Low-basicity 5-HT7 Receptor Agonists Synthesized Using the van Leusen Multicomponent Protocol

A series of 5-aryl-1-alkylimidazole derivatives was synthesized using the van Leusen multicomponent reaction. The chemotype is the first example of low-basicity scaffolds exhibiting high affinity for 5-HT7 receptor together with agonist function. The chosen lead compounds 3-(1-ethyl-1H-imidazol-5-yl)-5-iodo-1H-indole (AGH-107, 1o, K i 5-HT7 = 6 nM, EC50 = 19 nM, 176-fold selectivity over 5-HT1AR) and 1e (5-methoxy analogue, K i 5-HT7 = 30 nM, EC50 = 60 nM) exhibited high selectivity over related CNS targets, high metabolic stability and low toxicity in HEK-293 and HepG2 cell cultures. A rapid absorption to the blood, high blood-brain barrier permeation and a very high peak concentration in the brain (Cmax = 2723 ng/g) were found for 1o after i.p. (5 mg/kg) administration in mice. The compound was found active in novel object recognition test in mice, at 0.5, 1 and 5 mg/kg. Docking to 5-HT7R homology models indicated a plausible binding mode which explain the unusually high selectivity over the related CNS targets. Halogen bond formation between the most potent derivatives and the receptor is consistent with both the docking results and SAR. 5-Chlorine, bromine and iodine substitution resulted in a 13, 27 and 89-fold increase in binding affinities, respectively, and in enhanced 5-HT1AR selectivity.


Results and Discussion
Chemistry. Compounds 1a-1q, 1s-1w, 1y-1za, 2b-2f and 3a-3d (Fig. 2) were synthesized using a concise multicomponent protocol (Fig. 1). The van Leusen reaction proceeds via the stepwise cycloaddition of TosMIC to a polarized double bond of the preformed imine. The elimination of p-toluenesulfonic acid from the cyclic intermediate affords imidazole. Aromatic aldehydes were combined with appropriate amines in methanol. When the imine formation equilibrium was achieved, TosMIC and potassium carbonate were added to yield the desired products. Most of the indole-3-carboxaldehydes used were synthesized by the Vilsmeier-Haack formylation of the appropriate indoles. Derivatives 1r and 1x were synthesized by reductive debenzylation of 1y and 1 s, respectively. Compound 1za was synthesized from 1p by the controlled hydrolysis of the cyano group in basic conditions.

Pharmacology.
All the synthesized compounds were tested in a radioligand binding assays to determine their affinity for five receptors: 5-HT 1A , 5-HT 2A , 5-HT 6 , 5-HT 7 and D 2 . The assays were performed via the displacement of the respective radioligands from cloned human receptors, all stably expressed in HEK293 cells (except for 5-HT 2A  Ligands with a high affinity for the 5-HT 7 receptor were selected for a functional assay (as hydrochloride salts). The experiment was performed in HEK293 cells overexpressing 5-HT 7 R, and the ability of tested compounds to increase cAMP production was measured. The cAMP concentration was measured using TR-FRET and a cAMP specific antibody. Lead compounds 1e (AGH-44) and 1o (AGH-107) were screened toward several CNS targets (Eurofins Cerep, Le Bois l'Eveque, 86600 Celle L'Ev-escault, France, www.cerep.fr): α 1 , α 2C , CB 1 , H 1 , D 3 , 5-HT 1B , 5-HT 2B and 5-HT 5A . The intrinsic clearance of 1e and 1o were determined using human liver microsomes according to Obach et al. 45 (see Supporting Information for detailed results). The HEK-293 and HepG2 cell lines viabilities were tested after 72 h of incubation with tested compounds. Doxorubicin (DX), and carbonyl cyanide 3-chlorophenylhydrazone (CCCP) were used as internal standards (see Supporting Information). The pharmacokinetics of compound 1o was tested on CD-1 mice after intraperitoneal injection at 5 mg/kg dose (see Supporting Information). The compound 1o was administered at 0.5, 1 and 5 mg/kg to investigate the ability of the compound to reverse MK-801 Structure-Activity Relationship. After acquiring highly encouraging initial results for compounds 1a and 1b (K i 5-HT7 = 385 nM and 50 nM, respectively, selectivity over 5-HT 1A R > 11-fold, Fig. 2) a literature search made us realize that an arylimidazole 5-HT 7 R ligand was recently described, although no SAR studies were reported 46 . The compound, 5-(4-([ 11 C]methoxyphenyl)-1-methyl-4-nitro-1H-imidazole, 2a, has been developed as a PET radioligand. The high and rapid uptake of 2a to the rat brain after injection was reported which would suggest at least decent ADME properties. The finding further encouraged us to explore the series, so 2a was resynthesized to serve as a reference compound. However, we were not able to reproduce the results presented by Tiwari et al. 46 . In our radioligand displacement binding assays compound 2a did not bind to 5-HT 7 R and 5-HT 1A R (binding curves of compound 2a can be found in Supporting Information). Moreover, a very close analogue of 2a (the only structural difference being the lack of the nitro group) -2b -was also found to be completely inactive. A compound based on the 2,5-dimethoxy-4-iodophenyl core, which is a scaffold common to many 5-HT 2 R ligands 47,48 , 2f was expected to bind to 5-HT 2A R, but this assumption was not validated. Compounds synthesized from unsubstituted and differently methoxy-substituted benzaldehydes (entries 2c-2e) were also not active. Compounds 3a-3d were synthesized to answer the question of whether methylimidazoles coupled to different aromatic cores (3-pyrrole, 2-indole, 5-indole, or 6-indole) could show any 5-HT 7 R activity. This group of compounds did not bind to 5-HT 7 R at all, even 3a which is a close, simplified analogue of 1a did not possess any activity. It was thus evident that indole-3-imidazole (the core structure of series 1) is the molecular scaffold of choice.
