First Synthesis of Racemic Trans Propargylamino-Donepezil, a Pleiotrope Agent Able to Both Inhibit AChE and MAO-B, with Potential Interest against Alzheimer’s Disease

Alzheimer’s disease (AD) is a multifactorial neurodegenerative disease towards which pleiotropic approach using Multi-Target Directed Ligands is nowadays recognized as probably convenient. Among the numerous targets which are today validated against AD, acetylcholinesterase (ACh) and Monoamine Oxidase-B (MAO-B) appear as particularly convincing, especially if displayed by a sole agent such as ladostigil, currently in clinical trial in AD. Considering these results, we wanted to take benefit of the structural analogy lying in donepezil (DPZ) and rasagiline, two indane derivatives marketed as AChE and MAO-B inhibitors, respectively, and to propose the synthesis and the preliminary in vitro biological characterization of a structural compromise between these two compounds, we called propargylaminodonepezil (PADPZ). The synthesis of racemic trans PADPZ was achieved and its biological evaluation established its inhibitory activities towards both (h)AChE (IC50 = 0.4 µM) and (h)MAO-B (IC50 = 6.4 µM).


Introduction
Among the therapeutic agents getting to clinical trials, more and more examples illustrate the concept of Multi-Target Directed Ligands (MTDLs) [1]. Such compounds display several activities by interacting with different biological targets that are implied in a given disease, in order to obtain a synergy of action. MTDLs appear to be particularly interesting against multifactorial diseases such as Alzheimer's Disease (AD). As a matter of fact, the classical pharmacological approach consisting in interacting very selectively with a single target has shown clinical limitations, failing to restore such complex biological systems. Thus, the association of drugs has been favored for many years now. Comparatively to this approach, the advantages of MTDLs, in addition to the synergistic action they theoretically display, are the absence of drug-drug interactions, and a better patient compliance by reducing the number of intakes [2].
Basically, there are two main types of MTDLs: conjugates, containing two different active moieties linked by a chemical bond, and molecules that are able to interact indiscriminately with several therapeutic targets, combining the structural elements needed for both activities. As an example, our group recently described donecopride, first MTDL inhibiting acetylcholinesterase (AChE) and activating serotoninergic receptors 5-HT 4 , and now in preclinical evaluation [3][4][5].
Ladostigil ( Figure 1) is a novel type of MTDL currently in phase II clinical trials. It is the first compound able to release an active compound, the hydroxyrasagiline, a activities directed towards both AChE and MAO-B in AD [8], we aimed at conceiving a new MTDL targeting these enzymes but contrarily to ladostigil in an indifferently manner, i.e., able to bind to each of the active site of these targets. To achieve such a goal, the latter have to share sufficient common features liable to be recognized by a same agent. In support of such allegation, we consider the fact that both donepezil (DPZ), a marketed AChE inhibitor (AChEI), and rasagiline, a marketed, MAO-B inhibitor (MAO-BI), possess an indane core, and we designed propargylaminodonepezil (PADPZ) as a perfect structural compromise between these two drugs. This paper describes our efforts to achieve the first synthesis of PADPZ and its preliminary biological evaluation towards these two enzymes.

Chemistry
The synthetic strategy we followed for the access to PADPZ (Scheme 1), involved an aldolisation reaction between a properly N-protected dimethoxyindanone and para-N Besides the originality of the mechanism of action of ladostigil, the first results of its evaluation seem to validate the interest lying in targeting both AChE and MAO-B.
AChE inhibition is the mechanism of action of the main anti-AD drugs, leading to a symptomatic effect by preventing acetylcholine (ACh) degradation, responsible for AD symptoms. The efficiency of these drugs, however, decreases with the evolution of the pathology. Associating the latter with a neuroprotective effect appears theoretically able to express a disease-modifying effect while maintaining a long-acting symptomatic benefit.
MAO-B is already an identified target for Parkinson's disease. The enzyme recycles dopamine in the brain, leading to the formation of hydrogen peroxide, particularly neurotoxic. Its inhibition then provides a neuroprotective effect, which is currently evaluated [7]. More recently, MAO-B was also identified as implied in the amyloid cascade, comforting the therapeutic interest of this target for AD [8][9][10].
Taking into consideration the interest lying in the association into a sole structure of activities directed towards both AChE and MAO-B in AD [8], we aimed at conceiving a new MTDL targeting these enzymes but contrarily to ladostigil in an indifferently manner, i.e., able to bind to each of the active site of these targets. To achieve such a goal, the latter have to share sufficient common features liable to be recognized by a same agent. In support of such allegation, we consider the fact that both donepezil (DPZ), a marketed AChE inhibitor (AChEI), and rasagiline, a marketed, MAO-B inhibitor (MAO-BI), possess an indane core, and we designed propargylaminodonepezil (PADPZ) as a perfect structural compromise between these two drugs. This paper describes our efforts to achieve the first synthesis of PADPZ and its preliminary biological evaluation towards these two enzymes.

