Novel Xanomeline-Containing Bitopic Ligands of Muscarinic Acetylcholine Receptors: Design, Synthesis and FRET Investigation

In the last few years, fluorescence resonance energy transfer (FRET) receptor sensors have contributed to the understanding of GPCR ligand binding and functional activation. FRET sensors based on muscarinic acetylcholine receptors (mAChRs) have been employed to study dual-steric ligands, allowing for the detection of different kinetics and distinguishing between partial, full, and super agonism. Herein, we report the synthesis of the two series of bitopic ligands, 12-Cn and 13-Cn, and their pharmacological investigation at the M1, M2, M4, and M5 FRET-based receptor sensors. The hybrids were prepared by merging the pharmacophoric moieties of the M1/M4-preferring orthosteric agonist Xanomeline 10 and the M1-selective positive allosteric modulator 77-LH-28-1 (1-[3-(4-butyl-1-piperidinyl)propyl]-3,4-dihydro-2(1H)-quinolinone) 11. The two pharmacophores were connected through alkylene chains of different lengths (C3, C5, C7, and C9). Analyzing the FRET responses, the tertiary amine compounds 12-C5, 12-C7, and 12-C9 evidenced a selective activation of M1 mAChRs, while the methyl tetrahydropyridinium salts 13-C5, 13-C7, and 13-C9 showed a degree of selectivity for M1 and M4 mAChRs. Moreover, whereas hybrids 12-Cn showed an almost linear response at the M1 subtype, hybrids 13-Cn evidenced a bell-shaped activation response. This different activation pattern suggests that the positive charge anchoring the compound 13-Cn to the orthosteric site ensues a degree of receptor activation depending on the linker length, which induces a graded conformational interference with the binding pocket closure. These bitopic derivatives represent novel pharmacological tools for a better understanding of ligand-receptor interactions at a molecular level.


Introduction
In the last few years, different receptor sensors based on the fluorescence resonance energy transfer (FRET) were generated for various G protein-coupled receptors (GPCRs) and represented a valuable tool for investigating real-time receptor activation as well as ligand-receptor interactions [1][2][3][4]. The use of GPCR-based biosensors allows for the study of GPCRs in their native membrane and the exploration of receptor dynamics [5,6]. Indeed, GPCRs behave as a highly dynamic system, transitioning among distinct conformational states, which dictate their signaling pathways through the cell membrane [7]. Analysis of GPCR structural dynamics is therefore essential for understanding their physiology and is quite informative in light of the rational design of GPCR-targeted drugs.
FRET receptor sensors made a decisive contribution to the overall understanding of GPCR dynamics in the context of ligand binding and functional activation [5,8]. Moreover, these sensors have been used not only to detect ligand agonistic properties but also to investigate the conformational changes produced by receptor allosteric modulators [8]. The allosteric binding regions on GPCRs are outside the orthosteric binding domain and were found to be less conserved among receptor subtypes [9]. GPCR allosteric modulators have a great potential for selective ligand development. Additionally, in the attempt to achieve subtype selectivity, allosteric building blocks may be fused with orthosteric receptor activators to give dual-steric (i.e., orthosteric/allosteric) GPCR ligands [10], which couple the high affinity receptor activation from the orthosteric site with the allosteric control of subtype and signaling pathway selectivity [11].
FRET sensors based on muscarinic acetylcholine receptors (mAChRs) have been used to study dual-steric ligands, leading to the detection of different kinetics and distinguishing between partial, full, and super agonism [12][13][14][15]. In particular, this analysis was performed on the two series of bitopic ligands, 6-Cn and 7-Cn (Figure 1), designed for a selective interaction with M 1 mAChRs by using an M 1 receptor FRET sensor [12]. In these hybrids, the orthosteric moieties of the endogenous ligand acetylcholine (ACh) 1 and of the super agonist Iperoxo 2 [16][17][18][19] were respectively connected with the highly selective M 1 positive allosteric modulator benzyl quinolone carboxylic acid (BQCA) 3 [20] by means of alkyl chains of different length. An optimal linker length of six methylene groups was found to engender a maximal response at M 1 mAChRs, whereas an unprecedented conformational change for GPCRs was evidenced when these bitopic ligands were endowed with a longer spacer. Moreover, FRET biosensors were employed to investigate the conformational changes induced at M 1 and M 2 mAChRs by the dual-steric compounds 8-Cn and 9-Cn (Figure 1), which are composed of Iperoxo 2 and the fragments, respectively, of the two negative allosteric modulators W84 4 [21] and Naphmethonium 5 [22] covalently connected by a polymethylene chain of varying length. Previously, compounds 8-Cn and 9-Cn were used as chemical probes to investigate the conformational transitions in the allosteric vestibule of the M 2 mAChR by using an M 2 FRET sensor. A graded receptor activation from the orthosteric binding site was evidenced, with a simultaneous restriction of spatial flexibility of the allosteric vestibule, which was found to affect the extent of receptor movement and the activation of differential signaling pathways [13]. Recently, both series of bitopic ligands, 8-Cn and 9-Cn (Figure 1), were employed to investigate also the M 1 mAChR activation. Using a biosensor for the M 1 mAChR, the intracellular conformational changes in the receptor at the G protein-coupling interface were governed by the linker length of the bitopic ligands and allowed selective G protein signaling. Indeed, the decrease in length of the spacer gradually hampered the extracellular binding pocket closure and induced different coupling in distinct G protein families [14].  (1,2), allosteric modulators (3,4,5), and dualsteric hybrid compounds ACh/BQCA (6-Cn), Iperoxo/BQCA (7-Cn), Iperoxo/W84 (8-Cn), and Iperoxo/Naphmethonium (9-Cn) investigated by using muscarinic FRET receptor sensors.
As a continuation of our efforts towards the development of innovative bitopic ligands of mAChR subtypes [23][24][25][26][27][28] and to further deepen ligand-receptor interactions at a molecular level, in this study, we designed and synthesized a new set of derivatives, 12-Cn and 13-Cn (Figure 2), that integrate with the same molecular skeleton as the pharmacophoric moieties of Xanomeline and 77-LH-28-1 (1-[3-(4-butyl-1piperidinyl)propyl]-3,4-dihydro-2(1H)-quinolinone). Xanomeline is a well-known M1/M4preferring orthosteric agonist that ameliorated cognitive impairments in Alzheimer's disease patients and showed activity in various models of schizophrenia, with a potential benefit for the treatment of positive, negative, and cognitive symptoms [29]. On the other hand, 77-LH-28-1 was characterized as an M1-selective, positive allosteric modulator, thus representing an interesting pharmacological tool with cognition-enhancing properties [30]. As illustrated in Figure 2, we designed the novel bipharmacophoric derivatives as merged structures, with the tetrahydropyridine nucleus of Xanomeline as the central core. The target hybrids were designed as tertiary amines, 12-Cn, and methyl tetrahydropyridinium salts, 13-Cn, and the distance between the tetrahydropyridine ring and the dihydroquinoline moiety was modulated by varying the length of the chain, from three to nine methylene units, that connects the nitrogen atoms of the two fragments. Herein, we report the synthesis of the new bitopic ligands 12-Cn and 13-Cn and their pharmacological characterization by means of FRET-based mAChR sensors.  (1,2), allosteric modulators (3,4,5), and dualsteric hybrid compounds ACh/BQCA (6-Cn), Iperoxo/BQCA (7-Cn), Iperoxo/W84 (8-Cn), and Iperoxo/Naphmethonium (9-Cn) investigated by using muscarinic FRET receptor sensors.
As a continuation of our efforts towards the development of innovative bitopic ligands of mAChR subtypes [23][24][25][26][27][28] and to further deepen ligand-receptor interactions at a molecular level, in this study, we designed and synthesized a new set of derivatives, 12-Cn and 13-Cn (Figure 2), that integrate with the same molecular skeleton as the pharmacophoric moieties of Xanomeline and 77-LH-28-1 (1-[3-(4-butyl-1-piperidinyl)propyl]-3,4-dihydro-2(1H)-quinolinone). Xanomeline is a well-known M 1 /M 4 -preferring orthosteric agonist that ameliorated cognitive impairments in Alzheimer's disease patients and showed activity in various models of schizophrenia, with a potential benefit for the treatment of positive, negative, and cognitive symptoms [29]. On the other hand, 77-LH-28-1 was characterized as an M 1 -selective, positive allosteric modulator, thus representing an interesting pharmacological tool with cognition-enhancing properties [30]. As illustrated in Figure 2, we designed the novel bipharmacophoric derivatives as merged structures, with the tetrahydropyridine nucleus of Xanomeline as the central core. The target hybrids were designed as tertiary amines, 12-Cn, and methyl tetrahydropyridinium salts, 13-Cn, and the distance between the tetrahydropyridine ring and the dihydroquinoline moiety was modulated by varying the length of the chain, from three to nine methylene units, that connects the nitrogen atoms of the two fragments. Herein, we report the synthesis of the

