Stereoselective Activity of 1-Propargyl-4-styrylpiperidine-like Analogues That Can Discriminate between Monoamine Oxidase Isoforms A and B

The resurgence of interest in monoamine oxidases (MAOs) has been fueled by recent correlations of this enzymatic activity with cardiovascular, neurological, and oncological disorders. This has promoted increased research into selective MAO-A and MAO-B inhibitors. Here, we shed light on how selective inhibition of MAO-A and MAO-B can be achieved by geometric isomers of cis- and trans-1-propargyl-4-styrylpiperidines. While the cis isomers are potent human MAO-A inhibitors, the trans analogues selectively target only the MAO-B isoform. The inhibition was studied by kinetic analysis, UV–vis spectrum measurements, and X-ray crystallography. The selective inhibition of the MAO-A and MAO-B isoforms was confirmed ex vivo in mouse brain homogenates, and additional in vivo studies in mice show the therapeutic potential of 1-propargyl-4-styrylpiperidines for central nervous system disorders. This study represents a unique case of stereoselective activity of cis/trans isomers that can discriminate between structurally related enzyme isoforms.


■ INTRODUCTION
Specific interaction patterns and molecular geometry are critical factors that influence ligand binding affinity and ligand selectivity toward a target. 1 Numerous examples of "chiral switching" in drug development have highlighted the effects of spatial geometry on not only pharmacodynamics but also pharmacokinetics and toxicology of drugs. 2,3 In addition to optical isomers, cis and trans isomers can have different pharmacological activities, as demonstrated by the well-known cases of cisplatin, 4 tamoxifen, 5 and combretastatin A4 (CA-4). 6 This last includes a stilbene motif in the cis conformation ( Figure 1A), which efficiently binds to the colchicine binding site of tubulin and acts as a microtubule-destabilizing agent. 6 The binding of trans-combretastatin A4 (trans-CA-4, Figure  1A) to this site is thermodynamically less stable than that of cis-combretastatin A4 (cis-CA-4), and consequently the trans conformation does not prevent microtubule assembly. 6 However, examples of stereoselective activities of cis/trans isomers are rare in comparison with optical isomers.
Human monoamine oxidase A (hMAO-A) and human monoamine oxidase B (hMAO-B) are flavoenzymes that are encoded by separate genes on the X chromosome, and they share 70% sequence identity. 7 They are differentiated by their substrate specificities and inhibitor sensitivities. This arises from the differences in the active site size and shape, and the altered amino acid residues between these isoforms. The hMAO-A and hMAO-B active sites are hydrophobic cavities with volumes of about 550 Å 3 and 700 Å 3 , respectively. 8 The active site of hMAO-B is flat and elongated, with a bipartite configuration, where Ile199 and Tyr326 are the gating residues that determine the shape of the cavity and the specificity of substrate/inhibitor binding. 9 hMAO-A has a monopartite and more spherical active site, with Ile335 and Phe208 as the gating residues. MAOs catalyze the oxidative deamination of primary and some secondary amines, and they are responsible for neurotransmitter metabolism in peripheral tissues and in the central nervous system. 10 This reaction produces an imino intermediate, which is spontaneously hydrolyzed to the corresponding aldehyde, with ammonia and hydrogen peroxide as side-products. 11 Reduced levels of neurotransmitters, increased activity and levels of MAOs, and the production of toxic metabolites are linked with various neurological, 12,13 psychological, 14 and cardiovascular disorders 15 and cancers. 16 MAO-A inhibitors have been widely used for several decades in the therapy of depression and affective disorders. 17 On the other hand, selective MAO-B inhibitors are used as monotherapies in the early stages of Parkinson's disease and as add-on therapy to levodopa in advanced forms of Parkinson's disease. 18 MAO-A and MAO-B are recognized as validated targets. In terms of their contribution to increased cellular oxidative stress, 19 novel inhibitors are emerging as important disease-modifying therapeutic agents in such devastating neurodegenerative diseases as Alzheimer's disease and a number of others. 20 To the best of our knowledge, cis/trans isomerisms of double bonds have never been used to achieve stereoselective inhibition of MAO isoforms. Here, we demonstrate a unique case study whereby configuration of the double bond in stilbene-like derivatives can define the selectivity for either MAO-A or MAO-B. We identify 1-propargyl-4-((E)-styryl)piperidines as potent, selective, and irreversible hMAO-B inhibitors. Cis analogues with small substituents on the phenyl ring selectively inhibit hMAO-A. These inhibitors were evaluated for their MAO inhibitory activities and pharmacological properties.

■ RESULTS
Identification, Design, Synthesis, and Selectivity Profile of Potent and Selective hMAO-A and hMAO-B Inhibitors. Compound 1 ( Figure 1B) contains a stilbene-like motif, and it was identified as a hit inhibitor in an in-house library screening campaign. Indeed, compound 1 is a potent selective hMAO-B inhibitor, with no significant inhibition of hMAO-A at 100 μM ( Figure 1C).
The active site of hMAO-B is also significantly more curved than the active site of hMAO-A, with a bend of 90°, while in hMAO-A, the cavity bends at an angle of ∼60°(compare Figure 2A,B). The narrow entrance of hMAO-B and the higher curvature might prevent the relatively wider cis isomer 6 from entering the hMAO-B binding site. Indeed, 6 has a maximum width of 4.1 Å when measured as the height of the triangle composed of the fluorine atom, one of the sp 2 carbons, and the carbon bonded to the nitrogen, while the maximum width of the trans isomer 1 is only 2.5 Å when measured as the distance between two opposing carbon atoms in the piperidine ring. The cis isomer thus appears to inhibit hMAO-A but not hMAO-B because it is too wide to enter the narrower hMAO-B entrance. Its maximum width is only slightly greater than the 3.8 Å diameter of the hMAO-A entrance, while it is significantly greater than the 3.2 Å diameter of the hMAO-B entrance, which suggests that the hMAO-B entrance would need to undergo more substantial conformational changes to accommodate the cis isomer 6. The active site loop in hMAO-A contains Gly (i.e., Gly110), which is replaced by Pro102 in hMAO-B. This produces much higher rigidity in comparison to hMAO-A, where the active site loop is more flexible. This lack of flexibility of hMAO-B adds to the restricted access of the cis isomer to the hMAO-B active site. The cis isomer 6 inactivity against hMAO-B thus appears to be caused by steric hindrance in the active site, compound 6 being unable to fit into the elongated and rigid cavity ( Figure 2).
