Synthesis, Biological Evaluation, and Docking Studies of Antagonistic Hydroxylated Arecaidine Esters Targeting mAChRs

The muscarinic acetylcholine receptor family is a highly sought-after target in drug and molecular imaging discovery efforts aimed at neurological disorders. Hampered by the structural similarity of the five subtypes’ orthosteric binding pockets, these efforts largely failed to deliver subtype-selective ligands. Building on our recent successes with arecaidine-derived ligands targeting M1, herein we report the synthesis of a related series of 11 hydroxylated arecaidine esters. Their physicochemical property profiles, expressed in terms of their computationally calculated CNS MPO scores and HPLC-logD values, point towards blood–brain barrier permeability. By means of a competitive radioligand binding assay, the binding affinity values towards each of the individual human mAChR subtypes hM1–hM5 were determined. The most promising compound of this series 17b was shown to have a binding constant towards hM1 in the single-digit nanomolar region (5.5 nM). Similar to our previously reported arecaidine-derived esters, the entire series was shown to act as hM1R antagonists in a calcium flux assay. Overall, this study greatly expanded our understanding of this recurring scaffolds’ structure–activity relationship and will guide the development towards highly selective mAChRs ligands.


Introduction
Muscarinic acetylcholine receptor (mAChR) M 1 , as the most prevalent subtype and in accordance with its abundant expression in all major forebrain areas, has been implicated in the pathophysiology of various neurological diseases such as Alzheimer's disease, Parkinson's disease, and schizophrenia [1]. Hence, the muscarinic receptor subtype M 1 has emerged as an attractive drug target for the treatment, among others, for the aforementioned disorders [2][3][4]. In spite of the spurred interest in this receptor, success stories of the development of therapeutic or diagnostic agents targeting M 1 are lacking. One major hurdle to overcome is the high degree of sequence homology of the orthosteric binding site among the five subtypes (M 1 -M 5 ) of mAChRs [5], rendering subtype-specific ligand design a challenging task. As a result, candidates entering the clinical stage are commonly plagued by dose-limiting side effects due to poor subtype selectivity profile [6]. Despite the vast body of research addressing muscarinic signaling in the central nervous system (CNS), our current understanding is still limited [7]. Positron emission tomography (PET), a molecular imaging technique allowing extensive, noninvasive studies in humans, has tremendous potential to not only advance this understanding but also to guide the development of novel brain-penetrable subtype-selective therapeutics targeting mAChRs with the possibility to quantify target engagement and occupancy [8,9].
As a continuation of our group's research efforts towards the development of subtype selective CNS PET radiotracers for the mAChRs, we recently have shown the benzhydrol ester of arecaidine benzhydryl 1-methyl-1,2,5,6-tetrahydropyridine-3-carboxylate (DPMA, Figure 1) to display an affinity and subtype selectivity profile towards M 1 theoretically suited for such applications [10]. Unfortunately, the excessive extent of nondisplaceable binding (NDB) renders DPMA unsuitable for PET applications. To potentially overcome this hurdle, we replaced the benzhydrol motif with a hydrobenzoin substituent, leading to slightly inferior binding properties but markedly lower lipophilicity [11]. Building on this prior research, herein we report our most recent efforts to refine our structure-activity relationship (SAR) understanding of arecaidine-derived esters. Spurred by the promising hydrobenzoin esters, we sought to synthesize and biologically evaluate a range of systematically hydroxylated arecaidine esters and subsequently study their physicochemical and thermodynamic binding properties towards the five subtypes of mAChRs. In a last step, we sought to computationally evaluate selected ligands from this series in a set of docking studies to identify whether the hydroxylated motifs were able to capture additional hydrogen bonding interactions in the orthosteric binding site.

