Impact of N-Alkylamino Substituents on Serotonin Receptor (5-HTR) Affinity and Phosphodiesterase 10A (PDE10A) Inhibition of Isoindole-1,3-dione Derivatives

In this study, a series of compounds derived from 4-methoxy-1H-isoindole-1,3(2H)-dione, potential ligands of phosphodiesterase 10A and serotonin receptors, were investigated as potential antipsychotics. A library of 4-methoxy-1H-isoindole-1,3(2H)-dione derivatives with various amine moieties was synthesized and examined for their phosphodiesterase 10A (PDE10A)-inhibiting properties and their 5-HT1A and 5-HT7 receptor affinities. Based on in vitro studies, the most potent compound, 18 (2-[4-(1H-benzimidazol-2-yl)butyl]-4-methoxy-1H-isoindole-1,3(2H)-dione), was selected and its safety in vitro was evaluated. In order to explain the binding mode of compound 18 in the active site of the PDE10A enzyme and describe the molecular interactions responsible for its inhibition, computer-aided docking studies were performed. The potential antipsychotic properties of compound 18 in a behavioral model of schizophrenia were also investigated.

The phosphodiesterase 10A (PDE10A) enzyme is highly expressed in the striatal medium spiny neurons (MSNs), where it modulates both dopamine D 2 -and D 1 -dependent signaling. Striatal MSNs pathways are divided into direct dopamine D 1 receptor-mediated and indirect dopamine D 2 receptor-mediated striatal output pathways [19]. Dopamine D 1 receptors in direct pathway MSNs are coupled with G s proteins, which activate adenylate cyclase to increase cyclic adenosine monophosphate (cAMP) concentration, whereas the stimulation of D 2 receptors inhibits the activity of adenylyl cyclase, resulting in a decrease in the level of cAMP. On the other hand, inhibition of PDE10A in the dendritic membranes of MSNs results in suppression of D 2 -mediated signaling, which is similar to the effects of D 2 antagonists. Upregulation of cyclic nucleotide levels in indirect pathway MSNs could be an alternative approach for novel antipsychotics [20,21], whereas activation of direct pathway MSNs could be associated with improvement of cognitive functions. Thus, in contrast to the marketed antipsychotics, phosphodiesterase inhibitors (PDEIs) influence both the indirect and direct pathways, and the additional action of these compounds on the direct pathway could help distinguish them clinically from D 2 antagonists [22]. Moreover, PDEIs may have potential therapeutic utility in the treatment of cognitive impairment associated with schizophrenia or may be useful as an adjunctive treatment for negative symptoms in patients with schizophrenia [19].
In complex diseases such as schizophrenia, single-target drugs have turned out as failures, whereas multi-target drugs have been found to be much more efficacious. In clinical practice, many combinations of psychotropics have been tried for schizophrenia, based on a trial-and-error paradigm guided by clinical experience and patient response. An overall superior and safer solution seems to be the modern concept of designer multiple-targeting ligands (DMLs). The DMLs are also described as "polypharmacology by design", which claims parallel modulation of defined multiple biological targets by designer small molecule drugs [23]. An exemplification of this idea is atypical antipsychotics, which target a number of G protein-coupled receptors (GPCRs) simultaneously. Currently, novel strategies in designing drugs effective in treatment of schizophrenia focus on targets beyond the dopaminergic hypothesis of the disease [24].
The series of designed ligands were synthesized and evaluated in adequate biological assays, a luminescence PDE inhibition assay, as well as a radioligand binding assay for 5-HT 1A and 5-HT 7 receptors. Computer-aided studies were undertaken to identify the structural elements responsible for in vitro activity. Finally, for the most active compound, Compound 18, in vitro safety studies were performed and preliminary in vivo activity was tested in an animal model of schizophrenia.
In addition, the relevance of the introduction of a 4-methoxyl group to the 4-methoxy-2,3-dihydro-1H-isoindole-1,3-dione fragment was also verified. For this purpose, a small phthalimide series without a 4-methoxy group at position 4 ( Table 1) connected via an alkylene linker to the 1H-benzimidazole moiety (20-23) was synthesized. The activity of compounds 20-23 was significantly lower than that of their 4-methoxy analogs (15,(17)(18)(19). Moreover, similar to the 4-methoxy series, the inhibitory activity of compounds 20-23 increased with the length of the linker up to four methylene units and then slightly decreased.
In the next step, the five most active compounds chosen (15,(17)(18)(19)22) were quantitatively tested in the luminescence PDE inhibition assay, alongside papaverine as the reference drug. The effectiveness of inhibition was expressed as the half maximal inhibitory concentration (IC 50 ) values. The chosen compounds displayed a percentage of inhibition above 70% at the concentration of 10 µM. All investigated compounds (15,(17)(18)(19)22) displayed satisfactory IC 50 values (886 nM-6.71 µM, Table 1 and Figure 3) in comparison with papaverine (5.755 µM). It is noteworthy that the most potent PDE10A inhibitor, compound 18 (IC 50 = 886 nM), was in-line with the predictions stipulated by Lipinski's Rule of Five, which describes "drug-like" parameters of small molecules in terms of lipophilicity, size, polarity, solubility, flexibility, and saturation [35] (see graph, Figure S1, in Supporting Materials).
Furthermore, because compound 18 contains structural moieties such as benzimidazole and phthalimide, which are also present in serotonin 5-HT 2A and dopamine D 2 receptor antagonists [14], it was decided to test the affinity of Compound 18 to these receptors. The obtained results (appendix below Table 1) revealed that at a concentration of 1 µM compound 18 displayed negligible percentage of binding for 5-HT 2A and none for D 2 receptors (8.8% and 0%, respectively).