Synthesis of Hybrid Ligands
The synthetic approach for the preparation of target hybrids, such as tertiary amines 12-C3, 12-C5, 12-C7, and 12-C9 and ammonium salts 13-C3, 13-C5, 13-C7, and 13-C9 are illustrated in Scheme 1. We initially prepared the orthosteric pharmacophoric moiety Xanomeline 10 according to a known procedure [25,31]. In the presence of acetic acid, 3pyridinecarboxaldehyde 14 underwent a Strecker-type reaction with trimethylsilyl cyanide to afford cyanohydrin 15. The reaction of 15 with ammonium chloride in ammonium hydroxide yielded the corresponding aminonitrile 16, which was then cyclized by treatment with disulfur dichloride in DMF and afforded the 3-chloro-4-(pyridine-3-yl)-1,2,5-thiadiazole intermediate 17. The latter was functionalized by nucleophilic substitution with 1-hexanol using sodium hydride in THF to provide intermediate 18. Next, Xanomeline 10 was obtained after the conversion of 18 to the corresponding quaternary pyridinium iodide 19 by reaction with methyl iodide in acetone, followed by the reduction in the aromatic positively charged heterocyclic ring to the corresponding 1,2,5,6-tetrahydropyridine by using sodium borohydride in methanol. The monobromo-3,4-dihydroquinolinone allosteric fragments 22-Cn were obtained through the reaction of the commercially available 3,4-dihydroquinolinone 20 with alkyl dibromides 21-Cn in the presence of sodium hydride, followed by heating at 50 °C in DMF.