The role of the indole nitrogen atom in the binding of the active compounds to 5-HT 7 R was determined by the synthesis of two derivatives. Entry 1c (K i 5-HT7 = 5623 nM) which is a methylated analogue of 1a, and 1d (K i 5-HT7 > 10000 nM) which is a methylated counterpart of 1b clearly showed that the unsubstituted pyrrole fragment plays an important role in receptor binding. Homologues 1b, and 1e-1 g and compounds 1 h, and 1i were synthesized to optimize the length and shape of the amine component. Apparently, the ethyl substituted derivative exhibited optimal binding affinity (1e, K i 5-HT7 = 30 nM). Interestingly, 1 h (R1 = cyclopropyl) and 1 s (R1 = benzyl) were found to be non-active which would suggest a very narrow space in the binding pocket reserved for an alkyl fragment. The modification of the indole containing component resulted in compounds 1j-1 v. Forbidden positions 2-, 6-, and 7-(R7, R5, and R6) can be clearly observed in entries 1k, 1 u, and 1 v, respectively. Compound 1t exhibited moderate activity toward 5-HT 7 R (K i = 122 nM) indicating that appropriate substituents are acceptable at the fourth position of the indole nucleus. Although 1j and 1k (4-bromo-and 6-bromoindole cores, respectively) were found to possess low affinity for 5-HT 7 R, 1n (5-bromo) was a hit compound with K i = 20 nM. Successively, a series of 5-substituted indole-containing derivatives were synthesized. The activity of compounds 1l-1q and 1w showed a very probable involvement of halogen bonding in the binding of the most active derivatives 1m-1o. 5-HT 7 R affinities of methyl and fluoro derivatives 1q and 1 l (44 nM and 302 nM, respectively) were considerably lower than for the halogen series (42 nM, 20 nM, and 6 nM for Cl, Br and I, respectively), indicating the hydrophobicity to be of secondary importance. Compound 1o (5-iodo) became the lead in the series exhibiting K i = 6 nM at the human 5-HT 7 receptor, and 176-fold selectivity over 5-HT 1A R. A very close serotonin analogue 1r bound to 5-HT 7 R with K i = 39 nM, whereas an analogue of 5-CT, 1za displayed K i = 5 nM.
In vitro pharmacology. Functional assay results for the chosen compounds are shown in Table 1.
The receptor profile of the tested compounds shows high selectivity over the tested CNS targets. All compounds from chemotype 1 showed K i > 1000 nM at the 5-HT 2A , 5-HT 6 and D 2 receptors. 1e and 1o were found to be selective toward a set of CNS targets (Table 2). Within the extended receptor panel tested, 1e and 1o showed affinity only for 5-HT 2B (% of inhibition at 1 µM = 93 and 87, respectively), but at a 10 nM concentration, the level of receptor binding was very low (6 and 7%, respectively). Additionally, compounds 1e and 1o showed very promising metabolic stability with metabolic clearance 3.69 and 6.3 ml/min/kg, respectively.
