Novel Potent Muscarinic Receptor Antagonists: Investigation on the Nature of Lipophilic Substituents in the 5- and/or 6-Positions of the 1,4-Dioxane Nucleus

A series of novel 1,4-dioxane analogues of the muscarinic acetylcholine receptor (mAChR) antagonist 2 was synthesized and studied for their affinity at M1–M5 mAChRs. The 6-cyclohexyl-6-phenyl derivative 3b, with a cis configuration between the CH2N+(CH3)3 chain in the 2-position and the cyclohexyl moiety in the 6-position, showed pKi values for mAChRs higher than those of 2 and a selectivity profile analogous to that of the clinically approved drug oxybutynin. The study of the enantiomers of 3b and the corresponding tertiary amine 33b revealed that the eutomers are (2S,6S)-(−)-3b and (2S,6S)-(−)-33b, respectively. Docking simulations on the M3 mAChR-resolved structure rationalized the experimental observations. The quaternary ammonium function, which should prevent the crossing of the blood–brain barrier, and the high M3/M2 selectivity, which might limit cardiovascular side effects, make 3b a valuable starting point for the design of novel antagonists potentially useful in peripheral diseases in which M3 receptors are involved.


■ INTRODUCTION
Muscarinic acetylcholine receptors (mAChRs) are proteins with seven transmembrane domains separated by intracellular and extracellular loops. Acetylcholine binds to the extracellular region of mAChRs and thereafter activates GTP-binding regulatory proteins in the intracellular compartment. The mAChR family consists of five closely related members (M 1 − M 5 ). M 1 , M 3 , and M 5 mAChRs are associated with G q/11 proteins to trigger phospholipase-C activation. Their activation increases neuronal excitability through the opening of nonspecific cation channels, mobilization of intracellular Ca 2+ , or inhibition of small-conductance Ca 2+ -activated K + channels. M 2 and M 4 subtypes couple to G i/o proteins, inhibiting adenylate cyclase and reducing the levels of intracellular adenosine 3′,5′-cyclic monophosphate (cAMP). 1 mAChRs mediate several functions in the central nervous system (CNS), where they play a crucial role in cognitive functions 2 and pain circuits. 3 Moreover, in the periphery, M 2 and/or M 3 subtypes are involved in smooth muscle contraction, 4 cardiovascular function, 5 and glandular secretion. 6 Acetylcholine is not only a neurotransmitter but can also act on non-neuronal cells, and the muscarinic system is involved in the regulation of stem 7 and cancer cells, 8 in immunity and inflammation, 9 and in the mucocutaneous epithelial barrier. 10 Moreover, muscarinic signals have been demonstrated to be transmitted by mesenchymal stem cells (MSCs) from different tissues. 11, 12 The 1,4-dioxane nucleus has been demonstrated to be a versatile scaffold for the development of compounds interacting with different receptor systems, 13−20 including mAChRs. 21−25 We have demonstrated that the size of the substituent in the 6-position affects the functional activity of 1,4-dioxane ligands directed to mAChRs. 24 Indeed, a methyl group in this position led to the effective agonist (2R,6S)-1, 23 whereas aromatic rings characterized potent antagonists, such as the 6,6-diphenyl derivative (S)-2 24 (Figure 1). In in vivo studies in anesthetized rats, compared to oxybutynin ( Figure  1), an antagonist clinically used for overactive bladder (OAB) treatment, 26 (S)-2 more efficaciously reduced the volumeinduced contractions of the urinary bladder.
In the effort to obtain novel potent mAChR antagonists preferentially targeting peripheral M 3 subtype and potentially useful for the treatment of OAB, the diphenyl group in the 6position of compound 2 has been replaced by different lipophilic groups (compounds 3−8, Figure 2). Furthermore, the lipophilic moiety has been moved from the 6-to 5-position (compounds 9−17, Figure 2) or introduced in both 5-and 6positions of the 1,4-dioxane ring (compounds 18−19, Figure  2). All the substituents are aromatic groups, except for the aliphatic cyclohexyl ring in compounds 3 and 11, which has been selected because it is present in several potent mAChR antagonists, including oxybutynin ( Figure 1).
Owing to the lack of M 3 subtype selectivity, the muscarinic compounds used in therapy for OAB show cardiovascular side effects, due to the interaction with peripheral M 2 subtype, and/ or cognitive side effects, due to the blockade of central mAChRs. 27,28 Among these drugs, only trospium bears a hydrophilic quaternary ammonium head (Figure 1) that prevents the crossing of the blood−brain barrier (BBB), thus minimizing CNS side effects. 29 Because of the high degree of amino acid sequence homology in the orthosteric site of the M 1 −M 5 mAChR subtypes, it is very difficult to obtain orthosteric ligands selective for M 3 mAChR over all the other subtypes. For this reason, the aim of the present study was to confine the activity of the novel compounds to peripheral tissues, minimizing CNS side effects and, hopefully, to limit cardiovascular side effects by improving the M 3 /M 2 selectivity ratio. Therefore, the new molecules have been designed to have a quaternary ammonium function that should prevent the crossing of the BBB.
The novel compounds will provide further information on the role played by the lipophilic moiety in the interaction with the five mAChR subtypes.