Synthesis of Hybrid Ligands
The synthetic approach for the preparation of target hybrids, such as tertiary amines 12-C3, 12-C5, 12-C7, and 12-C9 and ammonium salts 13-C3, 13-C5, 13-C7, and 13-C9 are illustrated in Scheme 1. We initially prepared the orthosteric pharmacophoric moiety Xanomeline 10 according to a known procedure [25,31]. In the presence of acetic acid, 3pyridinecarboxaldehyde 14 underwent a Strecker-type reaction with trimethylsilyl cyanide to afford cyanohydrin 15. The reaction of 15 with ammonium chloride in ammonium hydroxide yielded the corresponding aminonitrile 16, which was then cyclized by treatment with disulfur dichloride in DMF and afforded the 3-chloro-4-(pyridine-3-yl)-1,2,5thiadiazole intermediate 17. The latter was functionalized by nucleophilic substitution with 1-hexanol using sodium hydride in THF to provide intermediate 18. Next, Xanomeline 10 was obtained after the conversion of 18 to the corresponding quaternary pyridinium iodide 19 by reaction with methyl iodide in acetone, followed by the reduction in the aromatic positively charged heterocyclic ring to the corresponding 1,2,5,6-tetrahydropyridine by using sodium borohydride in methanol. The monobromo-3,4-dihydroquinolinone allosteric fragments 22-Cn were obtained through the reaction of the commercially available 3,4dihydroquinolinone 20 with alkyl dibromides 21-Cn in the presence of sodium hydride, followed by heating at 50 • C in DMF.
The synthesis of the desired hybrids, 12-C3, 12-C5, 12-C7, and 12-C9, was achieved through the Menshutkin reaction between intermediate 18 and the bromo derivatives 22-Cn. The reaction was performed in refluxing acetonitrile and led to the corresponding Molecules 2023, 28, 2407 5 of 24 pyridinium bromides 23-Cn, which were subsequently reduced with NaBH 4 in ethanol at room temperature to produce the final compound, 12-Cn, as tertiary amines. The free base 12-C3 was then reacted with oxalic acid in methanol to provide the related crystalline 12-C3 oxalate; the same treatment of homologs 12-C5, 12-C7, and 12-C9 did not afford the corresponding oxalates. The desired positively charged analogs 13-C3, 13-C5, 13-C7, and 13-C9 were obtained through the Menshutkin reaction between Xanomeline 10 and the bromo intermediates 22-Cn in refluxing acetonitrile. The synthesis of the desired hybrids, 12-C3, 12-C5, 12-C7, and 12-C9, was achieved through the Menshutkin reaction between intermediate 18 and the bromo derivatives 22-Cn. The reaction was performed in refluxing acetonitrile and led to the corresponding pyridinium bromides 23-Cn, which were subsequently reduced with NaBH4 in ethanol at room temperature to produce the final compound, 12-Cn, as tertiary amines. The free base 12-C3 was then reacted with oxalic acid in methanol to provide the related crystalline 12-C3 oxalate; the same treatment of homologs 12-C5, 12-C7, and 12-C9 did not afford the corresponding oxalates. The desired positively charged analogs 13-C3, 13-C5, 13-C7, and 13-C9 were obtained through the Menshutkin reaction between Xanomeline 10 and the bromo intermediates 22-Cn in refluxing acetonitrile.

Muscarinic Receptor FRET Sensors and Ligand Characterization
In this work, we aimed to investigate the effects of our novel compounds on the muscarinic receptor subtypes that are more abundantly expressed in the brain (M1, M2, M4, and M5). For this purpose, we made use of M1, M2, M4, and M5 mAChR FRET sensors, which were not truncated but were modified by fusing a cyan fluorescent protein (CFP) to the C-terminus of the receptor and, additionally, by insertion of the fluoresceine arsenical hairpin (FlAsH) binding motif, consisting of the six amino acids CCGPCC sequence, into the third intracellular loop (IL3) region. For FRET experiments, the mAChR sensors were stably expressed in HEK 293 cells. Due to ligand binding, a structural rearrangement of the transmembrane vestibule occurs, with a change in the relative distance and orientation of the fluorophores towards each other that was detected with a millisecond resolution. For ligand application, a pressure-based perfusion system was used. By superfusing the cells with a physiological buffer, a straight baseline occurred. Upon ligand addition (indicated by black bars), a sharp peak was detected with a concentration-dependent intensity. After the signal reached a constant level, superfusion was switched back to buffer, and the signal returned to the baseline. The signal direction is sensor-specific and probably due to the fluorophore orientation within the receptor