The cytotoxicity of the compounds 1e and 1o was tested by checking the viability of HEK-293 and HepG2 cells after incubation with different concentrations of the compounds in the medium. 1o was found toxic for the cells only in the highest concentration (100 μM), while with 1e the cells remained viable in all tested concentrations (Fig. 3). Preliminary pharmacokinetics and behavioural study. The pharmacokinetic analysis performed by a non-compartmental approach has shown a rapid absorption of 1o after i.p. administration (t max = 15 min) with maximum plasma concentration of 749 ng/mL (~2.2 μM) (Fig. 4). Half of the compound was eliminated from the body within 90 min but brain concentrations of compound 1o decreased slowly with t 0.5 = 130 min. Large volume of distribution (V d /F = 46.6 L/kg) indicated the penetration of 1o into the peripheral compartments. The compound 1o was measurable to the last sampling time (6 h) both in plasma and brain. The compound penetrated the blood-brain barrier after i.p. administration and peaked in the brain and plasma at the same time but the maximum concentration in the brain was higher (C max = 2723 ng/g, approx. 8 μM). The mean residence time (MRT) for compound 1o in brain was longer than in plasma and equalled 60 min and 27 min, respectively. The brain to plasma ratio was high and equal 6.34. The detailed description of the pharmacokinetic experiment can be found in the Supporting Information. In a recent paper by Leopoldo et al. 31 a disposition study for 5-HT 7 R agonists LP-44 and LP-211 after i.p. administration at a dose of 10 mg/kg in mice was reported. Despite 1o reached the C max in approx. 15 min, which is roughly the same for both LP compounds, the maximum concentrations of LP-44 and LP-211 in the brain were found several-fold lower than for 1o, despite the dose used was twice as high as that of 1o. The LP compounds were measurable in plasma for a very short time compared to over 6 hours for 1o which can be attributed to their very low metabolic stability. The blood/plasma ratio for 1o was three-fold higher compared to LP-44 and 9 times higher than for LP-211. Considering the PK study and suitable brain penetration we concluded that 1o has desirable pharmacokinetic profile to be characterized as a compound of choice to study the in vivo pharmacology. The compound dose-dependently reversed MK-801-induced disruption in novel object recognition (NOR) in mice (Fig. 5). This finding is in line with the results obtained earlier for 5-CT 49 . MK-801-induced disruption of novel object recognition in rodents reflects impairment of working memory, regarded as one of the cognitive symptoms of schizophrenia. For Table 2. Anti-target profiles of selected compounds 1e and 1o. a Experiments performed at Eurofins Cerep. logP, logD) was performed. MPO parameter which was calculated according to Wager et al. 50 was found to lie between 4-6 for most of the compounds in the series (Fig. 2). A very promising group of 5-HT 7 receptor agonists has been developed. To make the search for highly active molecules more rational and explain the unusually high selectivity, a binding mode model had to be explored. The proposed binding mode was based on the analysis of the mutual spatial arrangement of particular ligands at the 5-HT 7 receptor homology models. The final analysis and visualizations were performed using a model built on a 5-HT 1B R (4IAR, complex with agonist, ergotamine) template, because it showed the most coherent binding mode, combined with low binding free energies (see Experimental Section). The observed structure-activity relationships were in agreement with the docking results. Quantum chemical calculations of pK a for 1w (performed in Jaguar using DFT functional B3PW91 and the cc-pVTZ basis set) showed an imidazole fragment pK a = 6.9, indicating that the molecules at physiological pH = 7.4 occur in protonated (33%) as well as deprotonated (67%) forms. Two protonation states for each ligand were considered in docking study. The virtual ligand-receptor complexes of the protonated states did not follow the observed SAR, in addition, a more coherent binding mode was found for the unprotonated forms of the compounds. Thus in discussion and visualizations only deprotonated ligand-receptor complexes were used. The indole nucleus in all the docked ligands formed a hydrogen bond with Asp3.32 and aromatic interactions (CH-π or π-π stacking) with the Phe6.51 or Phe6.52 residues (Fig. 6A), whereas the imidazole ring was frequently hydrogen bonded to Arg6.58. It should be emphasized, that among all targets tested in vitro, the arginine in position 6.58 is a unique feature of 5-HT 7 R (Table 3) 51 , which could explain the high selectivity of the discussed compounds.

Molecular modelling.
