The benzofuran ligands were generated with the 2DSketcher of Maestro (Maestro, 2025, Schrödinger, LLC, New York, NY, USA) and converted to the corresponding 3D structures with the LigPrep module in Maestro. The former Schrödinger module added the hydrogen atoms, the bond orders were assigned, and energy was minimized with the OPLS4 force field. The ionization states were generated using the physiological pH values.

The crystal structure of human MAO-B in complex with safinamide (PDB ID: 2V5Z, resolution 1.60 Å) and that of human MAO-A in complex with harmine (PDB ID: 2Z5X, resolution 2.20 Å) were retrieved from the Protein Data Bank (PDB). Both structures were processed with the Protein Preparation Wizard in Maestro and were minimized with the OPLS4 force field. Bond orders were assigned, hydrogen atoms were added, and active water molecules were retained. The FAD cofactor was retained and was modeled as non-covalently bound.

The docking simulations were carried out with Glide’s XP (Schrödinger Maestro Suite) scoring function. The grid boxes were arranged around the co-crystallized ligand (Safinamide for 2V5Z and Harmine for 2Z5X), retaining the active water molecules. The inner box, which defines the region accessible to the ligand centroid, was set to 10 × 10 × 10 Å, whereas the enclosing outer box was set to 30 × 30 × 30 Å; a ligand length cut-off of 20 Å was applied. The active-site water molecules retained during protein preparation were included in the grid. The van der Waals radii of receptor atoms with partial charges below 0.25 were scaled by a factor of 1.00, and no positional, hydrogen-bond, or metal-coordination constraints were imposed.

Another docking software, GOLD 5.3 (CCDC, Cambridge, UK), was used, considering the implemented Genetic Search algorithm. As a scoring function, ChemPLP was applied. The grid box was set around the co-crystallized ligand with an 8 Å grid box. The active water molecules were involved in the docking calculations. Genetic algorithm runs were carried out per ligand at a search efficiency of 100%, with early termination disabled; the top-ranked pose of each ligand was retained for analysis. The two programs were applied sequentially rather than as a consensus-scoring protocol. Compounds ranked favorably by both programs were prioritized; where the two rankings diverged, the GOLD ranking was used for the final selection because it was obtained on the reduced, visually validated set. Following docking, the top-ranked poses were subjected to structural filtering. Compounds were retained when they displayed a sterically feasible orientation within the MAO-B active site, occupied the entrance and/or substrate cavity appropriately, positioned an aromatic or hydrophobic moiety near the Tyr398/Tyr435 aromatic “cage” and FAD cofactor, and formed plausible contacts with key residues, including Tyr326, Ile198, Ile199, Ile316, Leu171, or Cys172. Poses showing severe steric clashes, incomplete occupancy of the binding cavity, absence of meaningful contacts with the catalytic-region residues, or an orientation incompatible with the MAO-B cavity architecture were excluded.

Induced-fit docking (IFD) (Schrödinger) was used to enhance the robustness of the active conformations. The IFD enables full side chain flexibility and scores each new active solution with Glide’s XP score. The extended IFD protocol was used. In the initial Glide docking stage, the van der Waals radii of receptor and ligand atoms were scaled by 0.45, and up to 50 poses per ligand were retained. Receptor residues with at least one atom within 5.0 Å of a given ligand pose were subsequently refined with Prime, allowing full side-chain flexibility, and the side chains were optimized with backbone relaxation.

The free binding energies of the top-scored compounds with XP Glide were calculated using Prime’s Molecular Mechanics/Generalized Born Surface Area (MM-GBSA) calculations. The latter method incorporates the OPLS4 force field and VSGB solvation model to carry out the calculations (Pattar et al. 2020). The binding free energy was calculated as ΔG_bind = G_complex – (G_protein + G_ligand), where the energy of each species was evaluated in the minimized complex geometry.

Molecular dynamics (MD) simulations of the protein-compound complex obtained from the docking were performed using the GROMACS 2020 package to measure their stability (Abraham et al. 2015). Energy minimization was performed using the steepest-descent algorithm until convergence was achieved. Subsequently, the systems were equilibrated under the NVT ensemble and then the NPT ensemble, each for 100 ps, while applying position restraints to the protein and ligand heavy atoms. The temperature was maintained at 310 K using the V-rescale thermostat, and pressure was maintained at 1 bar using the Parrinello–Rahman barostat. Long-range electrostatic interactions were calculated using the particle mesh Ewald method, and a 1.0 nm cut-off was applied for short-range electrostatic and van der Waals interactions. All bonds involving hydrogen atoms were constrained using the LINCS algorithm, allowing a 2 fs integration time step. Production MD simulations were performed for 200 ns under periodic boundary conditions. The trajectories were analyzed for root-mean-square deviation (RMSD), root-mean-square fluctuation (RMSF), radius of gyration (Rg), and the number of protein-ligand hydrogen bonds using standard GROMACS analysis tools. The resulting plots were generated using QtGrace and subsequently analyzed (Muhammed and Aki-Yalcin 2023).

