The employed dataset comprised recently synthesized pyrrole-based hydrazide-hydrazones with various biological activities (Fig. 2) (Bijev and Georgieva 2010; Tzankova et al. 2020; Mateev et al. 2022b).

Figure 2. 

General core structure of the employed in-house database.

The exact structures of the utilized database are provided in Supplementary Table 1. The compounds were generated with the 2D Sketcher (Maestro), and converted to the corresponding 3D structures with the LigPrep module in Maestro. Utilizing the former module, addition of hydrogen atoms, bond order assignment and energy minimization with the OPLS4 force field was performed. Moreover, the tautomerization and the stereoisomer forms for all compounds were generated. The ionization states were generated in physiological pH values. Consequently, the size of the final chemical library consisted of 139 molecules (including the generated tautomers and enantiomers). The major interactions between the most active pyrrole-based ligand and the active site the enzyme were visualized employing Discovery Studio Visualizer (BIOVIA Dassault Systèmes, Pharmacopeia, Inc) and Maestro (Schrödinger Release 2021-3: Maestro, Schrödinger, LLC, New York, NY, 2021).

The docking simulations were performed with both Glide (Schrödinger Release 2021-3: Glide, Schrödinger, LLC, New York, NY, 2021) and GOLD 5.3 on an AMD Ryzen 9 5950× 16-core 4.0 GHz CPU, NVIDIA GeForce RTX 3060 12GB GPU, 64 GB RAM installed memory and 64-bit Operating system Windows 11 Pro. The default GOLD 5.3 docking protocol was constructed out of a 8 Å binding gorge around the co-crystallized ligand, ChemPLP as a scoring function, addition of active waters in the active site and no rotatable side chain residues. For Glide, we selected the Extra-Precision (XP) option and created the enzyme’s active site using Receptor Grid Generation based on the conformation of the co-crystallized ligand. The crystal structure with PDB code 1Q84 was selected, which includes the highly active AChE co-crystallized ligand TZ4 (Bourne et al. 2004) due to the absence of a pyrrole-based co-crystallized AChE inhibitor. Moreover, the TZ4 inhibitor was suitable considering the bulkiness of the compounds included in our database. The latter structure was retrieved from the Protein Data Bank (PDB) (https://www.rcsb.org) with a reliable resolution of 2.45 Å. The initial preparation of the crystallographic structure was carried out with the protein preparation module in Schrödinger (Schrödinger Release 2021-3: Protein Preparation Wizard; Epik, Schrödinger, LLC, New York, NY, 2021; Impact, Schrödinger, LLC, New York, NY; Prime, Schrödinger, LLC, New York, NY, 2021).

In this study, Molecular Mechanics/Generalized Born Surface Area (MM-GBSA) recalculations with Prime were employed to assess the free binding energies of the obtained through the docking studies complexes. The calculations were performed by the incorporation of the OPLS4 force field and VSGB dissolvable model (Sahakyan et al. 2021).

Donepezil, Dimethyl sulfoxide (DMSO), 5,5′-dithiobis-(2-nitrobenzoic acid (DTNB), acetylthiocholine iodide (ATChI), and acetylcholinesterase were purchased from Sigma-Aldrich in analytically pure or chemically pure grades. No further purifications were applied.

The inhibitory AChE potential of the top ranked pyrrole-based agents was measured according to a modified Ellman’s method (Chigurupati et al. 2016). Stock solutions (1 mg/ml) of test agents were diluted in DMSO. Subsequently, working solutions (10 μM concentration) were prepared by dilutions. The compounds (10 μM) were incubated with sodium phosphate buffer (0.1 M; pH 8.0; 200 μL), and AChE solution (0.1 U/mL; 40 μL) for 10 min at 36.5 °C. The reaction was initiated by addition of DTNB (10 mM; 20 μL) and ATChI (14 mM; 20 μL). The absorbance was measured employing a microplate reader at 412 nm wavelength against blank DMSO probe. The % inhibition was calculated against blank probe. Donepezil was applied as a positive control.