Chemistry
Before diving into the specifics of the followed synthetic sequences, the esterification step shall be put into the spotlight. Previous studies from our lab have shown CDI to be a suitable coupling agent for the esterification of arecaidine with bulky alcohols such as benzhydrol or hydrobenzoin albeit suffering from rather poor yields [10,11]. Hence, we sought to address this issue. Gratifyingly, there was no need for lengthy optimization efforts, since we found that treating a mixture of arecaidine and 4,4 -difluorobenzhydrol under Steglich conditions using EDC-HCl furnished 4-FBA with a 74% yield compared to the 9% yield using the CDI-mediated procedure [10]. Equipped with a promising esterification protocol, we focused our attention on the synthesis of the desired hydroxylated arecaidine esters. The reaction sequences followed for their synthesis are outlined in Schemes 1-5. For some of the compounds, the presence of a hydroxy or hydroxymethyl leads to the loss of a plane of symmetry in comparison to 4-FBA, and thus induces chirality. In this work, however, only racemates were synthesized to speed up subsequent preliminary biological testing.    The hydroxymethyl group-bearing benzyl esters 8a-c were derived in good yields (73-83%) from arecaidine and their commercially available benzenedimethanol precursors following the above-mentioned Steglich esterification using EDC-HCl as coupling reagent and 4-DMAP as nucleophilic catalyst (Scheme 1). The synthesis of the benzhydrol esters 10a-c followed a slightly longer synthetic sequence. Following quantitative TBS protection, 3-and 4-bromobenzyl alcohol (1b and 1c) underwent lithium-halogen exchange upon treatment with n-BuLi. Quenching the lithium species with benzaldehyde furnished the desired hydroxymethyl-substituted benzhydrols 6b and 6c in excellent yields (95 and 97%). The ortho-substituted benzhydrol 6a, however, could not be obtained via this route. Treating 2a with n-BuLi led to a complex mixture containing a newly formed silicon species. While we were not successful in characterizing this species, we presume some kind of retro-Brook rearrangement to take place upon lithiating 2a. Thus, aldehyde 5 was prepared from 1,2-benzenedimethanol (3) via TBS protection and subsequent benzylic oxidation with activated manganese dioxide (87% over two steps) [12]. Reaction with PhMgBr as a nucleophilic phenyl source then gave the desired benzhydrol 6a with a 94% yield. A Steglich esterification followed by a TBAF-mediated TBS removal provided the sought for hydroxylated arecaidine esters 10a-c (64-68% over two steps). In case of the ortho hydroxymethyl-substituted ester 10a, multiple purification attempts, including preparative HPLC and the attempt at forming a hydrochloride salt, were insufficient to increase the compound's purity above 95%. Hence, this compound was excluded from further biological evaluations.
The TBS-protected hydroxybenzhydrols 13a-c were synthesized in a similar fashion as 6a (Scheme 2), starting with the silylation of the corresponding hydroxybenzaldehydes 11a-c (97-99%) followed by phenylation with PhMgBr (93-96%). The silyl-ether-protected hydroxybenzaldehydes 12a-c were additionally reduced with sodium borohydride to produce the TBS-protected hydroxybenzyl alcohols 14a-c with a 96-98% yield. The esterification of 13a-c as well as 14a-c employing the previously highlighted conditions smoothly furnished the arecaidine esters 15a-c (78-86%) and 16a-c (80-89%), respectively. While subsequent desilylation of meta-hydroxylated 15b and 16b led to the desired free alcohols 17b (90%) and 18b (82%), respectively, the orthoand para-substituted derivatives decomposed upon treatment with TBAF. In accordance with our findings, Fang et al. have reported the 4-OTBS benzyl group to be a fluoride-labile protective group for carboxylic acids [13]. That the constitutional isomer 2-OTBS benzyl as well as the structurally similar 2-and 4-OTBS benzhydrol groups share the same fluoride lability is mechanistically plausible, since all of them meet the electronic requirements for a 1,4-or 1,6-elimination under liberation of an intermediary orthoor para-quinone methide. Different attempts at synthesizing 17a, 17c, 18a, and 18c, such as buffering TBAF with acetic acid, switching to other fluoride sources, using acid-labile protective groups or foregoing the use of a protective group altogether, were not fruitful (see Scheme S1 for details).
Compound 25 was synthesized from diol 24, which itself was prepared from methyl benzilate (23) via a hydride reduction using NaBH 4 (67% over three steps; Scheme 4). Despite being an unsymmetrical diol, 24 did not need protection for the esterification step since the steric hindrance of the benzhydrolic hydroxy group rendered it largely unreactive as compared to the primary alcohol; hence, only the desired ester was isolated.