DFT calculations of compounds 10 and 16 were undertaken with the Gaussian program as per the method explained previously (Dennington et al. 2008; Muhammed and Aki-Yalcin 2023). The geometries were fully optimized at the B3LYP/ 6-311⁺⁺G(d,p) level of theory. Dispersion corrections were not applied, and the calculations were performed in the gas phase, and no continuum solvation model was employed. The molecular electrostatic potential of the compounds was mapped onto the total electron density surface at an isovalue of 0.0004 a.u. The frontier molecular orbitals were plotted at an isovalue of 0.02 a.u. From the energies of the highest occupied and lowest unoccupied molecular orbitals, the global reactivity descriptors were derived within Koopmans’ approximation as ionization potential I = –E_HOMO, electron affinity A = –E_LUMO, energy gap ΔE = E_LUMO – E_HOMO, chemical potential μ = –(I + A)/2, electronegativity χ = (I + A)/2, chemical hardness η = (I – A)/2, softness S = 1/2η, electrophilicity index ω = μ²/2η and maximum charge transfer ΔN_max = (I + A)/2(I – A). Mulliken atomic charges were extracted from the same calculations.

The distribution of the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) orbitals was also drawn. The total energy, HOMO energy, and LUMO energy of the compounds were obtained from the Gaussian computation. Based on these values, the other related parameters were calculated, and then the resulting values were analyzed.

The in silico study on the pharmacokinetic properties - absorption, distribution, metabolism and excretion (ADME) of the most promising compounds from the molecular docking study (10 and 16) was carried out using the SwissADME online software.

The ligand dataset consists of 37 benzofuran derivatives that have been previously synthesized and characterized within our research group. The benzofuran core was paired with terminal aromatic and heterocyclic motifs selected to promote productive occupation of the MAO-B binding cavity and engagement of the aromatic “cage”, with the overall design aimed at achieving favorable MAO-B inhibition and selectivity. In this context, the present in silico study uses the existing benzofuran series as a structurally coherent platform to explore how systematic variation of substituents around the scaffold translates into differences in predicted MAO-B binding (Fig. 1).

Extra-precision (XP) docking

The docking protocols were previously validated by the authors through re-docking, cross-docking, and ligand-enrichment studies (Mateev et al. 2022).

Initially, docking simulations of the 37 benzofuran derivatives in MAO-B and MAO-A with Glide (XP scoring function) were performed. Thirteen of the derivatives returned valid poses within the MAO-B grid, and their scores are listed in Table 1; the remaining 24 compounds could not be accommodated within the grid and returned no pose. The validation of the docking protocols is described in the Materials and Methods section.

Most of the ligands were successfully situated in the active site of MAO-B. The top-scoring MAO-B inhibitor was compound 33 with –11.31 kcal/mol. The selective MAO-B inhibitor safinamide demonstrated a higher score of –15.25 kcal/mol. The second-best benzofuran derivative was 16, which showed a result of –10.14 kcal/mol. Interestingly, the majority of the ligands could not fit in the active site of MAO-A. Only three compounds returned active docking poses – compounds 12, 11 and 3. A possible explanation for the examined result could be found in the difference between the active sites of MAO-A and MAO-B. Although both isoenzymes share 72% sequence identity, the active site of MAO-A consists of a single hydrophobic cavity that is smaller than the bipartite gorge of MAO-B. Additionally, a specific “loop” conformation formed by the hMAO-A residues 210–216 differs from that of hMAO-B and has been used as a basis for the development of novel and selective MAO-A inhibitors (Mateev et al. 2022). The aromatic “cage” itself is conserved between the two isoforms, with the MAO-B residues Tyr398 and Tyr435 corresponding to Tyr407 and Tyr444 in MAO-A. The determinants of selectivity lie instead in the residues that shape the cavity: Ile199, which in MAO-B separates the entrance from the substrate cavity, and Tyr326, which lines the entrance cavity, are replaced by Phe208 and Ile335 in MAO-A. As a result, the elongated benzofuran derivatives examined here could not fit in the smaller, monopartite active site of MAO-A.