Moreover, considering the pivotal role played by stereochemistry in the interaction of both 1,4-dioxane agonists and antagonists with the five mAChR subtypes, 22,24,30 the enantiomeric resolution of the most potent compound 3b was performed. The absolute configuration of the enantiomers of 3b was determined by quantum mechanical simulations of electronic circular dichroism (ECD). To elucidate the binding  ■ RESULTS AND DISCUSSION Chemistry. Compounds 3−8 were synthesized following the procedure reported in Scheme 1 and were obtained as racemates. 4-Benzylbenzaldehyde 20 31 was converted into the oxirane 22 by reaction with sodium hydride and trimethylsulfonium iodide in dimethyl sulfoxide (DMSO), according to the procedure reported by Corey and Chaykovsky. 32 The opening of oxiranes 21, 33 22, and 23 34 with allyl alcohol in the presence of Na gave alkenes 24−26, respectively. The mixtures of diastereomers 28−30 were obtained by the oxymercurationreduction reaction with mercury(II) acetate and subsequent treatment with an aqueous solution of potassium iodide and iodine. The cis and trans isomers of 28 and 30 were separated by column chromatography, while attempts to obtain the pure diastereomers of 29 failed. The iodo derivatives 27a and 27b were synthesized as previously reported. 16 The phenyl thioethers 30a and 30b were oxidized with meta-chloroperbenzoic acid (m-CPBA) to give the sulfoxides 31a and 31b after 30 min at room temperature (r.t.) with one equivalent of m-CPBA or the sulfones 32a and 32b after 2 h at r.t. with 2 equivalents of m-CPBA. Concerning the sulfoxide derivatives 31a and 31b, a further center of chirality was introduced into the molecule. In both cases, only one of the two diastereomers was obtained. The amination of the intermediate iodo derivatives 27−32 with dimethylamine afforded the corresponding free amines 33−38, which were transformed into the methiodides 3−8 by treatment with methyl iodide.
The relative configuration between the substituents in 2-and 6-positions of diastereomers 3a and 3b was determined by Xray diffraction analysis performed on 3b (Figure 3), which confirmed the structure of intermediates 27a and 27b previously assigned by 1 H NMR studies. 16 The relative configuration between the substituents in 2-and 6-positions of the 1,4-dioxane ring of diastereoisomers 4a/b was assigned based on the 1 H NMR spectra of intermediates 28a/b ( Figure 4A). Because of the steric bulk, one may suppose that both the substituents in 2-and 6-positions of the cis isomers are equatorially oriented, whereas in the trans isomers, only one of the two substituents adopts the equatorial position and the other substituent is axially oriented. In the 1 H NMR spectrum of the iodo derivative 28b, precursor of the final methiodide 4b, the protons of the CH 2 I chain (3.55 ppm) are deshielded compared to the same protons of diastereomer 28a (3.22 ppm), precursor of methiodide 4a. This deshielding effect for CH 2 I protons of diastereomer 28b (see Supporting Information, Figure S4) suggests an axial position for the side Scheme 1. Reagents: (a) (CH 3 ) 3 SI, NaH, DMSO; (b) Na, CH 2 CHCH 2 OH; (c) (CH 3 COO) 2 Hg; H 2 O, KI, I 2 ; (d) 1 equiv m-CPBA, CH 2 Cl 2 , 30 min; (e) 2 equiv m-CPBA, CH 2 Cl 2 , 2 h; (f) (CH 3 ) 2 NH, C 6 H 6 ; (g) CH 3 I, (CH 3 CH 2 ) 2 O a chain, as already evidenced in 1,4-dioxane analogues bearing a CH 2 I chain 35 and, consequently, the relationship between the biphenyl substituent and the chain is trans ( Figure 4A). Similar considerations can be made for diastereomers 30a/b, precursors of the final methiodides 6a/b. Indeed, the signals for CH 2 I protons of 30a and 30b are positioned at 3.18 and 3.52 ppm, respectively, indicating a trans configuration between the substituents in the 1,4-dioxane nucleus of 30b.
The novel compounds 9−17 were prepared following the procedure depicted in Scheme 2 and were obtained as racemates. The opening of oxiranes 39, 36 21, 33 22, and 23 34 with allyl alcohol in the presence of perchloric acid yielded compounds 40, 42, 43, and 44, respectively. The olefine 41 was prepared starting from 3,3-diphenylpropane-1,2-diol (45), 37 whose primary hydroxyl group was selectively protected with tert-butyldimethylsilyl chloride (TBDMSCl) to give compound 46, which was treated with allyl bromide in the presence of NaH affording olefine 47. The cleavage of the silyl ether with tetrabutylammonium fluoride (TBAF) yielded the corresponding primary alcohol 41. The intermediates 48 and 49 were obtained as previously described in the literature. 13 The mixtures of diastereomers 50−54 were obtained starting from olefins 40−44 in the same reaction conditions used for the preparation of 28−30. The diastereomers were separated by column chromatography. The thioethers 54a and 54b were oxidized to give sulfoxides 55a and 55b, respectively, and sulfones 56a and 56b as above described for 31 and 32. Similarly to what was observed for 31a and 31b, also for 55a and 55b, only one of the two diastereomers was obtained. The amination of 48−56 with dimethylamine afforded the corresponding amines 57−65, which were transformed into the methiodides 9−17 by treatment with methyl iodide (Scheme 2).
