Promising Non-cytotoxic Monosubstituted Chalcones to Target Monoamine Oxidase-B

A library of monosubstituted chalcones (1–17) bearing electron-donating and electron-withdrawing groups on both aromatic rings were selected. The cell viability on human tumor cell lines was evaluated first. The compounds unable to induce detectable cytotoxicity (1, 13, and 14) were tested using the monoamine oxidase (MAO) activity assay. Interestingly, they inhibit MAO-B, acting as competitive inhibitors, with 13 and 14 showing the best profiles. In particular, 13 exhibited a potency higher than that of safinamide, taken as a reference. Docking studies and crystallographic analysis showed that in human MAO-B 13 binds with the halogen-substituted aromatic ring in the entrance cavity, similar to safinamide, whereas 14 is accommodated in the opposite way. The main conclusion of this cell biology, biochemistry, and structural study is to highlights 13 as a chalcone derivative that is worth consideration for the development of novel MAO-B-selective inhibitors for the treatment of neurodegenerative diseases.


Materials and Methods S6
Table S1 S14 Table S2  S15 Table S3 S16 Figure S1 S17 Figure S2 S18 NMR spectra S19 General. 1 H and 13 C NMR spectra were recorded in CDCl 3 on Bruker DRX Avance (300 and 75 MHz) equipped with a non-reverse probe. Chemical shifts (in ppm) were referenced to residual solvent proton/carbon peak.

Synthesis of α,β-Unsaturated ketones
An aqueous solution of sodium hydroxide (30% p/p, 20 mL) was slowly added to a methanolic solution of the chosen acetophenone (1 eq in 20 mL) at 0°C. The solution was cooled to room temperature and the appropriate benzaldehyde (1.2 eq.) was added. The mixture was stirred at room temperature overnight and was then poured into water (50 mL). The obtained solid was filtered, washed with water until neutral pH and recrystallized twice from ethanol.