Physicochemical Properties
Considering the underlying rationale for the synthesis of the present set of hydroxylated arecaidine esters-the development of brain-penetrable mAChRs ligands for the potential application as PET probes-we decided to investigate the important physicochemical properties governing blood-brain barrier (BBB) permeability. On top of its significant role in mediating brain penetration, lipophilicity, frequently approximated by logD measurements, is an important metric in the context of PET tracer development as it allows for the preliminary assessment of the extent of NDB to be expected. While the correlation of logD and NDB is weak at best [14,15], certain hard lipophilicity cutoffs, e.g., shake-flask logD < 3.5, continue to be successfully employed in PET tracer development [16]. However, the design of brain-penetrable PET tracers displaying low NDB is a multiparameter optimization effort: relying too much on individual physicochemical parameters may be misleading since it does not give a full picture. To address this multifariousness, which is equally important in the context of classical drug design, a group of medicinal chemists around Travis Wager from Pfizer developed the CNS MPO (multiparameter optimization) score [17]. This algorithm takes into account six fundamental physicochemical properties (clogP, clogD, MW (molecular weight), tPSA (topological polar surface area), HBD (hydrogen bond donor count), and pKa) and gives a numeric output between 0 and 6.0 while avoiding hard cutoffs. The validity of the CNS MPO score was underpinned by a study correlating increasing scores with increasing experimentally determined unbound fractions in the brain [18]. Hence, we chose to use this score to assess the present compound series and compare the newly designed compounds with the parent molecule DPMA (Table 1). Overall, the CNS MPO scores of all hydroxylated arecaidine esters fall within a narrow range of 4.83-5.75 skewed towards the upper end of the scale. According to a recent study, compounds with CNS MPO scores >3 have a lower probability of high NDB, which as such is a first indication for potential PET suitability [20]. While the parent molecule DPMA met this criterion as well and still exhibited excessive NDB in autoradiography studies, the entirety of the present compounds, in fact, had a higher score than DPMA.
Since we did not want to solely rely on the computational realm for the analysis of the synthesized compounds' physicochemical property profiles, we experimentally approximated their lipophilicities. For this, logD values were estimated in an HPLC-based high-throughput assay employing an octadecyl-poly(vinyl alcohol) stationary phase [21]. The absence of silanol groups renders this type of column especially suited in case of analytes with hydrogen bond donor capabilities as it avoids overestimating their lipophilicity [22,23]. Notably, all compounds of this series showed a distinct decrease in HPLC-logD in comparison to its parent compound DPMA [10], with the majority falling within a range (1.2-3.1) used as a rule of thumb to develop centrally acting drugs [24]. Compounds 8a-c and 22, on the other hand, were borderline cases with HPLC-logD values of roughly one or slightly below.
Taken together, the computationally predicted CNS MPO score as well as the experimentally determined HPLC-logD render the physicochemical property profile of the hydroxylated arecaidine esters potentially suitable as imaging probes in PET applications.

Biological Evaluation