The cis and trans configurations between the substituents in 2-and 5-positions of diastereoisomers 9a and 9b, respectively, were assigned based on the previously published structures of compounds 48a and 48b. 13 The structures of diastereoisomers 11−15 were assigned by 1 H NMR spectroscopy. Because of the steric bulk, it may be supposed that in the trans isomers both the substituents in 2-and 5-positions of the 1,4-dioxane nucleus are in the equatorial position, whereas in the cis isomers, only one of the two substituents is equatorially oriented and the other is axially oriented. Analogous to what occurs for the diastereomers 48a and 48b, 13 in the 1 H NMR spectrum of the iodo derivative 52a, precursor of the final methiodide 13a, the protons of the CH 2 I chain (3.57 ppm) are deshielded compared to the same protons of diastereomer 52b (3.19 ppm), precursor of methiodide 13b (see Supporting Information, Figure S5). This deshielded effect for CH 2 I protons of diastereomer 52a suggests an axial position for the side chain, as also evidenced in 1,4-dioxane analogues bearing a 2-CH 2 I chain 35 and, therefore, a cis configuration between the chain and the biphenyl substituent ( Figure 4B). Similar considerations can be made for diastereomers 51a/b, 53a/b, and 54a/b, precursors of the final methiodides 12a/b, 14a/b, and 15a/b. Indeed, the CH 2 I protons of diastereomers 51a, 53a, and 54a are more deshielded (3.38, 3.58, and 3.52 ppm, respectively) compared to those of diastereomers 51b, 53b, and 54b (3.08, 3.17, and 3.16 ppm, respectively), demonstrating a trans configuration between the substituents in the 1,4-dioxane nucleus for 51a, 53a, and 54a.
The relative orientation between the CH 2 I fragment and the 5-substituents of 11a and 11b was assigned by 1 H NMR analysis (NOESY studies, see Supporting Information, Figure  S7). In particular, evident NOEs were observed between the axial proton in the 3-position and the hydrogen atoms of the phenyl ring in the 5-position and between the axial protons in 2-and 6-positions (4.21 and 3.88 ppm, respectively) of 11a, indicating that the 5-phenyl nucleus and the 2-side chain are axially and equatorially oriented, respectively. Therefore, the relative configuration between the 2-side chain and the 5phenyl substituent is cis in 11a and, consequently, trans in 11b ( Figure 5).
Compounds 18a−c were synthesized following the procedure described in Scheme 3 and were obtained as racemates. The alcohol intermediates 68a and 68b, synthesized as previously described, 38 and 68c, obtained by treatment of olefine 66 39 with m-CPBA and subsequent reaction of oxirane 67 with trifluoroacetic acid, were reacted with p-toluenesulfonyl chloride followed by treatment with dimethylamine to give 69a, 69b, and 69c, whose reaction with methyl iodide afforded the final methiodides 18a, 18b, and 18c, respectively.
The stereochemical relationship among the substituents of the diastereomers 18a and 18b was determined based on the previously assigned structure of the alcohol intermediates 68a and 68b. 38 The stereochemical relationship among the substituents in 2-, 5-, and 6-positions of 18c was assigned by  . Structure of (A) compounds 28a and 28b, precursors of 4a and 4b, respectively, and of (B) compounds 52a and 52b, precursors of 13a and 13b, respectively. 1 H NMR analysis (NOESY studies, see Supporting Information, Figure S8). In the 1 H NMR spectrum of 18c, the axial hydrogen atom in the 3-position at δ 3.69 ppm showed two large coupling constants (J = 11.2 Hz and J = 10.0 Hz), one with the geminal equatorial hydrogen atom and the other with the axial hydrogen atom in the 2-position. Hence, the chain in the 2-position is equatorially orientated. Moreover, NOEs were observed between the axial proton in the 3-position and the proton in the 5-position at 3.69 and 5.22 ppm, respectively, and between the axial proton in the 2-position at 4. 24   the phenyl ring in the 6-position, indicating that the 2-side chain is trans oriented with both phenyl substituents ( Figure  6).
In the effort to obtain the fourth diastereomer, in which the stereochemical relationship among the three substituents is cis, the olefine 66 39 was treated with mercury(II) acetate, followed by an aqueous solution of potassium iodide and iodine. However, also in this case, only one diastereomer (70) was obtained. The amination with dimethylamine and subsequent reaction with methyl iodide yielded the same diastereomer (18c) obtained following the previously described procedure.
Compounds 19a and 19b were prepared following the procedure described in Scheme 4 and were obtained as racemates. Olefine 72, obtained by reaction of the α-allyloxy ketone 71 40 with phenylmagnesium chloride, was treated with m-CPBA in CH 2 Cl 2 , affording oxirane 73, whose treatment with trifluoroacetic acid in CHCl 3 led to alcohols 74a and 74b, which were separated by flash chromatography. Treatment of the alcohols with p-toluenesulfonyl chloride followed by reaction with dimethylamine afforded the amines 75a and 75b, whose treatment with methyl iodide gave 19a and 19b, respectively.