The significant physicochemical and pharmacokinetic properties of the most prominent AChE inhibitor in the current database were calculated with the QikProp module in Schrödinger (Schrödinger Release 2021-2: QikProp, Schrödinger, LLC, New York, NY, 2021.). The simulation provides ranges based on the properties of 95% of the known drugs and also evaluates outliners based on the Lipinski’s rule of five.

The docking protocols of GOLD 5.3 and Glide were initially validated through re-docking simulations. The re-docking procedures are essential for the preliminary assessments of the softwares’ reliability and robustness (Mateev et al. 2021). During the latter simulations, the co-crystallized ligand of the protein is removed, and without any minimizations, re-docked back into the original protein. A root-mean-square-deviation (RMSD) is calculated, and values under 2 Å are considered as optimal (Mateev et al. 2022c).

The re-docking of the co-crystallized ligand TZ4 back into the active site of 1Q84 was carried out with the ChemPLP docking score of GOLD 5.3 and the XP score of Glide. Notably, GOLD 5.3 demonstrated RMSD value of 1.21 Å, whereas Glide re-docked the native ligand with RMSD of 0.54 Å, which indicated the slightly better performance of Glide. The re-docking conformations obtained with both docking softwares are provided in Fig. 3.

Figure. 3. 

Superimposed native conformation of TZ4 and the re-docking conformations acquired with Glide (A) and GOLD 5.3 (B).

To increase the reliability of the structure-based drug design simulations, a consensus docking technique was implemented. Several studies have reported the superiority of simulations with multiple programs when compared to a single software docking (Houston et al. 2013; Ren et al. 2018). The main justification behind the process is that all docking softwares have limitations, and the simulations with various searching and scoring algorithms could improve the overall hit rate (Ren et al. 2018).

Initial molecular docking simulations were carried out with the XP option in Glide, and the ChemPLP scoring function in GOLD 5.3 for the whole dataset (Supplementary Table 1). The consensus docking acquired similar docking scores for the applied dataset, thus an increased experimental correlation could be expected. The docking scores of the co-crystallized ligand were dissimilar. Glide provided significantly lower docking scores of TZ4 and Donepezil (-20.18 Kcal/mol and -18.41 Kcal/mol, respectively), therefore, higher affinity towards the AChE enzyme. However, GOLD 5.3 identical fitness score of all employed ligands, including the native inhibitor TZ4, which should correspond to close experimental blocking capacities (Table 1).

The final recalculations of the binding free energies of the complexes were conducted with the MM/GBSA method considering the enhanced hit rate of the latter (Table 1) (Sahakyan et al. 2021). Moreover, focusing on the top-scored docking poses implies that significant computational costs could be saved (Sun et al. 2014). The MM/GBSA scores of the top ten pyrrole-based hydrazide-hydrazones were ranging from -60.44 to -70.93 Kcal/mol. The most prominent AChE inhibitor after the rescoring was 11b.

However, the major concern is the immense gap of the binding free energy related to the native co-crystallized ligand – TZ4. In the former case the complex protein-ligand demonstrated drastically elevated stability (MM/GBSA score of -192.06). Thus, the expected in vitro experimental values of the title compounds may not achieve the IC₅₀ value of TZ4. Nevertheless, the docking simulations suggest that pyrroles condensed with a short, non-bulky aminoacids could potentially be used as AChE inhibitors.

Pharmacologically active drugs acting as acetylcholinesterase (AChE) inhibitors are frequently employed for patients suffering from AD. Drugs such as Donepezil, Rivastigmine and Galanthamine are registered as AChE inhibitors (Sitaram et al. 1978).

The in vitro inhibitory capacity of the top ranked through docking compounds was measured against eeAChE (electric eel acetylcholinesterase) according to the method of Ellman et al. (Chigurupati et al. 2016). Donepezil was used as a standard. The compounds were applied at 10 μM concentrations (Fig. 4).

