Virtual screening, ADMET prediction, molecular docking, and dynamic simulation studies of natural products as BACE1 inhibitors for the management of Alzheimer’s disease Authors: Davood Gheidari, Morteza Mehrdad, Zahra karimelahi Published: 2024 ABSTRACT AbstractAlzheimer’s disease (AD) is a degenerative neurological disorder that chronically and irreversibly affects memory, cognitive function, learning ability, and organizational skills. Numerous studies have demonstrated BACE1 as a critical therapeutic target for AD, emphasizing the need for specific inhibition of BACE1 to develop effective therapeutics. However, current BACE1 inhibitors have certain limitations. Therefore, the aim of this study was to identify potential novel candidates derived from natural products that can be utilized for the treatment of AD. To achieve this, 80,617 natural compounds from the ZINC database were subjected to virtual screening and subsequently filtered according to the rule of five (RO5), leading to the identification of 1,200 compounds. Subsequently, the 1,200 compounds underwent molecular docking studies against the BACE1 receptor, utilizing high-throughput virtual screening (HTVS), standard precision (SP), and extra precision (XP) techniques to identify high-affinity ligands. Of the 50 ligands that exhibited the highest G-Scores in HTVS, further analysis was conducted using SP docking and scoring methods. This analysis led to the identification of seven ligands with enhanced binding affinities, which were then subjected to additional screening via XP docking and scoring. Finally, the stability of the most promising ligand in relation to BACE1 was assessed through molecular dynamics (MD) simulations. The computational screening demonstrated that the docking energy values for seven ligands binding to target enzymes ranged between − 6.096 and − 7.626 kcal/mol. Among these, ligand 2 (L2) exhibited the best binding energy at -7.626 kcal/mol with BACE1. MD simulations further confirmed the stability of the BACE1-L2 complex, emphasizing the formation of a robust interaction between L2 and the target enzymes. Additionally, pharmacokinetic and drug-likeness evaluations indicated that L2 is non-carcinogenic and able to permeate the blood-brain barrier (BBB). The findings of this study will contribute to narrowing down the selection of candidates for subsequent in vitro and in vivo testing.Subject terms: Biochemistry, Biophysics, Computational biology and bioinformatics AD is a prevalent neurodegenerative disorder characterized by a progressive deterioration in memory and cognitive abilities, ultimately leading to dementia. This condition is marked by cognitive impairment, neuropsychiatric disturbances, the accumulation of amyloid plaques, neurofibrillary tangles, and reduced cholinergic transmission1–3. Despite significant advancements in the treatment of AD, the efficacy of current pharmacological interventions remains limited. Projections indicate that the burden of AD will continue to escalate in the coming years due to the lack of effective strategies to delay, prevent, or cure this condition at present4. AD has a multifaceted etiology, with extensive research demonstrating that its pathogenesis is largely driven by four primary pathophysiological mechanisms: oxidative stress5, aggregation of neurofibrillary tangles following tau protein degeneration6, and amyloid-beta (Aβ) plaque formation leading to senile plaques7. BACE1, the key enzyme driving the amyloidogenic pathway, represents a promising target for pharmacological interventions aimed at reducing cerebral Aβ levels, with the potential to treat or prevent AD8. The BACE1 enzyme is highly expressed in the brain and is categorized as an aspartyl protease. A literature review revealed that the catalytic dyad residues Asp32 and Asp228 are present in BACE1 proteins and are essential for regulating the optimal acidic pH and sequence specificity required for enzymatic