The substitution of hydrogen to a halogen atom in position 5 of the indole ring, proportionally increased the 5-HT 7 R affinity to the size of the halogen atom. The analysis of binding modes, showed that a significant contribution to the binding can be attributed to the halogen bond created between the ligand and carbonyl oxygen of one or several amino acids of transmembrane helix 5 (TMH5). The geometric parameters for the 5-iodine derivative showed the significant contribution of Ser5.42 (d(I···O) = 3.1 Å, sigma hole angle (C-I···O) = 155°) and Thr5.39 (d(I···O) = 4.4 Å, sigma hole angle (C-I···O) = 171°) (Fig. 6A). Additionally, the order of the binding free energy changes (after chlorine, bromine and iodine substitution ∆∆G was −6.6, −7.2 and −10.1 kcal/mol, respectively) and the interaction sphere plotted onto the relevant backbone carbonyl oxygen (all X···O=C interactions were within the energetically favourable areas of the sphere, Fig. 6A and B) confirmed the role of halogen bonding  in the formation of the L-R complexes. The interaction spheres plotted onto the backbone carbonyl oxygen of Ser5.42 were also used to explain the decreased activity upon shifting the bromine atom from position 5 to 4 or 6 in the indole ring (Fig. 6B). The obtained interaction sphere showed that 5-Br derivative (d(Br···O) = 3.2 Å, sigma hole angle (C-Br···O) = 159°) lies within the energetically favourable areas of the sphere, while the 4-and 6-Br derivatives bromine atoms were pointed outside the sphere, ruling out the formation of halogen bond. This observation is consistent with calculated binding free energy values: only 5-bromo derivative 1n showed a gain in the binding energy (ΔΔG = −7.6 kcal/mol), compared to the 4-bromo and 6-bromo analogues (ΔΔG = +0.9 and +1.8 kcal/mol, respectively).
The significance of the hydrogen bond formed between the indole NH group and the Asp3.32 side chain can be observed by comparing compounds 1e and 1d (Figs 2 and 6D). The substitution of hydrogen with a   −CH 3 group resulted in an approximately 580-fold decrease in binding affinity because no hydrogen bond can be formed with Asp3.32. The calculated ΔΔG for the methylated derivative (1d) showed the stabilization of the L-R complex which is in sharp contrast with the experimental K i value (>10000 nM).
Because the discussed compounds are structurally very similar to the class of tryptamines that are known to be promiscuous serotonergic agents, representative tryptamine derivatives were purchased and their binding affinities for 5-HT 7 , 5-HT 1A , 5-HT 2A , 5-HT 6 and D 2 receptors were determined (Fig. 7).
It is clear that tryptamines (tryptamine as an analogue for 1a; 5-methoxytryptamine for 1b, 1e; 5-chlorotryptamine for 1m, and 5-methyltryptamine for 1q) exhibit multi-receptor profiles, i.e. a high affinity for 5-HT 7 , 5-HT 1A , 5-HT 2A , and 5-HT 6 receptors. The introduction of halogen at the 5 th position of indole in tryptamines resulted in an enhancement of the affinity for 5-HT 7 R, but not nearly as much as the introduction of a methoxy group (introduction of the chlorine and methoxy groups had very similar impacts on the binding affinity of the title imidazoles). The methylation of the indole NH group in tryptamines resulted in a dramatic drop in potency for all four tested serotonin receptors. It was rather unusual that the tryptamines exhibited very similar 5-HT 7 R SARs while exhibiting distinctive multi-receptor profiles. The proposed binding mode for the imidazole compounds was compared with several tryptamines, which were docked to the best 5-HT 7 R homology models using the same procedure (Fig. 8). The binding mode of the tryptamine derivatives was similar to the findings of Vermeulen et al. 52 and was different from the one found for the title compounds. The tryptamines formed a salt bridge (charge assisted hydrogen bond) between the protonated amine moiety and Asp3.32, but did not interact with Arg6. 58 Fig. 6C) indicated the formation of a hydrogen bond between the hydroxyl group of Ser5.42 and the oxygen atom of substituents. Interestingly, shifting the -OCH 3 group from position 5 to 4, resulted in only a slight decrease in activity, which can be linked to creation of a hydrogen bond with Arg6.58 (Fig. 6C).
Apparently, it is possible to synthesize heterocyclic analogues of tryptamines that maintain high affinity and agonist function at 5-HT 7 R. Modelling studies revealed that the incorporation of an amine moiety into an imidazole ring resulted in a distinctive binding mode. The difference may be attributed partially to the lower basicity of imidazoles versus tryptamines. An interaction with Arg6.58, which is a residue unique for 5-HT 7 R may underlay the observed dramatic effect on the selectivity.
Possible application as a PET radioligand.  receptor is yet to be PET imaged in humans 53 . No suitable 5-HT 7 R agonist radioligand candidates were reported till date 54,55 . Serotonin receptors exist in high and low agonist affinity states. The antagonist ligands bind to the high affinity (HA) and low affinity (LA) conformations with similar affinity, whereas agonist ligands bind preferentially to the HA state of the receptor, which is coupled to G-protein and therefore agonist binding provides a more meaningful functional measure of the serotonin receptors activity. Although antagonist PET tracers can measure the total receptor binding, they cannot detect changes in the high affinity receptor binding in disease states. Antagonist binding is insensitive to the changes in the intra-synaptic 5-HT concentration and antagonist PET tracers are less sensitive for measuring agonist receptor occupancy in clinical studies to guide new drug development in dose-finding 56 .
