Abstract
Alzheimer’s disease (AD) is a progressive neurodegenerative disorder characterized by amyloid beta (Aβ) plaque accumulation, in which β-secretase (BACE1) plays a central role in the amyloidogenic pathway. Inhibition of BACE1 represents a promising therapeutic strategy; however, effective and safe inhibitors remain limited. This study aimed to identify novel coumarin-1,2,3-triazole hybrids as potential BACE1 inhibitors using comprehensive in silico approaches. Molecular shape and distance-based features derived from co-crystal ligands were used to generate and validate a pharmacophore model through ROC-based area under the curve analysis. A designed library of hybrids underwent pharmacophore-based virtual screening, yielding 92 hits with ≤0.6 RMSD. These compounds were further evaluated via structure-based molecular docking, from which six top candidates were selected, demonstrating superior docking scores relative to reported inhibitors. ADMET and density functional theory analyses indicated favorable pharmacokinetic profiles and electronic properties, highlighting potential reactive sites. The selected compound, CUM-0199, was subsequently validated through a 200 ns molecular dynamics simulation and MMGBSA binding free energy analysis, confirming structural stability and favourable binding affinity toward BACE1. Overall, the findings suggest CUM-0199 as a potential lead candidate for AD therapy, warranting further experimental validation and clinical investigation.
Graphical Abstract
1 Introduction
Alzheimer’s disease (AD) is a debilitating and progressive neurodegenerative brain disorder that damages the brain and impairs cognitive functions, particularly memory. The symptoms of AD deteriorate over time, ultimately leading to the loss of independence and the requirement for full-time care (). It is the major cause of dementia in the elderly, and its prevalence is increasing globally. The incidence of AD has a significant influence on the lives of the patient’s family, as well as a considerable financial cost to society (). The World Alzheimer’s Report 2021 estimates that 55 M people are living with Alzheimer’s disease and others with dementia (). The number is expected to rise to 152 M by the year 2050 (Li et al., 2022). It’s the sixth most common cause of mortality in the US and the fifth most common worldwide ().
The pathogenesis of AD is complex, and the precise etiology of Alzheimer’s disease is unclear. It is considered to be caused by a complex interplay of genetic, environmental, and lifestyle factors () along with inflammation, neurodegeneration, and neurofibrillary tangles (NFTs) in the brain. Key contributors to AD include BACE1, which cleaves amyloid precursor protein at the β-site within the extracellular domain, and the accumulation of two abnormal proteins, i.e., amyloid beta (Aβ) and hyperphosphorylated tau proteins (Silva et al., 2019) that interfere with the communication between nerve cells, leading to cell death and brain shrinkage (Selkoe, 2001). As the disease advances, affected individuals experience cognitive decline, memory loss, and behavioral changes. Some other variables, i.e., oxidative stress, lead to the formation of senile plaques and accumulation of NFTs (). Furthermore, increasing levels of inflammatory cytokines and related genes have been linked to the development of AD (). Mutations in genes such as amyloid precursor protein (APP), presenilin-1 (PSEN1), and presenilin-2 (PSEN2) have been identified as associated with the onset of familial Alzheimer’s disease, a rare form of the ailment that affects individuals with a family history of the condition (). Additionally, other identified risk factors encompass a history of head injury, elevated blood pressure, and increased cholesterol levels (Sperling et al., 2011).
Beta-secretase 1 (BACE1), a transmembrane aspartyl protease enzyme, is involved in the formation of Aβ peptides, which are one of the medical hallmarks of AD. To reduce the formation of amyloid beta peptides in AD patients, it’s necessary to inhibit BACE1 activity, as it is one of the key contributors involved in amyloidogenic processing. BACE1 is a challenging central nervous system (CNS) drug target. Understanding the structure and function of BACE1 provides valuable insights for designing potential inhibitors as drug candidates for the treatment of AD.
There is presently no proper cure for Alzheimer’s disease, but there are medications available that provide temporary symptomatic relief in some patients, but do not cure or halt the progression of AD, including cholinesterase inhibitors, memantine, N-methyl-D-aspartate (NMDA) receptor antagonists, and TRx0237, also known as HMTM (Hydromethylthionine mesylate) (). However, the results of some clinical trials have been inconsistent, while some have negative effects on cognitive functions, and some have been rejected by the FDA due to conflicting data in clinical trials (). Clinical success of BACE1 inhibitors has been limited by safety concerns and a lack of significant cognitive benefits. Moreover, some Aβ-targeting vaccines reduce mild amyloid burden in Alzheimer’s disease, including CAD106, ACI-24, and UB-311. None of these medications have yet received regulatory approval to treat AD, as they are still in the evaluation phase (Rafii et al., 2022; Winblad et al., 2012; Yu et al., 2023). Non-pharmacological interventions, such as cognitive training, physical activity, and social interaction, have also been found to improve cognitive function and quality of life in individuals with AD ().
As the global population ages, the incidence of AD is expected to rise. It is therefore important for researchers to continue studying such diseases and developing new treatments and interventions that can improve the standard of living for individuals with AD and their families. Several therapeutic approaches are currently being investigated, but further research is needed to develop effective treatments for this devastating disease (Prince et al., 2013). Most importantly, innovative therapeutic scaffolds with minimal side effects remain an urgent priority for medicinal chemists. The purpose of this study is to determine whether Coumarine-1,2,3-triazole hybrids with neuroprotective effects can inhibit BACE1 and help overcome AD.
The coumarin-1,2,3-triazole hybrids belong to a distinct class of benzopyrone compounds, which gained attention as potential candidates for their neuroprotective, anti-inflammatory, and anti-amyloidogenic properties (Lin et al., 2022). These synthetic compounds were designed to integrate structural features associated with both coumarin and 1,2,3-triazole moieties in order to increase their biological activity, stability, bioavailability, and therapeutic potential (). They have been the subject of extensive research due to their neuroprotective effects (; ), which are essential in combating neurodegenerative processes associated with AD. Moreover, some studies reported their effects as anti-diabetic, antioxidant, anti-tumor, antimicrobial, and antibacterial properties (; ; Shaikh et al., 2025). Reducing inflammation and inhibiting amyloid-beta aggregation are key strategies for addressing the underlying pathologies of the disease.
2 Materials and methods
The general flowchart of the methodology employed in the research project is depicted in Figure 1. Legends were given at the bottom of the flowchart to understand the proposed study.
FIGURE 1
2.1 Pharmacophore modelling and validation
40 cocrystal structures of BACE1 were retrieved from the RCS protein data bank (https://www.rcsb.org/), while their activity was obtained from ChEMBL (https://www.ebi.ac.uk/chembl/) database and literature review. Physicochemical properties of all cocrystal compounds were determined via the SwissADME (http://www.swissadme.ch/index.php) webserver. Poses and interaction information of cocrystal complexes were explored. The active ligands and their conformations were aligned via the flex align module to extract common shape and distance-based features that play a significant role in biological activity (Qing et al., 2014). Stochastic conformational search along iteration limit 20, failure limit 10, and energy cutoff value of 10 was set for alignment purposes. C27 depicts the key interaction pattern across the majority of co-crystalized ligands. Therefore, the co-crystal ligand with the highest activity was selected. The pharmacophore query was generated on template compound C27, which is a representative of all superimposed ligands. Common key features were added to generate a pharmacophore query with a 1.2 tolerance and 50% threshold value. Before virtual screening, the pharmacophore model was validated to assess its capability and specificity in discriminating between active and inactive compounds from a dataset comprising both active and decoy compounds. For this, a prepared dataset of 80 active compounds from the PDB and ChEMBL databases was used, based on their structural and physicochemical properties. Moreover, a library of 285 decoys from a Database of Useful Decoys Enhanced (DUD.E) was selected, which used similar physicochemical properties of C27, along with dissimilar 2D topology.
2.2 Library designing and virtual screening
An in-house library of 1840 coumarin-1,2,3-triazole hybrids was prepared by substitution of distinct functional groups at the 12th position to get potential compounds targeting beta-secretase 1. Coumarin is well known for its anti-inflammatory and neuroprotective properties, which are significant for the treatment of AD (). 1,2,3-triazoles are employed in medicinal chemistry as useful building blocks due to their stability, simplicity in synthesis, and potential for diverse interactions with distinct biological targets. When coupled with coumarin, they can display elevated pharmacological properties and target-specific interactions, making them attractive candidates for AD therapy (). Designed compounds depict good synthetic accessibility scores and fulfill the RO5 (Lipinski rule of 5) criteria. The pharmacophore model was employed to screen 1840 coumarin-1,2,3-triazole hybrids and resultantly get potential hits.