Induced-fit docking

The study continued with induced-fit docking (IFD) simulations of the ligands top-scored by Glide XP in the active site of MAO-B. The IFD module in Maestro provides full flexibility of the active-site residues situated within the grid box. None of the compounds acquired an IFD score comparable to that of safinamide (Table 2). However, all tested ligands passed the threshold of –10.00 kcal/mol. The top-scoring ligand was compound 2, with a score of –13.12 kcal/mol. It should be noted that the IFD score incorporates a fraction of the receptor strain energy and is therefore not directly comparable with the Glide XP scores reported in Table 1.

Visual inspection of the IFD results

To acquire more reliable in silico data from a virtual screening, it is essential to visualize the active conformations of each top-scored ligand. A recent study emphasized the major role of visual inspection after docking studies (Fischer et al. 2021); thus, each acquired MAO-B/ligand complex was evaluated. Given the importance of the aromatic “cage” for potent MAO-B inhibitors (Li et al. 2006), the assessment focused on whether the title structures fit tightly into the active pocket and interacted with the latter cavity.

After IFD simulations, compound 3 did not display an optimal active conformation. It was found that the 5-bromobenzofuran fragment was not situated in the aromatic “cage”, which is crucial for the formation of a stable complex (Li et al. 2006). The distance between the bromo-moiety and the N5 atom of the FAD cofactor was calculated to be 9.44 Å (Suppl. material 1: fig. S1A).

The active conformations of compounds 2 and 8 after the IFD calculations were also not optimal. The ligands were not correctly placed in the entirety of the enzyme’s active cavity, which led to odd orientations (Suppl. material 1: fig. S1B, C).

The lack of halogen substituents in the benzofuran ring led to better localization of compound 14 in the active site of MAO-B. The benzofuran moiety was placed in the aromatic “cage” where it formed hydrophobic interactions with the amino acids Tyr398 and Tyr435 (Suppl. material 1: fig. S1D).

The active conformation of compound 5 had a different orientation. The anilide group of the former was located in the aromatic “cage” where Tyr398 formed a stable hydrogen bond with the amino group (Suppl. material 1: fig. S1E). A similar active conformation was formed after the docking simulations of compound 1 (Suppl. material 1: fig. S1F).

The chlorophenyl moiety of compound 16 was detected in close proximity to the FAD co-enzyme in MAO-B (distance 2.97 Å) (Suppl. material 1: fig. S1G). Thus, the essential amino acids forming the aromatic “cage” (Tyr398 and Tyr435) were interacting with ligand 16.

Compounds 33 and 12 were lacking a terminal benzene ring that could bind to the aromatic “cage” (Suppl. material 1: fig. S1H, I). Therefore, the morpholinyl moieties were facing away from the FAD co-enzyme, and lower MAO-B activities for both ligands could be expected.

Compound 10 comprises a tertiary amino moiety which was facing the FAD structure (Suppl. material 1: fig. S1J). No stable bonds were formed between the amino acid residues forming the aromatic “cage” and the amide group of 10.

The MAO-B/compound 37 complex is shown in Suppl. material 1: fig. S1K. The benzene fragment of the compound was located in the aromatic “cage”; however, Tyr326, which is essential for MAO-B selectivity, was not involved in any stabilization processes.

Compound 13 interacted with the active site of MAO-B. However, most of the amino acid residues essential for MAO-B activity did not form stable bonds with the ligand (Suppl. material 1: fig. S1L).

After the visual inspection of the active conformations of all top-ranked compounds, MM/GBSA re-calculations of the most prominent ligands were conducted. Compounds 3, 11 and 12 were not examined further because of the potential MAO-A inhibitory activity and were therefore not expected to be isoform-selective.

The remaining top-ranked compounds were excluded because they failed at least one of the structural criteria defined in the Materials and Methods section.

MM/GBSA re-calculations

Compounds 1, 5, 10, 14 and 16 were further examined for their MM/GBSA free binding energies. The results demonstrated that the most stable complex was formed between compound 14 and the active site of MAO-B (Table 3). It was found that most of the compounds possess similar VdW, Electrostatic solvation, lipophilic and Coulomb energies. The major stabilizing force was Van der Waals energy.