The relative configuration between the 2-substituent and the 5-phenyl group of the diastereomers 19a and 19b was assigned by 1 H NMR spectroscopy. In the 1 H NMR spectrum of the tertiary amine 75a, precursor of 19a, the axial hydrogen atom in the 3-position at 3.78 showed two large coupling constants (J = 11.2 Hz and J = 10.4 Hz), one with the geminal equatorially located proton and one with the axial proton in the 2-position. Hence, the CH 2 N(CH 3 ) 2 fragment in the 2-position assumes the equatorial position. Analogously, as shown by the 1 H NMR spectrum of 75b, precursor of 19b, the CH 2 N(CH 3 ) 2 fragment in the 2-position is equatorial because the axial proton in the 3-position at 3.58 showed two large coupling constants (J = 11.5 Hz and J = 10.3 Hz), one with the geminal equatorially positioned hydrogen atom and one with the axially oriented hydrogen atom in the 2-position. Moreover, the proton in the 5-position of 75b (5.82 ppm) is deshielded compared to the same proton of 75a (4.95 ppm) (see Supporting Information, Figure S6). The observation that in the 1 H NMR spectra of the cis and trans diastereomers of 5phenyl-1,4-dioxane-2-carboxylic acid and 6-phenyl-1,4-dioxane-2-carboxylic acid, whose structure had previously been determined by NOE measurements, 13 the equatorially oriented protons are deshielded compared to the axially oriented protons allows us to hypothesize that the proton in the 5position is axially oriented in 75a and equatorially oriented in 75b. Therefore, the relative configuration between the 2-CH 2 N(CH 3 ) 2 chain and the 5-phenyl ring is trans in 19a and cis in 19b ( Figure 7).
The enantiomers (+)-3b and (−)-3b were separated by preparative HPLC performed on the intermediate amine (±)-33b using a Regis Technologies Whelk-O 1 (R,R) H (25 cm × 2 cm) column as the chiral stationary phase and nhexane/2-propanol 85/15 v/v as the mobile phase at a flow rate of 18 mL/min. The enantiomeric excess (e.e.), determined by analytical HPLC using a Regis Technologies Whelk-O 1 (R,R) H (25 cm × 0.46 cm) column as the chiral stationary phase and n-hexane/2-propanol 85/15 v/v as the mobile phase at a flow rate of 1 mL/min, proved to be >99.5% for both enantiomers.
The absolute configuration of 33b was determined by quantum mechanical simulations of ECD. The ECD spectra of the two enantiomers of the tertiary amine 33b (Figure 8), measured on the oxalate salt dissolved in acetonitrile, contain the typical bands of a benzene chromophore attached to a chiral moiety: 41 the 1 L b band between 240 and 280 nm, which is electric-dipole forbidden, shows the characteristic vibrational fine structure, and the 1 L a band between 210 and 225 nm, which is electric-dipole allowed, is more intense. 42,43 Time-dependent density functional theory (TDDFT) calculations have been shown to be practical means to simulate the CD spectra of this series of ligands, especially with reference to the 1 L a band. 43 In fact, the 1 L b band is more  problematic because of some known issues of TDDFT for aromatic hydrocarbons 44 and because vibronic calculations are needed to reproduce the vibrational pattern. 45 This fact practically limits the comparison between experiment and calculation to a single band, namely, 1 L a . To exclude possible pitfalls, the present computational protocol based on DFT calculations 46,47 was first validated on the tertiary amines of (S)-2 and (R)-2, whose absolute configuration is known. 24 Conformational analysis and DFT geometry optimizations run on the ammonium ion of (2R,6R)-33b led to only two populated conformers at r.t. (Figure 9). They mainly differ in the rotation around the cyclohexyl-C2 bond, while the rest of the structure is preserved. The preferential orientation of the 2side chain is dictated by an intramolecular NH−O1 hydrogen bond.
TDDFT calculations were run with several different DFT functionals and basis sets (see ECD and NMR Calculations), either in vacuo or including a solvent model for acetonitrile. All    . Two populated conformers of the ammonium ion of (2R,6R)-33b. explored combinations predicted a negative rotational strength for the 1 L a band of both conformers, in a very consistent way. Thus, the prediction of the diagnostic ECD band is very robust. In Figure 8, the experimental spectrum is compared with the spectrum calculated at the CAM-B3LYP/def2-SVP/ PCM level. As can be seen, the relative energy of the 1 L b band is overestimated by calculations and the vibrational pattern is missing. Still, the correct negative rotational strength is reproduced for this band too. The agreement between experimental and calculated ECD spectra is satisfactory. Therefore, the absolute configuration is (2R,6R) for (+)-33b and (2S,6S) for (−)-33b.
Binding Studies. The pharmacological profile of methiodides 3−19 was assessed by radioligand binding assays with human recombinant hM 1 −hM 5 receptor subtypes stably expressed in Chinese hamster ovary (CHO) cell lines using [ 3 H]N-methylscopolamine ([ 3 H]NMS) as a radioligand to label mAChRs, following previously described protocols. 48,49 The affinities, expressed as pK i , are shown in Table 1 along with those of compound 2, oxybutynin and trospium, which are included for useful comparison.
The analysis of data reveals that among all the modifications, the replacement of one of the two phenyl rings of 2 with a cyclohexyl group, affording 3, proved to be the most favorable for the interaction with mAChRs. In particular, the diastereomer 3b, with a cis configuration between the CH 2 N + (CH 3 ) 3 chain in the 2-position and the cyclohexyl fragment in the 6-position of the 1,4-dioxane ring, shows pK i values for all mAChR subtypes, except for M 2 , higher than those of the 6,6-diphenyl derivative 2. Compound 3b displays a selectivity profile analogous to that of the clinically approved drug oxybutynin, with affinities for M 1 , M 3 , and M 4 higher than those for M 2 and M 5 subtypes. Interestingly, the M 3 /M 2 selectivity ratio of 3b (14.5) is significantly higher than those of the lead 2 and trospium (1.6 and 1.1, respectively). The M 3 / M 2 selectivity profile of 3b is noteworthy because the presence of a quaternary ammonium head, enhancing the charge transfer interactions that it elicits with the surrounding aromatic residues, generally increases the pK i values for all muscarinic subtypes at the expense of the selectivity ratios. Indeed, these aromatic side chains, and in particular four tyrosine residues, represent a structural signature which is completely conserved by all mAChR subtypes.
