2.1. Acetylcholinesterase

Acetylcholinesterase (AChE, E.C. 3.1.1.7) is an enzyme involved in the termination of impulse transmission by rapid hydrolysis of acetylcholine into choline and acetic acid ^[13]. In AD, the patient’s cholinergic systems endure extensive degeneration changes, leading to a hypofunction of the cholinergic neurons and a decline in the endogenous levels of acetylcholine ^[14]. Hence, AChE inhibitors are administered to counteract these effects in an attempt to decrease the breakdown rate of acetylcholine and restore its synaptic levels ^[15]. To date, only three AChE inhibitors are used in AD therapy, donepezil, rivastigmine, and galantamine (Figure 1); however, they only offer symptomatic relief and are mostly used to treat mild to moderate dementia ^[7]. Thus, new and more effective AChE inhibitors are needed ^[16].

Figure 1. Structure of the clinically approved AChE inhibitors.

Many researchers took advantage of the available 3D human AChE structures deposited in the PDB database and the information regarding the interactions with several ligands to carry out in silico methodologies to design potential inhibitors ^[17]. Structurally, the target has two distinct binding sites: one is peripheral and is situated at the entrance of the gorge and the other is located in the catalytic site ^[15]. Currently, researchers are focused in molecules that can occupy both binding sites and inhibit acetylcholine hydrolysis ^[17]. Grafov et al. studied the affinity of several naturally occurring alkaloids for both binding sites of AChE (PDB#6H12) using neostigmine, a known AChE inhibitor, as positive control ^[18]. The molecular docking results demonstrated that the alkaloid 5-N-methylmaytenine (Figure 2) could simultaneously bind to both binding sites, mainly by hydrophobic interactions, and therefore, could have a high pharmacological potential towards the design of novel AChE inhibitors ^[18]. A similar analysis was carried out by Ortiz and his research group for other alkaloid compounds extracted from plants of the Hieronymiella genus, but employing an additional step of molecular dynamic simulations to evaluate the binding modes to better understand in vitro results ^[19]. In this study, the compound sanguinine (Figure 2), structurally similar to galantamine (Figure 1), showed the most promising binding energy and interacted with residues Trp84, Gly117, Glu199, Ser200, Phe330, and His440. In addition, it presented high in vitro inhibitory potency for AChE (PDB#1DX6) as well, which can indicate that these amino acids are essential to promote higher inhibition rates ^[19]. Mughal and co-workers synthetized a series of 4-thioflavonols that displayed very promising in vitro outputs and studied their interactions with the AChE (PDB#4BDT) active site through molecular docking ^[20]. Particularly, compound 1 (Figure 2) had the highest affinity and also formed identical interactions with the amino acids of the catalytic site of AChE similar to the observed for donepezil, especially with residues Trp86 and Tyr337 ^[20]. An identical strategy was performed to study other natural products, namely canadine derivatives ^[21], cinnamic acid derivatives, indolinones and cycloartane triterpenoids ^[22], phenolic acid derivatives ^[23], and synthetic compounds, such as arylisoxazole-phenylpiperazine derivatives ^[24], dipropargyl substituted diphenylpyrimidines ^[25], quinoline chalcone derivatives ^[26], and N-(4-methylpyridin-2-yl)thiophene-2-carboxamide analogs ^[27], as displayed in (Figure 2). Ranjan’s research group ^[28] studied the affinity of several organophosphate derivatives against AChE (PDB#1B41) using docking-based virtual screening combined with molecular dynamics simulations. The compounds were selected based on the interaction with the main residues of the catalytic triad, Ser203, Glu334, and His447 ^[28]. The top ranked ligand was phoxim ethyl phosphonate (Figure 2), displaying the highest binding energy with these residues and it was advanced for further in vitro studies ^[28].

Castro-Silva et al. studied the affinity of fucosterol (Figure 3) towards both binding sites of AChE (PDB#4EY7) and compared with the inhibitor neostigmine by docking and molecular dynamics to improve the analysis ^[29]. The results demonstrate that fucosterol has affinity towards both binding sites, by specifically interacting with the residues Trp286, Leu289, and Tyr341 of the peripheric site, and with the Trp86, Glu202, and Tyr449 of the AChE catalytic site, indicating that this compound might be a promising compound to advance for further studies ^[29]. Other researchers, namely Gurjar and co-workers, performed an in silico analysis in 2-substituted-4,5-diphenyl-1H-imidazole analogues (Figure 3) effecting a prediction of the compounds’ ADMET properties in addition to the molecular docking of the best ranked compounds, avoiding further testing of compounds with potential toxicity and unfavorable pharmacokinetics ^[30]. For instance, compound 2 (Figure 3) demonstrated the best results, being a potential candidate for further structural optimization for even better AChE inhibition ^[30]. Rocha and her group built and validated a machine learning model using pharmacophores based on the structures of more than 500 compounds with known and no inhibitory activity against AChE to predict the potential inhibitory activity of multiple indole alkaloids ^[31]. Of these, uleine (Figure 3) was predicted as being the alkaloid with the highest probability to present AChE inhibitory activity based on the in silico results ^[31]. The most promising compounds were further tested in vitro confirming the computational predictions regarding the AChE inhibition ^[31]. A distinct methodology was applied by Niu et al., building 2D- and 3D-QSAR models to classify molecules based on their potential to inhibit AChE, from a library of compounds that included known AChE inhibitors and non-inhibitors, with a predicted accuracy of 89.63% ^[32]. The most promising compounds were further tested by molecular docking to evaluate their affinity towards the target active site (PDB#1QTI) and the interaction with the residue Ser124 was demonstrated to be crucial for a higher affinity ^[32].