Chemistry
The synthetic strategy we followed for the access to PADPZ (Scheme 1), involved an aldolisation reaction between a properly N-protected dimethoxyindanone and para-N benzylpiperidinecarboxal-dehyde. The methylene derivatives would have then to successively undergo, in various order, a reduction of its alkene moiety, a deprotection of its amino group and a substitution of the latter by a propargyl group.

Molecules 2020
3 of 17 benzylpiperidinecarboxal-dehyde. The methylene derivatives would have then to successively undergo, in various order, a reduction of its alkene moiety, a deprotection of its amino group and a substitution of the latter by a propargyl group. Scheme 1. Synthetic strategy for the access to PADPZ.
In a first attempt, aldolisation of 3 was undertaken in acidic medium but failed to give the methylene derivative (12) (Scheme 4). The latter was obtained in alkaline medium, quasi exclusively under its E form. The latter was attributed through a selective 1D NOE experiment during which a response in the signal of the proton, located in the para position of the nitrogen atom of the piperidine ring, was observed upon excitation of H3, accounting for a close position between the two protons. Further, a significant deshielding of the signal of the methylene proton was also observed for the E form (6.70 ppm) of 12, due to the cone-shaped shielding zone of the carbonyl group, versus those of the Z form (6.25 ppm) observed as a trace. At the same time, para N-benzylpiperidine carboxaldehyde (11) was obtained in 45% overall yield, through the N-benzylation of ethyl 4-piperidinecarboxylate (8), followed by a reduction of the ester (9) into the alcohol (10) and a final Swern oxidation of the latter (Scheme 3) [11][12][13].   At the same time, para N-benzylpiperidine carboxaldehyde (11) was obtained in 45% overall yield, through the N-benzylation of ethyl 4-piperidinecarboxylate (8), followed by a reduction of the ester (9) into the alcohol (10) and a final Swern oxidation of the latter (Scheme 3) [11][12][13].  All attempts, however, to hydrolyze the trifluoroacetylamino group of 12 in alkaline medium failed. Such a chemical behavior was already observed with aminoindanones which, in these conditions, are suitable for an internalization of their double bond likely to lead to a degradation [14]. This can also explain the low yield observed in the preparation of 12 (19%).
All attempts aiming at synthesizing the propargyl derivative (13) starting from 5 and according to a similar manner, failed, probably for the same reasons. In a similar manner, the N-acetyl and N-Boc methylene derivatives (14) and (15) were obtained starting from 6 and 7, exclusively under their E form and in 51% and 46% yield respectively (Scheme 5). This geometry was attributed by analogy with the structure of 12. Reduction of the latter succeeded this time, using NaBH4 for 14 and H2 Pd/C for 15, respectively. Only one diastereoisomeric form of the N-acetylcompound 16 was recovered, explaining the moderate yield observed (35%). Its trans geometry was established by its X-ray diffractometry study ( Figure 2). The two diastereoisomeric forms of the N-boc derivative 17 were obtained, with a relative proportion of 3 trans for 1 cis and a global yield of 65%. Attribution of the two forms of 17 was achieved through 1D NOE experiment which showed, upon excitation of the NH proton, a response observed on the signal of H2 for the trans form and on the signal of the CH2 protons for the cis form, respectively. It was not possible to separate them. The coupling constant values between H2 and H3 signals appeared higher in the cis form (6.5 Hz) versus the trans one (3.5 Hz) and confirmed the major formation of trans-17 by comparison with the 1 H-NMR spectrum of trans-16. All attempts, however, to hydrolyze the trifluoroacetylamino group of 12 in alkaline medium failed. Such a chemical behavior was already observed with aminoindanones which, in these conditions, are suitable for an internalization of their double bond likely to lead to a degradation [14]. This can also explain the low yield observed in the preparation of 12 (19%).
All attempts aiming at synthesizing the propargyl derivative (13) starting from 5 and according to a similar manner, failed, probably for the same reasons.
In a similar manner, the N-acetyl and N-Boc methylene derivatives (14) and (15) were obtained starting from 6 and 7, exclusively under their E form and in 51% and 46% yield respectively (Scheme 5). This geometry was attributed by analogy with the structure of 12. Reduction of the latter succeeded this time, using NaBH 4 for 14 and H 2 Pd/C for 15, respectively. Only one diastereoisomeric form of the N-acetylcompound 16 was recovered, explaining the moderate yield observed (35%). Its trans geometry was established by its X-ray diffractometry study ( Figure 2). The two diastereoisomeric forms of the N-boc derivative 17 were obtained, with a relative proportion of 3 trans for 1 cis and a global yield of 65%. Attribution of the two forms of 17 was achieved through 1D NOE experiment which showed, upon excitation of the NH proton, a response observed on the signal of H2 for the trans form and on the signal of the CH 2 protons for the cis form, respectively. It was not possible to separate them. The coupling constant values between H2 and H3 signals appeared higher in the cis form (6.5 Hz) versus the trans one (3.5 Hz) and confirmed the major formation of trans-17 by comparison with the 1 H-NMR spectrum of trans-16.
All the attempts aiming at deprotecting the N-acetyl group of 16 failed. Compound 18, however, was obtained in 78% yield starting from the N-Boc derivative (17), which was easily N-deprotected using TFA. Compound 18 was recovered as a mixture of its trans/cis forms in a 90/10 ratio, from which the trans form was isolated.
Finally, the expected PADPZ (19) was successfully obtained under its trans form, through the N-substitution of trans-18 by a propargyl group. All the attempts aiming at deprotecting the N-acetyl group of 16 failed. Compound 18, however, was obtained in 78% yield starting from the N-Boc derivative (17), which was easily N-deprotected using TFA. Compound 18 was recovered as a mixture of its trans/cis forms in a 90/10 ratio, from which the trans form was isolated.
Finally, the expected PADPZ (19) was successfully obtained under its trans form, through the N-substitution of trans-18 by a propargyl group.