2.3 Structure-based molecular docking
Docking-based screening was performed to further screen hits and to identify potential ligands that depict a high probability of binding with beta-secretase 1. The macromolecule was preprocessed by eliminating water molecules, ions, and ligands (). Furthermore, fix the side chain atoms of residues. Then, polar hydrogens and charges were added to ensure that the receptor is properly prepared to capture important molecular interactions and accurately predict preferred binding modes. Coumarin-1,2,3-triazole hybrids were optimized and minimized by employing the UFF force field along the conjugate gradient optimization algorithm over 200 iterations. However, the energy difference was less than 0.1, and then the files were converted into pdbqt format. For blind docking, the size of grid coordinates was X = 64, Y = 76, Z = 57, which covers the whole protein. Blind docking helps to determine whether the hit compounds bind to the active site, other binding sites, or allosteric sites. Prepared receptor and ligands were subjected to molecular docking via PyRx 0.8. x software (). The software’s protocol was validated via redocking and cross-docking analysis, demonstrating its robustness and liability. Moreover, 2D and 3D depictions were subjected to Discovery Studio and PyMOL, respectively. The binding poses were analyzed by considering factors such as binding scores, RMSD values between native and docked poses, as well as distances among ligands and binding residues of BACE1.
2.4 ADMET analysis of top hits
Physicochemical properties, pharmacokinetic features (i.e., absorption, distribution, metabolism, excretion), and safety profile of potential hits were studied via SwissADME (), ADMETlab 3.0 (), and Osiris Property Explorer, which employs integrated machine learning and deep learning models trained on experimental datasets. ADME screening predicts how effectively the drug will be administered by the body in its active form. How it will be distributed with the ability to cross biological barriers, such as the blood-brain barrier, and reach target tissues. How rapidly it will be metabolized and excreted from the body, with special focus on renal, biliary, and fecal excretion. Toxicity testing predicts the potential adverse effects and safety issues associated with selected compounds. ADMET analysis provides critical data for assessing safety, minimizing the likelihood of late-stage drug failure, and guaranteeing patient safety. Multiple rules and scoring systems have been employed to assess the viability of chemical compounds for further exploration, with an emphasis on factors such as synthetic accessibility, logP values, physicochemical features, and bioavailability.
2.5 Bioactivity score prediction
To access the biological activity and efficacy of hit compounds, the PASS online webserver (https://way2drug.com/PassOnline/index.php) was utilized. Top hits against human receptors, i.e., GPCRs, ion channel modulators, numerous kinases, proteases, nuclear receptors, and enzyme inhibitors were explored. Bioactivity scores help to prioritize and predict potential off-target effects. Additionally, these scores assist in observing contrary interactions with unintended biological targets and resultantly minimize the chances of unexpected adverse effects ().
2.6 DFT study of coumarin-1,2,3-triazole hybrids
Density functional theory (DFT) and time-dependent density functional theory (TD-DFT) analysis were used to discover the best stable theoretical configuration for the compounds under study. The absence of imaginary frequencies served as an indication for these computational methods that led to the discovery of the energetically ideal state. To do this, optimal computational configurations were used to perform optimization at the ground-state level. Electrostatic potential (ESP) mapping and Density of State (DOS) analyses were also performed, in addition to the standard studies of Frontier molecular orbitals (FMOs), inhibitor characteristics, global chemical reactivity descriptors, and so on (). These elements helped to differentiate the system’s different contributions. Finding the characteristics and findings of the hybrid functional, the Becke-3 exchange functional method, provided by Lee, Yang, Parr (B3LYP), was employed (Russo, Martin, and Hay, 1994) with the basis set 6-31G** required the use of the DFT Package Gaussian 16, Revision C. 01 (), a software package in conjunction with the GaussView 6.0 program to visualize the findings. All the global reactivity parameters for the studied compounds were computed by utilizing Equations 1-5, where η, µ, S, represent hardness, chemical potential, Softness, electronegativity, and electrophilicity index.
The highest occupied molecular orbital (HOMO), the lowest unoccupied molecular orbital (LUMO), and the HOMO-LUMO energy gaps (E.g.,) are used to describe electronic energy levels of molecules. These energy levels play a crucial role in understanding the interaction between drugs and their target molecules, as well as in predicting various properties of the potential drug molecules themselves. The energy difference between HOMO and LUMO can provide insights into the chemical reactivity of a molecule. Molecules with smaller gaps are generally more reactive because electrons can be more easily excited from the HOMO to the LUMO, facilitating chemical reactions (). The HOMO and LUMO orbitals are important variables that show how excited simulated compounds are. The energy gap (E.g.,) is a good way to figure out how chemically reactive the compounds being studied are. The HOMO level acts as the source of electrons, whereas the LUMO level serves as the less effective electron acceptor. Due to the significant energy gap between them, the (HOMO-LUMO) interaction has an adverse impact, leading to the limited efficiency of charge transfer for the inhibitor.
2.7 MD simulation study
MD simulations were carried out to understand ligand-receptor binding mechanisms at the atomic level, along with their conformational alterations, utilizing the Desmond software (). The best binding poses of lead and reference compound were utilized as a starting conformation to run MD simulations. Both studied complexes were preprocessed via Maestro’s protein preparation wizard. Both complexes were minimized to remove steric clashes and unfavorable contacts. A 10Å orthorhombic box of TIP3P was utilized as a solvent model. Complexes further neutralized by the addition of 0.15 M Na+ and Cl− ions with the OPLS 2005 force field (Rasheed et al., 2021). The system builder tool was employed to prepare both systems and parameterize them for accurate and efficient molecular dynamics simulations. Both systems employed NPT (constant Number of particles, Pressure, and Temperature) ensemble to stabilize the system at 300 K temperature and 01 atmospheric pressure using Nosé-Hoover chain thermostat. After equilibrium, each MD trajectory was followed over 200 ns. The receptor-ligand interactions were investigated through the simulation interaction diagram tool. MD simulations for both systems were effectively performed on an Intel Core i9-12900 KF processor, equipped with 128 GB RAM, and an MSI GeForce RTX 3080 Ti GPU on a specially built liquid-cooled PC, installed Ubuntu 22.04.1 LTS.
2.8 Binding free energy calculation
The binding free energy (ΔGbind) for both systems was calculated via prime MM-GBSA to estimate the strength and stability of the studied complexes (). To calculate ΔGbind, complexes were optimized and minimized. Counter ions and solvent molecules were added. Default parameters, i.e., OPLS3 force field, dielectric constant along the VSGB solvent model, were applied to compute ΔGbind for both systems. The MD trajectory frames were selected at intervals of 10 ns after the simulation run. ΔGbind, Gcomplex, Gprotein, Gligand depict total binding free energy, binding free energy of complex, receptor, and ligand, respectively (Equation 6). ΔGbind is the combination of energy in gas and solution phase (Equation 7) (; ). Gas phase energy (ΔEgas) is the combination of van-der Waals, electrostatic, and internal energy (Equation 8). Solvation free energy (ΔGsol) is decomposed into polar (generalized born model) and nonpolar solvation free energy (solvent accessible surface area (SASA)), illustrated in (Equation 9) (). TΔS denotes conformational entropy expressed in (Equation 10), which was neglected due to high computational cost.
3 Results and discussion
3.1 Pharmacophore query generation and model quality assessment
High-resolution co-crystalized beta-secretase one structures were aligned and analyzed to investigate diverse structural features in its binding site along non-covalent interactions. The best 40 co-crystal structures of active compounds, along with their physicochemical properties and IC50 values, were displayed in Supplementary Table S1. Conserved residues of BACE1’s active site were labelled in Figure 2A, while aligned active compounds lay in the sub-pockets S1, S2, S3, and S4 of BACE1 (Figure 2B).
FIGURE 2
Based on biological activity, binding poses, common structural and shape-based features, C27 was selected as a reference compound to build a pharmacophore model. The final pharmacophore model contains four pharmacophoric features. The query features were analyzed based on four defined sub-pockets of the active site of BACE1, as displayed in Figure 2B. Both hydrogen bond donor and acceptor lie in the S1 sub-pocket, F1 aromatic feature lies near S2 subsite, while the aromatic ring, the F4 feature, lies near S3 and S4 sub-pockets of BACE1’s active site. Figure 3 depicts the template compound of highest activity along selected features, the distance among selected features, and the area under the ROC curve. The selected pharmacophoric features were F1: second ring (Aro), F2: N24 (Don), F3: N25 (Don/Acc), F4: first ring (Aro) with 100% score and 0.50 radius (Figure 3A). The distance between all selected features was less than 8Å (Figure 3B). Figure 3C displays top-aligned compounds with selected features. Further, the generated pharmacophore query was validated through a receiver operating characteristic (ROC) curve (Figure 3D). The best selected model with an area under the curve (AUC) value of 0.985 has excellent discriminatory power to differentiate active and decoys. Enrichment Factor (EF) for the top 1% was 14.7, while for the top 20% was 4.167, depicting that the pharmacophore model successfully identifies a high proportion of actives compared to random selection, which seems beneficial for virtual screening.