However, the selectivity of the trans isomer 1 is not explained only by the analysis of its access to the active site, as trans isomer 1 can bind to both hMAO-A and hMAO-B. To simulate the transport process of the inhibitor from the active site to the surface of the protein, these two isomers were undocked from hMAO-A and hMAO-B using the CaverDock software. 22 Starting from the docked conformations of both ligands at the active sites, the energetically most favorable trajectories of the ligands to the protein surfaces were investigated ( Figure 3). In hMAO-A, cis isomer 6 is oriented with the propargyl group facing the FAD cofactor, with a binding energy of −8.3 kcal/mol ( Figure 3A, movie file jm9b01886_si_002.mp4 in Supporting Information ). As cis isomer 6 moves toward the entrance, its energy peaks at 0.4 kcal/mol and then falls to −3.8 kcal/mol at the entrance ( Figure 3B). Trans isomer 1 in the correct orientation has a binding energy of −6.2 kcal/mol ( Figure 3C), which is higher than for 6. As the trans isomer is undocked, its binding energy peaks at −3.4 kcal/mol and then falls to −4.1 kcal/mol at the entrance ( Figure 3D). Further, the trans ligand in the "wrong" orientation in the active site (i.e., with the fluorophenyl group facing the FAD cofactor) has a much lower energy of −10.1 kcal/mol ( Figure 3D; lower-bounds energy curve). However, in this orientation, the trans ligand cannot form the covalent bond with FAD. Thus, in agreement with the experimental findings, the computational undocking simulation of hMAO-A showed that cis isomer 6 has lower binding energy at the active site than trans isomer 1, which indicates that the cis isomer is more likely to bind to hMAO-A than the trans isomer. In addition, the trans isomer might preferentially access the active site in an orientation that does not allow formation of the covalent bond, as can be seen by the lower binding energy of the "wrong" conformation compared to the energy of the correctly oriented trans isomer. This noncovalent only binding of the trans isomer might lead to the worse performance observed experimentally of this inhibitor with hMAO-A.
In hMAO-B, the docked cis ligand 6 in the correct orientation in the active site has energy of −6.5 kcal/mol ( Figure 3E; movie file jm9b01886_si_003.mp4 in Supporting Information). No energy peaks were observed along the active site for this isomer, whereby the energy reaches a plateau of 4.6 kcal/mol near the active site entrance ( Figure 3F). Trans isomer 1 in the correct orientation has the energy of −7.4 kcal/ mol ( Figure 3G; movie file jm9b01886_si_003.mp4 in Supporting Information), which is lower than the energy of the cis isomer, and then this falls further to −8.2 kcal/mol at the lowest point on the way out of the entrance ( Figure 3H). The energy barrier to reach the active site is lower for the trans isomer than it is for the cis isomer, with the highest peak of 1.5 kcal/mol near the entrance, which is lower than the 4.6 kcal/ mol energy barrier for the cis isomer. This lower energy barrier of the trans isomer at the entrance might be explained by the smaller maximum width of the trans isomer compared to the cis isomer, as discussed above. We also note that the docked trans inhibitor position in the active site has a root-meansquare deviation of 2.1 Å, compared to the cocrystal structure of inhibitor 1 in hMAO-B reported in the present study (PDB code 6RKB). This indicates that the docking successfully predicted the positioning of this inhibitor in the active site. In agreement with the experimental findings, the lower binding energies at the active site for the trans isomer and the lower energy barrier to reach the active site for the trans isomer indicate that the binding of trans isomer 1 to hMAO-B is favored compared to the binding of cis isomer 6.
The distinct activity of this geometric isomer pair prompted us to study the structure−activity relationships (SARs) through the construction of a focused library of substituted piperidines (19−26; Scheme 2, Table 1) to probe the effect of piperidine disubstitution pattern on the inhibitory potency.
1,4-Disubstituted piperidines were the most tolerated and showed the highest selectivity for both hMAO-A and hMAO-B, and thus several analogues with various substituents on the phenyl ring were synthesized (Scheme 3, compounds 67− 107).
Structure−Activity Relationships and Stability of Compounds. Hit inhibitor 1 and the synthesized analogues 6, 19−26, and 67−107 were evaluated for their inhibition of both hMAO-A and hMAO-B using a horseradish peroxidase (HRP)−Amplex Red coupled assay ( Table 1). The most important findings regarding the SARs are summarized in Figure 4. Scanning the 1,3-and 1,4-disubstitution patterns of piperidine derivatives with the trans configuration of the double bond revealed at least 5-fold superior inhibition of hMAO-B for the 1,4-disubstitution (e.g., 19 IC 50 = 1470 nM vs 1 IC 50 = 342 nM). Interestingly, reduction of the double bond retained this selective hMAO-B inhibition (IC 50 of 243 nM and 370 nM for hit compound 1 and analogue 25, respectively. Broadening the series of 1,4-disubsituted piperidines showcased selective inhibition of hMAO-A by cis analogues with small substituents on the phenyl ring (e.g., fluorine). The most potent was 69, a 3-fluorophenyl derivative, with an IC 50 of 68.4 nM and almost a 3-log unit selectivity over hMAO-B. On the other hand, the trans derivatives preferentially inhibited hMAO-B, with a general trend of improved inhibitory potencies with increasing hydrophobicity/size of the sub-   For additional experiments, compounds 1, 6, 67, 69, 84, 97, and 100 were used in the form of hydrochloride salts to avoid possible solubility issues. The thermodynamic solubility of the compounds in PBS (pH 7.4) was at least 17.6 g/250 mL, 10.5 g/250 mL, 8.6 g/250 mL, 12.5 g/250 mL, 14.5 g/250 mL, and 16.3 g/250 mL for 6, 67, 69, 84, 97, and 100, respectively, which is good solubility according to biopharmaceutical classification for any realistically expected dose of investigated compounds.