Materials. Human recombinant MAO-A and MAO-B were expressed in baculovirus infected
BT1 cells (5 mg/ml) and horseradish peroxidases were purchased from Fluka-Sigma-Aldrich s.r.l. (Italy). A Cary-Eclipse fluorimeter (Varian Inc., Palo Alto, CA, USA) was employed for fluorescence measurements. The stock solutions of chalcone derivatives and safinamide were prepared in dimethylsulfoxide.
Cell growth assay. A2780 (human ovarian carcinoma cells) and HT29 (human colorectal adenocarcinoma) were grown in RPMI-1640 (Sigma Chemical Co.), MSTO-211H (human biphasic mesothelioma cells) and Met-5A (human mesothelial cells) were grown in RPMI-1640 (Sigma Chemical Co.) supplemented with 2.38 g/L Hepes, 0.11 g/L pyruvate sodium and 2.5 g/L glucose. 10% Heat-inactivated fetal calf serum (Biowest), 100 U/mL penicillin, 100 g/mL streptomycin and 0.25 g/mL amphotericin B (Sigma Chemical Co.) were added to the media. The cells were cultured at 37 °C in a moist atmosphere of 5% carbon dioxide in air. To determine the growth inhibition, cells (3-410 4 ) were seeded into each well of a 24-well cell culture plate. After incubation for 24 h various concentrations of the test agents were added and the cells were then incubated in standard conditions for a further 72 h. A trypan blue assay was performed to determine cell viability. Cytotoxicity data were expressed as GI 50 values, i.e. the concentration of the test agent inducing 50% reduction in cell number compared with control cultures.
Monoamine oxidase inhibition assay. Monoamine oxidase activity was determined by measuring H 2 O 2 generation rate by a peroxidase-coupled continuous assay, using the Amplex Red reagent as fluorogenic substrate for horseradish peroxidase. 1 All experiments were carried out as previously reported, 2 in 100 mM potassium phosphate buffer, 0.1 mM EDTA, pH 7.4 and at 37°C. The assays were carried out in a final volume of 800 L, in the presence of Amplex Red (100 M) and horseradish peroxidase type II (5 UmL -1 ), and using kynuramine (20- Protein conformation selection. The selection of the most suitable protein conformations for docking was performed through the similarity with the co-crystallized ligands. In particular, 2D and 3D similarity assessments were performed by using the previously extracted MAO crystallographic ligands as reference queries. The chalcone inhibitor was prepared for the ligand-S8 based analyses, by firstly generating all the combinations of ionization states and tautomers potentially present at physiological pH with LigPrep. 7 Then, up to 600 conformers were generated for each of the pre-treated compounds with the OMEGA2 software (OpenEye) using default parameters. 8,9 ROCS (OpenEye) was used as an engine for the 3D similarity estimation, using default parameters. 10,11 The Tanimoto COMBO score (i.e., the sum of the ShapeTanimoto and ColorTanimoto similarity indexes) was used to estimate the 3D similarity. 2D similarity analyses were performed by using three different types of molecular fingerprints, namely MACCS and ECFP4, ECFP6, implemented into the OpenEye python toolkits. 12 Results of the analyses (see Table A) indicated a low degree of similarity, but allowed the identification of PDB ID 2Z5X (chain A) and PDB ID 2V5Z (chain A) as the most suitable protein conformations for docking into MAO-A and MAO-B isoforms, respectively. 13,14 In particular, safinamide, which is in complex with MAO-B in the 2V5Z crystal structure, resulted to be the most similar ligand to 13 according to the ECFP4 and ECFP6 indexes (Table A). As for MAO-A, the selection of the most suitable crystal structure (PDB code 2Z5X), was guided by visual inspection of the predicted 3D ligand-based alignments ( Figure A). In fact, none of the reported crystallographic MAO-A ligands resulted to be similar to 13 according to employed 2D similarity indexes, the evaluated similarities being below commonly accepted thresholds (Table A). 15 Interestingly, these crystal structures have already been widely used in previous molecular modelling studies involving these targets. 16,17 Moreover, in the case of MAO-B, the selected crystal structure is in complex with the same ligand (safinamide) 13 that was used as a control in the in vitro assays. Structure-based calculations. Docking calculations were performed into the 2V5X (MAO-A) and 2V5Z (MAO-B) crystal structures by using the Glide software. 18,19 In particular, receptor grids (outer box of 10 Å × 10 Å × 10 Å) were firstly built around the co-crystallized ligands by using the default settings. The flavin adenine dinucleotide (FAD) cofactor and conserved water molecules were considered as components of the MAO binding sites during the structure-based calculations. Then, redocking of the HRM (MAO-A) and SAG (MAO-B) ligands into their parent receptors was performed to assess the posing performance of the in silico models, obtaining root mean square deviation (RMSD) values below 2 Å. Afterwards, 13, which resulted to be the most active compound of the investigated series, was docked into the validated MAO-A and MAO-B models, and the predicted binding modes (two poses with the best docking scores for each complex) were visually inspected. Docking complexes obtained with Glide were also post-processed with BEAR (Binding Estimation After Refinement), an in-house developed postdocking processing procedure, which consists of three steps based on molecular mechanics (MM) minimization and molecular dynamics cycles for the complex refinement, and the application of the MM-PBSA and MM-GBSA methods for binding free energy estimations. 20 Further details about the BEAR procedure, which has extensively demonstrated to be suitable for correcting and overcoming limitations of docking procedures, can be found in references. [21][22][23] In silico ADME predictions. In silico ADME (absorption, distribution, metabolism and excretion) predictions were performed to evaluate the drug-likeness and blood-brain-barrier permeability of compound 13, which resulted to be the best candidate of the series. To this aim, the chalcone inhibitor was manually drawn in Maestro of the Schrodinger suite and prepared with the LigPrep software, 6,7 with default settings. Then, the prepared structure was analysed through the use of QikProp software available in the Schrödinger suite (release 2018-3). 24 Default settings were used in the ADME analyses, which allowed evaluating more than fifty different properties for compound 13, including drug-likeness and molecules descriptors. The results of the analyses are reported in Table S1.
X-ray crystallography. Recombinant hMAO-B were expressed in Pichia pastoris and purified as previously described. 25 A protein sample was gel-filtered by Superdex200 (GE Healthcare) in 50 mM potassium phosphate pH 7.5, 8.5 mM Zwittergent 3-12. Fractions corresponding to a single peak eluted from the column were concentrated by Amicon30K (Millipore) up to about 2 mg mL -1 and co-crystallized with each chalcone inhibitor (13 and 14) by the sitting-drop vapour diffusion method following published protocols. 26 X-ray diffraction data were collected on crystals flash-cooled in liquid nitrogen (100 K, in mother liquor solution with 18% (v/v) glycerol added) using the beamlines of the Swiss Light Source in Villigen (Switzerland) and European Synchrotron Radiation Facility in Grenoble (France). Data processing and scaling (Table S3) were carried out using XDS 27 and the CCP4 package. 28 The coordinates of hMAO-B in complex with safinamide (PDB code 2V5Z, 13 was used as initial model after removing water and inhibitor molecules. Model building and analysis were performed by Coot, 29 whereas the program REFMAC5 30 was used for structure refinement. Figures were generated by Pymol. 31 Table S1: ADME predictions of compound 13. The calculated molecular descriptors and drug-like properties are within the recommended ranges reported in the reference manual of the QikProp software. Of note, compound 13 showed good agreement with Lipinski's rule of five ("RuleOfFive") and Jorgensen's rule of three ("RuleOfThree"), the number of violations evaluated being 0. Moreover, the compound showed favorable values of brain/blood partition coefficient ("QPlogBB"), binding to human serum albumin ("QPlogKhsa"), and apparent Caco-2 ("QPPCaco") and MDCK ("QPPMDCK") cell permeabilities.