Muscarinic Receptor FRET Sensors and Ligand Characterization
In this work, we aimed to investigate the effects of our novel compounds on the muscarinic receptor subtypes that are more abundantly expressed in the brain (M 1 , M 2 , M 4 , and M 5 ). For this purpose, we made use of M 1 , M 2 , M 4 , and M 5 mAChR FRET sensors, which were not truncated but were modified by fusing a cyan fluorescent protein (CFP) to the C-terminus of the receptor and, additionally, by insertion of the fluoresceine arsenical hairpin (FlAsH) binding motif, consisting of the six amino acids CCGPCC sequence, into the third intracellular loop (IL3) region. For FRET experiments, the mAChR sensors were stably expressed in HEK 293 cells. Due to ligand binding, a structural rearrangement of the transmembrane vestibule occurs, with a change in the relative distance and orientation of the fluorophores towards each other that was detected with a millisecond resolution. For ligand application, a pressure-based perfusion system was used. By superfusing the cells with a physiological buffer, a straight baseline occurred. Upon ligand addition (indicated by black bars), a sharp peak was detected with a concentration-dependent intensity. After the signal reached a constant level, superfusion was switched back to buffer, and the signal returned to the baseline. The signal direction is sensor-specific and probably due to the fluorophore orientation within the receptor structure. A slight reduction in the FRET signal over time was detectable and may be attributed to photobleaching. To prevent artificial underestimation of ligand efficacy, reference and ligand compounds were measured alternately.
Initially, we characterized the functional response of the M 1 and M 2 receptor FRET sensors by applying the three muscarinic orthosteric ligands, namely, the endogenous neurotransmitter ACh 1 (Figure 1), Iperoxo 2 ( Figure 1), and Xanomeline 10 ( Figure 2), which represents the orthosteric moiety of the new bitopic ligands. At the M 1 receptor sensor, both ACh and Iperoxo showed 100% receptor activation, thus behaving as full agonists. We generated concentration-dependent response curves ( Figure 3A), and Iperoxo showed a lower EC 50 value (EC 50 = 0.5 µM) compared to that of Ach (EC 50 = 2.8 µM). Conversely, Xanomeline did not show any response. The concentration of Xanomeline was increased up to 300 µM with no effect (Figure 3B), and higher concentrations led to artifacts in the experimental setup. Compared to 1 and 2, this lack of response could be due to an alternative binding site for Xanomeline on the M 1 receptor, which could also suggest an alternative pattern of receptor activation. It was indeed evidenced that Xanomeline shows a noncompetitive binding profile when investigated together with atropine [32,33]. As we explored the mAChR dynamics by introducing a fluorophore below the fifth transmembrane region (TM5) and at the C-terminus, we detected a conformational change mainly in the third intracellular loop. It would be hard to detect ligands that bind but do not activate the receptor, and they are even harder to characterize because strongly biased ligands affect receptor conformations to a higher extent at TM6 than at TM5. neurotransmitter ACh 1 (Figure 1), Iperoxo 2 ( Figure 1), and Xanomeline 10 ( Figure 2 which represents the orthosteric moiety of the new bitopic ligands. At the M1 recept sensor, both ACh and Iperoxo showed 100% receptor activation, thus behaving as fu agonists. We generated concentration-dependent response curves ( Figure 3A), an Iperoxo showed a lower EC50 value (EC50 = 0.5 μM) compared to that of Ach (EC50 = 2 μM). Conversely, Xanomeline did not show any response. The concentration Xanomeline was increased up to 300 μM with no effect ( Figure 3B), and high concentrations led to artifacts in the experimental setup. Compared to 1 and 2, this lack response could be due to an alternative binding site for Xanomeline on the M1 recepto which could also suggest an alternative pattern of receptor activation. It was indee evidenced that Xanomeline shows a noncompetitive binding profile when investigate together with atropine [32,33]. As we explored the mAChR dynamics by introducing fluorophore below the fifth transmembrane region (TM5) and at the C-terminus, w detected a conformational change mainly in the third intracellular loop. It would be ha to detect ligands that bind but do not activate the receptor, and they are even harder characterize because strongly biased ligands affect receptor conformations to a high extent at TM6 than at TM5.
Reference compounds 1 and 2 also behaved as full agonists at the M2 FRET senso The known super agonism of Iperoxo appeared as a larger conformational chan compared to that induced by Ach, and again, the EC50 value shown by Iperoxo (EC50 = 0 μM) was lower than that of ACh (EC50 = 6.3 μM) ( Figure 3C). Xanomeline did not sho any effects at the M2 sensor ( Figure 3D), as in the case of the M1 FRET receptor.  (2), and 100 and 300 micromolar Xanomeline (10), were applied as indicated at the appropriate time points by black bars below or above the recorded signal.
Reference compounds 1 and 2 also behaved as full agonists at the M 2 FRET sensor. The known super agonism of Iperoxo appeared as a larger conformational change compared to that induced by Ach, and again, the EC 50 value shown by Iperoxo (EC 50 = 0.8 µM) was lower than that of ACh (EC 50 = 6.3 µM) ( Figure 3C). Xanomeline did not show any effects at the M 2 sensor ( Figure 3D), as in the case of the M 1 FRET receptor.