Possible chemical instability and photoisomerization of E/Z isomers were also addressed as these are known liabilities of  Inspection of the spectra confirmed that these compounds are relatively stable after more than 1 year of storage at room temperature (protected from sunlight) as only smaller amounts of impurities were detected for 84 (in the aromatic area), whereas compounds 69 and 100 showed no signs of chemical instability. The samples dissolved in the NMR tubes were then left on the benchtop exposed to natural daylight for 7 days, which could induce photoisomerization. No spectral changes were observed in comparison to spectra of freshly prepared samples. Altogether, these studies imply reasonable stability of the compounds.
Kinetic Evaluation. The exploration of SARs allowed the identification of a few compounds for more in-depth kinetic analysis with the purified enzyme samples, which allowed their inhibition mechanism to be better defined. On the basis of their inhibitory potencies and molecular features, compounds 1, 84, and 97 were selected for hMAO-B, whereas for hMAO-A the analysis was restricted to 69. The inhibitors were tested using UV−vis spectrum measurements ( Figure 5) and steadystate enzymatic assays ( Table 2). The propargyl unit is known to react covalently with the N5 atom of the flavin cofactor by rapidly forming a stable covalent adduct that produces a typical modification of the spectrum; this is seen as bleaching of the peak at 450 nm and appearance of a stronger peak at 415 nm. 24 All of these selected compounds were tested by adding them in 10-fold excess to the protein solution in the spectrophotometer cuvette and measuring the spectra over time until no further modification was detected. All of the irreversible inhibitors produced a clear change in the enzyme spectra ( Figure 5), similar to other classic acetylenic inhibitors, like L-deprenyl and clorgyline. 24 However, there were differences in the times required to complete the reactions for the adduct formation, i.e., to reach the final spectrum with no further modification. While 69 fully inactivated hMAO-A almost instantaneously ( Figure 5A), similar to clorgyline, the hMAO-B selective compounds were slower. In particular, 84 initiated the reaction with the flavin within 1 min, showing an intermediate spectrum profile that evolved into the final 415 nm peak in 30 min ( Figure 5B). Instead, 1 and 97 did not show any spectral modifications in the first few minutes and required half an hour to obtain an intermediate profile and more than 2 h to complete the reaction ( Figure 5C and Figure 5D, respectively).
Full kinetic analysis was then undertaken for hMAO-A and hMAO-B inhibition by the selected compounds, using purified enzyme samples and the spectrophotometric version of the assay coupled to HRP that was used for IC 50 determination, as described above. As enzyme inhibition by propargyl-based  inhibitors involves a first step in which the ligand is accommodated into the protein active site to be correctly oriented for flavin adduct formation, competitive inhibition can be measured under steady-state conditions using kynuramine and benzylamine as substrates for hMAO-A and hMAO-B, respectively. Thus, both the reversible and irreversible inhibition parameters were determined ( Table 2). Compound 69 is a potent hMAO-A inhibitor with a reversible K i of 0.23 μM, which is 5-fold higher than that reported for clorgyline. This difference is in line with the irreversible constant K I , and it is balanced by the irreversible inactivation velocity that is 10fold higher that for clorgyline. Overall, 69 is a very potent hMAO-A inhibitor with comparable activity to clorgyline. Similar to L-deprenyl, in the case of hMAO-B selective ligands, the reversible K i values are all in the micromolar range. The irreversible constant K I for covalent hMAO-B inhibitors is in general 5-fold the reversible K i ; however, the k inact /K I indicated comparable inhibitory potencies of the hMAO-B inhibitors to L-deprenyl.

Structure of hMAO-B in Complex with Inhibitors.
