Synthetic Approaches for Piperidone-Based Templates as Scaffolds to Access Chirally Enriched Donepezil Analogues

A concise and high-yielding double aza-Michael reaction is presented as an atom-efficient method to access chiral 2-substituted 4-piperidone building blocks from divinyl ketones. The piperidones were further converted into analogues of donepezil, an acetylcholinesterase inhibiting drug used in the treatment of Alzheimer’s disease. The donepezil analogues were obtained as inseparable diastereomeric mixtures with resolved stereochemistry in position 2 of the piperidine ring. Biological evaluation of the acetylcholinesterase inhibition by these analogues provides a new insight into structure–activity relationship studies with regard to donepezil’s piperidine moiety toward stereochemical enhancement.


■ INTRODUCTION
The piperidine ring is an important scaffold in pharmaceutical research and a ubiquitous structural motif which is present in several natural products. 1,2 In general, the 1,4-disubstitution pattern on the piperidine ring prevails among drug prototypes due to more accessible synthetic routes and limited stereochemical complications. Nonetheless, it has been shown that additional substitution on different positions of piperidine could show highly beneficial in terms of identifying compounds with increased biological activity. 3,4 The acetylcholinesterase (AChE) inhibiting drug donepezil 1 (Figure 1) contains such a 1,4-disubstituted piperidine core.
Donepezil is the most commonly prescribed medication for Alzheimer's diseases (Aricept). 5−7 There is a growing interest in the field for the development of analogues to allow for various therapeutic approaches, such as dual-or multifunctional targeting and AChE reactivation. 5−10 In this context, medicinal chemistry research efforts have been mainly centered on ligand-based and fragment-based drug design methodologies to generate "fully" organic compounds, 5−8 as well as chelators for metals 9 and organometallic complexes, 10 as donepezil-derived hits for medical applications. The X-ray resolved structure of AChE has also permitted structure-based drug design approaches. 11,12 Although these latter studies are reported in a limited number, they have established a structural baseline for efficient AChE inhibitor design. 10 In the search for a novel class of acetylcholinesterase inhibitors, studies conducted by the Japanese company Eisai Co., Ltd. demonstrate that the presence of a piperidine ring into the donepezil structure is essential for increased AChE inhibitory effects. 13,14 Despite piperidine's crucial role in the overall activity of donepezil and the various structure−activity relationship studies reported for this lead compound, 15−20 to the best of our knowledge, no other modifications on the piperidine ring have been reported in the literature, with the exception of substitutions in positions 1 and 4. Moreover, analysis of the crystal structure of donepezil bound to acetylcholinesterase suggests that additional substituents on the piperidine ring could be accommodated in the binding pocket, thereby leading to an improved pharmacological profile of derived analogue drugs. 18,21 In this context, we have used divinyl ketones to set up an efficient aza-Michael synthesis for the preparation of 4piperidone scaffolds, which are substituted in position 2 of the ring. We have then used the resulting 4-piperidones to develop a new series of donepezil-based derivatives, to biologically evaluate the impact of chiral modification on the piperidine moiety with regard to their acetylcholinesterase inhibition.
Synthesis of 2-Substituted 1-S-α-Phenylethyl-4-piperidones. The first step toward the synthesis of divinyl ketones 7 was the reaction of suitable vinyl aldehydes 5 with vinylmagnesium bromide under standard Grignard conditions (Scheme 2). Dienols 6a−e were obtained in high yield (Table  1) and sufficiently pure to be used in the next synthetic step without further purification. Also, no degradation was observed for 6a−e, if stored at 4°C for several weeks (as confirmed by 1 H NMR analysis). The desired divinyl ketones 7 (Scheme 2 and Table 1), substituted with aliphatic or aromatic groups, were obtained through the following oxidation reaction under mild conditions of intermediates 6, using manganese(IV) oxide (for 6a,b) or 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ) (for 6c−e) (Scheme 2). The oxidizing agent was selected according to the reactivity of the substrate and the stability of the product. In this regard, the conversion of methyl-and propyl-substituted 6a,b to ketones 7a,b was straightforwardly achieved using MnO 2 . In contrast, MnO 2mediated conditions were found not to be suitable to convert the aromatic analogues 6c−e, which were preferentially oxidized using DDQ. Due to the higher stability of ketones 7c−e, column chromatography was successfully performed in this case, allowing the removal of the oxidizing agent and the complete purification of the products, which were obtained in a good to moderate yield (Table 1).