FIGURE 3
3.2 Pharmacophore-based virtual screening
Further, the generated query was employed to screen a prepared in-house library of 1840 coumarin-1,2,3-triazole hybrids and resultantly get 92 hits with an RMSD value of ≤0.6 Å and ≤350 D molecular weight (MW). Generally, lower RMSD values are associated with high structural consistency and stable conformations within the active site (Trott and Olson, 2010), while a stricter MW criterion was used to retain chemical space for subsequent optimization. Hits were carried out for molecular docking to find plausible binding modes. 92 hits along their SMILES, RMSD scores, molecular weight, TPSA, H.B.A., H.B.D., and Lipinski rule of five are shown in Supplementary Table S2.
3.3 Validation of docking protocol
Before docking, validate the reliability and accuracy of the PyRx docking protocol via two methods. Initially, redock co-crystal compounds (obtained from co-crystal structures: 2OF0, 2OHQ, 2VA7, 2OHT) at the same binding site, then investigate their binding modes and compare RMSD values. Align all four redocked co-crystal ligands with their native poses, and they depict <1.0 Å RMSD values (Supplementary Table S3), which illustrate the accuracy of docking methodology (Figure 4).
FIGURE 4
Furthermore, cross-docking analysis was performed as an additional validation approach to check the robustness of the docking protocol. For this, 03 BACE1 complexes were utilized, i.e., 2VA7, 2VA6, 2VA5. Each cocrystal ligand was docked with a non-native protein conformation. Then, calculate the RMSD values for the best-selected poses based on binding affinity and interaction profile, and compare them with their corresponding native conformations. The RMSD values are less than 1Å (Table 1), indicating the protocol’s ability to replicate correct binding modes under the structural variability.
TABLE 1
| Ligand (protein) | 2VA7 | 2VA6 | 2VA5 |
|---|---|---|---|
| L1 (2VA7) | — | 0.17 | 0.26 |
| L2 (2VA6) | 0.17 | — | 0.27 |
| L3 (2VA5) | 0.17 | 0.26 | — |
Cross-docking results of 3 BACE1 structures. RMSD values less than 1Å are considered excellent, within the range of 1-2Å acceptable, while above 2Å are depicted as unreliable results.
3.4 Docking and binding pose analysis
After validation of the PyRx docking protocol, 92 hits were subjected to structure-based molecular docking with the BACE1 receptor to acquire a comprehensive understanding of interactions and energies. All docked hits showed favourable binding affinity values between −9.2 and −6.7 kcal/mol (Supplementary Table S4). 34 hits follow the cutoff criteria of −8.0 kcal/mol. The next cutoff was employed based on RMSD (<1.5 Å) along with hydrogen bond distance (<3.5 Å) to pick the best poses and to prioritize binding conformations. Resultantly, the best six hits were selected. These drug-like molecules underwent additional scrutiny through superimposition on the reference ligand (C27), followed by interaction analysis. The results indicate that these chemicals lie within the binding site and depict a similar binding space as observed with C27 (Figure 5).
FIGURE 5
Various studies reported that Asp32 and Asp228 residues are crucial ones in BACE1’s catalytic mechanism (). Gly34 was implicated in the glycine loop of BACE1, a pivotal residue involved in substrate recognition and binding. The active site is slightly covered by the flexible flap region of beta-secretase 1. Tyr71 and Phe108, two residues in the flap region, have been linked to substrate binding and catalytic activity ().
The binding scores of the best selected compounds, i.e., CUM-0199, CUM-0158, CUM-0196, CUM-0351, CUM-0421, CUM-0350 were observed as −9.2, −8.9, −8.7, −8.7, −8.7, and −8.6 kcal/mol, respectively. On the other hand, the reference compound displays a binding affinity of −8.1 kcal/mol. CUM-0199 displayed the highest binding energy of −9.2 kcal/mol when compared with others and made interactions similar to those of the reference compound. It was observed that polar functional groups of CUM-0199, improved the docking scores by forming stable hydrogen bonds with key catalytic residues, particularly Asp32 and Asp228. Additionally, it made H-bond with polar uncharged residue Thr232. In contrast, hydrophobic substitutions contributed to improved ligand stability through van der Waals interactions with Ser35, Ile118, Ile110, Gly11, Gly13, Gly230, and Trp76 residues. Moreover, it shows nonpolar contacts with the aliphatic Gly34 residue (Figure 5A). On the other hand, the 1,2,3-triazole linker enabled proper interactions, with balanced polar and hydrophobic features, improving docking performance. CUM-0158 makes hydrogen bond interactions with aromatic Phe108, polar uncharged Thr232, Ser10, and negatively charged Asp32 and Asp228 within 3.3Å distance (Figure 5B). CUM-0196 illustrates interactions that lie within 3Å with polar uncharged Ser10, nonpolar aliphatic Gly11, and Phe108 with −8.7 kcal/mol binding affinity (Figure 5C). In Figure 5E, CUM-0351 shows binding pose interactions with Gly11 (2.29 Å), Gly230 (2.3 Å), Tyr71 (2.3 Å), Ser35 (3.2 Å), and Asp32 (3.3 Å). As shown in Figures 5F,G, CUM-0421 and CUM-0350 display H-bonds with Phe108, Gly11, Ser10, and polar uncharged Thr232 within distance cutoff criteria of <3.5 Å. Based on H-bonds, hydrophobic, van der Waals, Pi-Alkyl, Pi-Cation, and anion interactions, the proposed molecules mainly show interactions with conserved active site residues, i.e., Asp32, Gly34, Thr232, Phe108, Tyr71, Ser10, Ser35, Lys107, and Asp228. Table 2 depicts interactions among the active site residues of the receptor and the best 10 hits, along with their distances.
TABLE 2
| I Ds | Binding affinities (kcal/mol) | RMSD (Å) | H-bond interacting residues (Å) | van der Waal’s interacting residues (Å) | Pi-Alkyl/Pi-Cation/Anion interacting residues (Å) |
|---|---|---|---|---|---|
| CUM-0199 | −9.2 | 0.3 | Thr232 (2.2), Asp32 (2.2), Gly34 (2.1), Asp228 (3.3) | Ser35, Ile118, Ile110, Gly11, Gly13, Gly230, Trp76 | Tyr71 (3.5), Tyr198 (5), Phe108 (5.2), Trp115 (4.9), Leu30 (5.0) |
| CUM-0158 | −8.9 | 0.4 | Asp32 (2.14), Asp228 (3.3), Gly34 (2.5), Phe108 (1.93) | Trp115, Thr231, Ile126, Arg128, Ser35, Ile118, Leu30, Ile110 | Gln12 (3.6), Ile110 (3.7), Ile118 (5), Tyr71 (5), Tyr198 (3.7) |
| CUM-0196 | −8.7 | 0.5 | Phe108 (2.6), Gly11 (2.5), Ser10 (2.12, 2.47) | Ile110, Gly230, Thr231, Gly34, Thr232, Asp32, Gly74 | Tyr71 (3.7.4.0), Gln12 (3.6), Phe108 (4.0) |
| CUM-0351 | −8.7 | 0.5 | Gly11 (2.29), Gly230 (2.3), Tyr71 (2.3), Ser35 (3.2), Asp32 (3.3) | Phe108, Ile110, Lys107, Ser10, Gln12, Gly13, Leu30, Thr231, Asp32, Ser35 | Ile118 (4.5) |
| CUM-0421 | −8.7 | 0.5 | Gly11 (3.2), Thr232 (2.3), Ser10 (2.01), Lys107 (2.7), Phe108 (3.0) | Gly13, Ala335, Thr231, Ile110, Gly230, Gly34, Ser35, Trp115 | Try71 (3.7, 3.9), Gln12 (3.6) |
| CUM-0350 | −8.6 | 0.6 | Lys107 (), Phe108 (2.42, 2.47), Thr232 (2.09), Ser10 (2.5), Gly11 (2.7) | Trp115, Gly13, Lys107, Gly230, Thr231, Asp32, Leu30 | Ile110 (3.8), Gln12 (3.7), Ile118 (5.0), Tyr71 (3.9) |
| CUM-1228 | −8.6 | 0.5 | Thr232 (2.7), Ser10 (2.2, 2.5), Gly11 (2.2), Phe108 (2.2, 2.7) | Asp32, Trp115, Lys107, Ser35, Gly230, Thr231, Gln12, Gly13, Ala335 | Ile110 (3.7), Ile118 (5.0), Phe108 (4.4) |
| CUM-0234 | −8.5 | 0.5 | Gly11 (2.2), Ser10 (2.2, 2.4), Thr232 (2.6), Gly230 (2.9), Phe108 (2.4, 2.75) | Trp115, Gly13, Gly13, Thr232, Lys107, Asp32, Ala335 | Phe108 (4.5), Ile110 (3.9), Tyr71 (3.9, 5.3) |
| CUM-0736 | −8.5 | 0.5 | Asp228 | Gly13, Phe108, Asp32, Ser35, Ile110, Thr232, Gly11, Gly13, Ser10 | Tyr71 (3.4), Ile118 (4.8), Leu30 (5.0), Gln12 (4) |
| CUM-1333 | −8.5 | 0.3 | Phe108 (2.3, 2.5), Gly11 (2.3), Ser10 (2.4), Thr232 (2.8), Tyr71 (3) | Lys107, Trp115, Leu30, Gln12, Gly13, Gly230, Ala335, Thr231, Asp32, Gly34 | Ile110 (3.8), Ile118 (5), Phe108 (4.4) |
| C27 (Reference ligand) | −8.1 | 0.2 | Gly230 (2.19), Ser10 (3.5), Asp228 (2.7), Asp32 (2.02) | Gly13, Thr232, Ser35, Tyr71, Gly11, Ile118, Thr231, Ile110, Ser229, Ala335 | Gln12 (3.9), Leu30 (5), Trp115 (4.9), Phe108 (5) |
Docking scores with RMSD and some other calculated features of hit compounds.