Molecular Modeling Studies
In line with in vitro results, the most active 4-methoxy-2,3-dihydro-1H-isoindole-1,3-dione derivative with a bezimidazol-2-yl butyl moiety (18) was selected to clarify and draw useful information regarding the nature of the interactions of PDE10A with its ligands, and insights into structure-activity relationships were investigated. We elucidated putative binding modes of several derivatives throughout the process of docking to the protein target, which was represented by the optimized crystal structure of the studied enzyme. As a representative example, binding interactions were presented for compound 18 ( Figure 4). . Crucial interactions of the compound in the catalytic site with Gln716 and Phe719, as well as its positioning in the selectivity pocket (Tyr683) predict inhibitory activity of the series of rationally-designed derivatives synthesized in this study. Amino acid residues engaged in ligand binding (within 4 Å from the ligand atoms) are displayed as sticks, whereas crucial residues, e.g., forming H-bonds (dotted yellow lines) or π-π stacking (dotted cyan lines), are represented as thick sticks. The 2,3-dihydro-1H-isoindole-1,3-dione moiety, bound in the catalytic active site, interacted with Gln716 (H-bond) and with Phe719 (π-π stacking). In the case of the most active compound (18), the 4-methoxy group acted as the H-bond acceptor, whereas unsubstituted analogs interacted through the carbonyl group of the imide moiety. The oxygen atom of the phthalimide was less electronegative than the one of the methoxy group, thus making the H-bond accepting properties weaker and disadvantageous in comparison with that of the 4-methoxy-substituted analogs. Moreover, the H-bond formed by the latter group attracted the heteroaryl ring of the ligand closer to Phe719/686, thereby facilitating the π-π interaction in the hydrophobic clamp. The 1H-1,3-benzodiazole fragment accepted the H-bond from Tyr683 in the selectivity pocket. The interaction was secured by the optimal length of the linker-4 carbon atoms and resulted in favorable scoring function value −8.885 (glide gscore) (Figure 4). Both longer and shorter aliphatic chains decreased the inhibitory potential (see Table 1), enforcing suboptimal arrangement of the terminal moieties in the active site, which resulted in the loss of one of the aforementioned key interactions (e.g., in the case of compound 19) or decrease in scoring function value (−8.254 for compound 15).

Cytotoxicity
In the subsequent stage of the study, the safety profile of the most active compound, 18, was analyzed. Such analysis requires the exclusion of adverse effects of organ toxicity before proceeding to further in vivo studies. According to the guidelines for the study of biologically active molecules, cell lines are a good alternative to animal models at the initial stage of the drug cytotoxicity evaluation process. The two cell lines used in this study, namely, the SHSY-5Y and HepG2 cell lines are approved in vitro models, which can be used to obtain preliminary results regarding possible neuroand hepato-toxicity [36].
The MTT assay, which measures the metabolic status of cells, was used to determine cytotoxicity. The results of the experiments clearly indicate that compound 18 is safe. Decreasing cell viability (HepG2/SHSY-5Y) is observed only at higher concentrations of compound 18, while the reference standard, doxorubicin (DOX), a well-described cytotoxic agent, demonstrated significantly decreased cell viability already at much lower concentrations ( Figure 5).