A double aza-Michael addition of primary amines (i.e., benzylamine or S-α-phenylethylamine) was carried out on ketones 7 as an atom-efficient method to access chiral 4piperidine-based building blocks 8, 2, and 2′ (Scheme 2). Initial investigations on the appropriate solvent for the double aza-Michael addition were performed with benzylamine and phenyl-substituted ketone 7c. Since neat acetonitrile or dichloromethane proved unsuccessful in these test reactions, we applied a slightly modified procedure in comparison to methods previously reported in the literature. 22−24 Specifically, divinyl ketone 7c was slowly added to a mixture of benzylamine in acetonitrile and aqueous sodium bicarbonate at 16°C over a period of 40 min and then refluxed for 1.5 h (entry 3, Table 2). The 2-phenyl-substituted piperidine 8c was obtained with 79% yield. With these reaction conditions, a range of aliphatically and aromatically substituted piperidones have been prepared ( Table 2, 8a−e). As expected, the yields for the methyl-and propyl-substituted piperidones 8a,b ( Table  2, entries 1 and 2) were lower due to the less stable nature of the crude starting ketones 7a,b. In contrast, when purified aromatic divinyl ketones 7c−e were used for the cyclization with benzylamine ( Table 2, entries 3−5), piperidones 8c−e were obtained in high yield (79−84%). S-α-Phenylethylamine was chosen as a chiral auxiliary to synthesize diastereomeric 2substituted-4-piperidones. 25,26 In line with the analogue series 8a−e, the combined yields for the aliphatically substituted piperidones 2a + 2′a (R = methyl) and 2b + 2′b (R = propyl) were lower ( Table 2, entries 1 and 2), while aromatically substituted piperidones 2c + 2′c (entry 3−5) were differently obtained in good, albeit with lower combined yields in   27 Synthesis of Donepezil Analogues. By modifying the reported synthesis of donepezil, 28 diastereomeric methyl-and phenyl-substituted 4-piperidones building blocks 2a + c and 2′a + c were subjected to a Wittig reaction using [(Ph 3 )-PCH 2 OCH 3 ]Cl and lithium diisopropylamide (LDA), 29 to generate the corresponding methoxymethylene derivatives 9a− d + 9′a−d in good to high yields ( Table 3). The chromatographic separation of the two isomers of all substrates allowed the unambiguous determination of their correct stereochemistry due to distinct nuclear Overhauser effect (NOE) correlations found in the isolated products (see Section S1 in the Supporting Information).
Sugimoto's conditions for the originally reported acidic hydrolysis of the methoxymethylene intermediate (in donepezil) lead only to a moderate yield (i.e., 54%), 28 due to the need of not trivial column chromatography for the purification of the product. Therefore, we set up a different protocol based on the mild hydrolysis of enol ethers 9 + 9′a−d with a mixture of tetrahydrofuran/1. 6 M HCl (1:1) 29 (Table 4). The corresponding aldehydes 3 + 3′a−d were obtained as diastereomeric mixtures in excellent yields and high purity. Nonetheless, it was observed (by 1 H NMR analysis) that the storage of aldehydes at low temperatures (i.e., −20°C) is necessary to prevent degradation over time. The diastereomeric ratio was determined by analysis of the 1 H NMR spectra (Table 4), and the chirality of the 4-formyl moiety was confidently assigned for all products from the observed nuclear Overhauser effect spectroscopy (NOESY) correlations (see Section S1 in the Supporting Information).
As an example, in Figure 2 are reported the correlations observed between H ax -2 and the proton in position 4 of aldehyde 3b (see Section S1 in the Supporting Information). This indicates an axial position for H-4 and, hence, R-chirality for the stereocenter in position 4. In contrast, no such NOESY correlation between H ax -2 and the proton in position 4 was found in diastereomer 3′b, while the interaction was observed between the protons of the methyl group in position 2 and H eq -4, which confirms S-chirality for this stereocenter in position 4 of the piperidine ring.