activity9. BACE1 plays a pivotal role in the pathogenesis of AD due to the elevated levels and increased activity observed in the brains and bodily fluids of individuals with AD. Extensive cell biology studies have been conducted to deepen the understanding of BACE1 regulation and identify novel therapeutic targets within its pathway10–15. Therefore, BACE1 is recognized as a pivotal therapeutic target for AD, and selective BACE1 inhibition is critically sought after for drug development. The pursuit of BACE1 inhibitors has been rigorously undertaken for over one to two decades. Currently, four therapeutic agents approved by the FDA, including donepezil, rivastigmine, galantamine, and memantine, are utilized to alleviate the symptoms of AD. However, these agents have not demonstrated efficacy in preventing or reversing the progression of the disease16. Each year, numerous drugs are developed with the aim of treating AD, yet the majority fail during the preclinical phase, prior to advancing to clinical trials. As of 2019, fewer than 2% of the drugs under investigation target the inhibition of BACE1 activity. Notably, the inhibition of BACE1 by natural products has shown considerable promise in AD treatment strategies. In 2003, the first series of natural product BACE1 inhibitors, specifically catechins17, was reported. Currently, two approved anti-AD drugs, galantamine and rivastigmine, are derived from plant-based natural products. Additionally, there are reports of natural compounds that target tau aggregation, a key factor in AD pathogenesis18. Computer-aided structure-based drug design is a specialized domain within drug discovery that employs computational and theoretical methodologies to efficiently identify and optimize lead compounds. Virtual screening, which serves as the in silico equivalent of high-throughput screening for large compound libraries, is a crucial component of the drug discovery process. This approach substantially reduces both the time and cost associated with identifying new drug candidates19. The availability of comprehensive databases is critical to this strategy. One of the largest among these is ZINC, a freely accessible web-based repository of natural compounds that contains over 80,617 molecules20. This study sought to identify natural products within the ZINC database using in silico approaches, including molecular docking, ADMET predictions, and MD simulations, with the aim of discovering compounds that are potentially more effective than current drugs and may be targeted for the treatment of AD. Materials and methods This study entailed the selection and preparation of protein targets and ligands, execution of molecular docking, and performing MD simulations. The 3D structure of the target enzyme was obtained from the RCB21, and the molecular library was constructed using ZINC. The 3D and 2D interactions were demonstrated using the Discovery Studio Visualizer22, while other computational analyses were performed on the Maestro platform, utilizing tools such as LigPrep, the Protein Preparation Wizard, GLIDE, and Desmond from Schrödinger Inc. Additionally, SwissADME and ADMET Lab 2.0 were used to assess the chemical and physical properties as well as the absorption, distribution, metabolism, excretion and toxicity (ADMET) profile. Protein preparation The BACE1 crystal structure (PDB ID: 6ej3) was obtained from the RCSB database at a resolution of 1.94 Å. Protein preparation involves the refinement of hydrogen bonds, inclusion of missing hydrogen atoms, optimization, and the removal of atomic conflicts and water molecules prior to docking. Ultimately, the protein structure underwent energy minimization, and optimization using the OPLS 2005 force field. Ligand preparation A total of 1,200 natural products was obtained from the ZINC database based on the RO5 criteria. This library was prepared using Maestro’s ‘LigPrep’ module from the Schrödinger suite. Ligands were integrated into the