FRET Measurements of Hybrid Ligands
The newly synthesized Xanomeline/77-LH-28-1 hybrid compounds, 12-Cn and 13-Cn, were investigated for their ability to induce a conformational change at the mAChRs by using M 1 , M 2 , M 4 , and M 5 muscarinic FRET sensors. Due to the ligand binding, a structural rearrangement of the receptor transmembrane vestibule occurs, the relative distance of the fluorophores to each other changes, and the dynamic conformational receptor changes are studied in real time with a millisecond resolution. Due to the lack of a receptor response with Xanomeline, the signal from a saturating concentration of Iperoxo was chosen as the reference signal in all the following measurements.
The results of the FRET investigation at mAChR sensors for the tertiary amine hybrid derivatives 12-Cn ( Figure 2) are illustrated in Figure 4. Figure 4A,B summarize the response of the M 1 FRET sensor. The FRET trace in Figure 4A shows that compound 12-C3, characterized by the shortest linker of three methylene groups, did not induce any conformational change at the receptor sensor. Hybrids 12-C5, 12-C7, and 12-C9 were instead able to induce a conformational change with a slightly increasing tendency with longer linker length. Thus, for the M 1 receptor, we found that a 9-carbon polymethylene spacer is the best linker for the compounds obtained via hybridization of 10 and 11. This can also be observed in Figure 4B, which shows the maximal ligand-induced changes in comparison with Iperoxo. Conversely, when tested at the M 2 receptor sensor ( Figure 4C), the hybrids 12-C5, 12-C7, and 12-C9 did not induce any conformational change, therefore, displaying a first indication of conformational receptor subtype selectivity. Surprisingly, we observed a small conformational variation upon testing 12-C3 at the M 2 receptor subtype. The same hybrid was also the only ligand of this series that induced a detectable FRET signal at the M 4 sensor ( Figure 4D), at variance with the related homologs 12-C5, 12-C7, and 12-C9. Moreover, the four compounds of the 12-Cn group were unable to induce a significant FRET signal at the M 5 receptor subtype ( Figure 4E).
The same experiments were performed with the corresponding set of quaternary tetrahydropyridinium salts, 13-Cn (Figure 2), and the results are displayed in Figure 5. When compounds 13-Cn were investigated via FRET for their ability to induce a conformational change at the M 1 receptor, a similar trend ( Figure 5A) to that of the uncharged analogs 12-Cn ( Figure 4A) was observed. Whereas hybrid 13-C3 with the shortest linker did not exhibit a FRET signal, the longer linker derivatives 13-C5, 13-C7, and 13-C9 induced a conformational change indicative of their agonistic properties. Signal quantification brought about a bell-shaped graph ( Figure 5B), since a signal increase was observed for 13-C5 and 13-C7, followed by a decrease for 13-C9, characterized by the longest linker. These results clearly evidenced a linker-length-dependent receptor response with an optimal spacer length of approximately seven methylene groups (≈10.8 Å). Hybrid 13-C7 induced the largest conformational change at the M 1 FRET sensor compared to all the other structural analogs, both of the quaternary ammonium 13-Cn series and of the tertiary amine 12-Cn series. The FRET investigation at the M 2 receptor sensor revealed that the four 13-Cn compounds were not able to induce any conformational change ( Figure 5C), displaying a similar trend to that of the corresponding uncharged set of derivatives 12-Cn ( Figure 4C), the analog 12-C3 excepted. Figure 5D displays a FRET trace of hybrid 13-Cn at the M 4 receptor sensor, and the bar graph in Figure 5F summarizes the corresponding FRET signals. Interestingly, at the M 4 receptor, antiparallel signals for derivatives 13-C5, 13-C7, and 13-C9 were detected, with a clear FRET effect. Similar to the results obtained at the M 1 FRET sensor, the activation pattern showed bell-shaped features. Once again, the signal intensity was dependent on the linker length, with the highest response corresponding to Molecules 2023, 28, 2407 8 of 24 the 13-C7 hybrid containing a spacer with seven methylene groups. Analogously to the tertiary amines, hybrids 12-Cn, also hybrids 13-Cn did not affect the M 5 subtype, and no significant FRET signal was obtained ( Figure 5E). The same experiments were performed with the corresponding set of quaternary tetrahydropyridinium salts, 13-Cn (Figure 2), and the results are displayed in Figure 5. When compounds 13-Cn were investigated via FRET for their ability to induce a conformational change at the M1 receptor, a similar trend ( Figure 5A) to that of the uncharged analogs 12-Cn ( Figure 4A) was observed. Whereas hybrid 13-C3 with the

FRET Measurements of Fragments
A fragment-based screening was applied to unravel the moieties of the hybrid ligands 12-Cn and 13-Cn contributing to the observed signals and figure out whether it is possible to reconstruct the signal derived from the hybrids by combining that of the individual orthosteric and allosteric fragments. We chose to perform this study at the M 1 FRET sensor, which displayed significant FRET signals with both sets of hybrids, 12-Cn and 13-Cn. Therefore, Xanomeline 10 and dihydroquinolinone 20 were superfused alone and applied at the same time to the M 1 receptor-expressing sensor cells ( Figure 6). None of the single fragments nor the mixture of the fragments exhibited a detectable effect at the FRET sensor, indicating that the activation patterns observed for hybrids 12-Cn and 13-Cn are not a combination of those induced by their molecular component parts. Therefore, we may conclude that the entire structure of ligands 12-Cn and 13-Cn is relevant to engendering receptor activation; to this end, the nature and length of the polymethylene spacer represent an essential feature of the molecular skeleton.
Molecules 2023, 28, x FOR PEER REVIEW 11 of 25 Figure 6. Fragment-based analysis to evaluate the contributions of the single molecular moieties, dihydroquinolinone (20), and Xanomeline (10), of the Xanomeline/77-LH-28-1 hybrids to the M1 FRET signal. A single-cell FRET recording of the M1 receptor sensor stably expressed in HEK293 cells is shown. One hundred micromolar Iperoxos were used as a reference ligand throughout the recording, as indicated by black bars above the recorded signal. One hundred micromolars of 20 and 10 were applied separately or together at the appropriate time points, as indicated by black bars below the recorded signal.