Behavioral Evaluation
For further behavioral characterization, compound 18, with the highest PDE10A inhibitory activity and satisfactory cytotoxicity profile, was selected. Thus, the potential antipsychotic properties of compound 18 as an inhibitory effect on PDE10A in a behavioral animal model was determined. The obtained results revealed that treatment of mice with d-amphetamine (AMPH) significantly increased horizontal locomotor activity throughout the duration of the test. Papaverine dose-dependently antagonized this response at both tested doses of 40 and 60 mg/kg for 60 min. Compound 18 administered at a dose of 60 mg/kg decreased AMPH-induced hyperactivity, but only for the first 10 min after injection ( Figure 6). The lower doses of compound 18, as well as papaverine, did not change the stimulating effect of AMPH (data not shown). To assess if a nonspecific suppression of locomotion contributes to the ability of the tested compounds to oppose AMPH-induced hyperactivity, the influence of these compounds on spontaneous locomotor activity was measured. The locomotor-suppressing effects were observed after the administration of papaverine (40 and 60 mg/kg) as well as 18 (60 mg/kg) (see Table S1 in Supporting Materials). In summary, the suppression of AMPH-induced hyperlocomotion suggests antipsychotic-like properties of compound 18, as well as of papaverine. However, these effects are transient and may only reflect the decrease in spontaneous locomotor activity. Our findings are consistent with previous studies conducted with papaverine in rats [19,37], showing that the period of papaverine-induced suppression of spontaneous locomotor activity fully included time of reversal of hyperlocomotion produced by AMPH injection. Similar findings are reported by Schmidt et al. [38] for compound TP-10, which is a PDE-10 inhibitor with improved potency and selectivity. The aforementioned results suggest that antipsychotic-like properties of PDE10A inhibitors detected with locomotor-based paradigms need to be treated with caution, due to concurrent inhibition of spontaneous locomotor activity. Moreover, at this stage of our preliminary study, we can only speculate that the cause of compound 18 s transient effect on AMPH-induced hyperlocomotor activity may be a result of either rapid redistribution to the other brain structures or rapid metabolism. In order to overcome possible pharmacokinetic and metabolic stability problems, we plan to further optimize the leading structure of compound 18 and conduct extended in vitro and in vivo investigations.

Chemistry
All chemicals and solvents were purchased from commercial suppliers (Aldrich, Poznań, Poland and Chempur, PiekaryŚląskie, Poland) and were used without further purification. Melting points were measured in open capillaries on an Electrothermal 9300 apparatus. Thin-layer chromatography (TLC) was run on Merck silica gel 60 F 254 aluminium sheets (Merck; Darmstadt, Germany), using the following mixtures of solvents: (S 1 ) dichloromethane (9)/methanol (0.3), (S 2 ) dichloromethane (9)/methanol (0.7), (S 3 ) dichloromethane (9)/methanol (1). Analytical HPLC was conducted on a Waters HPLC instrument with a Waters 485 Tunable Absorbance Detector UV, equipped with a Symetry column (C18, 3.5 µm, 4.6 × 30 mm) using a water/acetonitrile gradient with 0.1% trifluoroacetic acid (TFA) as the mobile phase at a flow rate of 5 mL/min. Additionally, the liquid chromatography/mass spectrometry (LC/MS) analysis was performed on a Waters Acquity TQD system, with a Waters TQD quadrupole mass spectrometer with detection by UV (DAD) using an Acquity UPLC BEH C18 column (1.7 µm, 2.1 mm × 100 mm). A water/acetonitrile gradient with 0.1% TFA was used as a mobile phase at a flow rate of 0.3 mL/min. The UPLC/MS purity of the investigated compounds  proved to be over 98%. NMR spectra were recorded on a Varian Mercury 300 MHz spectrometer (Varian Inc., Palo Alto, CA, USA) using the solvent (CDCl 3 or DMSO-d 6 ) signal as an internal standard; chemical shifts are expressed in parts per million (ppm). Signal multiplets are represented by the following abbreviations: s (singlet), brs (broad singlet), d (doublet), t (triplet), m (multiplet).

