Multitargeted comparative evaluation suggests 2-Aoeobenoxmide shows favourable in silico binding compared to Tucatinib against ERα, HER2, AKT1, EGFR, and PIK3CA in breast cancer Authors: Mohammed H. Al-Qahtani, Mohammad Alam Jafri, Mourad Assidi, Abdelbaset Buhmeida, Peter Natesan Pushparaj, Maha Khalid Abdullah, Nofe Ateq Alganmi Published: 2026 ABSTRACT Breast cancer is a leading cause of cancer-related morbidity and mortality globally, with the WHO reporting approximately 2.3 million new cases and 685,000 deaths annually. Drug resistance in breast cancer complicates treatment, with mutations in critical proteins contributing to therapy failure. Key oncogenic proteins involved in breast cancer progression—namely ERα (a ligand-activated nuclear receptor; PDB: 1A52) and the kinase domains of HER2 (PDB ID: 3PP0), AKT1 (PDB ID: 4EJN), EGFR (PDB ID: 4I23) and PIK3CA (PDB ID: 7R9V)—are pivotal in tumour progression and resistance mechanisms. Targeting these proteins using multitargeted therapeutic strategies may overcome resistance by disrupting key signalling pathways involved in cell proliferation, survival, and metastasis. Such combinatorial approaches promise to improve treatment efficacy and patient outcomes in cases of resistant breast cancer. In this study, we performed multitarget docking on prepared and validated protein structures against the ZINC natural compound library using HTVS, SP, and XP, with pose validation using MM-GBSA. We identified 2-Aoeobenoxmide (2-[1-(2-amino-2-oxo-ethoxy)-6-oxo-benzo[c]chromen-3-yl]oxyacetamide, ZINC134008) with docking and MM-GBSA scores ranging from –8.162 to –10.327 kcal/mol and from –47.18 to –57.62 kcal/mol, respectively, and compared the results with the FDA-approved drug Tucatinib, which exhibited lower binding affinity scores. We further evaluated pharmacokinetic properties using QikProp and electronic properties using DFT (Jaguar) and compared the descriptors of 2-Aoeobenoxmide with those of Tucatinib and with accepted reference ranges. We also performed the WaterMap for 5 nanoseconds (ns), computed various energies, interactions and hydration sites, and the comparison suggests that 2-Aoeobenoxmide shows more favourable hydration-site displacement and binding interactions than Tucatinib. Additionally, a 100 ns MD Simulation has resulted in far less deviation, fluctuations, and intermolecular interactions than Tucatinib, suggesting stable protein–ligand interactions, while the binding free energy and total complex energy computed across 0–1000 frames of the MD trajectories indicate that 2-Aoeobenoxmide is a promising in silico candidate. Importantly, because the entire study is computational, the findings should be interpreted as in silico hypotheses, and experimental validation through in vitro and in vivo assays is warranted before any clinical translation is considered.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1038/s41598-026-57604-0. Introduction Breast cancer is a malignant disease characterised by the uncontrolled proliferation of cells originating primarily from the epithelial lining of the mammary gland’s ducts or lobules. It represents one of the most prevalent cancers globally, particularly among women, with an estimated 2.3 million new cases diagnosed annually and causing approximately 685,000 deaths worldwide, according to recent GLOBOCAN data1,2. The disease manifests in several subtypes, primarily classified based on the expression of hormone receptors such as estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2), giving rise to distinct molecular classifications, including ER-positive, HER2-positive, triple-positive, and triple-negative breast cancers (TNBC)3,4. The aetiology of breast cancer is multifactorial, involving genetic predispositions (e.g., BRCA1/2 mutations), hormonal imbalances, prolonged estrogen exposure, lifestyle factors like obesity and alcohol consumption, radiation exposure, and environmental carcinogens. Therapeutic interventions for breast cancer are determined based on tumour subtype, stage, and patient health status, encompassing a combination of surgical resection, radiation therapy, cytotoxic chemotherapy, hormone therapy, targeted molecular therapy, and, more recently, immunotherapy5,6. Among targeted agents, Tucatinib—a potent, selective oral tyrosine kinase inhibitor of HER2—has demonstrated clinical efficacy, particularly in HER2-positive metastatic breast cancer, even in cases with central nervous system involvement7,8. It is often combined with trastuzumab and capecitabine to enhance therapeutic outcomes. Other notable agents include Palbociclib, a CDK4/6 inhibitor used primarily in ER-positive, HER2-negative advanced breast cancer, and Everolimus, an mTOR inhibitor employed in hormone receptor-positive settings to overcome endocrine