Figure 4. 

Inhibitory activity of the top ranked ligands against AChE (10 μM concentrations). * P < 0.1; *** P < 0.001 vs control (pure eeAChE). Data are presented as means from three independent experiments ± SD.

The most active pyrrole-based compound was the hydrazide-hydrazone condensed with 2-nitrofuran 12d which inhibited the enzyme with 55% at 10 μM. In comparison, the standard drug Donepezil revealed blocking capacity of 93% at the same concentration. The enhanced AChE inhibitory effects could be related to the nitro moiety (Parveen et al. 2016). The hydrazide-hydrazone substituted with a benzaldehyde moiety inhibited the enzyme with 28%. The top ranked compound 11b showed only 26% blocking capacity, which underlines the current drawback of the structure-based drug design techniques – low capacity in differentiating true inhibitors from false-positive hits.

Subsequently, the major intermolecular interactions between 12d and the active site of 1Q84 were examined (Fig. 5). One stable halogen bond between Ser293 and the bromo atom from p-bromophenyl fragment was formed. A hydrogen bond between the carbonyl moiety from the ester group and the active amino acid Thr75 was also detected. The amino acid Tyr72, which is included in the active loop in the substrate pocket of MAO-B, was interacting with the pyrrole structure through a hydrophobic π−π stacking. Moreover, the pyrrole and the benzene aromatic rings were involved in hydrophobic interactions with Trp286 and Tyr341 amino acids. Trp286 has been reported to be part of the quaternary ammonium binding locus (Geromichalos et al. 2021).

Figure 5. 

Major intermolecular interactions between 12d and the active site of AChE (PDB: 1Q84). The interactions are provided in 2D (A) and 3D (B) forms. The AChE enzyme is depicted in grey while the active inhibitor – 12d, is presented as green sticks with its electrostatic potential.

The molecular docking simulations showed that 12d is located near the opening of the hydrophobic pocket. Hydrophobic interactions with the active amino residues Tyr286 and π−π bond with Tyr72 (5.25 Å) indicate that the most active pyrrole-base inhibitor was located near the PAS site (above the active site triad and near the gorge entrance) of the protein (Geromichalos et al. 2012). The probable reason for the incomplete binding to the active site of AChE might be the narrow entrance of the catalytic gorge (Lu et al. 2011). A hydrogen bond with Thr75 (2.13 Å) and the carbonyl moiety from the ethyl ester fragment was formed. A halogen bond between the bromine atom from the p-bromophenyl moiety and the active amino acid Ser293 was detected. Tyr72, Trp86, Trp286, Leu289, Ile294, Phe297, Tyr337, Phe338, and Tyr341 were involved in hydrophobic interactions with the active pyrrole-based AChE inhibitor 12d (Table 2).

The residues Trp86 and Tyr341 were also found to stabilize the active conformation of recently synthesized pyrrole-based AChE inhibitor (Pourtaher et al. 2022). The visualizations of the docking simulations revealed that the future design of pharmacologically active pyrrole-based AChE inhibitors should be targeted against smaller pyrrole molecules. The former ligands will suite better the active pocket of AChE.

As a final stage of our study, we carried out an in silico ADME analysis to examine the pharmaceutically relevant properties of the most prominent compound in our dataset – 12d. The QikProp module in Maestro 11.8 was employed for the virtual determination of the absorption, distribution, metabolism and excretion (ADME) (Table 3).

Notably, 12d exerted excellent physicochemical properties related to 95 % of the existing drugs. None of the calculated descriptors felt out of range during the conducted simulations. However, the observed molecular mass of 517.35 violated one of the Lipinski’s Rule of 5. Importantly, the calculated brain/blood partition coefficient (QPlogBB) of the examined compound was in the optimal range of -3.0–1.2, which is essential for potential AChE inhibitors (Thomas 2000).