workstation, and their energy was minimized using the OPLS 2005 force field, with each chemical’s ionization state neutralized. Additionally, tautomers and a minimum of 10 conformations per ligand were generated23. Molecular docking and validation studies Molecular docking on BACE1 was performed using Schrödinger’s GLIDE module in XP mode. Glide XP employs flexible docking to efficiently bind ligands, minimizing false positives and improving the alignment between accurate conformations and corresponding scores. Before the final docking, the selected complexes were re-docked for validation. The co-crystallized inhibitor B7T of 6ej3 was re-docked into its binding site. From these analyses, the docking score of the lowest root mean square deviation (RMSD) poses was used as the threshold for selecting docked candidates for BACE1, with acceptable RMSD values being ≤ 2 Å24,25. After validation, Schrödinger’s Glide Grid Generation was employed around the co-crystallized ligand, enabling precise positioning and interaction analysis within the specified spatial coordinates. MD simulation MD simulations were performed on the top-ranked docked candidate with the BACE1 (6ej3) enzyme using the Desmond module of the Maestro Schrödinger suite for 100 ns each. The best pose of the complex was positioned in an orthorhombic box, solvated with TIP3P water molecules, and 0.15 M NaCl was added for neutralization. The energy of the system was minimized using the OPLS 2005 force field. Simulations were executed at 300 K and 1.01325 bar of pressure. The MD trajectory was analyzed using metrics such as RMSD, root mean square fluctuation (RMSF), radius of gyration (rGyr), intramolecular hydrogen bonds (intraHB), molecular surface area (MolSA), and polar surface area (PSA). In silico physico-chemical parameters The ADMET Lab 2.0 was utilized to evaluate the pharmacokinetic and physicochemical properties of the substances. Lipinski’s RO5 was applied to evaluate the drug-likeness of the molecules; it indicates that the molecular weight (MW) must be less than 500 Da, LogP must be less than 5, and the number of hydrogen bond acceptors (HBA) and donors must be fewer than 1026,27. Results and discussion Molecular docking of ligands with BACE1 receptor Molecular docking procedures were employed to perform computational analyses of potential compounds targeting the BACE1 protein. The docking procedure was initially validated by re-docking the co-crystallized ligand B7T to verify reliability. The results demonstrated a good superposition between the re-docked and original co-crystallized ligands, with an RMSD value of 0.77 Å (Fig. 1). The superimposition of the re-docked pose (yellow) of ligand B7T with the BACE1 protein. Subsequent to the validation of docking, 1200 natural products underwent docking in the active site of BACE1 using HTVS. Fifty ligands that demonstrated the highest G-Score with HTVS were further assessed using SP docking. This was performed to verify the reliability and appropriateness of the docking postures. To minimize false-positive results, seven ligands that demonstrated significant potency were screened using XP docking. An analysis was performed on the interactions of these protein-ligand complexes, and the results are presented in Table 1. The amino acid residues of BACE1’s active site, including both bonding and non-bonding interactions with ligands L1–L7, are ASN37, SER35, LEU30, GLN73, GLY230, ASP228, GLY34, ILE226, TYR198, THR329, THR72, ARG128, VAL69, SER36, ALA39, ASP32, TRP115, ILE110, LYS107, ILE118, ILE126, PRO70, ALA127, TRP76, TYR71, LYS107, GLN12, VAL332, LYS224, and THR231. Table 1Docking scores, and interaction of each ligands L1-L7.LigandDocking scoreInteraction residueHydrogen BondVan der wallsHydrophobicUnfavorable