resistance. Despite these advances, the emergence of drug resistance remains a formidable challenge in breast cancer management9. Resistance mechanisms include mutations in target receptors (e.g., HER2 mutations that impair binding of inhibitors), activation of compensatory signalling pathways such as PI3K/AKT/mTOR or MAPK, overexpression of efflux transporters like P-glycoprotein, epithelial-to-mesenchymal transition (EMT), tumour microenvironment remodelling, and immune evasion strategies1,9,10. These resistance phenotypes underscore the need for combinatorial regimens, biomarker-driven therapies, and continuous molecular monitoring to tailor treatment strategies and improve clinical outcomes in breast cancer care. Drug resistance in breast cancer represents one of the most critical barriers to successful long-term treatment outcomes, particularly in advanced or metastatic cases. It refers to the ability of cancer cells to withstand the cytotoxic or cytostatic effects of chemotherapeutic agents, targeted therapies, or endocrine treatments, leading to therapeutic failure and disease progression. Resistance can be intrinsic—present before therapy—or acquired after initial treatment efficacy11. Mechanistically, drug resistance in breast cancer arises from a complex interplay of molecular, genetic, and epigenetic alterations that enable tumour cells to evade apoptosis, rewire survival signalling, enhance DNA repair capabilities, and adapt their metabolism. Single-target therapies often become ineffective over time due to compensatory activation of parallel or downstream signalling pathways, the emergence of drug-tolerant persister cells, and modifications in drug transport, such as increased expression of ATP-binding cassette (ABC) transporters that actively efflux therapeutic agents12. Additionally, the heterogeneity of breast tumours—both inter and intra-tumoural—drives the coexistence of diverse cellular clones with variable sensitivity to therapy, fostering evolutionary dynamics under drug pressure and further promoting resistance13. In this context, multitargeted drug design offers a robust and rational strategy to overcome drug resistance in breast cancer by modulating multiple molecular targets and signalling nodes critical for tumour survival and progression. Rather than focusing on a single oncogenic driver, multitargeted agents or polypharmacological compounds are designed to interfere with several interlinked pathways, thereby reducing the probability of compensatory pathway activation and resistance emergence14–16. Key oncogenic nodes in breast cancer biology include the estrogen receptor alpha (ERα, a ligand-activated nuclear receptor; PDB ID: 1A52) and the kinase proteins HER2 (human epidermal growth factor receptor 2; PDB ID: 3PP0), AKT1 (RAC-alpha serine/threonine-protein kinase; PDB ID: 4EJN), EGFR (epidermal growth factor receptor; PDB ID: 4I23) and PIK3CA (phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha isoform; PDB ID: 7R9V), and they are frequently involved in resistance to targeted therapies17–20. HER2 is a receptor tyrosine kinase belonging to the EGFR family and is overexpressed in approximately 20–25% of breast cancers, correlating with aggressive tumour behaviour and poor prognosis. HER2 activation initiates downstream signalling through the PI3K/AKT/mTOR and RAS/RAF/MEK/ERK cascades, promoting cellular proliferation, survival, and angiogenesis17. The protein structure of HER2, as elucidated in PDB ID: 3PP0, provides critical insights into its active conformation and druggable sites, which have been targeted by monoclonal antibodies (e.g., trastuzumab), tyrosine kinase inhibitors (e.g., lapatinib), and antibody-drug conjugates. AKT1, represented by PDB ID 4EJN, is a serine/threonine kinase that acts downstream of PI3K and plays a central role in regulating metabolism, cell cycle progression, apoptosis suppression, and DNA repair. Its aberrant activation, often driven by upstream mutations in PIK3CA or loss of PTEN function, contributes to resistance to endocrine therapies and HER2-directed drugs20. PIK3CA, encoded by the gene frequently mutated in breast cancer, especially in ER-positive subtypes, leads to constitutive activation of the PI3K/AKT/mTOR axis. The crystal structure (PDB ID: 7R9V) reveals key catalytic and regulatory domains, which are essential for designing selective or dual inhibitors. EGFR, which shares structural homology with HER2, is overexpressed or mutated in triple-negative breast cancers (TNBC) and contributes to uncontrolled proliferation and resistance to standard chemotherapeutics. Targeting EGFR has shown limited success in breast cancer due to the rapid development of resistance, often involving activation of AKT1 or MET amplification, underscoring the importance of dual or pan-receptor