3.5 LE, LLE, and LELP prioritize hit selection
In rational drug design and optimization, ligand efficiency (LE) plays a crucial role in prioritizing hits and identifying potential drug candidates. Commonly, a higher magnitude (more negative) is considered worthy. LE of reference compound was −0.35 kcal/mol. Comparatively, the potency of CUM-0199 relative to its size was −0.37, while for other hits it lies within an acceptable range of −0.34 to −0.36 kcal/mol (Figure 6A). Lipophilic ligand efficiency (LLE) and Ligand Efficiency Lipophilic Penalty (LELP) prioritize lipophilic molecules with favorable drug-like properties, resulting in increased selectivity and overall success in the drug development pipeline. LLE and LELP values for the reference compound were six and -1.01 kcal/mol, respectively. On the other hand, selected compounds contain higher LLE and lower LELP, which are considered beneficial. CUM-0199 contains higher LLE (7.50) and −0.45 LELP, indicating that it achieves binding potency with minimal lipophilicity, which reduces off-target effects and toxicity, better selectivity, and BBB penetration (). These matrices are used as preliminary indicators in early-stage drug discovery and further require experimental validation. Figure 6B displays a positive correlation between ligand efficiency and drug score, indicating that high values of LE tend to have better drug scores.
FIGURE 6
3.6 ADME and toxicity profile of potential compounds
To investigate potential therapeutic candidates, the approach of drug-likeness evaluates hits based on their physicochemical properties, structural features, ADMET analysis, and specific rules, i.e., Lipinski, Egan, Veber, Muegge, and Ghose. As seen in Figure 7, physicochemical properties of all the selected hits lie within the acceptable range as stated in the rule of five (RO5), and no compound violates these rules.
FIGURE 7
CNS (central nervous system) drugs have stringent criterion than peripheral targets. Therefore, rigorously inspect all the checkpoints as listed in Table 3. Absorption level of hits was predicted via Caco-2 (log cm/s), MDCK, skin, and GI (gastrointestinal) absorption parameters, along with 20% and 30% human oral bioavailability. Hits considered Caco-2 Permeable with predicted value > -5.15 log cm/s. CUM-0199, CUM-0351, CUM-0421 compounds showed excellent absorption levels. All hits demonstrate moderate to good MDCK Permeability between the range of 4e-06 to 6.7e-06 cm/s. The drug should have adequate penetration through the skin to attain therapeutic concentration while avoiding excessive systemic absorption. According to predicted data, all hits showed skin permeation. High GI absorption is frequently required for oral medications, and all hits fulfill this criterion. Human oral bioavailability is an essential pharmacokinetic parameter that should be ≥ F20% and ≥ F30%, within the range of 0–one (excellent to poor). CUM-0199 and CUM-0421 compounds depict higher F20% and F30% values, indicating better oral bioavailability, while CUM-0158 and CUM-0196 compounds show moderate bioavailability. Contrastingly, CUM-0350 displays lower F20% and F30% values. The CNS drugs should be BBB (blood-brain barrier) permeant to reach their targeted receptor, and the value should lie within 0–one (excellent to poor) range. All hits displayed BBB permeability except CUM-0158 and CUM-0421. Other distribution parameters, such as PPB (plasma protein binding) and volume distribution (VD), describe In vivo drug uptake and distribution. Hits with higher PPB and VD above the range of 0.04–20L/kg exhibit a lower therapeutic index. CUM-0199 shows a lower PPB value than others, along with an excellent 0.765L/kg VD. Phase I human cytochrome P450 (CYPs) enzymes are drug metabolizing enzymes (oxidative reactions) that bio-transform and detoxify >80% xenobiotics and values lie within the range of 0–1. All hits are prime CYP1A2 inhibitors within the range of 0.7–1.0. Moreover, 50% hits, i.e., CUM-0199, CUM-0351, and CUM-0421 were moderate to effective CYP2C19, CYP2C9, CYP2D6, CYP3A4 inhibitors, and could be used as chemotherapeutic agents. Another significant pharmacokinetic metric is drug clearance (CL), which is crucial for reducing toxicity and improving the therapeutic efficacy of hits. Predicted CL value exceeding 5 mL/min/kg signifies excellence, while CL < 5 mL/min/kg indicates poor clearance. Reported data for all compounds denoted moderate CL values < 15 mL/min/kg, which clearly showed that compounds eliminated from excessive clearance criteria and did not demonstrate toxicity. Half-life of all hits along the reference system ranges from 0 to one (excellent to poor), indicating reduced risk of accumulation and sustained therapeutic effect.
TABLE 3
| Pharmacokinetic properties | | CUM-0199 | CUM-0158 | CUM-0196 | CUM-0351 | CUM-0421 | CUM-0350 |
|---|---|---|---|---|---|---|---|
| Absorption | Caco-2 permeability | −5.121 | −5.230 | −5.311 | −5.097 | −5.080 | −5.351 |
| | MDCK permeability | 5.4e-06 | 4e-06 | 4.7e-06 | 6.7e-06 | 5.3e-06 | 5.1e-06 |
| | Skin permeation (cm/s) | −6.74 | −6.37 | −7.12 | −6.86 | −6.86 | −7.23 |
| | GI absorption | High | High | High | High | High | High |
| | F20% | 0.1 | 0.2 | 0.5 | 0.7 | 0.1 | 0.7 |
| | F30% | 0.1 | 0.5 | 0.8 | 0.9 | 0.2 | 1.0 |
| Distribution | BBB penetration | 0.3 | 0.5 | 0.3 | 0.3 | 0.7 | 0.3 |
| | PPB | 94.168% | 95.540% | 95.233% | 95.229% | 95.166% | 95.055% |
| | Volume distribution | 0.765 | 0.894 | 1.040 | 0.656 | 1.425 | 0.816 |
| Metabolism | CYP1A2 inhibitor | +++ | +++ | ++ | +++ | +++ | ++ |
| | CYP2C19 inhibitor | + | + | -- | + | ++ | - |
| | CYP2C9 inhibitor | -- | --- | - | --- | + | -- |
| | CYP2D6 inhibitor | + | -- | - | + | + | - |
| | CYP3A4 inhibitor | - | - | -- | -- | + | -- |
| Excretion | Total clearance (mL/min/kg) T1/2 | 8.975 0.192 | 8.074 0.258 | 7.642 0.413 | 10.000 0.277 | 6.895 0.282 | 8.529 0.779 |
| Toxicity | hERG blockers | --- | -- | -- | - | -- | - |
| | Human hepatotoxicity | -- | +++ | - | +++ | + | - |
| | Drug-induced liver injury | ++ | +++ | +++ | +++ | +++ | +++ |
| | AMES toxicity | -- | - | -- | -- | -- | --- |
| | Rat oral acute toxicity | -- | ++ | -- | ++ | - | ++ |
| | Eye irritation | -- | -- | --- | --- | --- | --- |
| | Eye corrosion | --- | --- | --- | --- | --- | --- |
| | Respiratory toxicity | + | +++ | + | ++ | - | + |
| | Skin sensitization | -- | + | --- | + | -- | -- |
| | Carcinogenicity | +++ | +++ | +++ | +++ | +++ | ++ |
| | Mutagenic | No | No | No | No | No | No |
| | Tumorigenic | No | No | No | No | Yes | No |
| | Reproductive effect | No | No | No | No | Yes | No |
| | Irritant | No | No | No | No | No | No |
| Toxicophoric rules | Acute toxicity rule | 0 alert | 0 alert | 0 alert | 0 alert | 0 alert | 0 alert |
| | Genotoxic carcinogenicity rule | 0 alert | 2 alerts | 5 alerts | 1 alert | 1 alert | 1 alert |
| | Nongenotoxic carcinogenicity rule | 0 alert | 0 alert | 1 alert | 0 alert | 1 alert | 1 alert |
| | Aquatic toxicity rule | 0 alert | 0 alert | 3 alerts | 0 alert | 0 alert | 1 alert |
| | Non-biodegradable rule | 0 alert | 0 alert | 0 alert | 0 alert | 1 alert | 1 alert |
| | SureChEMBL rule | 0 alert | 1 alert | 0 alert | 0 alert | 0 alert | 0 alert |
ADMET analysis of the best selected compounds along with some toxicophoric rules’ alert. The prediction values transformed into six symbols: 0-0.1(---), 0.1-0.3(--), 0.3-0.5(−), 0.5-0.7(+), 0.7-0.9(++), and 0.9-1.0(+++). MDCK (Madin−Darby Canine Kidney cells), Caco-2 (human colon adenocarcinoma cell lines), BBB (blood-brain barrier), PPB (plasma protein binding), F20%, F30% (oral bioavailability).