Donor-DonorH-bondingC-bondingL1-6.096ASN37,SER35LEU30, GLN73, GLY230, ASP228, GLY34, ILE226, TYR198, THR329, THR72, ARG128, VAL69, SER36, ALA39, PHE108, ASP32ILE118,TYR71,TRP76L2-7.626ASP32PHE108, TRP115, ILE110, LYS107, ILE118, GLN73, SER35, GLY34, TYR198, SER36, ILE126, PRO70TYR71,TRP76,ARG128,VAL69L3-6.535ASN37,LYS107SER35,PHE108GLY34, ARG128, SER36, VAL69, ALA127, TRP76, ILE118, GLY230, LEU30, ILE110, GLN73, TRP115, ASP32TYR71,TYR198,ILE126L4-7.105TYR71,ARG128SER35TYR198, SER36, ASN37, TRP76, ILE118, PHE108, TRP115, ASP32, LS107VAL69,ARG128L5-6.468LYS107PHE108,LYS107GLN73, GLN12, GLY230, TRP115, LEU30, VAL69, ARG128, SER35, ASN37, ASP32ILE118,TRP76,ILE110,TYR71TYR71L6-6.935TYR198GLY34ILE110, TRP115, LEU30, LYS107, THR329, VAL332, ILE226, LYS224, ASP228, THR231, ASP32, SER35, ARG128, VAL69, TRP76PHE108,ILE118,TR71L7-6.568TRP76,ASP32VAL69, SER35, ARG128, SER36, ASN37, ALA39, ILE118, GLY230, LEU30, ILE110, GLN73, TRP115, TYR71, PHE108 Docking scores, and interaction of each ligands L1-L7. Ligand L2 demonstrates the most favorable binding energy, with an affinity of -7.626 kcal/mol. These interactions can be attributed to its favorable binding within the active site of the BACE1 enzyme, indicating that it is the most efficacious ligand-protein binding configuration. L2 forms one hydrogen bond with ASP32 and four hydrophobic interactions with TYR71, TRP76, ARG128, and VAL69 in BACE1. Furthermore, it exhibits twelve van der Waals interactions with the residues PHE108, TRP115, ILE110, LYS107, ILE118, GLN73, SER35, GLY34, TYR198, SER36, ILE126, and PRO70. Figure 2 provides the 3D and 2D details of the binding interactions between L2 and the active site of BACE1. The 3D and 2D binding’s mode of L2 into the active site of BACE1. Following L2, L4 exhibits the strongest binding affinities with the active site of the BACE1 protein. L4 forms two hydrogen bonds with the amino acid residues TYR71 and ARG28, with bond lengths of 2.88 Å and 2.70 Å, respectively. It forms a carbon-hydrogen bond with SER35, two hydrophobic interactions with the BACE1 protein, as well as nine van der Waals interactions with the residues TYR198, SER36, ASN37, TRP76, ILE118, PHE108, TRP115, ASP32, and LYS107. Figure 3 illustrates the 2D and 3D binding interactions of L4 within the active site of BACE1. The 3D and 2D binding’s mode of L4 into the active site of BACE1. MD simulations Recent advancements in drug development have leveraged MD simulations to analyze the behavior of proteins and other biomolecules with atomic-level precision and high temporal resolution. This approach has significantly influenced the design and optimization of small molecules, peptides, and proteins, as well as the understanding of the structural basis of diseases. An MD simulation was performed on the highest-scoring complex of L2 over a 100 ns period. The RMSD plots of the ligand (Lig Fit Prot) and the protein backbone (Cα) suggest that following initial fluctuations, the simulation reached a state of convergence and equilibrium. According to Fig. 4, the RMSD of L2 in the binding pocket varied from 1.70 to 6.40 Å over the 100 ns simulation. The BACE1 protein’s RMSD increased from 0.60 Å in the first frame to 2.40 Å. This value, which is below 3 Å, is considered within the acceptable range for small, globular proteins. Notable deviations were observed between 45 and 90 ns, after which the RMSD remained stable until the conclusion of the 100 ns simulation. The convergence of RMSD values suggests that L2 and BACE1 consistently maintained their interaction throughout the experiment. RMSD values for L2 protein and ligand during MD simulation. The RMSF serves as an effective metric for illustrating specific molecular interactions, providing a distance measure that quantifies the average deviation of each atom from its reference position throughout a simulation or within a specific set of structures. Figure 5 offers an in-depth analysis of the interactions between L2 and the residues within the BACE1 binding site. The vertical blue bars highlight the protein residues, including amino acids 90, 160, 270, and 300, which interacted with L2. RMSF plot for