Structural studies were performed by cocrystallization of hMAO-B with the inhibitors selected for the kinetic analyses, using previously published procedures. 26 Structures for inhibitors 1, 84, and 97 were solved at resolution ranging Table 2 Table S1. Crystallographic studies on hMAO-B in complex with 1, 84, and 97 showed that these inhibitor molecules are certain to be bound to the enzyme active site ( Figure 6). Ligand binding does not cause any major conformational changes to the overall protein dimeric structure ( Figure S4). As for all of the inhibitors, the structure of the active site is identical in the two monomers present in the asymmetric unit that forms the protein dimer, and here we will refer to chain A of each structure in the following discussion. The electron density map clearly showed that all of these inhibitors form a covalent adduct with the flavin ( Figure 6). All of these inhibitors bind to the hMAO-B active site in a similar fashion, from the covalent adduct with the flavin to the 4-substituted phenyl ring at the other end of the molecule ( Figure 6). They adopt an extended conformation, as they occupy both the substrate and the cavity spaces (Ile199 in the open conformation). The 4-substituted phenyl ring is constrained by the flat hydrophobic shape of the hMAO-B cavity in a position perpendicular to the flavin plane, as for other wellknown inhibitors, such as L-deprenyl and safinamide ( Figure  7A). The site occupied by the halogen substituent on the aromatic ring of 1 is also conserved, which is a common feature of many hMAO-B inhibitors. 27 The more polar substituents that have lower inhibitory activities (e.g., 82, 105) would be much less favored in their binding to this highly hydrophobic site. The enzyme active site architecture does not undergo any conformational change with respect to that of other hMAO-B structures except for minor modifications to the side chains of Cys172 and Tyr188 ( Figure 7A). This is due to the position adopted by the piperidine ring that represents the hallmark of this series of inhibitors. Despite the flat shape of the hMAO-B cavity, the nonplanar conformation of the piperidine and its pyramidal nitrogen that is involved in the covalent adduct constrain the piperidine ring to bind perpendicular to the plane of the cavity ( Figure 6). Instead, with other inhibitors such as L-deprenyl and safinamide, there is an aromatic ring that binds coplanar with the cavity plane and with any substituted aromatic at the other end of the molecule, if present ( Figure  7A). Therefore, the position of the piperidine in the upper part of the cavity allows Cys172 and Tyr188 to move slightly upward. In the structure of the complex with 97, the conformation of the Cys172 and Tyr188 is conserved to some extent with respect to the previous structures. This is due to the linker that is devoid of the double bond, and the consequent tetrahedral conformation of its carbon atoms results in a more contracted conformation of the inhibitor molecule with respect to 84 ( Figure 7B) and 1. As a result, the piperidine collapses toward the bottom of the cavity, and the covalent linkage is forced to adopt a slightly different conformation with respect to the other propargyl inhibitors ( Figure 7).

. Kinetic Parameters for the Reversible and Irreversible Inhibition of the hMAOs by the Selected 4-Styrylpiperidine Inhibitors
Cell-Based Assays. Following the chemical-space exploration and kinetic and structural evaluations, several of these structurally diverse hMAO-A and hMAO-B inhibitors were also tested in vitro in cell models. First, their cytotoxicity profiles were defined using the SH-SY5Y human neuroblastoma cell line. By use of the MTS assay, no significant cytotoxic effects were seen for 24 h treatments of these cells with the compounds 1, 6, 67, 69, 84, 97, 100, and 102 at concentrations of 5 μM and 10 μM ( Figure S5). Indeed, at 50 μM, the cell viabilities remained >80% for almost all of the compounds, which is roughly 3 log units higher than the concentrations needed to achieve 50% in vitro inhibition of either of the two MAO isoforms.
The experimental in vitro cell model with the neurotoxin 6hydroxydopamine (6-OHDA) is commonly used to study neurodegenerative processes, as it induces toxicity that mimics the biochemistry and neuropathology of Parkinson's disease. 28 To determine the possibility that MAO inhibitors can abrogate 6-OHDA-induced toxicity, we examined lactate dehydrogenase (LDH) release and propidium iodide (PI) staining in the SH-SY5Y cells after 30 min pretreatment with the compounds at the noncytotoxic concentration of 5 μM, followed by 6-OHDA treatment for 24 h (Figure 8). The active treatment with 6-OHDA significantly increased LDH release (135% vs control), while pretreatment with compounds 1, 84, 97, and 102 protected against this increased LDH release (76%, 87%, 82%, 82% vs 6-OHDA, respectively), thus indicating improved cell viability. In the confirmation using flow cytometry, the proportion of PI-positive cells increased after the active 24 h treatment with 6-OHDA and was significantly reduced following the pretreatments with the hMAO-B inhibitors 84 and 97 (0.65 and 0.68, respectively), compared to the 6-OHDA-treated cells. Altogether, this indicates that the selective hMAO-B inhibitors 84 and 97 reduced the 6-OHDA-induced cytotoxicity and increased the viability of these SH-SY5Y neuronal-like cells.
In Vivo MAO Inhibition in Mouse Brain Homogenates and in Vitro Permeability. The inhibition of the MAOs in mouse brain homogenates was also investigated both in vitro and ex vivo (Table 3). Here, the more potent of the inhibitors against hMAO-A (i.e., 69) and hMAO-B (i.e., 100) were tested based on their IC 50 values and structural features. Brain homogenates from male Swiss mice were first characterized in terms of their MAOs activity and substrate specificity (Table S2, Figure S6).
Compound 69 completely abolished MAO-A activity in vitro at 50 μM when 5-HT was used as the selective MAO-A substrate. Compound 100 at 50 μM inhibited mice MAO-B when benzylamine was used as the selective MAO-B substrate ( Table 3). When the dual hMAO-A/B substrate p-tyramine was used, the inhibition was lower (i.e., higher residual activities; RAs). The same profile of inhibition was established ex vivo for brain homogenates from mice treated with these compounds, which were injected intraperitoneally (ip) 30 min before preparation of the brain homogenates. Again, more potent inhibition was seen in these brain homogenates when the isoenzyme-selective substrates were used. Selective MAO-A inhibitor 69 was the most active compound when administered at 60 mg/kg ip (RA, 1.6% ± 0.02%), while 60 mg/kg ip 100 resulted in similar MAO-B inhibition (RA, 24.3% ± 0.5%). Compounds 69 and 100 showed dosedependent inhibition of these corresponding MAO isoforms (ED 50 of 0.43 ± 0.26 mg/kg and 1.38 ± 0.87 mg/kg, respectively).
MAO inhibitors are in clinical use for treatment of psychiatric and neurological disorders, where they are used in chronic, low-dose regimes. 29 These MAO-A and MAO-B activities in ex vivo mouse brain homogenates were therefore studied after chronic 10-day administration to the mice of 69 and 100 at the low dose of 0.3 mg/kg ip. This chronic administration of selective MAO-A inhibitor 69 significantly reduced MAO-A activity compared to 0.9% saline ip in the control mice. After the single administration of 69 at 0.3 mg/ kg ip, the ex vivo MAO-A RA was reduced to 67.8% ± 31.2% (i.e., compared to 0.9% saline), whereas the 10-day treatment resulted in MAO-A RA of 9.5% ± 23.4%. Similar to 69, one single ip injection of 100 at 0.3 mg/kg showed MAO-B RA of 83.4% ± 19.2%, while chronic 10-day treatment resulted in RA of 28.3% ± 13.8%, which demonstrated accumulation of inhibition over time.