The compounds frequently fail and lead to costly termination of the drug development process due to their toxicity profile. Thus, it is preferable to foresee it at the early stages. None of the hits acts as hERG blockers, indicating that they do not inhibit the normal function of the potassium channel. Three hits demonstrated excellent hepatotoxicity (H-HT) while two exhibited moderate toxicity, and only CUM-0199 was nontoxic. One of the primary safety concerns of hit withdrawal from the market is DILI (drug-induced liver injury), and the probability of DILI was very low for all hits. None of the compounds demonstrated Ames toxicity, eye irritation, eye corrosion, or mutagenicity, while 50% depicted oral rat acute toxicity. In addition to this, CUM-0421 exhibited tumorigenic and reproductive effects. Further hits filtered based on toxicophoric rules, i.e., acute toxicity rule, genotoxic carcinogenicity rule, nongenotoxic carcinogenicity rule, nonbiodegradable rule, aquatic toxicity rule, and Sure-ChEMBL rule (). Solely CUM-0199 was identified as the potential hit among the top compounds as it passes all these rules.
3.7 Bioactivity profile analysis of potent hits
Bioactivity scores better comprehend compound’s poly-pharmacological potential and identify potential hits along with a favorable safety profile. The compounds with poor bioavailability may be unable to access their target site in sufficient concentrations to achieve a therapeutic effect and may require modifications to improve their bioavailability and activity. A compound is deemed active if its bioactivity score exceeds 0.0, slightly active if it falls within −5.0 and 0.0, and inert if it falls below −5.0. Table 4 displayed that all hits had bioactivity scores ranging from −5.0 to 0.45, clearly indicating their potential as drugs.
TABLE 4
| Parameters of bioactivity score | ||||||
|---|---|---|---|---|---|---|
| Compounds | GPCR ligand | Ion channel modulator | Kinase inhibitor | Nuclear receptor ligand | Protease inhibitor | Enzyme inhibitor |
| CUM-0199 | −0.35 | −0.54 | −0.03 | −0.70 | −0.48 | −0.10 |
| CUM-0158 | −0.33 | −0.32 | 0.06 | −0.50 | 0.45 | −0.10 |
| CUM-0196 | −0.20 | −0.63 | 0.08 | −0.80 | −0.55 | −0.07 |
| CUM-0351 | −0.35 | −0.54 | −0.03 | −0.70 | −0.48 | −0.10 |
| CUM-0421 | −0.20 | −0.56 | 0.18 | −0.83 | −0.38 | −0.12 |
| CUM-0350 | −0.06 | −0.21 | 0.18 | −0.70 | −0.33 | 0.04 |
Bioactivity score of the top six compounds with different human receptors.
3.8 Structural stability of hits via density functional theory
In comparison to other investigated compounds, CUM-199 exhibits a significantly lower energy gap (E.g.,). The interactions between ligands and proteins in a complex primarily depend on FMOs. Favourable FMO distribution facilitates H-bonding and electrostatic interactions with BACE1 active site residues such as Asp228 and Asp32. The decreasing order of energy gap values was STD > CUM-0350 > CUM-0351 > 0158 > CUM-0421 > CUM-0196 > CUM-0199. Specifically, the energy gap values for CUM-0199, CUM-0158, CUM-0196, CUM-0351, CUM-0421, CUM-0350, and STD were 2.68 eV, 3.60 eV, 3.57 eV, 3.60 eV, 3.58 eV, 3.65 eV, and 5.19 eV, respectively. CUM-0199 exhibited the greatest reactivity in terms of charge excitation, excitation values, energy gap, and the smallest LUMO level, owing to the inverse correlation between the amount of energy difference and ligand reactivity. Higher reactivity may enhance the ligand’s ability to form satisfactory interactions with the BACE1 active site, thus establishing stable protein-ligand complex formation. The negative values of the chemical potential (µ), as presented in Table 5, signify that all compounds are stable. The chemical potential reflects the tendency of the acceptor to acquire a negative charge, which indicates the stability of the compounds and their capacity to form enduring complexes with the receptor (). In comparison to the other compounds, CUM-0199 stands out with a hardness (ղ) value of 1.34, making it well-suited for further analysis. This observation aligns with its docking trend and potential connectivity. The electron density is depicted graphically in Figure 8 using the HOMO-LUMO orbitals. Red and cyan were used to represent the HOMO and LUMO states, respectively. These colors aid in locating areas rich in electrons. It becomes clear that CUM-0199 has the best inhibitory qualities after an extended examination of HOMO, LUMO, and complete reactivity indicators. This is because it has a small energy gap, which improves local and global attributes.
TABLE 5
| Compounds | HOMO | LUMO | E.g., | |||||
|---|---|---|---|---|---|---|---|---|
| CUM-0199 | −4.625 | −1.938 | 2.68 | 1.34 | 3.28 | 0.74 | −3.28 | 4.01 |
| CUM-0158 | −5.478 | −1.870 | 3.60 | 1.80 | 3.80 | 0.55 | −3.80 | 4.00 |
| CUM-0196 | −5.403 | −1.825 | 3.57 | 1.78 | 3.61 | 0.56 | −3.61 | 3.65 |
| CUM-0351 | −5.453 | −1.844 | 3.60 | 1.80 | 3.64 | 0.55 | −3.64 | 3.67 |
| CUM-0421 | −5.407 | −1.824 | 3.58 | 1.79 | 3.61 | 0.55 | −3.61 | 3.57 |
| CUM-0350 | −5.466 | −1.808 | 3.65 | 1.82 | 3.63 | 0.54 | −3.63 | 3.61 |
| STD | −6.011 | −0.812 | 5.19 | 2.59 | 3.41 | 0.38 | −3.41 | 2.23 |
Calculated energy of HOMO and LUMO, energy gap (E.g.,), Hardness (η), Softness (S), electronegativity (X), chemical potential (µ), and electrophilicity index (ɷ) of all investigated compounds.
FIGURE 8
3.8.1 Molecular electrostatic potential (MEP)
Understanding the distribution of electrons in drug discovery systems and studying how molecules react to electrophilic and nucleophilic assaults are both aided by MEP maps. As shown in Figure 9, the MEP surface map provides a visual representation of the regional distribution of molecular charges. These representations draw attention to areas of changing charge density inside molecules, shedding information on their interactions with biological entities through hydrogen bonding and exposing changes in electron density. It is now feasible to predict molecular interactions by evaluating data on charge distributions. In addition, the amount of positive and negative charges a molecule carries may be deduced from the charge density areas of the MEP map (). This property is widely used in a variety of contexts, including the prediction of electrophilic attack sites, the measurement of relative reactivity, the analysis of biological recognition, and the investigation of interactions enabled by hydrogen bonding or electrostatic interactions. Polar and negatively charged regions are indicated in red for molecular electric density, whereas non-polar and positively charged regions are shown in blue. On the many faces of the (MEP) maps, different colors represent different electrostatic potentials (). Our developed compound, CUM-0158, demonstrates a robust capacity for charge transfer, as assessed through the global reactivity parameter known as chemical potential (3.80 eV). The charge transfer properties of the molecule are heavily impacted by its chemical potential value. It should be noted that the chemical’s potential value is proportional to the amount of charge transfer. Compound CUM-0158 shows a high electronegativity (X = 3.80), low hardness, high softness, ideal nucleophilic attacking index, and the most favorable level of energy (LUMO level −1.870 eV), hence indicating that it is electrophilic, according to the investigation. Another important reactivity measure, hardness, is calculated by comparing the energy gaps among the HOMO and LUMO orbitals. A large energy gap represents a stiff molecule, whereas a small one is indicative of a flexible molecule linked to very reactive chemical compounds.
FIGURE 9
The MEP maps exhibit a color scheme where blue signifies advantageous positions for nucleophilic attacks, red denotes areas unfavorable for electrophilic attacks, and green indicates sites with zero potential or neutral regions between the highly negative and positive regions within the studied compounds.
3.8.2 Density of state analysis (DOS)
Figure 10 depicts the results of a density of state analysis, which measures the total number of states in a system over a specific energy range. This evaluation is important for analyzing the interactions between fragments and their impact on the energy levels of molecular orbitals in the complex. DOS was evaluated using the B3LYP/6-31G** theoretical framework to understand the influence of each segment on the charge density distribution order. The FMO and DOS are Frontier molecular orbitals. The density of electron distribution in these orbitals alters when acceptor components with varying electron-withdrawing capacities are used, according to further investigations on the ratios of different components in HOMO and LUMO (). Occupying the charge distribution surrounding the donor and acceptor groups are bonding and antibonding orbitals, as seen in the DOS diagram. Negative energy values represent bonding orbitals, and positive energy values represent antibonding orbitals. A score of 0 means there is no contact between the bonding orbitals. DOS calculations confirmed the FMO observations. Each molecule exhibits a specific arrangement of the acceptor groups HOMO and LUMO, which results in a change in the energy difference between these two orbitals. The extent of this interaction can be measured by examining the DOS spectrum.