The arecaidine esters' affinities towards the human mAChR subtypes hM 1 -hM 5 were studied in a competitive radioligand binding assay using CHO-K1 cell membrane homogenates expressing a single mAChR subtype and tritiated N-methyl scopolamine methyl chloride ([ 3 H]NMS). To speed up our affinity testing regime, we subjected all test substances to preliminary single-concentration displacement assays, as described previously [25]. Compounds with percent displacements below 75% at any of the subtypes were dropped from further investigations (Table S1). For compounds meeting the percent displacement threshold (8c, 10b, 10c, 17b, 18b, and 25), inhibition constants (K i ) were assessed in dosedependent displacement experiments ( Table 2). The three benzhydrolic esters 10b, 10c, and 17b displayed the highest affinities of the present set of arecaidine esters towards hM 1 R with K i values in the single-and double-digit nanomolar range. Attaching a hydroxy group to one of DPMA's phenyl rings, however, did not increase the affinity towards any of the muscarinic subtypes. Compound 17b, the mono meta-hydroxylated version of DPMA, almost matched DPMA's affinity towards hM 1 R but the selectivity over the other subtypes was inferior to varying degrees. While its selectivity against hM 2 R was, being over 30-fold, just slightly below that of DPMA, the added hydroxy group led to an almost fivefold decreased selectivity against hM 3 R. Compound 10c followed the selectivity trend observed for 17b: decreasing affinities in the order hM 1 R > hM 5 R > hM 4 R > hM 3 R > hM 2 R. However, the actual affinity values of the para-hydroxymethyl bearing compound 10c were roughly 4-13× higher, leading to lower selectivity levels. Its meta-hydroxymethyl-substituted isomer 10b followed a slightly different selectivity trend, displaying its second highest affinity towards hM 4 R instead of hM 5 R. Aligning with our previously published studies of structurally related ligands targeting mAChRs, all of the herein-evaluated compounds exhibited the highest binding constant for hM 2 R. The selectivity trend of 18b was especially noteworthy-mentally detaching one phenyl substituent from 17b and arriving at 18b leads to a unique to this series, significant hM 5preferrring tendency which considerably differs from 17b's selectivity profile.
As a prerequisite for our planned functionality testing efforts of the herein-presented hydroxylated arecaidine esters, we assayed the compounds' cell viability in living CHO-hM 1 cells using MTT. Gratifyingly, we found cytotoxicity to be of no concern for our purposes with the corresponding IC 50 values being in the double-digit micromolar range and above ( Figure S1). Having ruled out any interference by the esters' hypothetical cytotoxicity, we sought to explore the functionality of the synthesized arecaidine esters towards the muscarinic receptor subtype M 1 by discriminating between agonistic and antagonistic behavior. As reported previously [11], by using Fluo-4, the calcium mobilization in CHO-hM 1 cells was assayed in response to the stimulation with the hydroxylated arecaidine esters. At none of the assayed concentrations did any of the tested esters induce a significant calcium flux, as can be observed from known agonist carbachol (Figures 2a and S2a). Yet, all of them were able to inhibit a carbachol-induced flux in a similar fashion as the known antagonist scopolamine (Figures 2b and S2b), thereby indicating the antagonistic binding of the entire set of assayed compounds.