blockade18,19. Integrating the inhibition of HER2, EGFR, AKT1, and PIK3CA into a unified therapeutic framework, multitargeted drug design offers a strategic avenue to achieve more durable responses and delay or prevent resistance. This can be accomplished by rationalising small molecules that bind multiple kinases within the same pathway, across parallel pathways, or by combining structurally distinct agents that co-target these proteins in synergistic ways. For instance, dual HER2/EGFR or PI3K/AKT dual inhibitors have demonstrated preclinical and clinical potential by mitigating feedback loops and compensatory upregulation of survival signals. Furthermore, structure-based drug design using high-resolution crystallographic data from PDB structures enables the identification of conserved and allosteric binding pockets, allowing the design of multitarget ligands with optimised affinity and selectivity. Multitargeted approaches may also modulate the tumour microenvironment, reduce immune evasion, and improve pharmacokinetic profiles compared with combination therapies10,21,22. Collectively, multitargeted drug design represents a paradigm shift in breast cancer therapy by addressing the multifactorial nature of resistance and increasing the likelihood of achieving sustained therapeutic efficacy in heterogeneous, evolving tumour landscapes. Extensive studies in multitargeted drug design have adopted a comprehensive, integrative in silico pipeline to accelerate the identification and optimisation of potential therapeutics that can overcome drug resistance in breast cancer by engaging multiple oncogenic targets. The drug design process typically begins with rigorous data collection and curation, involving the retrieval of high-resolution crystal structures of relevant protein targets such as HER2, AKT1, EGFR, and PIK3CA from the Protein Data Bank (PDB), along with their co-crystallised ligands and relevant mutational data18,20. These structures undergo thorough preparation using protein preparation tools that assign correct bond orders, add missing side chains and loops, optimise hydrogen bonding networks, and remove crystallographic artefacts or water molecules not involved in binding. Concurrently, chemical libraries comprising FDA-approved drugs, natural products, or de novo-designed ligands are curated and optimised using LigPrep or similar tools to generate tautomers, stereoisomers, and protonation states under physiological conditions10,23. Following data preparation, molecular docking is used to evaluate ligand binding affinity and orientation within the active sites of each target protein. Rigid and flexible docking protocols—often utilising software such as Glide, AutoDock Vina, or GOLD—are performed to account for induced fit and conformational plasticity. Scoring functions rank ligands based on their predicted binding energies, and top-ranked hits are further subjected to interaction fingerprint analysis to elucidate key pharmacophoric features and noncovalent interactions such as hydrogen bonding, π–π stacking, salt bridges, and hydrophobic contacts that stabilise the protein-ligand complex. These fingerprints are compared across multiple targets to identify ligands with conserved interactions, indicative of true multitargeting potential. Subsequently, pharmacokinetic and drug-likeness assessments are conducted using ADMET prediction tools such as QikProp, SwissADME, and pkCSM to ensure that the selected compounds possess favourable absorption, distribution, metabolism, excretion, and toxicity profiles21,24. Parameters such as Lipinski’s Rule of Five, oral bioavailability, blood-brain barrier permeability, cytochrome P450 inhibition, and hepatotoxicity are evaluated to eliminate candidates with poor clinical translatability22,25–28. To explore the electronic properties of promising compounds and their propensity for charge transfer and reactivity, Density Functional Theory (DFT) calculations are performed using B3LYP or M06-2X functionals. These quantum chemical analyses yield frontier molecular orbitals (HOMO-LUMO), electrostatic potential maps, ionisation potential, and global reactivity descriptors, aiding in understanding how molecular electronics influence protein interaction and stability. Another crucial thermodynamic aspect is explored through WaterMap or hydration site analysis, which identifies high-energy water molecules within the binding pocket that contribute favourably to binding enthalpy when displaced by a ligand. This guides the optimisation of ligand structures for enhanced desolvation energetics and better accommodation within hydrophobic or solvent-exposed regions. Molecular Dynamics (MD) simulations are conducted, typically for 100–200 ns, using software such as GROMACS, AMBER, or Desmond to validate the dynamic stability of