Because of rapid penetration to the brain and quick elimination from the CNS, 1o showed desirable properties as perspective PET radiotracer. Therefore, a potential synthetic pathway leading to the 11 C-labeled version of one of the lead compounds (1e) starting with commercially available substrates with only two concise steps involving "hot" chemicals has been developed (Fig. 9). The developed synthetic scheme included the sequential introduction of protecting groups to achieve the desired regioselectivity (Fig. 9). 5-Benzyloxyindole-3-carboxaldehyde was subjected to the MCR protocol to yield 1y. This compound was N-Boc substituted to protect the indole NH group from alkylation during one of the last steps. The derivative was reductively debenzylated yielding tert-butyl 3-(1-ethyl-1H-imidazol-5-yl)-5-hydroxy-1H-indole-1-carboxylate, ready for alkylation with 11 CH 3 I. After alkylation with a non-radioactive CH 3 I, Boc group was easily cleaved using either hydrochloric or trifluoroacetic acid. A similar strategy, involving synthesis of a N-Boc protected indole derivative, ready for the introduction of an isotopically labelled alkyl fragment to the 1-position of imidazole, may be employed for the synthesis of 1o, or other 5-HT 7 R agonists.

Conclusions
To date there have been no examples of low-basicity agonists of serotonin receptors. We report a series of highly drug-like, low-basicity 5-HT 7 R agonists synthesized using MCR imidazole synthesis. The procedure described by van Leusen proceeded under various conditions: a range of solvents and bases are tolerable. Notably, the scope of the carbonyl and amine reagents is very broad. Hydroxy-, amino-, amido-, ester and other reactive groups do not interfere with the reaction.
There are several highly active, selective 5-HT 7 R ligands within the presented series. Hit compound 1o exhibits high affinity towards the receptor (K i = 6 nM), intrinsic activity (EC 50 = 19 nM), low intrinsic clearance (Cl int = 6.3 ml/min/kg, human liver microsomes) and high predicted drug-likeness. These facts combined with an exceptional ease of synthesis may contribute to the use of radiolabeled 1o and 1e as potential molecular probes. 1e has 5-HT 1A /5-HT 7 receptor selectivity comparable to known 5-HT 7 R agonists, whereas 1o is 176-fold selective. Both compounds are very selective towards 5-HT 5A which is not a common feature of 5-HT 7 R agonists. The compounds are metabolically stable in human liver microsomes, exhibit very low toxicity in HEK-293 and HepG2 cells and are water soluble. The blood-brain barrier permeation of 1o was found very high, which is consistent The proposed binding mode was developed by SAR analysis and homology modelling and involves the formation of a hydrogen bond between the indole NH group and Asp3.32. The imidazole fragment is proposed to interact with Arg6.58 which could explain the unusually high selectivity that is not observed in tryptamines. The similar 5-HT 7 R SAR patterns of tryptamines and the described compounds can be attributed to the interaction of the substituents of indole at the 5 th position with Ser5.42 leading to comparable binding affinity gains for the same substituents in the two series.
Structures of NCE's in the area of medicinal chemistry have become increasingly complex due to the widespread application of the hit-to-lead strategy and the rising palette of structural modifications that enable ADME optimization. We suppose that a better strategy for radioligand searches may be to find simple scaffolds exhibiting moderate to high activity that upon decoration with substituents would retain or improve their affinity for a given target, rather then finding sophisticated structures via virtual or high-throughput screening that upon simplification may not yield any active scaffolds. Because the available chemical space of compounds with low molecular mass (MW < 250 Da) that pass drug likeness filters can be considered inexhaustible taking current synthetic methods into account, we suppose that there are still numerous privileged structures to be discovered. Quoting Richard Feynman: "There's plenty of room at the bottom".

Methods
Chemicals. All organic reagents were purchased from Sigma-Aldrich, Apollo Scientific Apollo Scientific Ltd, or Combi-Blocks and were used without purification. Solvents and inorganic reagents were acquired from Chempur. Reaction progress was monitored by TLC on Merck Silica Gel 60 F 254 on aluminium plates. Column chromatography was performed on Merck Silica Gel 60 (0.063-0.200 mm; 70-230 mesh ASTM).

Analytical Methods. UPLC/MS analysis was performed on Waters TQD spectrometer combined with
UPLC Acquity H-Class with PDA eLambda detector. Waters Acquity UPLC BEH C18 1.7 μm 2.1 × 100 mm chromatographic column was used, at 40 °C, 0.300 ml/min flow rate and 1.0 μL injection volume (the samples were dissolved in LC-MS grade acetonitrile, typically at a concentration of 0.1-1 mg/ml prior to injection). All mass spectra were recorded under electrospray ionization in positive mode (ESI+) and chromatograms were recorded with UV detection in the range of 190-300 nm. The gradient conditions used were: 80% phase A (water+ 0.1% formic acid) and 20% phase B (acetonitrile + 0.1% formic acid) to 100% phase B (acetonitrile + 0.1% formic acid) at 3.0 minutes, kept till 3.5 minutes, then to initial conditions until 4.0 minutes and kept for additional 2.0 minutes. Total time of analysis -6.0 minutes. 1 H and 13 C NMR spectra were recorded on a Bruker Avance III HD 400 NMR spectrometer. All samples were dissolved in DMSO-d 6 with TMS as the internal standard. The spectral data of the compounds refer to their free bases.