The ex vivo inhibitory activity of inhibitors 69 and 100 was also demonstrated in the mouse brain homogenates prepared 90 min after acute oral administration to the mice, via oral gavage at 30 mg/kg. Inhibitors 69 and 100 showed RAs of 45.3% ± 14.6% (vs 0.9% saline-treated mice) and 30.5% ± 10.4% (vs 0.9% saline treated mice), respectively. This prompted us to also study gut permeability of these representative inhibitors (i.e., 69, 84, and 100) in vitro on rat intestine. The apparent bidirectional permeability coefficient for the rat intestine in vitro was determined for 69. This was 108 ± 13 nm/s (as mean ± SEM) in the absorptive direction and 354 ± 99 nm/s in the eliminatory direction. For comparison, the apparent permeability coefficients of the internal low-and high-permeability standards furosemide and metoprolol in the absorptive direction were 64 ± 11 nm/s and 140 ± 18 nm/s, respectively. Data are mean values ± SD (n = 1, in triplicate). nd, not determined. b In vitro: 50 μM compounds (final concentrations) versus 0.9% saline with mouse brain homogenate from nontreated mice. c Ex vivo: brain homogenates from mice treated (n = 3, per group) with 60 mg/kg ip compounds or 0.9% saline 30 min before brain homogenate preparation. Journal of Medicinal Chemistry pubs.acs.org/jmc Article Behavioral Studies with the Selective MAO-A and MAO-B Inhibitors. MAO-A inhibitors are currently used in the therapy of treatment-resistant major depressive disorder and some specific types of depression. 30 Symptoms of depression correlate highly with those of anxiety. 31 These have been shown to be associated with 5-HT 1A receptors, 32 which are presynaptic and postsynaptic G-protein-coupled receptors that are involved in neuromodulation. 5-HT 1A receptor agonists, such as buspirone, show efficacy for the relief of anxiety and depression. 32 The binding affinities (K i values) for this 5-HT 1A receptor subtype were thus determined for these MAO inhibitors to potentially further explain their behavioral effects. These ranged from low micromolar to submicromolar K i values (Table S3). Following treatments with the selective irreversible MAO-A inhibitors 6, 67, and 69, the Swiss mice were subjected to the different behavioral tasks to determine whether these compounds have any antidepressant-like activity, any activity in an anxiety model and whether they affect locomotor activity. 33 The antidepressant-like activity was screened with the widely used tail suspension test. At 30 min before the behavioral assays, the mice were treated with 6, 67, and 69 at 3 mg/kg and 10 mg/kg ip, with 69 also at 1 mg/kg and 30 mg/kg ip, or were injected with 0.9% saline as control. Imipramine (30 mg/ kg ip) was used as the reference drug. Compounds 67 and 69 did not decrease the immobility time, whereas selective MAO-A inhibitor 6 and imipramine significantly reduced it, which demonstrated the antidepressant-like effects of these compounds under acute treatment ( Figure 9A).
The performance of mice in the "elevated plus maze test" was used here as a rodent model of anxiety, which was carried out 30 min after the ip treatments with 6, 67, and 69 (Table  S4). Compound 6 at the dose of 10 mg/kg ip induced a significant decrease compared to mice injected with 0.9% saline, for both the number of total arm entries (9.7 ± 2.9 vs 22.0 ± 1.5; p < 0.001) and the proportion of time spent in the center of the apparatus (13.4 ± 2.3 vs 43.5 ± 2.8; p < 0.001), and also an increase in the proportion of open-arm entries (2.0-fold; p < 0.001). Meanwhile, 69 at 3 mg/kg ip produced a significant increase in the proportion of open-arm entries (2.0fold; p < 0.001) and in the time spent in those entries (2.1fold; p < 0.001), similar to diazepam (reference drug) at 1 mg/ kg ip (2.1-fold, 2.4-fold, respectively; p < 0.001 for both).
The locomotor activity of the mice was also determined, as it can affect the data obtained in these behavioral assays. It was shown that 6 at 10 mg/kg ip significantly decreased spontaneous locomotor activity ( Figure 9B), which is correlated with the decrease in the total arm entries in the elevated plus maze test. This indicates that the effects of 6 after acute 10 mg/kg ip treatment are not associated with antidepressant-like or anxiolytic effects but rather with a decrease in the locomotion of the mice. On the other hand, the increased open arm parameters in the elevated plus maze for 69 (3 mg/kg ip) appear to be correlated with an anxiolytic-like effect, which would be mediated through the 5-HT 1A receptor, as locomotion was not altered at this dose of 69. Importantly, chronic 10-day treatment with 69 at a low dose (i.e., 0.3 mg/kg ip, daily) significantly decreased the immobility time in the treated group ( Figure 9C). No significant effects were seen for this chronic low dose of 69 in the elevated plus maze and locomotor activity tests (Table S5). This confirmed the antidepressant-like effects of 69 when administered chronically, which were not seen for the acute treatment regime.
Behavioral studies of the selective hMAO-B inhibitors were performed with compound 100 (irreversible MAO-B inhibitor). Although trans derivative 84 also showed potent irreversible MAO-B inhibition (IC 50 = 18.6 ± 3.3 nM; equipotent to 100), it was not studied here in vivo due to its poor permeability in the rat gut (see above) and its low binding potential for the 5-HT 1A receptor (Table S3).