With regard to the overall yield and purity, the present conditions for accessing N-benzyl-4-formyl-piperidine are superior to those of the procedures published earlier. 28,30,31 The subsequent aldol condensations of 5,6-dimethoxy-1indanone 10 with aldehyde mixtures 3a−d + 3′a−d were carried out according to a similar protocol developed by Imai   1, 4, and 5) or determined by analysis of the 1 H NMR spectra (for entries 2 and 3).   et al., 32 by avoiding the low temperatures (−78°C) and toxic solvents reported by Sugimoto et al. 13,28 (Table 5). When a mixture of 2S-methyl-substituted aldehydes 3 + 3′a and 2Rmethyl-substituted aldehydes 3 + 3′b was subjected to the aldol condensation reaction (Table 5, entries 1 and 2), not trivially separable methylene products were obtained and their diastereomeric ratios could be determined by analysis of the signals for the olefinic protons in the crude 1 H NMR spectra. The corresponding aldol products were obtained as inseparable diastereomeric species for both 2-phenyl-substituted aldehyde mixtures 3 + 3′c and 3 + 3′d (Table 5, entries 3 and  4).
Hydrogenation of alkenes 11a−d + 11′a−d using palladium on activated carbon was investigated according to a protocol by Sugimoto 13 and mixtures of the final donepezil analogues 4 + 4′a−h were obtained in a moderate to good yield (Table 6). Standard reversed-phase high-performance liquid chromatography (HPLC) was attempted by using various chromatographic conditions (for details, see representative method development in General procedures, Experimental Section), to isolate the single diasteroisomers from each mixture. All of the screened conditions gave a single peak for 4 + 4′a−h mixtures (see Section S3 in the Supporting Information), suggesting that the separation of chirally resolved single species was not possible in standard reversed-phase HPLC chromatography.
As the 2-substituted donepezil analogues also include the chiral N-α-phenylethyl moiety, we envisaged that an unambiguous evaluation of the biological effects (i.e., acetylcholinesterase inhibition) produced by a substituent on the piperidine ring (in comparison to the unsubstituted donepezil) is only feasible with the reference analogues 19 + 19′ (Scheme 3). Indeed, compounds 19 + 19′ bear the chiral N-α-phenylethyl moiety, but no additional substituent is present on the piperidine ring, allowing more accurate structure−activity relationship analyses for this class of compounds. Analogues 19 + 19′ were synthesized in six steps, according to the conditions depicted in Scheme 3. Briefly, intermediates 14 and 15 were synthesized using reported methods. 33,34 Piperidone 15 was transformed into methoxymethylene-based compound 16 through an analogous Wittig reaction (with [(Ph 3 )PCH 2 OCH 3 ]Cl and LDA) as reported above for 9 + 9′a−d. Intermediate 17 was produced from 16 by mild acidic hydrolysis with THF/1. 6 M HCl (1:1) and converted into 18 by aldol condensations of 5,6dimethoxy-1-indanone 10. The final step is the classical  reduction of the alkene group in 18 with H 2 and Pd/C (Scheme 3), to afford diastereomeric mixtures of the final analogues 19 + 19′. Biological Activity Studies. It has been reported that the R and S enantiomers of donepezil 1 interconvert in an aqueous solution at 37°C due to ketoenol tautomerism, with a racemization half-life of 78 h. 35 A similar spontaneous racemization of the stereocenter on the indanone moiety is also anticipated for the synthesized donepezil analogues 4 + 4′a−h. Also, it is worth mentioning that donepezil (Aricept) is not chirally resolved and is clinically used as a racemic mixture. 18,36 In this view, we opted for a preliminary biological screening of the diastereomeric mixtures 4 + 4′a−h, before engaging in laborious, as well as lengthy, preparative chiral HPLC separations, to obtain initial evidence of the AChE inhibitory activity for this series.
The electric eel AChE (eeAChE) inhibitory activity of the synthesized donepezil analogues was evaluated in comparison to donepezil. A UV−vis spectrophotometry-based assay was conducted to calculate the IC 50 values for the diastereomeric mixtures (Table 6), according to Ellman et al. 37 and Mohamed et al. 38 The half-inhibitory concentrations of all compounds possessing a substituent in position 2 on the piperidine ring (4 + 4′a−h; entries 1−6, Table 6) were evaluated in comparison to the diastereomeric pair of compounds lacking a substituent on the piperidine ring, 19 + 19′ (entry 7,Table 6), and the Nbenzyl-substituted reference 1 (i.e., donepezil; entry 8, Table  6).