Cα of BACE1 residues with L2. Figure 6 shows the ligand interaction profiles with each of the key protein residues over the simulation time of 100 ns. The X-axis denotes the specific residues involved in the interaction at the ligand-active site, while the Y-axis indicates the proportion of simulation time during which the interaction occurs. Additionally, the hydrogen bonds are shown in green, hydrophobic interactions in purple, and water bridges in blue columns. During approximately 20% of the simulation period, VAL69 engages in hydrophobic interactions with ligands. Furthermore, for 30% of the simulation time, GLN73 formed a hydrogen bond with the ligand. During the simulation, the most significant interactions were observed with TYR71, GLN73, and TRP76. TYR71 and TRP76 demonstrated interactions with L2, including water bridging, hydrophobic interactions, and hydrogen bonding. As a result, these residues underwent several interactions over the duration of the simulation. Protein-ligand interactions during MD simulation. Figure 7 depicts six parameters of L2 during a 100-ns MD simulation. The RMSD of L2 fluctuated between 0.2 and 2.2 Å, with a steady value of around 1.6 Å. The rGyr exhibited a range of 3.80 to 4.60 Å, with an equilibrium point of approximately 4.25 Å. Intense intra-HB interactions were detected between 0 and 22 ns and approximately 87 ns. The MolSA fluctuated between 315 and 338 Ų, with an average value of around 328 Ų. The SASA showed significant fluctuations, ranging from around 70 to 240 Ų, with the PSA fluctuating from 60 to 95 Ų. Properties of the L2 ligand during MD simulation. The rotatable torsional bond of the L2 ligand is depicted by a radial plot and bar charts of the same color (Fig. 8). The ligand torsion plot summarizes the structural changes of each rotatable bond in the ligand throughout the simulation, from 0 to 100 ns. Radial plots represent the torsion during the simulation, where the simulation begins at the center of the radius, and temporal evolution is plotted radially outward. The Y-axis of the bar charts displays the rotational bond potential expressed in kcal/mol. Along with the torsional potential, the plot also illustrates the cumulative potential of the corresponding twists for each rotatable bond. Radial and bar plots were used to assess the interactions of torsional potential and structural strain within the L2 ligand structure. Details of L2 ligand rotatable bonds during MD simulation. Drug-likeness prediction Drug-likeness pertains to the extent of similarity between specific compounds and established pharmaceuticals. This concept is based on a precise balance of molecular and structural properties. Assessing drug-likeness requires a comprehensive evaluation of various molecular properties, including hydrophobicity, electronic distribution, hydrogen bonding potential, MW, and the presence of pharmacophoric groups, bioavailability, reactivity, toxicity, and metabolic stability28. Lipinski’s rule is extensively applied to evaluate the solubility and permeability of compounds, facilitating the prediction of their potential as drug candidates. According to this rule, compounds that violate Lipinski’s (RO5) are more likely to exhibit suboptimal absorption or penetration. Ligands L1–L7 were thoroughly evaluated using SwissADME26. The analysis demonstrated that none of these ligands violate Lipinski’s rule, as their parameters fall within the permissible range, suggesting adequate absorption properties. Additionally, it is noteworthy that these ligands occupy a favorable position within the physicochemical space, which supports the potential for their classification as lead compounds. Pharmacokinetic analysis indicated that all ligands possess favorable absorption characteristics in the gastrointestinal (GI) tract upon oral administration. Most ligands, with the exception of L2, L3, and L6, are effluxed by plasma-glycoprotein (P-gp), which may be attributed to