Molecular Docking Studies
We performed molecular docking studies to better our understanding of the observed binding affinities in atomic detail. Due to the high degree of sequence homology between the muscarinic subtypes, docking alone frequently does not meet the requirements to predict/explain subtype selectivity, hence the herein-presented docking studies are focused on giving qualitative insights into the compounds' binding mode against the M 1 muscarinic receptor structure (PDB 5CXV). As a first step, for each of the docked ligands a representative pose was chosen among the nine modes returned. Hence, following our previously presented approach, in each case we selected the highest ranked pose featuring a salt bridge between the positively charged N-methyl amine and the-among aminergic GPCRs-conserved-aspartic acid residue Asp105 3.32 (superscript numerals refer to the Ballesteros-Weinstein numbering scheme for GPCRs [26,27]). Since none of the predicted modes of ester 8c showed the aforementioned interaction, simply the highest ranked pose was chosen. The 2D pharmacophores corresponding to the six fully, biologically evaluated ligands 8c, 10b, 10c, 17b, 18b, and 25 are highlighted in Figure 3, while those corresponding to the ligands 10a, 17a, 17c, 18a, and 18c whose syntheses were not successful are shown in Figure S3. Unlike benzylic ester 8c, which is predicted to adopt a pose merely engaging in hydrogen bonding with Ser109 3.36 paired with phenyl-based hydrophobic interactions (Figure 3a), the cationic head of its closely structurally related ester 18b is anchored in proximity to Asp105 3.22 (Figure 3e). However, the positively charged group also captures interactions with Tyr404 7.39 and Trp378 6.48 -a feature which is shared with compound 25 (Figure 3f). In addition, the carbonyl oxygen of 18b serves as a hydrogen bond acceptor for both Cys407 7.42 and Asn382 6.52 . Noteworthy, with the exception of 8b, the hydrogen bonding interaction with Cys407 7.42 is shared among all docked compounds, and has been observed in previous studies with structurally similar compounds [10,11,25]. Taken together, the sum of these pharmacophoric features explains 18b's higher affinity towards hM 1 R compared to 8c. Compound 25 also engages in hydrogen bonding with Asn382 6.52 , however, it does so via its hydroxy group, which serves as a hydrogen bond donor (Figure 3f). The benzhydrolic esters 10b, 10c, and 17b are of particular interest since they showed the highest affinities of all compounds of this study towards hM 1 R. Importantly, these compounds bear a chiral center, and for the sake of simplicity only the (R)-enantiomers' 2D pharmacophores are depicted in Figure 3b-d; for a depiction of the (S)-enantiomers see Figure S4). Evidently, the interactions with the orthosteric binding site for the arecaidinepart of the molecules are predicted to be identical, which holds true for the (S)-enantiomers as well-with the exception that the ammonium groups of (S)-10b and (S)-17b not only interact with Asp105 3.32 but also with Tyr404 7.39 ( Figure S5). The two phenyl rings fill two lipophilic pockets, hydrophobically engaging with similar amino acid residues as are characteristic for the receptor's cocrystallized inverse agonist tiotropium [28]. Neither the (R)-nor the (S)-enantiomer of 10c exhibits noteworthy interactions between its hydroxy groups and the receptor, which might explain the somewhat lower binding affinity towards hM 1 R compared to 10b and 17b. The phenolic hydroxy group of 17b and the benzylic hydroxy group of 10b, however, act both as a hydrogen bond acceptor and donor towards different amino acid residues (Figure 3b,d). Overlaying the poses of these two compounds reveals their very close resemblance ( Figure 4). In fact, merely the two hydroxy groups adapt slightly different positions and as a result, they both feature a mutual hydrogen bonding interaction with Tyr381 6.51 as well as separate interactions with Tyr106 3.33 and Thr189 5.39 , respectively. The hydroxy groups of the corresponding (S)-enantiomers point in the opposite lipophilic pocket. While the polarized hydrogen of the hydroxymethyl group of (S)-10b reaches the backbone carbonyl oxygen of Tyr106 3.33 and subsequently undergoes a hydrogen bond, (S)-17b's phenolic hydroxy group falls short of capturing any meaningful interactions. Due to the fact that 18b was the only compound of this series showing a significant preference for the hM 5 R in affinity evaluations, we felt compelled to compare this ligand's interactions with the orthosteric binding site of M 1 and M 5 . A side-by-side comparison of the 2D representations of the key interactions corresponding to the highest ranked docking pose in each case is shown in Figure 5. The interactions with the binding site of M 1 (Figure 5a) have mostly been discussed above-with the difference that 18b is only able to fill one of the binding site's lipophilic pockets and its hydrogen-bond-donating role towards Ala196 5.46 , its binding mode shows close resemblance with that of tiotropium. As opposed to this, none of the nine sampled binding modes of 18b feature an ionic interaction between the ester's cationic head and the aspartic acid residue Asp110 3.32 in the binding site of M 5 . Instead, the highest ranked pose is predicted to adopt a conformation in which its phenyl ring is placed such that the phenolic hydroxy group engages in hydrogen-bonding with Asp110 3.32 and Ser114 3.36 . The tetrahydropyridine-part of the molecule, on the other hand, does not seem to capture any meaningful interactions with the receptor. While this docking experiment gives first insights into the differing interactions in the M 1 and M 5 orthosteric binding site, it does not explain the almost threefold preference of 18b for hM 5 R over hM 1 R in in vitro experiments. For this, more sophisticated computational studies will be necessary.