Evaluation of Linker Elongation
FRET signals obtained by the investigations of the Xanomeline/77-LH-28-1 hybrids evidenced that an optimal linker length of seven methylene groups induced a remarkable conformational change at the M1 mAChR. Indeed, hybrid 13-C7 displayed the most pronounced receptor activation, likely because its linker length best matches the distance between allosteric and orthosteric binding sites, thus favoring the dual-steric (active) binding pose over the purely allosteric pose (inactive). A spacer of shorter or longer length results in a gradual decrease in activation. Shorter linkers likely forbid a productive ligand-receptor binding, and therefore, the related hybrids are not able to trigger a conformational variation in the receptor sensor or they give rise to a reduced activation pattern. Conversely, longer linkers probably cause steric hindrance with the aromatic lid structure of the receptor because of the fixed positive charge of the orthosteric moiety, thus resulting in a lower affinity and a lower receptor response. These findings confirm related data collected in previous investigations performed at the M1 mAChR with quinolone-based bitopic ligands 7-Cn [12] and Iperoxo/W84 hybrids 8-Cn [14] (Figure 1). A comparable chain of six methylene groups was reported for compounds 7-Cn as the optimal linker length for conformational changes at the M1 mAChR due to the defined distance between the allosteric and the orthosteric binding domains [12,34]. Similarly, the dual-steric compound 8-C8 induced a significantly larger FRET change than the related shorter derivatives (8-C7 and 8-C6), which accounts for a reduced conformational interference at the receptor/G protein-coupling interface, resulting in a greater G-protein signaling capability [14].
Moreover, a linker-length-dependent graded activation was evidenced by bitopic ligands 9-Cn (Figure 1), containing Iperoxo and a fragment of the allosteric modulator  (20), and Xanomeline (10), of the Xanomeline/77-LH-28-1 hybrids to the M 1 FRET signal. A single-cell FRET recording of the M 1 receptor sensor stably expressed in HEK293 cells is shown. One hundred micromolar Iperoxos were used as a reference ligand throughout the recording, as indicated by black bars above the recorded signal. One hundred micromolars of 20 and 10 were applied separately or together at the appropriate time points, as indicated by black bars below the recorded signal.

Evaluation of Linker Elongation
FRET signals obtained by the investigations of the Xanomeline/77-LH-28-1 hybrids evidenced that an optimal linker length of seven methylene groups induced a remarkable conformational change at the M 1 mAChR. Indeed, hybrid 13-C7 displayed the most pronounced receptor activation, likely because its linker length best matches the distance between allosteric and orthosteric binding sites, thus favoring the dual-steric (active) binding pose over the purely allosteric pose (inactive). A spacer of shorter or longer length results in a gradual decrease in activation. Shorter linkers likely forbid a productive ligandreceptor binding, and therefore, the related hybrids are not able to trigger a conformational variation in the receptor sensor or they give rise to a reduced activation pattern. Conversely, longer linkers probably cause steric hindrance with the aromatic lid structure of the receptor because of the fixed positive charge of the orthosteric moiety, thus resulting in a lower affinity and a lower receptor response. These findings confirm related data collected in previous investigations performed at the M 1 mAChR with quinolone-based bitopic ligands 7-Cn [12] and Iperoxo/W84 hybrids 8-Cn [14] (Figure 1). A comparable chain of six methylene groups was reported for compounds 7-Cn as the optimal linker length for conformational changes at the M 1 mAChR due to the defined distance between the allosteric and the orthosteric binding domains [12,34]. Similarly, the dual-steric compound 8-C8 induced a significantly larger FRET change than the related shorter derivatives (8-C7 and 8-C6), which accounts for a reduced conformational interference at the receptor/G protein-coupling interface, resulting in a greater G-protein signaling capability [14].
Moreover, a linker-length-dependent graded activation was evidenced by bitopic ligands 9-Cn (Figure 1), containing Iperoxo and a fragment of the allosteric modulator Naphmethonium 5, which were investigated by FRET at the M 2 mAChR [13]. The best activation profile was displayed by hybrid 9-C8, characterized by a spacer chain of eight methylene groups. We re-examined the behavior of hybrids 9-C6, 9-C7, and 9-C8 [13] by adding new data obtained on the further elongated bitopic homolog 9-C9 [35]. Even for this group of ligands, FRET measurements evidenced a bell-shaped activation pattern (Figure 7), which resembles that of the positively charged methyl tetrahydropyridinium salt 13-Cn at the M 1 mAChR.
These data further confirm that the simultaneous activation of the M1 as well as M2 receptors by hybrid ligands is modulated/optimized by the length of the spacer connecting the orthosteric and allosteric molecular fragments. The two subtypes may favor variable linker lengths for dual-steric ligands, due to differences in the protein shape and in the distance of their mutual recognition sites [12][13][14].

Receptor Activation Patterns
The two sets of Xanomeline/77-LH-28-1 hybrid ligands, 12-Cn and 13-Cn (Figure 2), were investigated by FRET for their interaction at the M1, M2, M4, and M5 mAChRs and evidenced a certain degree of M1 subtype selectivity. Interestingly, 12-Cn and 13-Cn displayed a different activation pattern at the M1 subtype. Whereas the tertiary amine derivatives 12-Cn showed an almost linear response ( Figure 4B), the methyl tetrahydropyridinium derivatives 13-Cn evidenced a bell-shaped activation response ( Figure 5B). The different trend could be explained by taking into account the only structural difference between the two groups of ligands, namely the presence in hybrid 13-Cn of the permanently positively charged nitrogen, which is missing in the tertiary amine derivatives 12-Cn.
The crystal structure of the superagonist Iperoxo 2 bound to the M2 muscarinic subtype receptor evidenced a fundamental interaction between the trimethyl ammonium group of 2 and the conserved Asp103 residue within the orthosteric binding site (PDB 4MQS) [36]. A comparable residue, Asp112, served as the counterion for the interaction with the positively charged inverse agonist, Tiotropium, which crystallized in the orthosteric site of the M1 receptor [37]. The two Asp residues play a crucial role in muscarinic ligand binding, as well as receptor activation through the recognition of the agonist's cationic head. The M1 and M2 receptor activation by bitopic ligands 8-Cn and 9-Cn is primarily regulated by their Iperoxo moiety [14,38], which occupies the orthosteric binding site and adopts the same orientation as Iperoxo alone [36]. The length of the polymethylene linker connecting Iperoxo with the allosteric moiety controls the position of the latter within the allosteric vestibule and the degree of conformational receptor Figure 7. Evaluation of the Iperoxo/Naphmetonium hybrid ligands 9-Cn at the M 2 muscarinic FRET receptor sensor stably expressed in HEK293 cells [13].
These data further confirm that the simultaneous activation of the M 1 as well as M 2 receptors by hybrid ligands is modulated/optimized by the length of the spacer connecting the orthosteric and allosteric molecular fragments. The two subtypes may favor variable linker lengths for dual-steric ligands, due to differences in the protein shape and in the distance of their mutual recognition sites [12][13][14].