Molecular Modeling Study
The aim of the present molecular modeling work was to check whether PADPZ (19)

Molecular Modeling Study
The aim of the present molecular modeling work was to check whether PADPZ (19) can bind to the active sites of the two targeted proteins and which of the two enantiomers of its trans form will be the more prone to display the more effective interaction with both AChE and MAO-B. The initial 3D model of compound 19 was built using the structurally closed X-ray structure of compound 16, its acetylamino analog. Two enantiomers were present in the crystal of 19, and therefore, the two enantiomers of trans-19 (R,R and S,S) were built. The models, further, were protonated on their piperidine nitrogen.
Firstly, the two enantiomers of 19 were docked separately into the (h)AChE active site and in a comparative manner with donepezil ( Figure 3A). In a similar manner as in our previous studies on (h)AChE [4], a water molecule interacting with protonated piperidine ring of donepezil was conserved (residue number 931) during docking. The results suggested different binding affinities for the two enantiomers. The enantiomer R,R ( Figure 3B) took an orientation within the active site of AChE closed to donepezil one, accounting for a possible inhibitory activity towards the enzyme similar to those of the reference compound. This enantiomer, indeed, reproduced crucial donepezil interactions: (i) the charged nitrogen of the piperidine ring was oriented in a position suitable for an interaction with the water molecule in the proximity of Tyr337 and Tyr341, (ii) the carbonyl group formed a hydrogen bond with NH of the Phe295 backbone and (iii) both benzene rings are placed in parallel to the Trp279 and Trp84 indole rings to favor a π-stacking interaction. An additional interaction is further observed between the piperidine NH of 19 and the OH of Tyr124. To check the ability of the enantiomers of trans-19 to bind to the (h)MAO-B binding site, a covalent docking was carried into the enzyme. The docked enantiomers were previously modified, engaging their propargylamino group into a covalent bond with the FAD co-factor of MAO-B and modifying the bond order in a same manner as observed in rasagiline-FAD complex ( Figure 4A). The covalent docking was then carried out and both enantiomers were placed into the MAO-B cavity without any steric clash ( Figure 4B,C). The interactions established by the two enantiomers with MAO-B are very similar: a hydrogen bond between their carbonyl group and the Gln206 side chain and another one between their methoxy groups and a tyrosine residue, Tyr326 for R,R and Tyr188 for S,S. From these observations, we hypothesized that the two enantiomers should bind to MAO-B with similar affinities and inhibitory activities. On the other hand, the position of the enantiomer S,S of 19 within the AChE groove did not appear as favorable for an inhibitory interaction ( Figure 3C). It lost some crucial interactions through the protonated piperidine nitrogen of 19, the π-stacking with Trp84 and the hydrogen bond between the carbonyl group of 19 and the NH of Phe295. Contrariwise this carbonyl group established a hydrogen bond with Tyr124.
To check the ability of the enantiomers of trans-19 to bind to the (h)MAO-B binding site, a covalent docking was carried into the enzyme. The docked enantiomers were previously modified, engaging their propargylamino group into a covalent bond with the FAD co-factor of MAO-B and modifying the bond order in a same manner as observed in rasagiline-FAD complex ( Figure 4A). The covalent docking was then carried out and both enantiomers were placed into the MAO-B cavity without any steric clash ( Figure 4B,C). The interactions established by the two enantiomers with MAO-B are very similar: a hydrogen bond between their carbonyl group and the Gln206 side chain and another one between their methoxy groups and a tyrosine residue, Tyr326 for R,R and Tyr188 for S,S. From these observations, we hypothesized that the two enantiomers should bind to MAO-B with similar affinities and inhibitory activities.
To check the ability of the enantiomers of trans-19 to bind to the (h)MAO-B binding site, a covalent docking was carried into the enzyme. The docked enantiomers were previously modified, engaging their propargylamino group into a covalent bond with the FAD co-factor of MAO-B and modifying the bond order in a same manner as observed in rasagiline-FAD complex ( Figure 4A). The covalent docking was then carried out and both enantiomers were placed into the MAO-B cavity without any steric clash ( Figure 4B,C). The interactions established by the two enantiomers with MAO-B are very similar: a hydrogen bond between their carbonyl group and the Gln206 side chain and another one between their methoxy groups and a tyrosine residue, Tyr326 for R,R and Tyr188 for S,S. From these observations, we hypothesized that the two enantiomers should bind to MAO-B with similar affinities and inhibitory activities.