Re-docking with GOLD 5.3

The results obtained with the Maestro workflow were promising in terms of the discovery of a novel class of MAO-B inhibitors. However, the acquired in silico scores were lower than those of the standard selective MAO-B inhibitor safinamide. The work therefore continued with docking simulations of the top-scored compounds using GOLD 5.3, which employs a genetic search algorithm and may lead to different results (Mateev et al. 2023). The simulations were carried out with the ChemPLP scoring function, and the most potent MAO-B inhibitors from the MM/GBSA stage were considered (Table 4).

Interestingly, all compounds demonstrated good docking scores, comparable to the standard safinamide. Compound 10 achieved the highest results with a fitness score of 161.60, similar to the score of safinamide - 165.77. Further visualizations of compounds 10 and 16 were conducted.

The overall prioritisation of compounds 10 and 16 followed a stepwise, multi-stage filtering strategy rather than a single scoring criterion. Of the 37 benzofuran derivatives, 13 returned valid poses within the MAO-B grid at the Glide XP stage. Compounds 3, 11 and 12 were subsequently excluded because they also produced valid poses in MAO-A and were therefore not expected to display the desired isoform selectivity. The remaining ligands were subjected to induced-fit docking and to the structural filtering criteria described in the Materials and Methods section, which eliminated compounds 2, 8, 13, 33 and 37. Compound 33, despite obtaining the most favourable Glide XP score, lacks a terminal aromatic ring capable of occupying the aromatic “cage”, and its morpholinyl moiety was oriented away from the FAD cofactor, so that its high score was not supported by a plausible binding orientation. The five compounds that passed all structural filters (1, 5, 10, 14 and 16) were rescored by MM/GBSA and subsequently re-docked with GOLD using the ChemPLP scoring function, which served as the final ranking step because it was obtained with an independent search algorithm and scoring function applied to the reduced, visually validated set. At this final stage, compounds 10 and 16 obtained the highest fitness scores (161.60 and 159.45, respectively) and the values closest to that of safinamide, whereas compound 14, although displaying the most favourable MM/GBSA binding energy, ranked third. Compounds 10 and 16 were therefore the only structures ranked highly by two independent scoring functions after passing every structural filter, and they were selected on this basis for the subsequent molecular dynamics and DFT analyses.

Visualization of the active conformations of 10 and 16

In an attempt to gain further understanding of the in silico results and to obtain additional insights into the active binding conformation of the best hypothetical inhibitors, the major intermolecular interactions of 10 and 16 within the active site of MAO-B (PDB: 2V5Z) were visualized (Fig. 2).

Interestingly, the N,N-diethyl moiety of compound 10 was facing the FAD co-enzyme of MAO-B, and several hydrophobic interactions were formed between the cofactor and this fragment. Tyr326 interacted with the 1,3,4-oxadiazole ring through π–π stacking. Tyr326 plays an important role in the formation of stable complexes with active MAO-B inhibitors (Boulaamane et al. 2023), and its involvement as a stabilizing residue may be related to the MAO-B selectivity of compound 10, since the equivalent position in MAO-A is occupied by Ile335 (Toprakçı and Yelekçi 2005). Favorable contacts were formed with Ile198 and Ile199, which are part of the “loop” connecting the entrance and substrate cavities of MAO-B. Cys172 was also involved in the stabilization, forming a π–sulfur interaction with the 1,3,4-oxadiazole ring. Tyr398, which is part of the aromatic “cage” of MAO-B, formed two hydrophobic interactions with the N,N-diethylamino fragment.

In comparison, compound 16 bears a 2-chlorophenyl moiety instead of the N,N-diethylamino group. Stronger π–π interactions were therefore observed between the 2-chlorophenyl fragment of 16 and the aromatic “cage” of MAO-B, and both residues forming the “cage” (Tyr398 and Tyr435) interacted with the ligand. A further π–π interaction was observed between 16 and FAD. Tyr326 was not engaged in any interaction; however, Leu171, Ile199 and Ile316 formed stable contacts with the compound.