The trans configuration between the substituents in 2-and 6-positions of the diastereomer 3a is detrimental for the binding affinity for all the mAChR subtypes, confirming that stereochemistry plays a crucial role in the interaction of 1,4dioxane derivatives with the mAChRs. 22,24,30 The replacement of the 6,6-diphenyl group of 2 with a parabiphenyl group, affording the diastereomers 4a and 4b, induces a dramatic decrease in affinity for all the mAChR subtypes. The higher flexibility of the terminal phenyl group of 4 obtained by introducing a methylene button (mixture 5a/b) or a sulfur atom (diastereomers 6a and 6b) between the two phenyl groups does not improve mAChR affinity. Similar results are obtained by oxidizing the sulfur atom of 6 to sulfoxide and sulfone, affording compounds 7 and 8, respectively. In the pairs of diastereomers 6a/6b, 7a/7b, and 8a/8b, the trans isomers show pK i values slightly higher than those of the corresponding cis isomers.
The shift of the diphenyl group from the 6-to 5-position of the 1,4-dioxane ring of 2, affording compound 10, is also detrimental for the binding to the five mAChR subtypes. The removal of one aromatic group of 10, obtaining the diastereomers 9a and 9b, further decreases the mAChR affinity. Similar to what was observed for the 6-substituted ligands, the replacement of an aromatic group of 10 with a cyclohexyl ring is favorable for the binding to the five mAChRs. In this case, stereochemistry seems not to play a role in the binding at mAChRs, both diastereomers 11a and 11b showing similar pK i values, with a preference for the M 1 subtype. The increased distance between the diphenyl lipophilic moiety and the ammonium head of 10, yielding the diastereomers 12a and 12b, decreases the pK i values for all the mAChRs. Analogous to what was observed for the corresponding 6-substituted derivatives, all the other modifications performed on the 6,6-diphenyl group of 10 (i.e., its replacement with C 6 H 4 -4-C 6 H 5 , C 6 H 4 -4-CH 2 -C 6 H 5 , C 6 H 4 -4-S-C 6 H 5 , C 6 H 4 -4-SO-C 6 H 5 , and C 6 H 4 -4-SO 2 -C 6 H 5 ), affording 13−17, are detrimental for the affinity for mAChRs. Though with low affinity, the diphenylsulfone 17a shows selectivity for M 2 over the other subtypes. This selectivity profile agrees with what was reported for other muscarinic derivatives bearing the diphenylsulfone moiety. 51 Compared to the 5-mono-phenyl derivatives 9a and 9b and the previously described 6-mono-phenyl derivatives, 24 the presence of a phenyl substituent in both 5-and 6-positions of the 1,4-dioxane ring (18a, 18b and 18c) seems to be advantageous, especially when the two phenyl groups are in a cis stereochemical relationship (18c). Instead, the insertion of a phenyl substituent in the 5-position of the 6,6-diphenyl derivative 2, affording 19a and 19b, markedly reduces the binding affinities.
The well-established influence of chirality on the biological activity of mAChR ligands 22,24,30 prompted us to prepare and study the enantiomers of the most interesting ligand 3b. Moreover, considering that the basic function of mAChR antagonists can also be a tertiary amine, 24 the racemic 33b and its enantiomers were included in this study.
As expected, the data reveal how the racemic tertiary amine (±)-33b shows high affinity for all mAChRs, though with pK i values slightly lower than those of the corresponding ammonium salt (±)-3b. Moreover, it maintains the interesting selectivity for M 3  . Such a configuration is the same of the eutomer (S)-(−)-2, suggesting that these derivatives bind to the same mAChR sites. The eudismic ratios (ERs) between the enantiomers of the tertiary amine are significantly higher than those between the corresponding enantiomers of the methiodide for all mAChR subtypes, especially for M 3 , for which the eutomer (2S,6S)-(−)-33b shows a pK i value 195fold higher than that of the distomer (2R,6R)-(+)-33b.
Docking Studies. To investigate the factors influencing the observed enantioselectivity of the three pairs of enantiomers (R)-2/(S)-2, (2R,6R)-3b/(2S,6S)-3b, and (2R,6R)-33b/ (2S,6S)-33b, docking simulations were carried out on the human M 3 mAChR structure in complex with a selective antagonist (PDB Id: 5ZHP). 52 Figure 10A, showing the Table 2. Equilibrium Binding Affinity of (±)-2, (±)-3b, (±)-33b, and Their Enantiomers pK i a Compd hM 1 hM 2 hM 3 hM 4 hM 5 Journal of Medicinal Chemistry pubs.acs.org/jmc Article putative complex as computed for compound (2S,6S)-3b, endowed with the highest affinity, reveals the following set of interactions: (a) the charged ammonium head is engaged by a set of contacts comprising the key ion-pair with Asp147 3.32 plus several charge transfer interactions with surrounding aromatic side chains (e.g., Tyr148 3.33 , Trp503 6.48 , Tyr506 6.51 , Tyr529 7.39 , and Tyr533 7.43 ); (b) the O4 dioxane atom is involved in a key H-bond with Asn507 6.52 , while the O1 atom is shielded by the close ammonium head and cannot elicit significant interactions; (c) the phenyl ring can stabilize π−π stacking interactions with a set of surrounding aromatic residues such as Tyr148 3.33 , Trp199 4.57 and Trp503 6.48 ; (d) the cyclohexyl ring is accommodated within a subpocket in which it can contact alkyl side chains such as Leu225 ECL2 , Ala235 5.43 , and Ala238 5. 46 . On these grounds, one may argue that the observed enantioselectivity can be ascribed to four moieties, the arrangement of which is influenced by the chiral centers: (a) the O4 dioxane atom, a feature which involves all three pairs of enantiomers; (b) the cyclohexyl and (c) the phenyl rings which concern only the compounds 3b and 33b; (d) the ammonium head which seems to play a marginal role for 2 and 3 reasonably due to the symmetry of the trimethyl ammonium group, while the need to properly arrange the proton toward Asp147 3.32 , and the N-methyl groups toward the aromatic residues, may impact on the enantioselectivity of 33b.