All presented compounds were of at least 95% purity as determined by LC-MS. Syntheses and characterization details for intermediate products and final compounds as well as the spectral data for all compounds is included in the Supporting Information.  1a-1q, 1s-1 v, 1y-1za, 2b-2f, and 3a-3d. Aromatic aldehyde (3 mmol) was mixed with amine (15 mmol) in 20 ml dry methanol. Reaction mixture was left overnight to complete the imine formation although it can be TLC monitored (SiO 2 /CHCl 3 ). Anhydrous K 2 CO 3 (3 mmol) and TosMIC (tosylmethylisocyanide, 3 mmol) were subsequently added. The mixture was stirred for an additional 8 hours, diluted with 50 ml H 2 O, and extracted three times with 20 ml ethyl acetate. The combined extracts were washed twice with 20 ml H 2 O, and once with 20 ml brine, treated with anhydrous magnesium sulfate and evaporated. The final products were purified either by trituration under a 2:1 hexane:isopropanol mixture, or chromatography on a short silicagel bed. The unreacted aldehydes were eluted with ethyl acetate or chloroform, and then a mixture of AcOEt:MeOH or CHCl 3 :MeOH was applied to elute the product.

General Procedure 2 for the Synthesis of Substituted Indole-3-carboxaldehydes. The
Vilsmeier-Haack reagent was generated by the addition of 2 ml of POCl 3 over the course of 15 minutes to 8 ml of dry DMF cooled in an ice-salt bath. After the addition was complete, the ice bath was removed and the contents of the flask were left to warm to room temperature over approx. 30 minutes. The substituted indole (21.9 mmol) was dissolved in 10 ml of DMF and added over a period of 15 minutes to the formylating mixture. The stirring was continued for an hour during which the flask contents were heated to 40 °C in a hot water bath. A total of 20 ml of 5 M NaOH was added, and the mixture was diluted with H 2 O, quickly brought to a boil and left to cool slowly. The crystals were removed by filtration, washed with cold water and vacuum dried. The products thus obtained were in most cases sufficiently pure for the subsequent reactions. And the impure aldehydes were recrystallized from ethanol-water mixtures. The yields varied from 48 to 90%.

General Procedure 3 for the Synthesis of Substituted 1-Alkylindole-3-carboxaldehydes.
A total of 10 mmol of an appropriate indole-carboxaldehyde, 50 ml of toluene, 36 ml of 30% NaOH solution, 1 mmol (322 mg) tetrabutylammonium bromide (TBAB) and 10.5 mmol alkyl iodide (0.66 ml of methyl iodide, for instance) were placed in a stoppered flask. The mixture was stirred until the substrate disappeared on TLC (on average 12 h). After completion of the reaction, the phases were separated, next the organic extract was washed with 30 ml of H 2 O followed by 30 ml of brine, and was dried over MgSO 4 , and the toluene was stripped off in a rotary evaporator. The crude product was triturated twice with 5 ml of hexane:isopropanol 2:1 mixture, and was vacuum filtered and dried. The yields varied from 85 to 97%.
Unsubsituted indole-3-carboxaldehyde was synthesized at full scale according to the OrgSyn procedure 56 .

General Procedure 4 for the Synthesis of 1r and 1x (Reductive Debenzylation). The benzyl pro-
tected compound was dissolved in methanol in a pressure reactor, and 10 mol% Pd/C was added. The reactor was sealed and pressurized to 7 bar with H 2 , and the completion of the reaction was monitored with LC-MS. The mixture was then filtered through Celite which was then washed with MeOH. The solvent was stripped off on a rotary evaporator and the resulting solid was triturated under a hexane:isoporopanol:acetone 3:1:1 mixture. The yields were not lower than 90%.