FIGURE 10
3.9 Conformational stabilities of potent hit complexes via MD simulation
Through investigation of 200 ns MD simulation, assess the overall stability, flexibility, inter-molecular interactions, and accuracy of both complex conformations (CUM-0199 and STD systems). Figure 11 illustrates the RMSD for the best hit and the standard (STD) system’s carbon alpha atoms. Generally, the optimal range for stable complexes, in terms of RMSD, is considered to be between 2 and 3 Å. It is noteworthy that both complexes exhibit stable P-RMSD (Protein Root Mean Square Deviation) values, lie within 1.8 ± 0.05 Å, and remain steady with little variation over the simulation time. Protein RMSD of CUM-0199 system stabilizes till 100ns, followed by a little upsurge until 120ns, and subsequently re-establishes stability until 200ns. P-RMSD of the standard (STD) system remains steady over the period of simulation. Ligand RMSD varied within the range of 2 Å, indicating that the ligand retained its binding mode in the active site of BACE1 and did not undergo large conformational changes.
FIGURE 11
In order to check the mobility of both complex systems, the Root Mean Square Fluctuation (RMSF) of individual residues was calculated across the simulation time. Figure 12 indicates regions that experience significant fluctuations during simulation. An elevated RMSF value indicates flexible regions, i.e., loops, while a lower RMSF value demonstrates rigid segments such as alpha helices and beta sheets. Specific elevated regions, such as L1 to L4, demonstrate loops, H1 depicts helices, and T1 represents turns. The N-terminal region of the receptor in both complex systems exhibits heightened flexibility due to the existence of loops and exposure to the solvent environment. Apart from that, both CUM-0199 and STD systems maintain a lower RMSF value throughout the simulation time. Active site residues, i.e., Thr232, Asp32, and Gly34, depict minimal mobility, signifying that conformational alterations within these regions are minimal. These observations imply secure and firm binding of CUM-0199 compound within the active site of the targeted receptor. To get an understanding of mobility in the studied systems, P-RMSF was compared with the experimental B-factor value (). An average B factor value was reported as 33.39 Å2 (∼5.77 Å), indicating average thermal motion of atoms in the crystal structure. On the other hand, both simulated complexes display almost the same trend in RMSF values as they lie within an optimal range of ∼2 to 2.9 Å, except for one loop (L4) region. Residues in the range of 295–310 exhibited greater fluctuations compared to other residues. Overall, both complexes depict acceptable RMSF values, indicating reduced mobility and competent system stability.
FIGURE 12
3.9.1 Exploration of secondary structure elements and stability analysis
Analysis of secondary structure elements (SSEs) provides crucial insights about structural characteristics and stability profile of complexes. The distribution of SSEs in both complexes was explored to track conformational changes and the influence of the attached ligand on the secondary structure formation (α-helices and β-sheets), which is considered crucial for their functionality. The pattern of alpha helices formation remains almost stable with only minor variations, as seen in Figure 13, while the pattern of beta sheets fluctuates ∼2% during the entire simulation period in both CUM-0199 and STD systems. CUM-0199 system is made up of 6.23% alpha helices, 31.64% beta strands, and a total of 37.87% SSEs. On the other hand, the STD system constitutes 39.16% of total SSEs with 6.23% alpha-helices and 39.16% beta strands. Secondary structure analysis revealed that the percentage of β-sheet decreases in CUM-0199 as compared to the STD system. We noticed that the attached ligand somewhat impacted the formation of secondary structures and reduced the content of stable moieties like beta strands, but this little change in SSEs does not leave a huge impact on the 3D structure and function of the targeted receptor.
FIGURE 13
3.9.2 Comprehensive dynamic interactions
Figure 14 depicts a summary of several types of interactions that CUM-0199 and STD (C27) compounds had with the receptor over the simulation time. CUM-0199 and STD made H-bonds, hydrophobic, water bridges, and ionic interactions with BACE1’s active site residues. H-bonds hold paramount significance among other types of interactions. In the case of the current simulation study, hydrogen bonds and water bridges are the major types of interactions, while hydrophobic and ionic interactions contribute around 64% and 36% of the simulation time, respectively. Negatively charged Asp32 and Asp228 are important residues to their activity, and they make multiple interactions such as H-bonds, ionic, and water bridges, over 100% and 194% of simulation time. Based on the protein-ligand contact summary of both systems, an average H-bond formation capacity was higher in the Asp32 residue compared to others. Positively charged Lys107, aromatic Phe108, and Tyr71 depict H-bonds, water bridges, and hydrophobic interactions for more than 62% of the simulation period. Gly34, Ser35, Thr232 made water bridges while Ile110, Trp115, Ile118 made hydrophobic interactions with CUM-0199 and STD compound C27. Strength and number of interactions of the mentioned residues within the active site of BACE1 depict the stability of both complexes.
FIGURE 14
Chronological representation of the contacts established by CUM-0199 and STD compounds with residues of the target protein, illustrated in Figure 15. The cumulative count of receptor-ligand interactions was displayed in the top panel, while the bottom panel lists the most significant and persistent forms of interactions formed by BENZI-0660 and STD. Robust interactions with CUM-0199 and STD are indicated by the abundance of deep orange bands in the bottom panel, which correspond to certain residues, i.e., Asp32, Tyr71, and Asp228.
FIGURE 15
3.9.3 Ligand torsion profile of CUM-0199 and STD system
Ligand torsion profile of CUM-0199 and STD ligands illustrated in Figure 16 over 200 ns MD simulation, which gives insights about the conformational stretch that ligands undergo to retain receptor-bound conformation (). The top panel for both studied ligands displays five rotatable bonds (RBs) in distinct colors, whereas the bottom panel shows the dial and the bar plots in the same color. Average probability density for CUM-0199 lies near 180°, 70°, −120°, −80°, and −180° with 2.39, 5.07, 8.00, 3.93, and 10.00 kcal/mol torsion potential, respectively. On the other hand, STD depicts average probability density around 80°, −100°, 0°, 80°, 150° with torsion potential of 8.05, 0.17, 7.83, 3.12, 5.68 kcal/mol, respectively. Table 6 demonstrates both studied ligands’ properties. RMSD for CUM-0199 lies between 0.6–1.8 Å, while for STD ranged between 0.8–1.2 Å. The radius of gyration (rGyr) provides insights into the conformational flexibility of ligands. Larger rGyr fluctuations indicate greater conformational changes. rGyr for CUM-0199 and STD lies within the optimum range of 4.00–4.50 Å. Molecular surface area (MolSA) for CUM-0199 exhibits dynamic variations between 304–312 Å2, while for STD it diverges from 344 to 360 Å2. Solvent Accessible Surface Area (SASA) for CUM-0199 and STD demonstrates fluctuation spanning from 50–200 Å2 and 30–120 Å2, respectively. Polar Surface Area (PSA) varies from 240–258 Å2 and 108–126 Å2 for CUM-0199 and STD, respectively.
FIGURE 16
TABLE 6
| Ligand properties | RMSD (Å) | MolSA (Å2) | SASA (Å2) | PSA (Å2) | rGyr (Å) |
|---|---|---|---|---|---|
| CUM-0199 | 1.2 | 310 | 110 | 247 | 4.45 |
| C27 (Reference) | 1.2 | 355 | 70 | 120 | 4.25 |
Both CUM-0199 and STD compounds’ properties were determined via MD simulation, such as RMSD, Molecular Surface Area (MolSA), Radius of Gyration (rGyr), Polar Surface Area (PSA), and Solvent Accessible Surface Area (SASA) fluctuated near the equilibrated state, confirming the stability of both ligands.
3.10 Binding free energy analysis
The binding free energy of both complex systems was computed via Molecular Mechanics-Generalized Born Surface Area (MM/GBSA) analysis, which helps to assess the strength of protein-ligand interaction, confirm the accuracy of best ligand poses, and binding orientations. and rank ligands based on their binding free energies (ΔGbind). All these constraints are beneficial for the systematic design and development of a drug. MM/GBSA calculates ΔGbind to offer reliable estimates of binding affinity, as shown in Figure 17, which depicts the binding free energy landscape for six hits in comparison to the reference system. According to the energy landscape, CUM-0199 has a more pronounced negative ΔGbind = −46.03 kcal/mol value, comparable to STD’s ΔGbind = −51.76 kcal/mol value, indicating robust ligand-receptor interactions. The variation in ΔGbind values between reported and virtually designed hits could be due to size differences. Van der Waals and Coulombic interactions contribute more to all systems than other types of interactions. CUM-0158, CUM-0196, CUM-0351, CUM-0421 systems display −31.65, 24.74, −14.21, −25.69, −17.30 ΔGbind values, respectively. MM/GBSA results were consistent with molecular docking and MD simulation results, and the ΔGbind value supports CUM-0199 to inhibit the BACE1 receptor.