Hence, inspection of Figure 10B, comparing the computed poses for the two enantiomers of compound 2, reveals that they suitably and similarly arrange the phenyl rings and the ammonium head, while the pose of the dioxane ring differs in the two complexes. In detail, Figure 10B shows that the eutomer (S)-2 is able to establish a strong H-bond with Asn507 6.52 , while the distomer (R)-2 less suitably arranges the O4 atom (as defined by both N−H···O distance, 2.1 Å vs 2.7 Å, and corresponding angle, 171.5 vs 103.6, for (S)-2 and (R)-2, respectively) which, therefore, weakly contacts Asn507 6.52 .
Similarly, Figure 10C, comparing the best obtained complexes for the two enantiomers of compound 3b, shows that both of them are able to conveniently accommodate the dioxane ring (e.g., the N−H···O4 distance with Asn507 6.52 is equal to 2.3 Å in both complexes) and the ammonium head but unavoidably differ for the arrangements of the two rings in the 6-position. Indeed, while the eutomer (2S,6S)-3b properly accommodates the phenyl and the cyclohexyl rings as described above, the distomer (2R,6R)-3b is constrained to approach the phenyl ring toward the alkyl side chains with the cyclohexyl ring completely surrounded by aromatic residues. Notably, the capacity of both enantiomers of 3b to stabilize similar H-bonds with Asn507 6.52 suggests that the greater (despite always restricted) flexibility of the cyclohexyl ring with respect to the phenyl one allows the distomer (2R,6R)-3b to minimize the configurational effects on the pose of the dioxane ring.
Finally, Figure 10D, comparing the best poses as computed for the two enantiomers of 33b, highlights that they differ for the arrangement of both the O4 dioxane atom and the cyclohexyl/phenyl rings. In detail, while the eutomer (2S,6S)- Journal of Medicinal Chemistry pubs.acs.org/jmc Article 33b can elicit the key H-bond with Asn507 6.52 (N−H···O4 distance with Asn507 6.52 is equal to 2.6 Å) and to insert the cyclohexyl and phenyl rings within the suitable subpockets, the distomer (2R,6R)-33b cannot contact Asn507 6.52 (N−H···O4 distance with Asn507 6.52 is equal to 3.8 Å) and accommodates the two rings in the 6-position within the wrong subpockets. Notably, the unique difference between 3b and 33b involves the ammonium head which is a quaternary salt only in the former. Figure 10D indicates that both enantiomers of 33b are able to properly arrange the ammonium head even though the lack of the symmetric trimethyl group in 33b increases the relevance of the C2 configuration and can explain why the enantiomers of 33b are constrained to differ for the arrangement of the O4 dioxane atom, while both enantiomers of 3b are able to properly accommodate the dioxane ring by minimizing the effects of the C2 configuration. These observations find encouraging confirmations in the reported ERs, thus allowing for some meaningful considerations. First, the observed differences in the dioxane arrangement exert a conceivably greater impact on affinity compared to those in the cyclohexyl/phenyl rings as seen when comparing the ER values of 2 and 3b. Again, the combination of both factors (dioxane and cyclohexyl/phenyl rings) reveals a synergistic effect by showing an ER value for 33b markedly higher than the previous ones. Such a synergistic effect can be explained at an atomic level by considering that, while both enantiomers of 2 are able to stabilize the H-bond with Asn507 6.52 even though the distomer elicits weaker interactions (as seen in the reported geometrical parameters), the 33b distomer is substantially unable to approach Asn507 6.52 , thus missing this key interaction. Finally, similar trends can also be seen when analyzing the corresponding affinity values and, in particular, the affinities of the distomers. Indeed, while the eutomers show comparable affinity values with 3b and 33b which reveal slightly higher values probably due to the favorable hydrophobic interaction stabilized by the cyclohexyl ring, the distomers show greater differences in affinity which are ascribable to their reduced interactions. Hence, (2R,6R)-3b which only fails in properly arranging the rings in the 6position reveals the greatest affinity, followed by (R)-2 which elicits a weak H-bond with Asn507 6.52 . The lowest affinity is shown by (2R,6R)-33b, which does not stabilize the mentioned H-bond and unsuitably arranges the rings in the 6-position.