The hole−board test is commonly used to study stressrelated changes in exploratory behavior of animals. 34 This was combined here with the locomotor activity test. At acute ip doses of ≥10 mg/kg of 100 in the mice, the hole−board test parameters and locomotor activities were reduced (Table S6, Figure S7A−C). Due to the locomotor impairment at high doses, compound 100 was further studied in the haloperidolinduced catalepsy test at lower doses (see Supplementary Results and Discussion). Compound 100 was also tested in the same battery of behavioral tests after 10 days of low-dose ip treatment (0.3 mg/kg, daily), where it did not affect the locomotor activity of the mice ( Figure S7D) or the parameters of the hole−board test ( Figure S8). During these 10-day treatments (i.e., 69, 100 at 0.3 mg/kg ip, daily), the body mass of the mice was also monitored ( Figure S9). No significant differences were seen for the proportions of body mass variations versus the control mice (0.9% saline treated). Also, all of these mice survived these 10-day treatments without any apparent signs of changes in behavior, which supports the general safety of these compounds.

■ DISCUSSION AND CONCLUSIONS
Biological receptors differ greatly in shape, polarity, and conformational dynamics. 1 However, it is a challenging task to obtain selectivity between structurally related targets such as isoenzymes, which requires different ligand-design approaches. MAO-A and MAO-B are two closely related isoenzymes where numerous selective inhibitors have been described in the literature, with some now used in the clinical setting for treatment of neurological disorders. 35 Selective hMAO-B inhibitor 6 was identified in a screening campaign, and chemistry-driven exploration of the chemical space revealed that the trans isomers selectively inhibit hMAO-B over hMAO-A. On the other hand, cis isomers were identified as selective hMAO-A inhibitors. The inhibitory potencies in the SARs exploration phase are generally expressed as IC 50 values (Table 1), although these derivatives are irreversible. The IC 50 values are thus dependent on the assay conditions and particularly on the preincubation period. 36 Despite these limitations, the measured IC 50 values still allow relatively simple, yet not the most precise, comparisons between such compounds in a series. 11 The inhibitory activity against hMAO-B was increased by introducing bulkier and more lipophilic substituents onto position 4 of the phenyl ring, as previously reported in the literature. 37 A similar trend was also observed for the turnover of variously substituted substrates of the benzylamine and phenethylamine classes, 38 where the binding affinity improved with increasing hydrophobicity of the substituent.
Further examination of the binding kinetics for the selected compounds here using purified enzyme samples confirmed the variable nature of these IC 50 values. The overall higher affinity inferred from the IC 50 values (Table 1) compared to K i ( Table  2) is related to the covalent nature of these inhibitors, which fully inactivate the enzyme during the preincubation step in the absence of substrate (not included in competitive steady-state Journal of Medicinal Chemistry pubs.acs.org/jmc Article experiments used to determine IC 50 ). For the same reason, the variability of the IC 50 values among the different inhibitors that was not seen for K i has to be interpreted in light of the timedependent inactivation process that showed different kinetics depending on the inhibitor (Table 2). For example, 84 showed an IC 50 that is 0.05-fold that of 1 and a similar reversible K i , while its inactivation rate was 4-fold that of 1, which results in a k inact /K I ratio that is comparable between these two inhibitors. Similar observations can be reasoned for the other covalent analogues and for L-deprenyl. Crystallographic studies confirmed the covalent adduct formation with flavin for the inhibitors investigated here. A similar binding mode for all of these inhibitors was seen, with only minor modifications of the active site arrangements in comparison to other hMAO-B structures reported in the literature. These changes were most prominent with the trans inhibitors 1 and 84, which imposed a slight upward shift of Cys172 and Tyr188. This is related to the conformation adopted by the piperidine ring, which represents the hallmark of this class of inhibitors, as a novel element with respect to previous propargyl inhibitors (i.e., where the nitrogen atom is within a linear molecular structure rather than part of a ring).
The inhibitor series studied here were not cytotoxic to SH-SY5Y cells up to treatments of 50 μM. Remarkably, only the selective hMAO-B inhibitors decreased the LDH release induced by the neurotoxin 6-OHDA. The reduced proportion of PI-positive SH-SY5Y cells was also confirmed for the two MAO-B-selective ligands during 6-OHDA-induced toxicity. This indicated their neuroprotective nature, which might also be attributed, among others possibilities, to their selective MAO-B inhibition. Following ip and peroral administration, these compounds inhibited MAO-A and MAO-B activities ex vivo in the mouse brain homogenates. The extent of their inhibition varied depending on the inhibitor selectivity and the substrate used. Selective MAO-A inhibitor 69 completely inhibited MAO-A ex vivo when the MAO-A substrate 5-HT was used (RA, 1.6%), while its inhibition was lower when using p-tyramine (RA, 14.2%), which is a dual MAO-A/B substrate. The difference is even more evident for the MAO-B inhibitors, with decreased inhibition when switching from the MAO-B substrate benzylamine to dual substrate p-tyramine. Both MAO-A and MAO-B can metabolize p-tyramine; however, only one of them was inhibited by the selective inhibitor, which explained the impaired inhibition. This additionally confirmed the in vitro selectivity of these inhibitors not only on the isolated human enzymes but also in complex matrices, such as mouse brain homogenates. The MAO inhibition in these mouse brain homogenates after peroral single-dose administration also indicated the oral bioavailability of these analogues. The in vitro permeability of 69 studied for rat intestine was between that of the low-permeability and highpermeability standards, and as such, 69 should be classified with an intermediate/moderate permeability, with the possibility of interference by the intestinal efflux mechanisms, as seen by the higher apparent permeability coefficient determined for the eliminatory direction. The permeabilities of 84 and 100 were not determinable in vitro due to their high tissue binding. These findings do not correlate directly with the ex vivo effects; however, the ex vivo effects show that both 69 and 100 reached the brain tissue and inhibited the MAO activity after their oral administration.
