-11 to the R378A mAPA mutant were similar to the ones obtained with such mutation in glutamate receptor [40] showing that this cation- protein/ligand interaction was important (decrease of 62fold in inhibitory potency).
Interestingly, amastatin has an aspartate residue in P3' position (residue in yellow in Fig. S8) that based on the crystal structure, the carboxylates moieties of this residue are at distances with the guanidinium groups from Arg368 and Arg386 compatibles with salt bridges and hydrogen bonds interactions (Fig. S6). We also showed that Arg368 is involved in maintaining the structural integrity of the GAMEN motif, a key monozinc aminopeptidase sequence involved in exopeptidase specificity and optimal positioning of the substrate. More generally, this work deepens our knowledge of the structural organization of the active site and the common catalytic mechanism of monozinc aminopeptidases.
Clinical phase IIa and IIb trials with RB150, renamed firibastat by the World Health Organization, the first APA inhibitor prodrug of EC33, provided pharmacological proof of principle for the efficacy of brain APA inhibition for decreasing BP in hypertensive patients [12,13], especially in Black patients, who are often salt-sensitive and where monotherapy with ACE inhibitors or AT1 receptor antagonists is less effective. If the firibastat efficacy was confirmed in the pivotal phase III trial, FRESH, central-acting APA inhibitors could constitute a new class of antihypertensive agents to improve blood pressure control in patients with difficultto-treat or resistant hypertension. EC33 interacts only with the S1 subsite of the APA active site, its inhibitory potency towards APA is around 100 nM and it is 100-fold less active on APN [5].
In this context, our data, by providing a new structural framework for understanding the catalytic mechanism and binding site of APA inhibitors, could be very useful for the identification of new more potent and more selective APA inhibitors with better bioavailability than RB150 / firibastat and this could lead to significant improvements in current hypertensive therapy.        . Active site flexibility was affected by the R368A mutation. Trajectory frames were pooled from three replicate simulations and aligned with MDLovoFit, based on the most conserved Cα atoms (diverging by less than 1 Å from the initial structure). (A) RMSF analysis of the four simulations showed that, in wild-type APA, the residues of the GAMEN motif (residues 357 to 361) were stabilized by inhibitor binding. In the R368A-mutated APA, however, Gly357 and the loop preceding the GAMEN motif were strongly disturbed relative to wild-type APA. The bold horizontal bars on each RMSF graph correspond to the GAMEN (left bar) and HEXXHX18E (right bar) motifs. (B) Tube representation of the superposed trajectory frames after alignment, showing the flexibility of the pre-GAMEN loop. The position of the GAMEN motif backbone is indicated by letters, residue 368 (Arg or Ala) is labeled. The blue residues are the most strongly conserved, whereas the red residues diverge from the initial structure by more than 1 Å.