FIGURE 17
4 Conclusion
In the current study, shape and distance-based virtual screening techniques identified tentative hits featuring a 1,2,3-triazole linking group and a coumarin fragment. Designed coumarin-1,2,3-triazole hits docked well in BACE1’s active site and displayed tremendous pharmacokinetic properties. These compounds indicate substantial improvements in the development of non-peptidic, druglike BACE1 inhibitors with moderate to excellent LE, bioavailability, and bioactivity. DFT studies gave excellent insights into the steric and electrical properties of top-selected compounds, important in their biological activity, and assessed the possibility of hits as therapeutic candidates. MD simulation and binding free energy analysis suggested that CUM-0199 system depicts stability along acceptable conformational changes with −46.03 kcal/mol binding free energy. Asp32 and Asp228 residues play a significant role in the binding of identified hits with the BACE1 receptor. We proposed that such information could be employed to develop a medication that selectively targets BACE1 while not disrupting other biological functions. The results will be useful to researchers and may lead to the development of a novel medicine for the treatment of Alzheimer’s disorders.
Statements
Data availability statement
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.
Author contributions
MA: Software, Investigation, Writing – original draft, Resources, Visualization, Funding acquisition, Validation, Formal Analysis, Methodology, Writing – review and editing, Conceptualization. IA: Software, Formal Analysis, Writing – review and editing, Data curation, Methodology, Writing – original draft, Conceptualization, Validation, Investigation, Visualization. GM: Investigation, Visualization, Writing – review and editing, Writing – original draft, Methodology, Data curation. AS: Formal Analysis, Visualization, Project administration, Data curation, Validation, Conceptualization, Methodology, Supervision, Writing – review and editing, Writing – original draft, Investigation, Software.
Funding
The author(s) declared that financial support was not received for this work and/or its publication.
Acknowledgments
The authors acknowledge their research institutes and departments for technical and administrative support.
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem.2026.1824875/full#supplementary-material
References
1
AbbasF.MohammadiM. D.LouisH.AgwambaE. C. (2023a). High-performance non-fullerene acceptor-analogues designed from dithienothiophen [3, 2-b]-pyrrolobenzothiadiazole (TPBT) donor materials. J. Mol. Model.29, 31. 10.1007/s00894-022-05435-x
2
AbbasF.MohammadiM. D.LouisH.AmoduI. O.CharlieD. E.GberT. E. (2023b). Design of new bithieno thiophene (BTTI) central core-based small molecules as efficient hole transport materials for perovskite solar cells and donor materials for organic solar cells. Mater. Sci. Eng. B291, 116392. 10.1016/j.mseb.2023.116392
3
AbreudeDutraI.Vicente ForlenzaO.BarrosHÉLIO LAUAR D. E. (2005). Alzheimer disease: correlation between memory and autonomy. Archives Clin. Psychiatry32, 131–136. 10.1590/s0101-60832005000300005
4
AbubakarM. B.SanusiK. O.UgusmanA.MohamedW.KamalH.IbrahimN. H.et al (2022). Alzheimer’s disease: an update and insights into pathophysiology. Front. Aging Neuroscience14, 742408. 10.3389/fnagi.2022.742408
5
AlabbasA. B. (2023). Identification of promising methionine aminopeptidase enzyme inhibitors: a combine study of comprehensive virtual screening and dynamics simulation study. Saudi Pharm. J.31, 101745. 10.1016/j.jsps.2023.101745
6
AliU.AbbasF. (2022). An extension of electron acceptor sites around Thiazolothiazole unit for evaluation of large power conversion efficiency: a theoretical insight. Spectrochimica Acta Part A Mol. Biomol. Spectrosc.281, 121610. 10.1016/j.saa.2022.121610
7
AliI.IqbalM. N.IbrahimM.HaqI.UlAlonaziW. B.SiddiqiA. R. (2023). Computational exploration of novel ROCK2 inhibitors for cardiovascular disease management; insights from high-throughput virtual screening, molecular docking, DFT and MD simulation. PloS One18, e0294511. 10.1371/journal.pone.0294511
8
AliI.RasheedM. A.CavaluS.RahimK.IjazS.YahyaG.et al (2023). Identification of natural lead compounds against hemagglutinin-esterase surface glycoprotein in human coronaviruses investigated via MD simulation, principal component analysis, cross-correlation, H-bond plot and MMGBSA. Biomedicines11, 793. 10.3390/biomedicines11030793
9
AliI.HaqI.UlAkbarN. I.IqbalM. N.RahimK.Ali KhanA.et al (2024). Computational exploration and molecular dynamic simulation for the discovery of antiviral agents targeting Newcastle disease virus. Microb. Pathog.196, 106884. 10.1016/j.micpath.2024.106884
10
AliI.AlmostafaM.AbbasF.YounisN. S.Ali KhanA.YahyaG. (2025). Computational discovery of BRD4 inhibitors for neuroblastoma therapy using pharmacophore screening and molecular simulations. Sci. Rep.15, 36850. 10.1038/s41598-025-20714-2
11
Alzheimer's Association, Alzheimer's Association (2025). Alzheimer's disease facts and figures. Alzheimer's Assoc. 21 (4), e70235. 10.1002/alz.70235
12
BarnesD. E.YaffeK. (2011). The projected effect of risk factor reduction on Alzheimer's disease prevalence. Lancet Neurology10, 819–828. 10.1016/S1474-4422(11)70072-2
13
BasappaV. C.Hamse KameshwarV.KumaraK.Kumar AchuthaD.KrishnappagowdaL. N.Kumar KariyappaA. (2020). Design and synthesis of coumarin-triazole hybrids: biocompatible anti-diabetic agents, in silico molecular docking and ADME screening. Heliyon. 6 (10), e05290. 10.1016/j.heliyon.2020.e05290
14
BowersK. J.ChowE.XuH.DrorR. O.EastwoodM. P.GregersenB. A.et al (2006). “Scalable algorithms for molecular dynamics simulations on commodity clusters,” in Proceedings of the 2006 ACM/IEEE Conference on Supercomputing, 84–es.
15
BozorovK.ZhaoJ.AisaH. A. (2019). 1, 2, 3-Triazole-containing hybrids as leads in medicinal chemistry: a recent overview. Bioorg. and Medicinal Chemistry27, 3511–3531. 10.1016/j.bmc.2019.07.005
16
ClareL.WoodsR. T.CookM.OrrellM.SpectorA. (2003). Cognitive rehabilitation and cognitive training for early-stage Alzheimer’s disease and vascular dementia. Cochrane Database Syst. Rev.2003 (4), CD003260. 10.1002/14651858.CD003260
17
CongdonE. E.SigurdssonE. M. (2018). Tau-targeting therapies for Alzheimer disease. Nat. Rev. Neurol.14, 399–415. 10.1038/s41582-018-0013-z
18
DainaA.MichielinO.VincentZ. (2017). SwissADME: a free web tool to evaluate pharmacokinetics, drug-likeness and medicinal chemistry friendliness of small molecules. Sci. Rep.7, 42717. 10.1038/srep42717
19
DallakyanS.OlsonA. J. (2014). “Small-molecule library screening by docking with PyRx,” in Chemical Biology: Methods and Protocols (Springer).
20
EdwardsP. D.AlbertJ. S.SylvesterM.AharonyD.AndisikD.CallaghanO.et al (2007). Application of fragment-based lead generation to the discovery of novel, cyclic amidine β-secretase inhibitors with nanomolar potency, cellular activity, and high ligand efficiency. J. Medicinal Chemistry50, 5912–5925. 10.1021/jm070829p
21
FanY.-L.XingKeLiuM. (2018). Coumarin–triazole hybrids and their biological activities. J. Heterocycl. Chem.55, 791–802. 10.1002/jhet.3112
22
Flores-MoralesV.Villasana-RuízA. P.Garza-VelozI.González-DelgadoS.Martinez-FierroM. L. (2023). Therapeutic effects of coumarins with different substitution patterns. Molecules28, 2413. 10.3390/molecules28052413
23
FrischM. J. eaTrucksG. W.SchlegelH. B.ScuseriaG. E.RobbM.CheesemanJ. R.et al (2016). Gaussian 16. Wallingford, CT: Gaussian, inc.
24
FuLiShiS.YiJ.WangN.HeY.WuZ.et al (2024). ADMETlab 3.0: an updated comprehensive online ADMET prediction platform enhanced with broader coverage, improved performance, API functionality and decision support. Nucleic Acids Research52, W422–W431. 10.1093/nar/gkae236
25
GauthierS.Rosa-NetoP.MoraisJ. A.WebsterC. (2021). World Alzheimer report 2021: journey through the diagnosis of dementia. Alzheimer's Dis. Int. Available online at: https://www.alzint.org/u/World-Alzheimer-Report-2021.pdf.
26
GhoshA. K.OsswaldH. L. (2014). BACE1 (β-secretase) inhibitors for the treatment of Alzheimer's disease. Chem. Soc. Rev.43, 6765–6813. 10.1039/c3cs60460h
27
GuerreiroR.HardyJ. (2014). Genetics of Alzheimer's disease.