their high molar refractivity (MR). The application of pan-assay interference substances (PAINS) structural alerts in pharmaceutical chemistry has been employed to identify regions within a compound’s structure that are prone to instability, reactivity, and toxicity29,30. Except for ligand L1, which has a catechol_A alarm, all other ligands show zero alarms in PAINS descriptors. The synthetic accessibility (SA) score is a numerical metric used to assess the complexity involved in synthesizing drug-like molecules. Table 2 shows that L1–L7 exhibit favorable SA scores, suggesting that they can be easily synthesized. Table 2Physicochemical, pharmacokinetics, and medicinal chemistry properties of the L1-L7.ModeMW (g/mol)HBAHBDTPSAConsensus Log Po/w *MRGI AbsorptionP-gp SubstrateLipinskiBioavailability ScorePAINS (alert)SA score(Å2) L 1 314.384369.923.3190.54HighYesYes0.5513.64 L 2 306.363146.533.9792.39HighNoYes0.5503.51 L 3 326.345065.743.3090.46HighNoYes0.5503.89 L 4 299.295345.321.9482.58HighYesYes0.5503.72 L 5 283.332265.201.4386.61HighYesYes0.5502.85 L 6 354.356063.223.2491.23HighNoYes0.5503.53 L 7 299.295345.321.9382.58HighYesYes0.5503.72 Physicochemical, pharmacokinetics, and medicinal chemistry properties of the L1-L7. ADMET properties Comprehensive knowledge of pharmacology and toxicology is essential for advancing therapeutic drug development. Obtaining this information not only reduces the time required for drug development but also improves the success rate. ADMET is frequently used to evaluate the properties of a substance. The ADMET properties for L1–L7 were evaluated using ADMET Lab 2.027. A widely accepted approach to evaluating the absorption of pharmaceutical drugs in the human GI system is the use of cancer coli-2 (Caco-2) cells, derived from human colon epithelial cells. Additionally, since madin-darby canine kidney (MDCK) cells proliferate faster than CaCo-2 cells, they are particularly useful for determining the rapid permeability of drug compounds31. Excluding L4 and L7, the CaCo-2 cell permeability results were found to be within an acceptable range, indicating that these ligands have adequate membrane permeability. All ligands exhibited favorable permeability in MDCK cells, indicating a higher likelihood of being eliminated by renal cells. Regarding P-gp substrates and inhibitors, L2, L3, and L6 are considered P-gp substrates, whereas L1, L4, L5, and L7 are considered P-gp inhibitors. The estimated values for human intestinal absorption (HIA) suggest that all ligands are highly likely to be readily absorbed through the intestinal membrane. Drugs with a high plasma protein binding (PPB) value, over 90%, are often associated with a narrow therapeutic index. Conversely, drugs with a low PPB value are generally considered comparatively safer. The study demonstrated that L4, L5, and L7 exhibited favorable PPB values and are considered safe. Compounds with a favorable BBB value have enhanced lipophilicity features, allowing them to be readily absorbed from plasma membranes. The lipophilicity profiles of L1, L2, and L3 were shown to be enhanced. Regarding carcinogenicity, L2 and L6 were found to be non-carcinogenic, while other ligands exhibited potential carcinogenicity. The AMES toxicity profile predicts that L4 is likely to be non-toxic (Table 3). Table 3ADMET profile of the L1-L7.Absorption and DistributionModeCaco-2 permeabilityPgp-InhibitorPgp-substrateHIAPPBVolume of distribution (VD)BBB L 1 -4.824---0.010+++0.99---0.00597.704%1.290---0.019 L 2 -4.798+++0.988---0.002---0.005101.745%0.598---0.042 L 3 -4.710+ 0.538---0.001---0.00596.395%0.535---0.017 L 4 -5.442--0.120+++0.95--0.20650.434%1.383+ 0.688 L 5 -4.811---0.005+++0.986---0.00576.823%0.706-0.411 L 6 -5.040+++0.976---0.000---0.00198.476%0.947-0.335 L 7 -5.400--0.144+++0.989---0.02150.327%1.140-0.337MetabolismEliminationModeP450 CYP1A2 inhibitorP450 CYP1A2 SubstrateP450 CYP3A4 inhibitorP450 CYP3A4 SubstrateP450 CYP2C9 inhibitorP450 CYP2C9 