Under the acute treatment regime in these mice, the selective MAO-A inhibitor 69 did not affect the immobility time of the mice in the tail suspension test. The observed anxiolytic-like activity with acute treatment with 69, which can be attributed to 5-HT 1A binding, prompted us to also study 69 in a 10-day treatment regime. The antidepressant-like effects of 69 in this setup, together with no significant toxic effects, confirmed the therapeutic potential for this class of selective hMAO-A inhibitors. On the other hand, the selective MAO-B inhibitor 100 did not produce the expected results for haloperidol-induced catalepsy, in terms of what would be expected for a MAO-B inhibitor. 39 Nonetheless, the general safety of these compounds has been established in the chronic low-dose treatment regime.
In summary, we have generated the first example of selective hMAO-A and hMAO-B inhibitors where this selectivity is conferred by the configurational cis/trans isomerism of a double bond. Cis isomers with fluorine or unsubstituted benzene ring selectively inhibit hMAO-A. On the other hand, trans derivatives and their corresponding reduced ethyl analogues with small lipophilic substituents on benzene ring selectively inhibit hMAO-B. In addition, we provide the first in depth, and we believe critical, kinetic and structural characterization of these 1-propargyl-4-styrylpiperidines as potential fragments in the design of pharmaceuticals that can be used to target the MAOs.

■ EXPERIMENTAL SECTION
Chemistry. General Information. The reagents and solvents used were obtained from commercial sources (i.e., Acros Organics, Sigma-Aldrich, TCI Europe, Merck, Carlo Erba, Apollo Scientific) and were used as provided. Anhydrous THF was prepared by distillation over sodium and benzophenone under Ar(g). Analytical thin-layer chromatography was performed on silica gel aluminum sheets (60 F254, 0.20 mm; Merck). Flash column chromatography was performed on silica gel 60 (particle size 0.040−0.063 mm, Merck). 1  Purities of the tested compounds were established to be ≥95%, as determined by HPLC. All compounds described in Table 1 were checked by SwissADME 40 for the presence of PAINS substructures, 41 and no PAINS substructures were identified.
General Synthetic Chemistry Experimental Protocols. General Procedure A: Synthesis of Wittig Reagents. Appropriate benzyl halide (25.0 mmol, 1.0 equiv) was dissolved in MeCN (25−50 mL), and triphenylphosphine (6.56 g, 25.0 mmol, 1.0 equiv) was added at room temperature. The reaction mixture was stirred at 85°C for 16−24 h. The solvent was evaporated, and CH 2 Cl 2 (10 mL) and Et 2 O (50 mL) were added to the residue. The precipitated solid was filtered off, washed with Et 2 O (20 mL), and dried overnight at room temperature. The crude product was used in the following reaction step without further purification.
Journal of Medicinal Chemistry pubs.acs.org/jmc Article General Procedure B: Wittig Reaction. Appropriate Witting reagent synthesized according to general procedure A (1.1 equiv) was dissolved in anhydrous THF (30 mL) under argon atmosphere. NaHMDS (2 N in THF, 1.2 equiv) or KHMDS (0.5 M in toluene, 1.1 equiv) was added to the resulting suspension at room temperature. The orange-red colored reaction mixture was then stirred for 30 min under argon, followed by a dropwise addition of N-Boc protected 3-or 4-formylpiperidine (1.0 equiv) solution in anhydrous THF (10 mL). The reaction mixture was stirred at room temperature overnight (16−24 h) and then quenched by adding saturated aqueous NaHCO 3 solution (10−15 mL). The solvent was evaporated, and the white residue was resuspended in EtOAc (50 mL) and saturated aqueous NaHCO 3 solution (50 mL) and transferred into a separating funnel. The phases were separated, and the aqueous phase was additionally extracted with EtOAc (2 × 50 mL). Combined organic phases were washed with saturated brine (100 mL), dried over Na 2 SO 4 , and evaporated. The crude product was purified by flash column chromatography to yield (in a majority of cases) both pure cis and trans isomers.
General Procedure C: Reduction of the Double Bond. Alkene derivative (Wittig reaction product-mixture of cis and trans isomer, 1.0 equiv) was dissolved in EtOH (50 mL) and purged under a stream of argon for 10 min. Catalytic amount of Pd/C (10% load on carbon, 10−20% [w/w] calculated to the starting material) was added, and the resulting suspension mixture was stirred under H 2 (g) atmosphere at room temperature for 16−24 h. The catalyst was removed by filtration through Celite and evaporated to obtain crude product. The product was used without further purification unless stated otherwise.
General Procedure D: Boc Protection Removal (HCl Method). To the solution of Boc-protected derivative (1.0 equiv) in 1,4-dioxane or EtOH (30 mL), concentrated HCl (10.0 equiv) was added at room temperature. The reaction mixture was stirred at 80°C for 2 h. The solvent was evaporated under reduced pressure, and the crude product was used in the next step without further purification.
General Procedure E: Alkylation of Piperidine. Piperidine intermediate (in the form of the salt with HCl) was dissolved in MeCN (20−30 mL) or DMF (5−10 mL) at 0°C under an argon atmosphere. K 2 CO 3 or Cs 2 CO 3 (3.0 equiv) was added, followed by dropwise addition of propargyl bromide (80% solution in toluene, 1.2 equiv). The resulting suspension was stirred at room temperature (unless specified otherwise) for 16−72 h. The solvent was evaporated, the residue dissolved in a mixture of CH 2 Cl 2 (50 mL) and saturated aqueous NaHCO 3 solution (50 mL) and transferred into a separating funnel. The organic phase was further washed with saturated brine solution (50 mL), dried over Na 2 SO 4 , and evaporated. The crude product was purified by flash column chromatography.