Protocols for Measuring PDE10A Inhibition In Vitro
Test and reference compounds were dissolved in dimethyl sulfoxide (DMSO) at a concentration of 1 mM and further diluted in assay buffer in order to obtain the final concentration of compounds: 10 µM and 3 µM, respectively. All reactions were carried out at 37 • C in white, half-area 96-well plates (Perkin Elmer, Waltham, MA, USA). The inhibition of PDE 10A enzyme was measured using the PDElight HTS cAMP phosphodiesterase assay kit (Lonza) according to the manufacturer's recommendations. 2.5 U of PDE 10A enzyme was preincubated either with DMSO (vehicle control) or compound for 20 min before incubation with the substrate, cAMP (final concentration 1.25 µM), for 1 h. Then, PDELight AMP Detection Reagent was added. After 10 min incubation, the luminescence was measured in a multifunction plate reader (POLARstar Omega, BMG Labtech, Ortenberg, Germany). The results were expressed as percent of inhibition or, for the most active compounds selected, as the half maximal inhibitory concentration values.

Radioligand Binding Studies
Radioligand binding studies with serotonin 5-HT 1A and 5-HT 7 receptors were conducted according to methods previously described [41,42], whereas percent of inhibition of the control binding for 5-HT 2A and D 2 were performed according to protocols described online (www.cerep.fr). Briefly: binding experiments were conducted in 96-well microplates in a total final volume of 250 µL of appropriate buffers. Reaction mix included 50 µL solution of test compound, 50 µL of radioligand ([ 3 H]8-OH-DPAT for 5-HT 1A , and [ 3 H]-LSD for 5-HT 7 , respectively) and 150 µL of diluted membranes (see description and Table S2 in Supporting Materials). Specific assay conditions for each receptor are described elsewhere [41]. The radioactivity was measured in a MicroBeta2 scintillation counter (PerkinElmer, USA). Each compound was tested in the assay as duplicate samples at 1 µM final concentrations. Results were expressed as percent inhibition of specific binding.

Molecular Modeling
The phosphodiesterase 10A model was developed on the basis of the experimental structure of the enzyme (PDB ID: 3SNI) [43]. The structure was refined using default settings in the Protein Preparation Wizard. Water molecules (except for the buried water molecules) and hetero groups, other than the ligand, were deleted and the whole system was minimized (OPLS3 force field). The model was tested extensively with docking studies involving PDE 10A inhibitors. The resulting consistent binding modes of the reference compounds of experimentally proven affinity were used as the basis for verifying the accuracy of the model that served as molecular target in docking studies. Ligand structures were optimized using the LigPrep tool. The Glide SP flexible docking procedure was carried out using default parameters. H-bond constraint, as well as centroid of a grid box, were located on Gln716.
Glide, LigPrep, and Protein Preparation Wizard were implemented in the Small-Molecule Drug Discovery Suite (Schrödinger, Inc., NY, USA), which was licensed to the Jagiellonian University Medical College.

Cytotoxicity Analysis-MTT
The MTT assay was used to determine the cytotoxic effects of the analyzed compounds. Cells were seeded at a density of 2 × 10 4 in 96-well plates. Following overnight culture, cells were then treated with increasing concentrations of compound 18 (0.1-100 µM) and incubated for 24 h. Following cell exposure for 24 h, 10 µL MTT reagent (Sigma Aldrich) was added to each well and, after 3 h of incubation (37 • C, 5% CO 2 ), the medium was aspirated and the formazan produced in the cells appeared as dark crystals in the bottom of the wells. Next, Crystal DMSO was added to each well. Then, the optical density (OD) of each well was determined at 570 nm on a plate-reader (Spectra iD3, Molecular Devices, San Jose, CA, USA). The number of metabolically active and living cells is directly proportional to the absorbance of the samples. The results are presented in Figure 5 as the percentage of control ± SEM. Doxorubicin was used as reference standard (cytotoxic agent).