For completeness and even though the affinity values of the single enantiomers were not measured, docking simulations also involved other proposed derivatives by focusing attention on those with pK i values on M 3 mAChR greater than 6. While avoiding systematic analyses, the docking results allow for some general considerations. The lower affinity values of the ligands bearing cyclohexyl/phenyl rings in 5 (instead of 6, e.g., 10, 11a, and 11b) can be ascribed to the steric hindrance exerted by these rings on the O4 dioxane atom which weakens the key H-bond with Asn507 6.52 . In contrast, the reduced steric hindrance exerted on the O1 dioxane atom allows this to be engaged in additional H-bonds as seen for (2S,6S)-11a with Tyr148 3.33 . The low affinity of ligands bearing a 4-(phenylthio)phenyl moiety (15a and 15b) and similar diphenyl groups is explainable by considering that these bulky substituents constrain the ligands to assume inconvenient poses, where even the ammonium head assumes suboptimal arrangements, without adding any additional contacts. Finally, the lower affinity values of the ligands with substituents in both 5-and 6-positions (e.g., 18c) is ascribable to the same factors affecting the binding of compounds substituted only in 5, namely, the greater steric hindrance on the O4 dioxane atom which weakens the H-bond with Asn507 6.52 .
Functional Studies on MSCs from Mouse Bone Marrow. It is well established that bone marrow MSC behavior is influenced by a variety of signaling systems. In context, cholinergic intramural stimuli and, in particular, muscarinic signals orchestrate MSCs viability and commitment. 11,12 Considering also the pluripotent MSC nature and their contribution to bone, blood, and systemic homeostasis, 53 viability studies on MSCs were performed to determine the functional profile of 3b, the most interesting compound in this series. Namely, the effect of this compound was similar to that of the well-known mAChR antagonist atropine because it was able to down-regulate MSCs viability when used at high concentration (10 −4 M), while increased cell viability when used at low concentration (10 −10 M) ( Figure 11A). Successively, the efficacy of 3b in contrasting cell viability induced by the well-known mAChR agonist carbachol was evaluated. The data reported in Figure 11B indicate that, analogous to atropine, the new compound 3b is able to contrast the increase of carbachol-induced MSC viability,

■ CONCLUSIONS
In the present study, the 6,6-diphenyl structural element of the potent mAChR antagonist 2 was replaced by lipophilic substituents in 5-and/or 6-position of the 1,4-dioxane nucleus. Among the novel compounds, the 6-cyclohexyl-6-phenyl derivative 3b, with a cis configuration between the CH 2 N + (CH 3 ) 3 chain in the 2-position and the cyclohexyl ring in the 6-position, showed pK i values for all mAChR subtypes, except for M 2 , higher than those of 2. Moreover, its selectivity profile is similar to that of the therapeutically used drug oxybutynin, with pK i values for M 1 , M 3 , and M 4 subtypes higher than those for M 2 and M 5 subtypes. The study of the enantiomers of 3b and those of the corresponding tertiary amine 33b, whose absolute configuration was determined by quantum mechanical simulations of ECD, provided useful information about the role played by chirality in the interaction with mAChRs. In particular, the absolute configuration of the carbon atom in the 2-position of the eutomers (2S,6S)-(−)-3b and (2S,6S)-(−)-33b is the same as (S)-(−)-2, suggesting that these derivatives bind to the same mAChR sites. The ERs between the enantiomers of the tertiary amine 33b proved to be higher than those between the corresponding enantiomers of methiodide 3b for all mAChR subtypes, especially for M 3 . Docking studies on the M 3 mAChR-resolved structure allowed us to shed light on the binding mode of the proposed compounds. In particular, while the enantiomers of 33b differ for the arrangement of O4 dioxane atom, both enantiomers of 3b are able to properly accommodate the dioxane ring by minimizing the effect of the C2 configuration. Finally, the assays on MSCs from mouse bone marrow showed for 3b a functional profile similar to that of the mAChR antagonist atropine concerning both the dose−response effect produced on the metabolic activity of viable MSCs and the effect in contrasting the increase of carbachol-induced MSC viability.
Compared to the tertiary amine drugs clinically used for the treatment of OAB, 3b presents a quaternary ammonium function that should prevent the crossing of BBB, minimizing central anticholinergic activity and, therefore, limiting CNS side effects. The prediction by SwissADME that 3b is a potential P-gp substrate makes the profile of such a compound more and more interesting. 54 Not to mention that the transformation into a quaternary amine markedly enhances the metabolic stability of this compound. Indeed, the metabolic prediction based on the similarity analysis using the MetaQSAR database on the tertiary amine indicates the oxidation in alpha to the N atom as a truly probable metabolic reaction which is largely inhibited by the presence of a permanent positive charge. 55 Moreover, the M 3 /M 2 selectivity ratio of 3b (14.5), which is significantly higher than those of the quaternary ammonium compounds 2 and trospium (1.6 and 1.1, respectively), might limit cardiovascular side effects. Therefore, the methiodide 3b might represent a valuable lead compound for the design of novel antagonists potentially useful in peripheral diseases in which M 3 receptors are involved.