Molecular dynamics (MD) simulations of the complexes formed by the most active derivatives, 10 and 16, with the human MAO-B enzyme were undertaken in order to assess the stability of the binding modes predicted by docking. RMSD, RMSF, Rg and hydrogen bond plots were generated to analyze the simulation results (Fig. 3). The stability of each ligand inside the binding region of the enzyme was monitored through the ligand RMSD calculated after least-squares fitting of the protein. Compound 10 maintained a stable ligand RMSD throughout the 200 ns trajectory, with a mean value of approximately 0.3 nm, which implies that the compound remained inside the binding site of the enzyme. In contrast, compound 16 produced an RMSD profile with considerably larger variation. During the first 60 ns of the simulation, its RMSD was higher than that of compound 10; thereafter, the two profiles converged to comparable values, although compound 16 continued to fluctuate more extensively (Fig. 3A). The larger amplitude of these fluctuations indicates that compound 16 adopts a less firmly anchored binding mode than compound 10, while the convergence of the two profiles after approximately 60 ns indicates that the ligand was nevertheless retained within the active site for the remainder of the trajectory. The RMSF plot was used to assess per-residue mobility. The two complexes showed similar RMSF profiles for MAO-B/10 and MAO-B/16, respectively, with increases confined to a few narrow intervals and the most pronounced fluctuations at the N- and C-terminal ends of the enzyme (Fig. 3B).

The effect of ligand binding on the compactness of the enzyme was assessed from the radius of gyration (Rg) (Fig. 3C). Hydrogen bonding between a ligand and its target plays a crucial role in binding and in maintaining the ligand inside the binding site (Muhammed et al. 2023). Compound 10 formed predominantly two hydrogen bonds, followed by three; three hydrogen bonds were dominant during the first 50 ns and in the 110–188 ns interval, whereas single and quadruple hydrogen bonds were too sparse to be taken into account. Compound 16 formed predominantly one hydrogen bond, followed by two, during the simulation period; intermittent triple hydrogen bonds were observed after 90 ns, and four to five hydrogen bonds were densely populated in the 184–200 ns interval (Fig. 3D).

Molecular electrostatic potential

The electrochemical properties of the most active compounds were calculated by the DFT method. In this regard, the molecular electrostatic potential (MEP) of the compounds was mapped to figure out their reactivity as well as the responsible vicinity for the predicted reactivity (Politzer and Murray 2002). The electrophilic reactivity of compound 10 might arise from the regions around the oxygen and nitrogen atoms in the structure according to the MEP distribution map. A predominantly yellow color around the oxygen of the benzofuran ring and red color around the oxygen of the carbonyl were observed. Similarly, a predominantly yellow color was observed around the nitrogen of the oxadiazole ring (Fig. 4). As the yellow and red colors in the MEP distribution maps represent electrophilic reactivity, the mentioned vicinities are expected to contribute much for this reactivity. The blue region, which represents the nucleophilic reactivity, was mainly concentrated around the hydrogen atoms of the compound. For compound 16, the regions around the oxygen atoms of the carbonyl group and the benzofuran ring were in the electrophilic reactivity supplying vicinity. Similarly, the region surrounding the nitrogen atoms of the oxadiazole ring was in the electrophilic reactivity vicinity. On the other hand, the hydrogen atoms of the compound and the nitrogen atom of the acetamide functional group were in the predominantly blue region (Fig. 4). Hence, these vicinities are anticipated to contribute to the nucleophilic reactivity of the compound. The concentration of the nucleophilic vicinity around the chlorophenyl ring of the compound was high. This has led the ring to have a conformation different from that of the diethyl substituent in compound 10 (Fig. 4).

Frontier molecular orbitals and global reactivity descriptors

The HOMO and LUMO energies of the active compounds were obtained from the DFT calculation. The related parameters were then computed by using the appropriate formulas. The two compounds (10 and 16) were found to have similar electrochemical properties, as they gave similar values for the calculated parameters. In DFT studies, HOMO and LUMO energy values are employed to find out the relative electron-exchange ability of compounds. In this sense, HOMO energy is used to measure the electron-donating capacity, whereas LUMO energy is used to measure the electron-accepting capacity of compounds. As compound 10 had slightly higher HOMO and LUMO energy values than compound 16, it is anticipated to have slightly higher electron exchange capability (Table 5) (Akman et al. 2023). The energy gap between the LUMO and the HOMO energies is employed to measure the relative chemical stability of compounds (Ruiz-Morales 2002). In this study, compounds 10 and 16 gave similar energy gap values (Table 5). Therefore, the two compounds are expected to have similar chemical stability. Global hardness is employed to measure resistance for electron transfer. In this study, compounds 10 and 16 had similar global hardness values that implied a similar electron transfer resistance for them (Table 5). In short, the two compounds are anticipated to have similar electrochemical properties with a minor difference in their electron exchange capability.