SubstrateP450 CYP2C19 inhibitorP450 CYP2C19 SubstrateP450 CYP2D6 inhibitorP450 CYP2D6 SubstrateT1/2CL L 1 + 0.631+++0.955+ 0.523+ 0.647++0.802+++0.931-0.465-0.352-0.455++0.8760.70419.012 L 2 +++0.979--0.144++0.801--0.205+++0.907+++0.947+++0.944--0.104+++0.908+ 0.6920.2052.665 L 3 ++0.850++0.831++0.764--0.232++0.883++0.843+++0.924---0.090+ 0.617++0.8530.56010.812 L 4 + 0.549--0.102-0.385--0.129--0.246---0.034+ 0.655++0.701--0.227-0.3980.5944.498 L 5 --0.101--0.222+ 0.593--0.142--0.196++0.862+ 0.608---0.092---0.027+ 0.6130.7834.792 L 6 +++0.990--0.126+++0.983-0.424+++0.925++0.855+++0.983--0.173+++0.990+++0.9180.23617.762 L 7 + 0.616---0.100--0.293--0.129--0.233---0.033+ 0.642+ 0.678--0.280--0.2550.4954.897ToxicityModeAMES toxicityCarcinogenicityEye corrosionEye irritationhERGH-HTLD50Respiratory toxicity L 1 ---0.085--0.127---0.005+++0.914---0.02--0.1580++0.868 L 2 --0.279++0.823---0.004++0.801--0.162+ 0.6150++0.899 L 3 -0.334+ 0.66---0.004-0.308--0.224+++0.9060---0.045 L 4 ++0.822--0.127---0.003---0.015---0.002--0.2190--0.211 L 5 ---0.034---0.097---0.003---0.038---0.030+++0.9100--0.136 L 6 ---0.070++0.871---0.004+ 0.510-0.359--0.2840--0.245 L 7 -0.498---0.037---0.003---0.016---0.002-0.430--0.175CY: Cytochrome; hERG : Human ether-a-go-go-related gene; H-HT: Human hepatotoxicity; LD50: Median lethal dose. ADMET profile of the L1-L7. CY: Cytochrome; hERG : Human ether-a-go-go-related gene; H-HT: Human hepatotoxicity; LD50: Median lethal dose. HBD: Number of hydrogen-bond donors; TPSA: Topological polar surface area. Conclusion The study focuses on targeting the BACE1 enzyme, which is the primary enzyme responsible for the amyloidogenic cleavage of β-amyloid precursor protein (APP), leading to Aβ deposition and plaque formation in the human brain, a hallmark of Alzheimer’s disease. Through virtual screening and molecular docking, a total of seven lead natural products were identified. The docking results revealed that these ligands exhibit high binding affinities for the BACE1 enzyme, with L2 showing the strongest binding affinity, evidenced by a docking score of -7.626. L2 forms a hydrogen bond with ASP32, engages in 12 van der Waals interactions, and establishes 4 hydrophobic interactions with the BACE1 enzyme. The RMSD analysis demonstrated that L2 consistently maintained stable interactions with key residues of BACE1, indicating that the L2-BACE1 complex remained thermodynamically stable throughout the 100 ns trajectory. Furthermore, the ADME properties analysis showed that the newly screened inhibitors possess good pharmacological properties and do not violate Lipinski’s rule. The physiological properties indicated that all ligands exhibit high absorption through the GI tract, with L1, L2, and L3 also showing the ability to permeate the BBB. To confirm the anti-alzheimer’s potential of these discovered compounds and to overcome the limitations of in silico analysis, further laboratory and clinical studies are required. These investigations will provide a more comprehensive understanding of the safety and efficacy of the identified compounds in the treatment of Alzheimer’s disease. The study offers valuable insights into the binding mechanisms of the BACE1 enzyme in Alzheimer’s disease, suggesting that future research exploring the efficacy of these lead molecules in animal models of Alzheimer’s could pave the way for clinical trials and the development of novel therapeutic strategies for this debilitating condition. Author contributions All authors of this research paper have directly contributed to the design, conduct or analysis of this study; and all authors of this paper have read and approved the final version submitted. Data availability The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Davood Gheidari, Email: davoodgheidari@gmail.com. Morteza Mehrdad, Email: m-mehrdad@guilan.ac.ir. References Associated Data Data Availability Statement The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.