Piperidine ring unsubstituted diastereomers 19 + 19′ possess an IC 50 of 1.83 μM (entry 7, Table 6) and are approximately 27 times less active than the donepezil 1 (entry 8, Table 6), indicating that the additional methyl group has a detrimental effect on the drug's inhibitory activity. Interestingly, 2S-methyl-4S compounds 4 + 4′a possess a lower halfinhibitory concentration of 1.01 μM (entry 1), demonstrating that the additional methyl group on the piperidine partially compensates for some of the initial drop of activity caused by the N-S-α-phenylethyl moiety in 19 + 19′. Among the 2-methyl-substituted analogues (entries 1−4, Table 6), those with 2S-chirality (entries 1 and 2, Table 6) are more active than the ones with 2R-chirality (entries 3 and 4, Table 6). The reduced inhibitory potency of these derivatives suggests that donepezil is sensitive to stereoselective substitution in position 2 on the piperidine ring, possibly due to significant changes in the overall conformation within the enzyme's binding pocket and, therefore, affecting also the binding affinities. Furthermore, both diastereomeric mixtures possessing a phenyl ring in position 2 on the ring (entries 5, 6, Table 6) showed no activity in the tested concentration range (0.001−100 μM). In this regard, previous studies indicate that the bulky phenyl ring could prevent these compounds from entering the narrow, "swinging-gate"-controlled entry site of the binding gorge of the enzyme. 21,36 Finally, when comparing the IC 50 values of the synsubstituted methyl compounds (entries 1, 4, Table 6) to the anti-substituted methyl analogues (entries 2 and 3, Table 6), the syn-compounds are significantly more active in both cases. 2S,4S-syn-substituted compounds 4 + 4′a are approximately 24 times more active than the 2R,4S-anti-substituted analogues 4 + 4′c (Figure 3), demonstrating that the stereochemistry of the substituent in position 4 in relation to the stereochemistry of the substituent in position 2 (i.e., syn-or anti-conformation) significantly influences the overall inhibitory activity of the diastereomeric mixtures.

■ CONCLUSIONS
Chirally resolved 2-substituted 4-piperidones were prepared from commercially available starting materials in three steps, thereby providing easy access to building blocks for the assembly of biologically relevant piperidine-based scaffolds. The reaction conditions reported for the synthesis of unstable aldehyde-type intermediates from these piperidine-4-one scaffolds are significantly improved (in terms of both overall conditions and final yield), compared to other procedures found in the literature. Furthermore, the aldehydes have been used to produce novel analogues of donepezil, which are stereochemically enriched through substituents at position 2 of the piperidine ring. AChE inhibition studies strongly indicate that the stereochemistry of substituents on the piperidine ring could play an important role in the binding behavior of such compounds within AChE's active pocket.

■ EXPERIMENT SECTION
General. Infrared spectra were obtained on a PerkinElmer 100 Fourier transform infrared spectrometer operating in attenuated total reflection mode. Only significant absorptions (ν max ) are reported, and all absorptions are recorded in wavenumbers (cm −1 ). Melting points were measured with an electrothermal apparatus and are uncorrected. Proton magnetic resonance spectra ( 1 H NMR) were recorded at 400 MHz using a Bruker spectrometer. Chemical shifts (δ) are quoted in parts per million (ppm) and are referenced to the residual  protonated solvent peak. The order of citation in parentheses is (i) number of equivalent nuclei (by integration), (ii) multiplicity (s, singlet; d, doublet; t, triplet; q, quartet and m, multiplet), and (iii) coupling constant (J) quoted in Hertz (Hz) to one decimal place. Carbon magnetic resonance spectra ( 13 C NMR) were recorded at 100.6 MHz using a Bruker spectrometer. Chemical shifts (δ) are quoted in parts per million (ppm) and are referenced to the appropriate solvent peak. The assignment is quoted in parentheses. Where necessary, assignments were made with the aid of DEPT, correlation spectroscopy, heteronuclear single quantum coherence, heteronuclear multiple-bond correlation, or NOESY correlation experiments. Low-resolution mass spectra (m/z) were recorded using an LCQ DECA XP instrument by electron spray ionization (ESI). Only molecular ions and major fragments of