In Silico ADME Parameters
In silico ADME parameters for PADPZ have been calculated using SwissADME (www.SwissADME.ch). No alert was identified (Table 1).

In Silico ADME Parameters
In silico ADME parameters for PADPZ have been calculated using SwissADME (www.SwissADME.ch). No alert was identified (Table 1).

Cholinesterases and Monoamine Oxidases Inhibition
The inhibitory activity of trans-PADPZ (19) as well as its acetylamino analogue (16) towards (h)AChE and (eq)BuChE was evaluated according to the Ellman test [15]. DPZ was used as a reference. Results are depicted in Table 2. The inhibitory activity of trans-PADPZ (19) towards human recombinant MAO-A and -B was determined as already described [17], measuring the fluorescence of 4hydroxyquinoline produced by MAOs in the oxidative deamination of substrate kynuramine (50 µM). Rasagiline and pargiline were used as references, as well as the dimethoxyketo analogue of rasagiline (5) we synthesized. Results are depicted in Table 3.

Discussion
The modulation of DPZ and indanone moiety to develop novel AChEI has been largely explored in recent years [19]. Indeed, the binding site of the enzyme could accommodate to an important number of chemical structures, and several MTDL have been obtained starting from an indanone moiety. Indeed, several modulations have generated MTDL with Aβ-aggregation [20] or BACE1 [21] inhibition. During these SAR, the presence of α-β insaturated ketone has generally led to a decrease in potency compared to DPZ. If most of the modulation has been engaged by changing the nature of the piperidine or replacing the dimethoxy substituent, few examples have been described in introducing substituent on the indanone moiety. With this aim, we have prepared and focus our attention on the preparation of the reduced form of PADPZ. Trans-PADPZ (19) showed a noticeable inhibitory activity towards (h)AChE with an IC 50 = 440 nM and a relative selectivity towards (h)BuChE which was inhibited at a concentration 10-fold greater (IC 50 = 4.2 µM). This AChE inhibitory activity, however, appeared weaker than those of DPZ (IC 50 = 14 nM), but higher than those of the acetamide trans-16 (% inhibition at 10 −6 M = 4%) which almost totally lost its effect.
Tested against (h)MAO-B, trans-PADPZ exhibited again a sound activity with an IC 50 = 6.4 µM, in a same order of magnitude than those of pargiline (IC 50 = 2.7 µM), but weaker with respect to rasagiline (IC 50 = 14 nM) used as references. Interestingly, trans-PADPZ (19) appeared selective towards MAO-B with a weak inhibition of MAO-A (% inhibition at 10 −6 M = 13%). The dimethoxyketo analogue of rasagiline (5) displayed also a modest and selective inhibitory activity towards MAO-B (IC 50 = 13.4 µM). According to the docking study, this inhibition is not influenced by the stereochemistry of the substitution at the difference of the corresponding rasagiline or ladostigil.
Trans-PADPZ (19) appears therefore as a dual AChE/MAO-B with a good selectivity towards BuChE and MAO-A. These results are consistent with the predictive data issued from the in silico study, which further gave us an incentive to separate the enantiomers of Trans-PADPZ, since one of them (R,R) appeared, according to this study, more able to inhibit AChE than the other one (S,S).