Receptor Activation Patterns
The two sets of Xanomeline/77-LH-28-1 hybrid ligands, 12-Cn and 13-Cn (Figure 2), were investigated by FRET for their interaction at the M 1 , M 2 , M 4 , and M 5 mAChRs and evidenced a certain degree of M 1 subtype selectivity. Interestingly, 12-Cn and 13-Cn displayed a different activation pattern at the M 1 subtype. Whereas the tertiary amine derivatives 12-Cn showed an almost linear response ( Figure 4B), the methyl tetrahydropyridinium derivatives 13-Cn evidenced a bell-shaped activation response ( Figure 5B). The different trend could be explained by taking into account the only structural difference between the two groups of ligands, namely the presence in hybrid 13-Cn of the permanently positively charged nitrogen, which is missing in the tertiary amine derivatives 12-Cn.
The crystal structure of the superagonist Iperoxo 2 bound to the M 2 muscarinic subtype receptor evidenced a fundamental interaction between the trimethyl ammonium group of 2 and the conserved Asp103 residue within the orthosteric binding site (PDB 4MQS) [36]. A comparable residue, Asp112, served as the counterion for the interaction with the positively charged inverse agonist, Tiotropium, which crystallized in the orthosteric site of the M 1 receptor [37]. The two Asp residues play a crucial role in muscarinic ligand binding, as well as receptor activation through the recognition of the agonist's cationic head. The M 1 and M 2 receptor activation by bitopic ligands 8-Cn and 9-Cn is primarily regulated by their Iperoxo moiety [14,38], which occupies the orthosteric binding site and adopts the same orientation as Iperoxo alone [36]. The length of the polymethylene linker connecting Iperoxo with the allosteric moiety controls the position of the latter within the allosteric vestibule and the degree of conformational receptor movements, thus affecting the extent of the ligand binding closure, which is necessary for receptor activation.
The bell-shaped responses observed for the charged methylated set of new bitopic ligands, 13-Cn at the M 1 and M 4 mAChRs should be similarly modulated by the liganddependent allosteric coupling mechanism correlated with the linker length. The methyl tetrahydropyridinium positive charge anchors the compounds to the orthosteric binding site of the M 1 and M 4 receptors and ensures the degree of receptor activation depends on the ligand spacer length, which induces a graded conformational change/interference with the binding pocket closure. Interestingly, a parallel trend was observed for derivatives of 13-Cn, with hybrid 13-C7 causing the most relevant FRET receptor activation at both M 1 and M 4 mAChRs. This group of derivatives essentially retains the pharmacological profile of Xanomeline at the two target receptor subtypes.
Conversely, the tertiary amine derivatives of 12-Cn showed a linear receptor activation pattern at M 1 mAChRs. These hybrids lack the contribution imparted to their permanently charged analogs by a strong ionic interaction with the Asp residue in the orthosteric binding pocket. Indeed, we calculated a degree of protonation of~15% at physiological pH for these compounds according to the pK a values measured experimentally (pK a 6.6-6.8).
A weaker interaction at this recognition site implies for compounds 12-Cn less stringent geometric features in orienting the allosteric fragment, with an almost negligible weight of their linker length and comparable conformational variations. However, due to the combined orthosteric/allosteric effect, also for these tertiary amines, an activating response of our model M 1 FRET receptor was evidenced, at variance with Xanomeline alone, which was unresponsive in the same experimental conditions.

Construction of the Muscarinic FRET Sensors
Muscarinic ACh receptor constructs were fused to the enhanced variants of cyan fluorescent protein (eCFP; BD Bioscience Clontech, TaKaRa Bio Europe, Saint Germain en Laye, France) by a standard PCR extension overlap technique [39]. The muscarinic receptor FRET sensors were obtained as previously reported [8,12,40]. All the resulting constructs were cloned into pcDNA3 (Invitrogen, Thermo Fisher Scientific GmbH, Dreieich, Germany) and verified by sequencing, performed by Eurofins Medigenomix GmbH.

Stable Cell Line Generation
Cells were seeded into a culture dish with a confluency of 30% 3 h before transfecting the cells with the Effectene reagent ordered from Quiagene. Reagent concentrations and incubation times were applied in accordance with the manufacturer's instructions. Twentyfour hours after transfecting, the normal culture medium was replaced by a culture medium supplemented with 400 µg mL −1 G-418. After that, the medium was refreshed every day until all the untransfected cells died. Now the cells were counted, diluted, and applied to 48-well plates, resulting in a one cell to well distribution. This homogeneous cell population was characterized by fluorescence microscopy and was investigated concerning its cDNA content.