Figure 3. Structure of the studied AChE inhibitors.

2.2. N-Methyl-D-aspartate Receptor

The N-methyl-d-aspartate receptors (NMDAR) are a family of ligand-gated ionic membrane channels involved in non-selective cation transport and in the excitatory glutamatergic neurotransmission ^[33]. There are two types of NMDARs: the synaptic and the extra-synaptic receptors ^[33]. The synaptic are essential for synaptic plasticity and for the survival of neurons, while the extra-synaptic promote cell death and excitotoxicity, contributing for the etiology of AD ^[33]. The hyper-activation of the extra-synaptic NMDAR by glutamate, which is linked to an overproduction of free radicals and several enzymes that contribute to the deterioration of the CNS, can be controlled with NMDAR antagonists ^[34]. Currently, memantine (Figure 4) is the only NMDAR antagonist approved by the regulatory agencies for clinical use to treat moderate to severe dementia, selectively inhibiting the activity of extra-synaptic NMDARs ^[34]. In addition to its cognitive and functional pharmacological benefits, memantine has also demonstrated to slow down the emergence of further behavioral and psychotic manifestations. Furthermore, its administration resulted in significant increases in the extracellular concentrations of the neurotransmitters dopamine, norepinephrine, and their metabolites, demonstrating a biogenic amine neurotransmission enhancing effect useful in AD treatment ^[35].

Figure 4. Structure of the clinically approved NMDAR antagonist.

Nevertheless, over the years, numerous studies reported the design of novel active NMDAR antagonists using in silico methodologies ^[9]. For instance, Ivanova and co-workers used a virtual screening approach to discover new potential NMDAR antagonists ^[36]. Using a combination of various machine learning methods, including artificial neural networks and advanced multilinear techniques to build QSAR models, they screened over 13,000 natural compounds and ranked them based on their predicted affinity towards the target (PDB#5TP9) ^[36]. The best candidates were also analysed by docking and molecular dynamics simulations to identify essential structural moieties that could serve as basis for the design and development of novel and improved NMDAR antagonists ^[36]. A distinct strategy was carried out by Sharma et al. using pharmacophore modeling and a four-phased virtual screening study to identify potential drug candidates ^[37]. The pharmacophore model was generated and validated with a library of 40 known NMDAR antagonists, followed by screening, where the compounds were sorted based on the Lipinski’s rules and in terms of affinity towards the target ^[37]. Additionally, the hits were submitted to a docking analysis to fully validate the used methodology ^[37]. The main residues in the NMDAR active site are His88, Ser114, Thr116, Art121, Gly172, Ser173, Thr174, and Tyr214, of which the predicted compound HTS 00987 (Figure 5) interacts with His88, Thr174, and Tyr214, while memantine was not predicted to interact with none of these residues, indicating a potential pharmacological interest of this compound ^[37]. Waqar et al. built a homologue model of the 3D structure of NMADR based on the structure of the rat NMDAR (PDB#3JPW) and analyzed by molecular docking the affinity of several conantokins towards this target ^[38]. Moreover, most of the compounds interacted with residues Gln110 and Glu236, in the NR2B subunit of the NMDAR, and with Ile111, Phe114, and Pro177 by hydrophobic interactions ^[38]. A similar binding pattern was observed for the rat crystal structure, indicating that conankotins might be potential NMDAR antagonists ^[38]. In another study carried out by Hu and co-workers, the affinity of several tetramethylpyrazine derivatives, with known NMDAR antagonist activity, was studied towards its catalytic site (PDB#5UOW) by molecular docking ^[39]. Specifically, compound 3 (Figure 5) demonstrated the most promising binding energy, as well as favorable interaction by hydrogen bonds with the amino acid Asn602 ^[39]. Kumar et al., studied the affinity of antipsychotic drugs towards NMDAR (PDB#1PBQ) by molecular docking to evaluate its potential use in the treatment of AD ^[40]. In the hydrophobic pocket of NMDAR 3D structure, the main residues are the Phe16, Phe92, Trp223, and Phe250 ^[40]. For the antipsychotic drug anisopirol (Figure 5), the compound of this group with the highest affinity mainly interacted with Phe92, Pro24, Thr126, Ser180, Trp223, and Phe246 ^[40]. Considering that this drug shared some interactions with the known NMDAR antagonist 5,7-dichlorokynurenic acid (DCKA), the co-crystallized ligand of this crystal structure, it might indicate that could also biologically interact with this target ^[40]. On the other hand, Singh et al. developed pharmacophore models based on the structure of ifenprodil, also a known NMDAR antagonist, performed virtual screening, studied the affinity of the hits by molecular docking and molecular dynamics simulations using the 3D structure of the NMDAR (PDB#5EWJ), as well as analyzed the hits ADMET properties ^[41]. The proposed study revealed that the molecules ZINC25726161 and ZINC95977857 (Figure 5) displayed a better affinity towards the target than the NMDAR antagonist drug ifenprodil, indicating that these virtual hits could have pharmacological interest ^[41].

Figure 5. Structures of the studied NMDAR antagonists.