the docked complexes under physiological conditions. These simulations offer insights into the time-resolved behaviour of protein-ligand interactions, conformational fluctuations, and the persistence of critical contacts. Key analyses derived from MD include deviations, fluctuations and interactions. Molecular Mechanics/Generalised Born Surface Area (MM-GBSA) calculations are performed on MD-generated trajectories to quantify the thermodynamic favourability of binding. This method estimates the binding free energy by incorporating van der Waals forces, electrostatics, solvation energies, and entropy contributions, offering a more refined metric than docking scores alone29,30. Binding energies are compared across multiple targets to assess selectivity and the balance of affinity, which are essential for multitarget drug design. Additionally, per-residue decomposition analysis highlights the specific amino acid residues that contribute to binding, thereby identifying potential hotspots for chemical optimisation. Such integrative methodologies accelerate drug discovery and provide molecular-level mechanistic insights that inform subsequent in vitro and in vivo validations, ultimately advancing precision oncology interventions29,30. The present study is based on a structure-based computational (in silico) screening pipeline and aims to identify potent multitargeted therapeutic candidates for breast cancer by screening natural compounds from the ZINC database against key oncogenic proteins—ERα, HER2, AKT1, EGFR, and PIK3CA—that drive tumour progression and resistance. Using structure-based virtual screening, molecular docking was conducted to identify compounds with strong binding affinities across multiple targets, benchmarked against the FDA-approved HER2 inhibitor Tucatinib. Interaction fingerprint analysis was used to compare residue-wise interaction patterns between the lead compounds and Tucatinib across all proteins. Pharmacokinetic profiling and DFT calculations were performed to evaluate drug-likeness, absorption, and electronic reactivity. WaterMap analysis assessed hydration-site energetics and the enthalpic contributions of water displacement upon ligand binding. MD simulations were carried out to validate complex stability under physiological conditions, followed by MM-GBSA binding free energy calculations to determine thermodynamic favourability and binding strength. A detailed description of each computational step of this in silico pipeline is now provided in the Methods section. Methods To clarify the comprehensive nature of our study, which includes multiple analyses and comparisons with a standard drug compound, we have illustrated the overall workflow in Fig. 1. This visual representation simplifies the complex sequence of steps undertaken. A detailed description of each methodological step is provided in the sections below. Each computational step has a defined role in the pipeline: Glide HTVS/SP/XP docking and MM/GBSA rescoring are used to rank candidates against the active sites; molecular interaction fingerprinting compares binding modes against the Tucatinib control; QikProp provides ADME/Tox profiling for drug-likeness; DFT/TDDFT (Jaguar) reports electronic descriptors relevant to chemical stability and binding-pocket electrostatic complementarity; WaterMap quantifies hydration-site thermodynamics that contribute to binding enthalpy and entropy; and 100 ns Desmond MD followed by MM/GBSA validates the stability and binding free energy of the docked complexes under physiological conditions. The output of each step informs the next, and only candidates that satisfy the criteria at every stage are advanced.Fig. 1Schematic workflow of the present in silico investigation. The pipeline is grounded in the working hypothesis that a multitarget hit for ERα, HER2, AKT1, EGFR and PIK3CA may provide a more durable in silico binding profile than HER2-selective inhibition alone from the prepared ZINC natural compound library using the Schrödinger Virtual Screening Workflow (HTVS → SP → XP), followed by MM-GBSA pose rescoring; this hierarchical filter identified 2-Aoeobenoxmide (ZINC134008) as the lead candidate, which was subsequently compared with the FDA-approved drug Tucatinib through molecular interaction fingerprinting, ADME/Tox profiling (QikProp), DFT/TDDFT analysis (Jaguar), WaterMap hydration-site analysis and 100 ns MD simulations with MM/GBSA energy decomposition. Schematic workflow of the present in silico investigation. The pipeline is grounded in the working hypothesis that a multitarget hit for ERα, HER2, AKT1, EGFR and PIK3CA may provide a more durable in silico binding profile than HER2-selective inhibition alone from the prepared ZINC natural compound library using the Schrödinger Virtual Screening Workflow (HTVS → SP → XP), followed by MM-GBSA pose rescoring; this hierarchical filter identified 2-Aoeobenoxmide (ZINC134008) as the lead candidate, which was subsequently compared with the FDA-approved drug Tucatinib through molecular interaction fingerprinting, ADME/Tox profiling (QikProp), DFT/TDDFT analysis (Jaguar), WaterMap hydration-site analysis and 100 ns MD simulations with MM/GBSA energy decomposition. Ligand library protein structure collection and preparation The collection and preparation of ligand libraries are essential steps in drug discovery, as they involve selecting and optimising chemical compounds that will be screened for their potential to interact with a given target protein29,31,32. We downloaded the complete natural compound library from the Zinc database (https://zinc.docking.org/) to prepare it33. It includes removing duplicate compounds, structural optimisation, energy minimisation, and format conversion to ensure consistency across the dataset. Protonation states and tautomeric forms are adjusted to reflect the physiologically relevant conditions, and rotatable bonds are defined to account for conformational flexibility. Lipinski’s rules of five, which predict the drug-like properties of molecules, are used to screen the library, ensuring that only viable candidates proceed to docking studies27,28. We used the LigPrep tool in Schrödinger’s Maestro (v.2024-2) to prepare the ZINC natural compound library28,34. We retained compounds containing up to 500 atoms (the default LigPrep upper bound for typical drug-like organic molecules, applied in addition to Lipinski’s Rule of Five), employed the OPLS4 force field for energy calculations, and generated different protonation states at a target pH of 7 ± 2 using Epik28,35,36. The resulting refined library contained 187,116 compounds subsequently used for molecular docking studies. Protein structural data acquisition involves obtaining 3D models of target proteins from publicly available databases, such as the Protein Data Bank (PDB), or by homology modelling when no experimental structure is available. Protein preparation is crucial for ensuring that structural data are suitable for downstream docking studies37–39. This process typically includes removing water molecules, adding missing residues or loops, optimising protonation states, and energy minimisation to enhance structural stability. Cofactors, metal ions, and non-essential ligands are removed, and the refined protein structure is validated to ensure stability and reliability for subsequent computational simulations. For this study, we focused on key proteins involved in breast cancer, including Estrogen Receptor Alpha (ERα) (PDB: 1A52), HER2 (Human Epidermal Growth Factor Receptor 2) (PDB: 3PP0), AKT1 (Protein Kinase B, Isoform 1) (PDB: 4EJN), EGFR Kinase Domain (Epidermal Growth Factor Receptor) (PDB: 4I23), and PIK3CA (Phosphoinositide-3-Kinase Catalytic Subunit Alpha) (PDB: 7R9V)4,17,19,20. Estrogen Receptor Alpha is a nuclear receptor activated by the hormone estrogen and plays a crucial role in regulating genes that control breast cell growth and differentiation4. HER2 is a receptor in the epidermal growth factor receptor family that regulates cell growth and differentiation, with its overexpression linked to aggressive breast cancer17. AKT1 is a serine/threonine kinase that regulates several cellular processes like metabolism, proliferation, survival, and angiogenesis, functioning as a major component of the PI3K/AKT signalling pathway18. EGFR, a receptor tyrosine kinase, is essential for cell signalling that governs division, survival, and differentiation; mutations or overexpression of EGFR are frequently associated with cancer. Finally, PIK3CA encodes the p110α catalytic subunit of phosphoinositide 3-kinase, and mutations in this gene are commonly observed in various cancers, including breast cancer, contributing to aberrant cell survival and proliferation19. These structures were obtained from the RCSB PDB (http://rcsb.org/) and prepared using the Protein Preparation Workflow in Schrödinger’s Maestro (v.2024-2)28,40–43. The original structures included Chain A and B of the protein, along with ligands, solvents, and metals/ions in 1A52 and 3PPO, and only Chain A of the protein, ligands, and solvents in 4EJN, 4I23, and 7R9V. In this process, we selected entries from the project table, capped the termini, filled missing side chains, assigned bond orders to the CCD, replaced hydrogens, created disulfide bonds, set zero bond orders for metals, added oxygen atoms to metal termini, filled missing loops, and generated heteroatom statistics using Epik at a pH of 7.4 ± 228,36,44. For optimisation, we applied crystal symmetry, sampled water orientations, minimised the hydrogen atoms of altered species, and used PROPKA for pH calculations28,45. During