A suspension of 1p (3.4 mmol, 0.8 g) in 1 ml MeOH and 5 ml THF was stirred in an ice-salt bath for 15 minutes until 0 °C was reached. Hydrogen peroxide (30% solution, 5.4 ml) was added dropwise keeping the temperature below 10 °C. Stirring was continued for 15 minutes and sodium hydroxide (20% solution, 5.4 ml) was added dropwise keeping temperature below 10 °C. The mixture was allowed to warm to room temperature and stirred for 24 hours. The product was extracted from the reaction mixture with chloroform and purified by trituration with acetone. Yield: 81 mg (9%). (4). To a round bottom flask there was added: THF (15 ml), 1y (410 mg, 1.29 mmol), di-tert-butyl dicarbonate (590 mg, 2.7 mmol, 2.1 eq) and a catalytic amount of DMAP. The mixture was stirred until disappearance of the substrate on TLC. H 2 O (20 ml) and ethyl acetate (20 ml) were added, the phases separated, the aquous phase extracted twice with 10 ml ethyl acetate. The organic phase was washed with brine, dried over MgSO 4 and evaporated. Crude product was of >95% purity and was used in the next step without purification. Yield: 485 mg (85%). An analytical sample was obtained via column chromatography (SiO2, ethyl acetate:hexane 2:1, then ethyl acetate). (5). 4 (200 mg, 0.48 mmol) was dissolved in 50 ml MeOH in a glass reactor. 10% Pd/C (50 mg) was added and the reactor was pressurized to 7 bar with hydrogen. The mixture was stirred overnight, filtered through celite and the celite bed was washed thrice with 20 ml MeOH. The solvent was evaporated and the residue was subjected to column chromatography (SiO 2 , ethyl acetate). Yield: 57 mg (36%). (6). To a round bottom flask there were added: 3 ml toluene, 5 (100 mg, 0.305 mmol), CH 3 I (20 μL, 0.32 mmol, 1.05 eq), TBAB (10 mg, 0.0305 mmol, 0.1 eq.) and 1 ml 30% NaOH solution. The mixture was stirred and the progress of the reaction was TLC monitored. The mixture was diluted with 10 ml H 2 O, extracted three times with 10 ml ethyl acetate, the combined organic phase was washed with 10 ml H 2 O, 10 ml brine, dried over MgSO 4 and evaporated. The crude product was purified via column chromatography (SiO 2 , ethyl acetate). Yield: 26 mg (25%).
In Vitro Pharmacology. Cell Culture. HEK293 cells with stable expression of human serotonin 5-HT 1A R, 5-HT 6 and 5-HT 7b R or dopamine D 2L R (obtained using of Lipofectamine 2000, Invitrogen) or CHO-K1 cells with plasmid containing the sequence coding for the human serotonin 5-HT 2A receptor (Perkin Elmer) were maintained at 37 °C in a humidified atmosphere with 5% CO 2 and were grown in Dulbecco's Modified Eagle's Medium containing 10% dialyzed fetal bovine serum and 500 μg/ml G418 sulfate. For membranes preparations, cells were subcultured into 150 cm 2 cell culture flasks, grown to 90% confluence, washed twice with phosphate buffered saline (PBS) prewarmed to 37 °C, pelleted by centrifugation (200 g) in PBS containing 0.1 mM EDTA and 1 mM dithiothreitol, and stored at −80 °C 57-59 . All assays were incubated in a total volume of 200 μL in 96-well microtiter plates for 1 h at 37 °C, except for 5-HT 1A R and 5-HT 2A R, which were incubated at room temperature and 27 °C, respectively. The equilibration process was terminated by rapid filtration through Unifilter plates with a 96-well cell harvester and the radioactivity retained on the filters was quantified using a Microbeta plate reader (PerkinElmer, USA).

5-HT
For the displacement studies, the assay samples contained the following as radioligands (PerkinElmer, USA):  4 Ci/mmol). Non-specific binding was defined using 10 μM of 5-HT in 5-HT 1A R and 5-HT 7 R binding experiments, whereas 20 µM of mianserin, 10 μM of methiothepine or 1 μM of (+) butaclamol was used in the 5-HT 2A R, 5-HT 6 R and D 2L R assays, respectively. Each compound was tested in triplicate at 7-8 concentrations (10 -11 -10 -4 M). The inhibition constants (K i ) were calculated using the Cheng-Prusoff equation 62 and the results were expressed as the means of at least two independent experiments. Functional Assay. The functional properties of compounds in HEK293 cells overexpressing 5-HT 7 R were evaluated for their ability to increase cAMP production for the agonists or to inhibit 10 nM 5-CT at a concentration producing 90% (EC 90 ) of the maximum agonist activation for the antagonists.