28
HollingworthP.HaroldD.JonesL.OwenM. J.WilliamsJ. (2011). Alzheimer's disease genetics: current knowledge and future challenges. Int. Journal Geriatric Psychiatry26, 793–802. 10.1002/gps.2628
29
HopkinsA. L.KeserüG. M.PaulD. L.ReesD. C.ReynoldsC. H. (2014). The role of ligand efficiency metrics in drug discovery. Nat. Reviews Drug Discovery13, 105–121. 10.1038/nrd4163
30
HuY.-H.YangJ.ZhangY.LiuK.-C.LiuT.SunJ.et al (2019). Synthesis and biological evaluation of 3–(4-aminophenyl)-coumarin derivatives as potential anti-alzheimer’s disease agents. J. Enzyme Inhibition Med. Chem.34, 1083–1092. 10.1080/14756366.2019.1615484
31
HuangW.-J.ZhangX. I. A.ChenW.-W. (2016). 'Role of oxidative stress in Alzheimer's disease. Biomed. Reports4, 519–522.
32
ImaiH.HiraiT.KumazawaR.NakagawaS.YonezawaA.MatsubaraK.et al (2020). Prevalence of and risk factors for adverse events in Alzheimer’s patients receiving anti-dementia drugs in at-home care. PloS One15, e0231226. 10.1371/journal.pone.0231226
33
JawadB.PoudelL.PodgornikR.SteinmetzN. F.ChingW.-Y. (2019). Molecular mechanism and binding free energy of doxorubicin intercalation in DNA. Phys. Chem. Chem. Phys.21, 3877–3893. 10.1039/c8cp06776g
34
KahveciB.YılmazF.MenteşeE.ÜlkerS. (2017). Design, synthesis, and biological evaluation of coumarin–triazole hybrid molecules as potential antitumor and pancreatic lipase agents. Arch. Pharm.350, 1600369. 10.1002/ardp.201600369
35
KalasariyaH. S.PatelN. B.GacemA.AlsufyaniT.ReeceL. M.Kumar YadavV.et al (2022). Marine alga ulva fasciata-derived molecules for the potential treatment of SARS-CoV-2: an in silico approach. Mar. Drugs20, 586. 10.3390/md20090586
36
KhanS. A.AkhtarM. J.GogoiU.MeenakshiD. U.DasA. (2023). An overview of 1, 2, 3-triazole-containing hybrids and their potential anticholinesterase activities. Pharmaceuticals16, 179. 10.3390/ph16020179
37
KiranA.AliU.HussainA.AhmadH. M. R.AbbasF.TariqueA. M.et al (2021). Molecular designing of tetra-aryl-p-benzoquinones derivatives toward strong optical properties. Chem. Pap.75, 6661–6671. 10.1007/s11696-021-01834-1
38
KumarS.AbbasF.AliI.GuptaM. K.KumarS.GargM.et al (2023). Integrated network pharmacology and in-silico approaches to decipher the pharmacological mechanism of Selaginella tamariscina in the treatment of non-small cell lung cancer. Phytomedicine Plus3, 100419. 10.1016/j.phyplu.2023.100419
39
KumarS.ShuklaM. K.AliI.AbbasF.VermaR.ChandraG.et al (2023). In-silico and in-vitro identification of triazole based compounds as potential EGFR inhibitors targeting lung cancer. Mol. Simul.49, 1614–1631. 10.1080/08927022.2023.2256624
40
KuzmanicA.ZagrovicB. (2010). Determination of ensemble-average pairwise root mean-square deviation from experimental B-factors. Biophysical Journal98, 861–871. 10.1016/j.bpj.2009.11.011
41
LeeY. (2020). Gaussian Density of States Driven Numerical Modeling of Organic field-effect Transistors. 10.70675/0e93cd2dzdad4z422ezb14fzd4f63c735936
42
LiX.FengX.SunX.HouN.HanF.LiuY. (2022). Global, regional, and national burden of Alzheimer's disease and other dementias, 1990–2019. Front. Aging Neuroscience14, 937486. 10.3389/fnagi.2022.937486
43
LinT.-H.ChangK.-H.ChiuY.-J.WengZ.-K.SunY.-C.LinW.et al (2022). Neuroprotective action of coumarin derivatives through activation of TRKB-CREB-BDNF pathway and reduction of caspase activity in neuronal cells expressing pro-aggregated tau protein. Int. Journal Molecular Sciences23, 12734. 10.3390/ijms232112734
44
PrinceM.BryceR.AlbaneseE.WimoA.RibeiroW.FerriC. P. (2013). The global prevalence of dementia: a systematic review and metaanalysis. Alzheimer's and Dementia9, 63–75. e2. 10.1016/j.jalz.2012.11.007
45
QingX.LeeX. Y.De RaeymaekerJ.TameJ. R. H.ZhangK. Y. J.De MaeyerM.et al (2014). Pharmacophore modeling: advances, limitations, and current utility in drug discovery. J. Recept. Ligand Channel Res.2014 (7), 81–92. 10.2147/jrlcr.s46843
46
RafiiM. S.SolO.MobleyW. C.DelprettiS.SkotkoB. G.BurkeA. D.et al (2022). Safety, tolerability, and immunogenicity of the ACI-24 vaccine in adults with Down syndrome: a phase 1b randomized clinical trial. JAMA Neurol.79, 565–574. 10.1001/jamaneurol.2022.0983
47
RasheedM. A.IqbalM. N.SaddickS.AliI.KhanF. S.KanwalS.et al (2021). Identification of lead compounds against Scm (fms10) in Enterococcus faecium using computer aided drug designing. Life11, 77. 10.3390/life11020077
48
RussoT. V.MartinR. L.HayP. J. (1994). Density functional calculations on first‐row transition metals. J. Chemical Physics101, 7729–7737. 10.1063/1.468265
49
SelkoeD. J. (2001). Alzheimer's disease: genes, proteins, and therapy. Physiol. Reviews. 81 (2), 741–766. 10.1152/physrev.2001.81.2.741
50
ShaikhS. A.LabhadeS. R.ChobeS. S.GaikwadM. V.MoreR. A.PatilB. U.et al (2025). Coumarin-Triazole-Thiazole hybrids: a new avenue in Antitubercular agents. Bioorg. Chem.161, 108567. 10.1016/j.bioorg.2025.108567
51
SilvaM. V. F.LouresC. de M. G.AlvesL. C. V.Cruz De SouzaL.BorgesK. B. G. (2019). Alzheimer’s disease: risk factors and potentially protective measures. J. Biomedical Science26, 33. 10.1186/s12929-019-0524-y
52
SperlingR. A.AisenP. S.BeckettL. A.BennettD. A.CraftS.FaganA. M.et al (2011). Toward defining the preclinical stages of Alzheimer’s disease: recommendations from the national institute on aging-alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimer's and Dementia7, 280–292. 10.1016/j.jalz.2011.03.003
53
TrottO.OlsonA. J. (2010). AutoDock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. J. Computational Chemistry31, 455–461. 10.1002/jcc.21334
54
WinbladB.AndreasenN.MinthonL.FloesserA.ImbertG.DumortierT.et al (2012). Safety, tolerability, and antibody response of active Aβ immunotherapy with CAD106 in patients with Alzheimer's disease: randomised, double-blind, placebo-controlled, first-in-human study. Lancet Neurology11, 597–604. 10.1016/S1474-4422(12)70140-0
55
YuH. J.DicksonS. P.WangP.-N.ChiuM.-J.HuangC.-C.ChangC.-C.et al (2023). 'Safety, tolerability, immunogenicity, and efficacy of UB-311 in participants with mild Alzheimer's disease: a randomised, double-blind, placebo-controlled, phase 2a study. EBioMedicine94, 104665. 10.1016/j.ebiom.2023.104665
Summary
Keywords
ADMET analysis, alzheimer’s disease, amyloid beta (aβ), BACE1 (β-secretase), coumarin-1,2,3-triazole hybrids, molecular dynamics simulation
Citation
Almostafa M, Ali I, Yahya G and Siddiqi AR (2026) Unveiling the structural insights and inhibitory potential of coumarin-1,2,3-triazole hybrids against BACE1: a promising approach for Alzheimer’s disease therapy. Front. Chem. 14:1824875. doi: 10.3389/fchem.2026.1824875
Received
06 March 2026
Revised
18 May 2026
Accepted
20 May 2026
Published
07 July 2026
Volume
14 - 2026
Edited by
Robert W. Góra, Wrocław University of Science and Technology, Poland
Reviewed by
Ghada Elmasry, Cairo University, Egypt
Hari Madhav, International Centre for Genetic Engineering and Biotechnology, India
Updates
Copyright
© 2026 Almostafa, Ali, Yahya and Siddiqi.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Abdul Rauf Siddiqi, araufsiddiqi@comsats.edu.pk
† These authors have contributed equally to this work
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.