Binding Studies. Cell Culture and Membrane Preparation. CHO-K1 cells stably transfected with the human muscarinic receptor subtypes (hM 1−5 ) were grown in Dulbecco's modified Eagle's medium (DMEM) with nutrient mixture F12 (DMEM/F12, 50/ 50), containing 10% fetal bovine serum, penicillin (100 U/mL), streptomycin (100 U/mL), L-glutamine (4 mM), and geneticin (G-418, 50 μg/mL) at 37°C in a 5% CO 2 humidified incubator. In order to harvest the cells, the culture medium was removed;the cells were washed with PBS and then trypsinized by trypsin−EDTA treatment for 2−3 min. Serum (0.7 mL) was added to inactivate the trypsin, and the cells were spun down by centrifuging at 300g for 5 min. The cells were then resuspended in ice-cold 25 mM sodium phosphate buffer containing 5 mM MgCl 2 , pH 7.4 (binding buffer) and homogenized using a cell disrupter (Ultra-Turrax, setting 3, 30 s). The homogenate was sedimented by centrifugation (17,000g, 15 min). The supernatant was discarded, and the resulting membrane pellets were resuspended with Ultra-Turrax in the same buffer to give a final protein concentration of 1−2 mg/mL. The protein content was determined by the method of Bradford (1976) with bovine serum albumin (Sigma) as a standard and stored at −80°C.
Inhibition Radioligand Binding Assay. Inhibition radioligand binding assays were conducted as previously described 48,49 with 0.2 nM [ 3 H]NMS in binding buffer in a final volume of 250 μL. Nonspecific binding was defined in the presence of 10 μM atropine. Briefly, membrane fractions (about 25−70 μg/mL of protein) were incubated with radioligand and unlabeled test compounds for 2 h at r.t. Bound and free radioactivity were separated by filtering the assay mixture through UniFilter GF/B plates using a FilterMate Cell Harvester (PerkinElmer Life and Analytical Science). The filter bound radioactivity was counted by a TopCount NXT Microplate Scintillation Counter (PerkinElmer Life and Analytical Science). Data (cpm) were normalized to percentage-specific binding and analyzed using a four-parameter logistic equation in GraphPad Prism 5.02; IC 50 values were determined, and K i values were calculated. 59 The values reported in Tables 1 and 2 represent the arithmetic mean ± S.E.M. of at least three independent experiments, each one performed in duplicate.
Docking Studies. Docking simulations involved the ligands with pK i values on M 3 mAChR greater than 6 and the recently resolved M 3 mAChR structure in complex with a selective antagonist (PDB Id: 5ZHP). 52 The protein structure was completed by adding hydrogen atoms, and the ionizable groups were set to be compatible to physiological pH using the VEGA suite of programs. 60 The prepared structure was finally minimized by using the NAMD program 61 and keeping fixed the backbone atoms to retain the experimental folding. The structure of the considered ligands was optimized by PM7-based semi-empirical calculations. 62 Docking simulations were performed by PLANTS 63 by focusing the searches within a 8.0 Å radius around the bound resolved antagonist. The simulations were carried out using the ChemPLP primary score with speed equal to 1 and 10 poses were generated for each ligand. The obtained complexes were optimized by using NAMD and by keeping fixed all atoms outside a 10 Å radius sphere around the docked ligand and then rescored by ReScore+. 64 Functional Studies on MSCs from Mouse Bone Marrow. MSC Collection and Culture. The in vitro studies were performed by using bone marrow MSCs as a model. Male BALB/c mice (Harlan Italy SrL, Milano, Italy) (8 weeks old; body weight ∼24.5 g; n = 4) were kept in a laminar-flow cage in a standardized environmental condition. Food (Harlan, Italy), and water was supplied ad libitum. Mice were sacrificed by CO 2 narcosis and cervical dislocation in accordance with the recommendations of the Italian Ethical Committee and under the supervision of authorized investigators. Long bones (femurs and tibiae) were dissected and cleaned from skin, muscle, and connective tissues as much as possible. Bones were placed in a culture dish containing sterile PBS. Then, the bone cavity was flushed in DMEM with a syringe in order to collect the bone marrow cells into a 50 mL sterile tube. The procedure was repeated until all marrow was removed. Cell suspension was filtered through a cell strainer (70 μm size) to remove cell clumps or bone debris. Then, bone marrow cells were plated in 100 mm culture dishes in DMEM containing 10% heat-inactivated-fetal calf serum (HIFCS), penicillin, and streptomycin. In order to obtain a population of bone marrow MSCs, the protocol by Solimani and Nadri 65 was followed. Cells were incubated at 37°C with 5% CO 2 in a humidified chamber. After 3 h, the nonadherent cells that accumulate on the surface of the dish were removed by changing the medium and replacing it with a fresh complete medium. After 8 h of culture, the medium was further replaced with fresh complete medium. The last step was repeated every 8 h for up to 72 h of initial culture. Then, the adherent cells were washed with sterile PBS and added with a fresh medium every 3−4 days. After 2 weeks of initiating culture, cells were washed with PBS, detached by trypsinization, counted, and plated at the density of 5,000 cells/well in 96 culture plates (Costar Corp., Milano, Italy) in DMEM containing 10% HIFCS, penicillin, and streptomycin.
Experimental Protocol. MSCs were treated with compound 3b (from 10 −4 to 10 −10 M) for 24 h. Control cultures were performed by incubating the cells with the only vehicle (DMSO) or by untreated cells. Parallel other cultures were incubated with 3b from 10 −4 to 10 −10 M for 1 h, and then, the culture medium was replaced with a fresh medium. The MSCs were maintained in the presence of carbachol at 10 −10 M for 24 h. At the end of each procedure, the MSCs viability was measured by MTS assay. Specifically, cells were incubated with Cell Titer 96 Aqueous One Solution Reagent (Promega Italia, Milano, Italy) for 2 h in a humidified 5% CO 2 atmosphere. The quantity of the formazan product was directly proportional to the number of living cells in culture. The colored formazan was measured by reading the absorbance at 490 nm using a 6-well plate reader.