The distribution of the HOMO and LUMO orbitals was similar for the two compounds. In addition, the HOMO and LUMO orbitals were distributed similarly. The HOMO and LUMO orbitals had high concentration around the benzofuran and oxadiazol heterocyclic rings. The only difference between the HOMO and LUMO orbital distribution of compound 10 was the relatively higher concentration of HOMO orbitals around the methyl and carbonyl functional groups of the compound (Fig. 5). The HOMO and LUMO orbitals of compound 16 were similar to each other, with a slightly higher LUMO orbital concentration around its oxadiazole ring. Molecular orbitals were not observed around the diethyl group in 10 and the chlorophenyl moiety in 16, which are next to the acetamide group (Fig. 5). Compound 10 exhibited slightly higher HOMO and LUMO energies than compound 16, indicating a marginally greater propensity for electron donation and electron acceptance. This electronic feature may contribute to favorable interactions within the MAO-B binding cavity; however, it should be interpreted together with the docking and MD results rather than as a direct predictor of inhibitory activity.

When considered together with the docking results, the calculated electronic descriptors provide a consistent electronic rationale for the predicted binding modes of the two compounds. In both structures, the regions of most negative electrostatic potential are localised on the carbonyl oxygen and on the nitrogen atoms of the 1,3,4-oxadiazole ring, and these are precisely the fragments engaged in the polar and π-electron-mediated contacts observed after docking: the oxadiazole ring of compound 10 forms a π–π stacking interaction with Tyr326 and a π–sulfur interaction with Cys172, whereas the corresponding electron-rich regions of compound 16 are engaged by the aromatic “cage” residues Tyr398 and Tyr435 and by the isoalloxazine ring of the FAD cofactor. The predominantly positive potential over the benzofuran and terminal aryl hydrogen atoms is consistent with the hydrophobic contacts formed with Ile198, Ile199, Ile316 and Leu171, which line the entrance and substrate cavities of the enzyme. In agreement with this, the frontier orbitals of both ligands are concentrated on the benzofuran and oxadiazole fragments, that is, on the same portions of the molecules that occupy the substrate cavity and interact with the aromatic “cage”, whereas the N,N-diethylamino fragment of 10 and the 2-chlorophenyl group of 16 carry little orbital density and appear to contribute to binding mainly through steric and hydrophobic complementarity. Finally, the closely similar and relatively large HOMO–LUMO gaps of the two compounds (4.473 and 4.462 eV) indicate comparable chemical stability and a low propensity for electron transfer, which argues against a reactive or mechanism-based mode of inhibition and is consistent with the reversible, non-covalent binding modes predicted by the docking and molecular dynamics simulations.

Drug-likeness and ADME predictions

An in silico ADME study was performed on the best-rated ligands from the docking studies (compounds 10 and 16) using an accessible online tool - SwissADME. Table 6 lists the main physicochemical parameters as shown in the software’s “Bioavailability radar” (Fig. 6). These specific parameters should be within certain limits for the further evaluation of the selected compounds as suitable base molecules for the development of potential selective MAO-B inhibitors.

The radar plot of compound 10 lies entirely within the optimal region defined by SwissADME, indicating that all six physicochemical descriptors fall within the ranges associated with acceptable oral bioavailability, and supporting its evaluation as a potential lead structure. Compound 16 satisfies the criteria for size, polarity, solubility, lipophilicity and flexibility, and deviates only on the insaturation axis, where its fraction of sp³-hybridised carbon atoms (Fsp³ = 0.06) falls below the recommended threshold of 0.25, compared with 0.31 for compound 10. This difference is a direct consequence of the single structural modification that distinguishes the two compounds: the N,N-diethylamino fragment of 10 contributes four sp³ carbon atoms, whereas the 2-chlorophenyl fragment of 16 introduces six additional aromatic sp² carbon atoms and no sp³ centre, so that the substitution lowers Fsp³ both by removing sp³ carbons and by increasing the total carbon count. The same substitution replaces four rotatable bonds by a single N-aryl bond, which is reflected in the slightly lower flexibility value of 16 (six rotatable bonds against seven for 10); both compounds remain well within the accepted limit of nine. It should be emphasised that the low Fsp³ value of 16 reflects its predominantly aromatic architecture rather than an unfavourable drug-likeness profile, since planar aromatic systems are a recurring structural feature of potent and selective MAO-B inhibitors, in which flat hydrophobic moieties are required for productive occupation of the substrate cavity and for engagement of the aromatic “cage” formed by Tyr398 and Tyr435 (Binda et al. 2003; Li et al. 2006). Taken together, the predicted physicochemical descriptors of both compounds are compatible with those of orally available lead structures and support their further in vitro and in vivo evaluation.