In Vivo Studies
The experiments were performed on male CD-1 mice (in the accredited animal facility at the Jagiellonian University Medical College, Krakow, Poland), and mice were kept in groups of ten in Makrolon type 3 cages (dimensions 26.5 × 15 × 42 cm). The animals were kept in an environmentally controlled room (ambient temperature 22 ± 2 • C; relative humidity 50-60%; 12:12 light:dark cycle, lights on at 8:00). They were allowed to acclimatize with the environment for one week before commencement of the experiments. Standard laboratory food (Ssniff M-Z) and filtered water were freely available. All the experimental procedures were approved by the I Local Ethics Commission at the Jagiellonian University in Krakow (Approval Nos.: 125/2017, 123/2015, and 158/2017).
All the experiments were conducted in the light phase between 09.00 and 14.00 h. Each experimental group consisted of 7-10 animals per treatment dose. The animals were used only once.
3.5.1. d-Amphetamine-Induced Hyperlocomotor Activity in CD-1 Mice Locomotor activity was recorded with an Opto M3 multi-channel activity monitor (MultiDevice Software v.1.3, Columbus Instruments, Columbus, OH, USA). The CD-1 mice were individually placed in plastic cages (22 × 12 × 13 cm) immediately after drug administration, and then ambulation was counted during 1 h with data recording every 10 min. The cages were cleaned up with 70% ethanol after each mouse.

Spontaneous Locomotor Activity in CD-1 Mice
Locomotor activity was recorded according to the method described above.

Drugs
The following drugs were used: d-amphetamine (sulfate, Sigma-Aldrich, Poznań, Poland), papaverine (hydrochloride, Sigma-Aldrich, Poznań, Poland), and tested compound 18. D-amphetamine and papaverine were dissolved in distilled water; the tested compound (18) was suspended in a 1% aqueous solution of Tween 80 immediately before administration. Papaverine and 18 were administered intraperitoneally (ip) and d-amphetamine subcutaneously (sc) just prior to the test starting. All compounds were injected at a volume of 10 mL/kg. Control animals received a vehicle injection according to the same schedule. Figure 6 shows only the doses at which the antipsychotic-like activity of AMPH, papaverine, and compound 18 were observed.

Statistics
All the data are presented as the mean ± SEM. The statistical significance of the results was evaluated by a one-way ANOVA, followed by a Bonferroni's Comparison Test.

Conclusions
In this study, we describe the synthesis of a library of novel 4-methoxy-2,3-dihydro-1H-isoindole-1,3-dione derivatives with various aminoalkyl moieties, as potential inhibitors of PDE10A and serotonin receptor ligands. Among the synthesized and tested compounds, 1H-benzimidazole derivatives were identified as potent PDE10A inhibitors. Compound 18, the most potent PDE10 inhibitor (IC 50 = 886 ± 0.017 nM) of the group, with drug-like properties as defined by Lipinski's Rule of five, was selected for further pharmacological evaluation. In order to explain its inhibitory activity towards PDE10A, the binding mode of compound 18 was determined. Molecular modeling studies showed that H-bonds formed by the 4-methoxy group in the phthalimide moiety as well as the optimal length of the carbon linker are crucial for interactions with PDE10A. Preliminary in vitro neurotoxicity and hepatotoxicity studies revealed that the compound is safe and did not exhibit noteworthy cytotoxicity. Furthermore, significant antipsychotic properties of 2-[4-(1Hbenzimidazol-2-yl)butyl]-4-methoxy-1H-isoindole-1,3(2H)-dione (compound 18) was identified in the d-AMPH-induced hyperlocomotor activity test in mice, but further pharmacological studies are needed to explain the mechanism underlying its antipsychotic activity. This study successfully identified a number of compounds with significant PDE10A inhibitory activity and provided further information into the molecular basis of their interaction with PDE10A.
Supplementary Materials: The following are available online, Figure S1: The spider graph of drug-like parameters of compound 18, Figure S2. Effects of papaverine and compound 18 on d-amphetamine-induced hyperlocomotor activity in CD-1 mice, Table S1: Effects of papaverine and compound 18 on spontaneous locomotor activity in CD-1 mice, Table S2. Detailed conditions of the in vitro binding assays for the 5-HT 1A and 5-HT 7

Conflicts of Interest:
The authors declare no conflict of interest.