Each compound was tested at 8 concentrations in the range of 10 −11 -10 −4 M. Cells (prepared with the use of Lipofectamine 2000) were maintained at 37 °C in a humidified atmosphere with 5% CO 2 and were grown in DMEM containing 10% dialyzed (FBA) and 500 µg/ml G418 sulfate. For the functional experiments, the cells were subcultured in 25 cm 2 diameter dishes, grown to 90% confluence, washed twice with PBS prewarmed to 37 °C and centrifuged for 5 min (160 × g). The supernatant was aspirated, and the cell pellet was resuspended in stimulation buffer (1 × HBSS, 5 mM HEPES, 0.5 mM IBMX, and 0.1% BSA). The total cAMP was measured using the LANCE cAMP detection kit (PerkinElmer), according to the manufacture's directions. For the cAMP level quantification, cells (5 µl) were incubated with compounds (5 µl) for 30 min at room temperature in a 384-well white opaque microtiter plate. After incubation, 10 µl working solution (5 µl Eu-cAMP and 5 µl ULight-anti-cAMP) was added to stop the reaction and induced cells lysis. The assay plate was incubated for 1 h at room temperature and time-resolved fluorescence resonance energy transfer (TR-FRET) was detected by an Infinite M1000 Pro (Tecan, Männedorf, Switzerland) using instrument settings from the LANCE cAMP detection kit manual.  63 . Sequences of the modeled receptor and selected templates were aligned manually using Accelrys Discovery Studio 64 , making sure that the most conserved amino acid in each helix, and the motif characteristic for class A GPCRs were in equivalent positions. The ranges of helices were determined on the basis of the crystal structures, and the loop regions were modeled, but not refined. For each template, 20 models were generated using Modeller software 65 , and were tested by the docking of active and inactive compounds, and enrichment calculation. For every model, ROC curves were calculated based on the Glide Score values of the docked compounds (the undocked actives and inactives were considered as false negatives and true negatives, respectively). The quality of the models was determined by the Area Under the ROC curve (AUROC). Three models per template with the highest AUROC value were selected for further studies. The final validation of the selected receptor models was performed by using the QM/MM molecular docking (QPLD Protocol) of the whole set of studied molecules. The models for which binding modes coherent for the whole set of compounds and explaining the main structure-activity relationships were observed were kept.

CNS MPO Calculation. The Central Nervous System Multiparameter Optimization (CNS MPO) scores
were calculated based on the six basic physicochemical molecular descriptors: logP, logD, molecular weight (MW), topological polar surface area (TPSA), the count of hydrogen bond donor (HBD), and the strongest basic pKa. All descriptors were calculated using Chemaxon's Calculator Plugins 66 . A monotonically decreasing function was used to transform the values of the descriptors into dimensionless scale logP, logD, MW, HBD, and pKa whereas a hump function was used to transform TPSA. The most desirable and least desirable ranges for each physicochemical descriptor were adopted from the original paper 50 . Transformed values (desirability score) of the six descriptors were determined and summed up for each compound, to obtain the final CNS MPO, which can range from 0 (the worst CNS drug) to 6 (best CNS drug). QM/MM Docking Protocol. The Quantum Mechanics/Molecular Mechanics (QM/MM) docking of all synthesized compounds to 5-HT 7 R homology models were performed by means of the QM-Polarized Ligand Docking Protocol (QPLD) 67 from Schrödinger Suite. At the first stage of the QPLD procedure the ligands were initially docked into a rigid protein using Glide. The resulting binding modes of the ligands were then used for the calculation of the partial charges of the ligand by a single-point calculation in QSite treating the ligand with the ab initio (B3PW91/cc-pVTZ) and the receptor with MM (OPLS-2005) level of theory. The partial atomic charges of the ligand molecules were recalculated using the electrostatic potential fitting method under the influence of external fields exerted by the surrounding protein atoms within and near the binding site. At the final stage, Glide re-docked the ligands using new charges, and the 10 most energetically favorable poses were returned for each ligand.
Binding Free Energy Calculation. MM-GBSA (Generalized-Born/Surface Area) was used to calculate the binding free energy for all ligand-receptor complexes to select the correct poses between possible binding modes. The binding free energy was calculated using ligand charges obtained via the QM/MM calculations using Prime software from Schrodinger. To assess the influence of a given substituent on the binding, the ΔΔG was calculated as a difference between binding free energy (ΔG) of unsubstituted (1w) and substituted analogues.

Plotting Interaction Spheres for Halogen Bonding.
To visualize (plotting interaction spheres) the possible contribution of halogen bonding for the resulting ligand-receptor complexes, the halogen bonding webserver was used (access 1.4.2016, http://www.halogenbonding.com/).

Associated Content. Syntheses and characterization details for intermediate products and final compounds,
cell culture and transfection methods, pharmacological data for compound 2a, detailed description of: metabolic stability assays for compound 1e and 1o, cytotoxicity assays, metabolic stability assays, pharmacokinetic study, novel object recognition experiment, 1 H NMR, 13 C NMR and LC-MS spectra (PDF).