Cell Culture
HEK293 cells stably expressing the muscarinic receptor FRET sensors were maintained in DMEM with 4.5 g L −1 glucose, 10% (v/v) FCS, 100 U mL −1 penicillin, 100 µg mL −1 streptomycin sulfate and 2 mM L-glutamine, and 200 µg mL −1 G-418. The cells were kept at 37 • C in a humidified 7% CO 2 atmosphere and were routinely passaged every 2 to 3 days. Untransfected HEK cells were maintained in cell culture medium without G-418.

FlAsH Labeling
A labeling protocol was applied as described previously [2,41,42]. In brief, cells were grown to near confluency on poly-D-lysine-coated glass coverslips. Initially, cells were washed with labeling buffer (150 mM NaCl, 10 mM HEPES, 2.5 mM KCl, 4 mM CaCl 2 , 2 mM MgCl 2 , supplemented with 10 mM glucose (pH 7.3)). After that, cells were incubated with labeling buffer containing 500 nM FlAsH and 12.5 µM 1,2-ethanedithiol (EDT) for 1 h at 37 • C, followed by flushing with labeling buffer. To reduce nonspecific FlAsH binding, the cells were incubated for 10 min with labeling buffer containing 250 µM EDT. After flushing with labeling buffer, the cells were held in cell culture medium.

Ligand Application
The reference ligands were prepared from 1 mM stock solutions that were stored at −20 • C, taking into consideration that at least acetylcholine remains unstable in solution [43]. Used stock solutions have not been older than a couple of weeks. Bitopic ligands or analogs were stored at 4 • C and were weighed out directly before the experiment. Then, the ligands were solubilized in a measuring buffer (140 mM NaCl, 10 mM HEPES, 5.4 mM KCl, 2 mM CaCl 2 , 1 mM MgCl 2 (pH 7.3)) to a final concentration of 100 µM.

Single Cell FRET Experiments
FRET measurements were performed using a Zeiss Axiovert 200 inverted microscope endowed with a PLAN-Neofluar oil immersion 100 objective, a dual emission photometric system, and a Polychrome IV light source (Till Photonics, Gräfelfing, Bavaria, Germany) as described previously [1,2]. Samples were excited at 436 nm (dichroic 460 nm) with a frequency of 10 Hz. Emitted light was recorded using 535/30 nm and 480/40 nm emission filters and a DCLP 505 nm beam splitter for FlAsH and CFP, respectively. FRET was observed as the ratio of FlAsH/CFP, which was corrected offline for bleed-through, direct FlAsH excitation, and photobleaching. To investigate changes in FRET on ligand addition, cells were continuously superfused with FRET buffer complemented with various ligands at saturating concentrations, as indicated. Superfusion was performed using the ALA-VM8 (ALA Scientific Instruments).

Data Processing
Fluorescence intensities were acquired using Clampex (Axon Instruments, Molecular Devices, San Jose, CA, USA). Data are shown as means ± SD for n independent experiments. Statistical analysis and curve fitting were performed using Origin (OriginLab Corporation, Northampton, MA, USA), or Clampfit (Axon Instruments, Molecular Devices, San Jose, CA, USA).

Conclusions
In this study, two series of Xanomeline/77-LH-28-1 hybrid compounds, 12-Cn and 13-Cn, were designed, synthesized, and evaluated by FRET for their interaction with the four muscarinic M 1 , M 2 , M 4 , and M 5 mAChR subtypes. This investigation provided additional information on ligand-receptor interactions between mAChRs and bitopic ligands at the molecular level. Hybrid compounds 12-C5, 12-C7, and 12-C9 evidenced a selective activation of the M 1 mAChR, while hybrids 13-C5, 13-C7, and 13-C9 displayed a degree of selectivity for the M 1 as well as M 4 muscarinic receptors, thus showing a subtype activation pattern comparable to that described for Xanomeline. The optimal linker length of seven methylene groups detected for these pyridinium salts reflects a defined distance between the allosteric and the orthosteric binding regions at the M 1 mAChR. The presence of a fixed positive charge on the orthosteric moiety of the hybrid limits the ligand conformational mobility and modulates the degree of receptor activation as a function of the spacer length. This partial agonist profile qualitatively matches those previously assessed in both M 1 and M 2 subtypes for various bitopic hybrid derivatives containing the super agonist Iperoxo as the preferential orthosteric moiety. In the case of dual-steric compounds, their partial agonism has been rationalized by the existence of a dynamic equilibrium between the fraction of purely allosteric (inactive) versus dual-steric (active) binding modes.
On the other hand, owing to their weaker interaction at the orthosteric binding site, the tertiary amine hybrids 12-Cn adopted a less defined conformation within the M 1 receptor protein. As a consequence, their receptor activation pattern is almost independent of the spacer length. Notably, the tertiary amine hybrid 12-C3, which contains the shortest linker analyzed in this study, induced a detectable conformational change only at the G i -coupled M 2 and M 4 mAChRs, whereas the permanently charged analog 13-C3 did not show any receptor activation at all the subtypes.
In conclusion, our findings further account for the utility of bitopic molecular probes for an in-depth exploration of dynamic ligand-receptor interactions of mAChR subtypes. More generally, the structural features of rationally designed hybrid ligands, in particular the nature and length of the spacer moiety, may allow for the tuning of a GPCR activation profile, with a chance to accomplish functionally selective agonist/partial agonists of putative therapeutic value.