General Considerations
Unless otherwise stated, all reagents were purchased from commercial suppliers and used as received without further purification. All reactions were conducted under an inert atmosphere of argon, and commercially available anhydrous solvents were used. Flash column chromatography was performed on a Biotage ® Selekt Flash Chromatography System equipped with Biotage ® Sfär HC cartridges using either HPLC-grade or reagent-grade solvents. Reactions were monitored by TLC on precoated aluminum sheets (Polygram SIL G/UV254, 0.2 mm, with fluorescent indicator; Macherey-Nagel, Düren, Germany); the spots were visualized under UV light (λ = 254 nm) and/or KMnO 4 stain. 1 H and 13 C NMR spectra were recorded in deuterated chloroform (CDCl 3 ) at 298 K on a Bruker Avance III 400 or Bruker Avance III 500 spectrometer and are reported as follows: chemical shift δ in ppm (multiplicity, coupling constant J in Hz, number of protons, assignment) for 1 H NMR spectra and chemical shift δ in ppm (assignment) for 13 C spectra. For 1 H and 13 C NMR spectra, the residual solvent peaks of CDCl 3 (δ H = 7.26 ppm, δ C = 77.00 ppm) were used as internal reference. The chemical shifts of all signals are reported as the center of the resonance range. Unless stated otherwise, a full and unambiguous assignment of all resonances was performed by a combination of standard NMR techniques, such as APT, HSQC, HMBC, COSY, and NOESY experiments. IR spectra were recorded on a Bruker Alpha II FTIR spectrometer. Samples were prepared as a film by evaporation of a solution in CH 2 Cl 2 or CDCl 3 and selected absorption bands are reported in wavenumbers (cm −1 ). HRMS spectra were recorded on a Bruker maXis 4G instrument (ESI-TOF). Melting points were measured with an Electrothermal IA9200 melting point apparatus in open glass capillaries and are uncorrected. All biologically tested compounds exhibited ≥95% purity under the HPLC conditions reported hereafter. HPLC analyses were performed using a Shimadzu HPLC system consisting of a degassing unit (DGU-20A3R), a liquid chromatograph (LC-20ADXR), an autosampler (SIL-20ACHT), a diode array detector (SPD-M20A), a column oven (CTO-20AC) and a communication bus module (CBM-20A). An Eclipse Plus column (4.6 × 100 mm, 3.5 µm, Agilent, Santa Clara, CA, USA) and a gradient consisting of the following components were used: solvent A: 0.1% TFA in double distilled water; solvent B: 0.1% TFA in acetonitrile. Purities were measured with a gradient run by increasing solvent B from 10 to 100% within 9.4 min with a flow rate of 1.5 mL/min as well as with an isocratic run ( Figures S6-S15). Intermediates 2a−b [29,30], 4 [12], 5 [12], 6b [31], 12a−c [32][33][34], 13a [32], 13c [35], 14a−c [32,34,36], 20 [37], and 24 [38] have been described previously, and were synthesized following the steps and conditions outlined in Schemes 1-4. Arecaidine was synthesized according to a previously reported procedure [39]. (6a, 6c, 13b) (2-(((tert-butyldimethylsilyl)oxy)methyl)phenyl)(phenyl)methanol (6a). To an icecooled solution of aldehyde 5 (300 mg, 1.20 mmol, 1.0 equiv) in anhydrous THF (6 mL) was added dropwise a solution of PhMgBr (1.0 M in THF, 1.3 mL, 1.1 equiv) over a period of 10 min. It was allowed to warm slowly to ambient temperature and was stirred overnight. Then, the mixture was quenched by the addition of water and was extracted with EtOAc (3×). The combined organic layers were dried (Na 2 SO 4 ) and concentrated under reduced pressure. Purification by flash column chromatography (0-20% EtOAc in n-heptane) afforded the title compound 6a (368 mg, 94%) as a colorless oil. 1  (3-((tert-butyldimethylsilyl)oxy)phenyl)(phenyl)methanol (13b). To an ice-cooled solution of aldehyde 12b (284 mg, 1.20 mmol, 1.0 equiv) in anhydrous THF (6 mL) was added dropwise a solution of PhMgBr (1.0 M in THF, 1.3 mL, 1.1 equiv) over a period of 10 min. It was allowed to warm slowly to ambient temperature and was stirred overnight. Then, the mixture was quenched by the addition of water and was extracted with EtOAc (3×). The combined organic layers were dried (Na 2 SO 4 ) and concentrated under reduced pressure. To a heterogeneous mixture of arecaidine (1.0 equiv) and alcohol (1.5 equiv) in anhydrous CH 2 Cl 2 (0.5 M) were added EDC-HCl (2.0 equiv) and 4-DMAP (0.5 equiv). The resulting mixture was vigorously stirred for 2 d. Then, the turbid mixture was poured into sat. aq. NaHCO 3 and extracted with CH 2 Cl 2 (3×). The combined organic layers were dried (Na 2 SO 4 ) and concentrated under reduced pressure. The crude residue was purified by flash column chromatography to give the desired ester.