the molecular ions are reported. Accurate masses were determined using a quadrupole time-of-flight mass spectrometer at King's College London or a Thermo Fisher LTQ Orbitrap XL instrument at the EPSRC National Mass Spectrometry Facility in Swansea using nanospray ESI (NSI) and atmospheric pressure chemical ionization (APCI). Flash chromatography was carried out using silica gel (Aldrich, 230− 400 mesh) as the stationary phase. Thin-layer chromatography was carried out on aluminum plates precoated with silica (Merck silica gel 60 F 254 on aluminum), which was visualized by the quenching of ultraviolet fluorescence (λ max = 254 nm) and/or by staining with potassium permanganate solution followed by heat. All reactions were carried out at atmospheric pressure with stirring unless otherwise stated. All reagents were used as received unless otherwise stated. The fractions of light petroleum ether boiling between 40 and 60°C are referred to as "hexanes". Optical rotations ([α]D T = α/l.c) were measured by a Bellingham and Stanley ADP 220 polarimeter at 589 nm (sodium-D line). Concentration (c) is in g 100 mL −1 . The HPLC analysis was performed on a Hewlett-Packard 1050 system equipped with an autosampler, a reversed-phase HPLC column (Agilent Zorbax 300 Å, C-18, 2.1 mm × 100 mm, particle size 3.5 μm), and a diode-array detector set to monitor 281 nm. The flow rate was 0.2 mL/min, and the column was eluted using three different linear gradients: (i) 0−90% MeCN in 0.1% (v/v) trifluoroacetic acid aqueous solution in 20 min (t R1 ); (ii) 0−90% MeCN in 0.1% (v/v) trifluoroacetic acid aqueous solution in 30 min (t R2 ); and (iii) 0−50% MeCN in 0.1% (v/v) trifluoroacetic acid aqueous solution in 50 min (t R3 ) (see Section S3 in the Supporting Information).
Synthesis. Preparation of Manganese Dioxide. Manganese dioxide (MnO 2 ) was purchased from Alfa Aesar (Cat. No. 014340.22manganese(IV) oxide, activated, tech. Mn 58% min, 100 g) and further activated by treatment with dilute nitric acid. MnO 2 (50 g) was placed on a large Buchner funnel and 10% nitric acid (80 mL) was added slowly. After the addition was completed, the MnO 2 cake was washed with water (2−3 L) until the filtrate was neutral. The MnO 2 was subsequently dried at 105°C for 2 days.
E-1,4-Hexadien-3-ol (6a). Crotonaldehyde (747 mg, 10.70 mmol, 1.3 equiv) in THF (6 mL) was added dropwise to an ice-cooled solution of vinylmagnesium bromide in THF (8.00 mmol, 1.0 M in THF, 1.0 equiv) under an atmosphere of nitrogen. After stirring at room temperature for 1 h, the reaction mixture was poured into a mixture of saturated NH 4 Cl (10 mL) and ice (10 g) and stirred vigorously for 5 min. The aqueous solution was extracted with ether (3 × 15 mL), and the combined organic phases were dried over MgSO 4 .
Before running the assay, one of these aliquots was diluted with buffer (50 mM Tris−HCl, pH = 8.0, 0.1% w/v BSA) to obtain 25 mL of a solution with an enzyme concentration of 0.22 U/mL. A 15 mM solution of acetylthiocholine iodide (ATCI, Sigma-Aldrich) was prepared by dissolving 108.45 mg in ultrapure water (25 mL). All of these solutions were prepared freshly directly before the assay was run. In 96-well plates, 160 μL of the 1.5 mM DTNB solution was first added, followed by 10 μL of different concentrations of test compounds. Then, a 0.22 U/mL enzyme solution was added (50 μL) and incubated at room temperature for 7 min. The background absorbance was measured at a wavelength of 412 nm on a POLARstar OPTIMA (BMG Labtech) microplate reader. The addition of 30 μL of the 15 mM ATCI solution initiated the enzymatic reaction, and the time elapsed between the beginning of the pipetting and the first measurement (t = 0) was noted and kept consistently at 2 min. Further kinetic measurements were made at different time intervals (t = 1, 2, 3, 4, 5 min) after a short episode of shaking (10 s). Each experiment was carried out in triplicate. Various control incubations containing 10 μL of methanol were run alongside the test compounds on each plate. The IC 50 values were calculated using GraphPad Prism software (version 6.01) by applying a nonlinear regression analysis (sigmoidal dose− response fit with a variable slope).
HRMS spectra and full 1 H and 13 C NMR spectra for all of the new derivatives reported in the manuscript, stereochemical assignment for 2-substituted piperidones, and HPLC chromatograms for the analogues tested as AChE inhibitors (PDF)