X-ray Diffractometry
Single X-ray crystal analysis on compound 16 was carried out on BRUKER D8 Quest diffractometer (Bruker, Billerica, MA, USA) with a PHOTON II detector (Bruker, Billerica, MA, USA) using 1 µS micro focus X-ray source (Mo Kα, λ = 0.71073 Å). The crystal structure was solved by direct methods and refined employing full-matrix least-squares refinement against F2 using SHELX2014 package [22]. All non-hydrogen atoms were refined anisotropically and hydrogen atom positions were determined via difference Fourier maps and refined with isotropic atomic displacement parameters. The structural data of compound 16 have been deposited with Cambridge Crystallographic Data Center, the CCDC (Deposition Number 2052041).

In Silico Study
The 3D models of compound 19 were built using both enantiomers present in the solved X-ray structure of similar compound 16. Their protonation states at pH 7.4 were predicted using standard tools of the ChemAxon Package (http://www.chemaxon.com/) and the majority microspecies protonated on piperidine nitrogen at this pH were used for docking studies.
For docking into human AChE, its crystallographic coordinates were retrieved from X-ray structure of the donepezil/AChE complex (PDB ID 4EY7, a structure refined to 2.35 Å with an R factor of 17.7%) [23].
The docking of the compound 19 into AChE was carried out with the GOLD program (v5.7.2) using the default parameters [24,25]. This program applies a genetic algorithm to explore conformational spaces and ligand binding modes. To evaluate the proposed ligand positions, the ChemPLP fitness function was used. The binding site in the (h)AChE model was defined as a 7 Å sphere from the co-crystallized donepezil ligand and a water molecule interacting with protonated piperidine ring of donepezil was conserved during the docking (residue number 931).
The starting structure for our docking into the (h)MAO-B was the high-resolution (2.2 Å) X-ray structure of rasagiline-inhibited human monoamine oxidase B in complex with 2-(2-benzofuranyl)-2-imidazoline (PDB ID 2XFQ) [26]. The rasagiline and 2-(2benzofuranyl)-2-imidazoline were removed before docking. As the compound 19 should bind covalently to the FAD co-factor through the propargyl moiety in the same way as rasagaline, the structure of 19 was modified by adding the nitrogen atom on the propargyl group and modifying its double bonds repartition according to the one observed in the rasagaline-(h)MAO-B complex. The covalent docking constraint between the nitrogen atom of FAD group (N5) and added nitrogen on propargyl moiety was applied during the docking. The binding site in the MaO-B model was defined as a 10 Å sphere from OH atom of Tyr188, and ChemPLP fitness function was used to evaluate the proposed ligand positions.

In Vitro Tests of AChE and BuChE Biological Activity
Inhibitory capacity of compounds on AChE biological activity was evaluated using the spectrometric method of Ellman [18]. Acetyl-or butyrylthiocholine iodide and 5,5-dithiobis-(2-nitrobenzoic) acid (DTNB) were purchased from Sigma Aldrich. Lyophilized BuChE from equine serum (Sigma Aldrich) was dissolved in 0.2 M phosphate buffer pH 7.4 such as to have enzyme solutions stock with 2.5 units/mL enzyme activity. AChE from human

Conclusions
In conclusion, we succeeded, within the frame of this work, in the synthesis of the trans diastereoisomers of PADPZ, a structural compromise between DPZ and rasagiline. PADPZ appears endowed with both the activities displayed by these reference compounds and behaves, apparently, in an indifferently manner, as selective AChE and MAO-B inhibitors.
Even if the required level of activities for MTDL are often weaker than those of monotargeted ligands, due to the synergistic effect MTDL are supposed to carry on, improving the activity of trans-PADPZ appears possible through the stereoselective synthesis or separation of its enantiomers. After the proof of interest lying in this original compound this preliminary work brought according to us, this second step, as well as the in vivo evaluation of these enantiomers in animal models of AD, will be quickly undertaken.