High-Throughput HPLC-logD
High-throughput HPLC-logD values were determined as published previously using the HPLC system described above equipped with an apHERA C18 column (10 × 6 mm, 5 µm, Supelco, Bellefonte, PA, USA) [21,40]. Briefly, a mixture of toluene (≥98%, Sigma-Aldrich, St. Louis, MO, USA) and triphenylene (≥99.9%, Carl Roth, Karlsruhe, Germany) was used as internal standard. Samples were dissolved in the internal standard mixture. Using gradient elution, the injection volume was set to 7 µL, the flow rate was 1.5 mL/min, and the mobile phase consisted of a mixture of methanol and 0.01 M sodium phosphate buffer pH 7.4. The HPLC-logD values were derived from the measured retention times following the previously published equation [21,40].

Biological Evaluation
For the biological evaluation of the herein synthesized compounds, we essentially followed our previously described workflow and steps [25]

Cell Viability (MTT Assay)
CHO-hM 1 cells were harvested from culture flasks by trypsinization and seeded into 96-well microculture plates (Corning ® , Corning, NY, USA) in densities of 4000 cells/well (100 µL/well). After a 24 h preincubation, cells were exposed in triplicates for each concentration level to dilutions of the test compounds in complete culture medium (100 µL/well) for 72 h. At the end of the exposure period, the compound solutions were replaced with 100 µL of nonsupplemented RPMI 1640 medium and 3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazoliumbromid (MTT reagent in PBS, 5 mg/mL) mixed in a 6:1 ratio. After incubation for 4 h, the medium was removed, and the formazan product was solved in DMSO (100 µL/well). Optical densities at 490 nm were measured with a microplate reader (Tecan Infinite ® 200 PRO, Männedorf, Switzerland) using a reference wavelength of 690 nm to correct for unspecific absorption. The quantity of viable cells was normalized to untreated controls.

Radioligand Binding Experiments
Cell membranes bearing hM 1 -hM 5 receptors were prepared as described previously [10]. Briefly, stably transfected CHO-K1 cells were grown to at least 80% confluency in T175 flasks, washed with ice-cold DPBS, and scraped into a mixture of ice-cold 2 mL 10 mM Tris-HCl, 1 mM EDTA-buffer, pH 7.4 and 200 µL protease inhibitor. A cell homogenate was prepared by passing the cell suspension through a G29 needle. The cell homogenates corresponding to two T175 flasks were combined and subsequently centrifuged (10 min, 1000× g, 4 • C). Ultracentrifugation of the supernatant (1 h, 100,000× g, 4 • C) yielded a membrane pellet, which was suspended in 250 µL 50 mM Tris-HCl buffer, pH 7.4 and stored at −80 • C.
Inhibition constants (K i ) were determined by means of a competitive radioligand binding assay using 50 mM Tris-HCl, 10 mM MgCl 2 , 1 mM EDTA, pH 7.4 as assay buffer as described previously [11]. Then, 5 µL of test compound in DMSO, 50 µL of [ 3 H]NMS in assay buffer, and 445 µL of membrane suspension in assay buffer were incubated for 90 min at 23 • C in PP tubes. Maximum binding was measured by using 5 µL DMSO, and nonspecific binding was measured by using 5 µL of 1 µM scopolamine in DMSO.

Data Analysis and Statistics
Data analysis in general was performed using Prism 9.00 (GraphPad Software, San Diego, CA, USA) or Microsoft Excel ® 365. Data are presented as means ± standard deviation (SD) for at least 3 independent experiments unless indicated otherwise.

Molecular Docking
All docking experiments were carried out using AutoDock Vina 1.1 with default settings [41]. The crystal structure of the M 1 muscarinic acetylcholine receptor (PDB 5CXV) and M5 muscarinic acetylcholine receptor (PDB 6OL9) were used for docking [28,42]. The performance of the docking algorithm was validated in a redocking experiment, in which the cocrystallized ligand's binding pose was reproduced with an acceptable RMSD of 0.252 Å in the binding pocket of M 1 and 0.294 Å in the binding pocket of M 5 [43]. Before docking, molecules were set to their energetically most favorable ionization state at pH 7.4 using the FixpKa program from OpenEye [44]. Docking results and the corresponding receptor-ligand interactions were analyzed with LigandScout 4.4.5 [45]. To visualize the spatial arrangement of such interactions, 3D and 2D pharmacophores were generated using the same software tool. Docking poses were additionally visualized using PyMOL [46]. The highest ranked pose of each docked compound exhibiting an ionic interaction with Asp105 3.32 was selected as a representative.

Conclusions
In summary, we successfully synthesized a series of hydroxylated arecaidine esters displaying favorable physicochemical properties for potential BBB penetration. The most affine compounds of this series 10b, 10c, and 17b were characterized by binding constants towards hM 1 in the low double-digit and single-digit region-similar to DPMA, the parent compound of this series. While the challenge of prominent subtype selectivity for hM 1 over the other muscarinic subtypes remains unsolved, this study significantly expanded our understanding towards the SAR of arecaidine esters. A noteworthy finding of this study was compound 8b's almost threefold preference for the subtype hM 5 . The presented docking studies were not able to explain this selectivity switch in comparison to DPMA; however, it serves as a starting point for more sophisticated computational studies and might guide the way towards M 5 -preferring ligands.