Biology. Protein Production, Enzymatic Assays, and X-ray Crystallography. All of the reagents were purchased from Sigma-Aldrich unless specified otherwise. The effects of the compounds on hMAO-A and hMAO-B were first investigated by determining the IC 50 values using a previously described fluorimetric assay. 42 The inhibitory activities of the compounds were determined according to their effects on the generation of hydrogen peroxide (H 2 O 2 ) as a sideproduct of hMAO activity. For details on protein production, enzymatic assays, and X-ray crystallography, see Supporting Information.
Cell Culture and Treatments. The human neuroblastoma SH-SY5Y cell line was purchased from American Type Culture Collection (CRL-2266; VA, USA). These cells were cultured in Advanced Dulbecco's modified Eagle's medium (Gibco, Thermo Fisher Scientific, MA, USA) supplemented with 10% fetal bovine serum (Gibco), 2 mM L-glutamine, 50 U/mL penicillin, and 50 μg/mL streptomycin (Sigma, MO, USA) in a humidified atmosphere of 95% air and 5% CO 2 at 37°C and grown to 80% confluence. Prior to cell treatments, the complete medium was replaced with serum-depleted medium. Compounds were prepared as stock solutions of 20 mM in DMSO and were used at concentrations of 5−50 μM. For the cytotoxic stimuli, the 10 mM 6-OHDA (Sigma, MO, USA) stock was prepared in phosphate buffered saline, pH 7.4, with 0.01% ascorbic acid. For the detailed experimental procedures see Supporting Information.
Animals. Adult male Swiss mice (weight, 25−30 g) and adult male Wistar rats (weight, 200−300 g) were obtained from the Central Animal House of the School of Pharmacy and Biochemistry, University of Buenos Aires, Buenos Aires, Argentina. For behavioral assays, the mice were housed in groups of five in a controlled environment (20−23°C), with free access to food and water, and maintained on a 12 h/12 h day/night cycle, with light on at 07:00 h. The housing, handling, and experimental procedures for both the mice and rats complied with the recommendations set out by the National Institutes of Health Guide for Care and Use of Laboratory Animals (NIH Publication No. 8023, revised 1996) and the Institutional Committees for the Care and Use of Laboratory Animals of the Faculty of Pharmacy and Biochemistry, University of Buenos Aires, Argentina (Code 31682/2014). All efforts were taken to minimize animal suffering. The number of animals used was the minimum number consistent with obtaining significant data. The mice were randomly assigned to the treatment groups and were used only once. The pharmacological tests were evaluated by experimenters who were not aware of the treatments administered and were performed between 10:00 h and 14:00 h.
Administrations and Procedures. The test compounds (used as their hydrochloride salts) were dissolved in 0.9% saline. Haloperidol, diazepam, and imipramine were diluted by sequential additions of DMSO, an aqueous solution of 0.25% Tween 80, and 0.9% saline to their final concentrations of 5%, 20%, and 75%, respectively. In each session, a control group of mice received only vehicle, in parallel with the mice that received the drug treatments. The volumes of the ip injections and oral gavage (po) administrations were 0.10−0.20 mL/ 30 g body weight and 0.30 mL/30 g body weight, respectively. The protein concentrations in the mouse brain homogenates were determined using the method of Bradford, with bovine serum albumin as the standard.
For the acute behavioral assays, all of the compounds were tested 30 min after ip administrations. For the chronical treatments, the mice were injected ip with 0.3 mg/kg 69, 100 or 0.9% saline (control group) once a day between 09:00 and 10:00 h over 10 consecutive days. On days 1, 3, 5, 7, and 9, these mice were weighted. On day 8, at 30 min after the relevant ip treatment, the mice were submitted to the plus maze (for 69) or the hole−board tests (for 100), and their locomotor activity was recorded immediately afterward. On day 10, at 30 min after the relevant ip treatment, the mice were submitted to the tail suspension test (for 69) or the haloperidol-induced catalepsy assay (for 100) (Figures S10 and S11). For all of these behavioral assays, the arena/apparatus was cleaned with 60% ethanol between each trial with each mouse. Behavioral assays are described in detail in Supporting Information.
Mouse Brain Homogenate Preparation. The mice were killed humanely by decapitation at 30 min after drug or 0.9% saline treatment (for acute ip injections), at 60 min after drug or 0.9% saline treatment (for acute oral gavage), or on day 10 at 90 min after the last drug or 0.9% saline ip treatment after the tail suspension test (for the mice chronically treated with 69) or after haloperidol-induced catalepsy test (for the mice chronically treated with 100). Also, three mice that were chronically treated with 100 or 0.9% saline ip but without the haloperidol injection were also tested to determine any haloperidol effects in the ex vivo MAO-B activity assay (Figures S10 and S11). After removal of the olfactory bulb, the whole mouse brains were washed several times in ice-cold 50 mM potassium phosphate buffer (pH 7.4, 0.05% [v/v] Triton X-114). Then, each brain was independently homogenized using a blender homogenizer (PRO Scientific Inc., CT, USA) at 30 000 rpm for 1 min, in 1.5 mL of 100 mM potassium phosphate buffer (pH 7.4, 0.1% [v/v] Triton X-114) at 4°C (in ice). The homogenate was then centrifuged at 3000g for 20 min at 4°C. The supernatant was stored at −20°C until the following day for the MAO inhibition assays.
Ex Vivo MAO Inhibition Assay. The ex vivo effects of 69 on MAO-A and 100 on MAO-B enzymatic activities (ip or oral administration, 30 min before decapitation) in the mouse brain homogenates were investigated using a previously described fluorimetric assay, with minor modifications. 42 For details see Supporting Information.