the minimisation phase, we minimised all atoms with a maximum deviation of 0.30 Å, utilised the OPLS4 force field, and removed water molecules situated more than 5 Å from the ligand28,35. After preparation, we retained only Chain A of the proteins and ligands across all five structures—1A52, 3PPO, 4EJN, 4I23, and 7R9V—to generate the grids for docking studies. Receptor grid generation and multitargeted molecular docking studies and control comparison The generation of a receptor grid is a critical step in molecular docking, as it defines the search space within the target protein where potential ligands can bind. By identifying the protein’s active site or functional pockets, the grid-generation process ensures the docking simulation focuses on the relevant region, improving both efficiency and accuracy. Receptor grid generation involves selecting key active-site residues based on known ligand interactions or functional relevance. The gridbox dimensions are optimised to encompass the binding site, with the flexibility to accommodate various ligand conformations30. Electrostatic and van der Waals parameters are carefully assigned to ensure the realistic ligand placement and interaction analysis represent potential biological interactions. In this study, receptor grids were generated using Schrödinger’s Maestro (v.2024-2) Receptor Grid Generation tool28,46. Native ligands in the protein-ligand complexes were used as a reference to define the docking space, and parameters such as scaling factor and partial charge cutoff were carefully selected. Multitargeted molecular docking was performed to assess the potential of the natural drug candidates against multiple breast cancer-related proteins 32. Multitargeted molecular docking is a powerful approach for identifying potential therapeutic candidates that can simultaneously interact with multiple biological targets, particularly in complex diseases like breast cancer, where multiple signalling pathways are often dysregulated. This strategy increases the likelihood of identifying compounds with broader therapeutic efficacy and potentially improved resistance profiles47,48. In this study, multitargeted docking was employed to screen a curated ZINC natural compound library against five key breast cancer-related proteins: Estrogen Receptor Alpha (1A52), HER2 (3PPO), AKT1 (4EJN), EGFR (4I23), and PIK3CA (7R9V)4,17–20. The goal was to identify natural compounds with favourable binding profiles across multiple targets, suggesting their potential as multitargeted drug candidates. The prepared ZINC natural compound library was screened against each protein target using Schrödinger’s Maestro (v.2024-2) Virtual Screening Workflow (VSW), which employs a hierarchical docking protocol comprising High-Throughput Virtual Screening (HTVS), Standard Precision (SP), and Extra Precision (XP) modes28,49. This tiered approach efficiently narrows large libraries while maintaining accuracy in pose prediction and binding affinity estimation. In our protocol, the prepared library of 187,116 ZINC natural compounds was passed through HTVS (top 10% retained), SP (top 10% retained) and XP (top compounds retained per target). The top XP poses (up to 100 per target) were rescored uniformly with MM-GBSA, and the best-scoring poses per ligand were carried forward for downstream analysis. High-scoring ligand-protein complexes were also subjected to Extra Precision (XP) docking to improve pose prediction accuracy. MM-GBSA (Molecular Mechanics/Generalised Born Surface Area) calculations were performed to quantify further the binding affinity and stability of the docked complexes. The docking results were evaluated based on binding scores, molecular interactions, and docking poses (Supplementary Sheets). For benchmarking, the compounds’ performance was compared with Tucatinib, a clinically approved breast cancer inhibitor, to assess relative binding affinity8. These analyses provided insights into the thermodynamic favourability of ligand binding, allowing a comprehensive comparison between the natural compounds and the reference drug, Tucatinib7. This combined docking and post-docking analysis pipeline helped identify promising multitargeted natural compounds for further validation. To validate the docking protocol, the native co-crystallised ligands of all five PDB entries (1A52, 3PP0, 4EJN, 4I23 and 7R9V) were re-docked into their respective grids and the predicted poses were compared with the experimental binding modes; the re-docked RMSD values were below 2 Å in every case, supporting the accuracy of the grid definitions and the Glide scoring functions used for the virtual screen (full re-docking results are provided in the Supplementary Sheets). The MM/GBSA rescoring step was applied uniformly to the top XP poses of every target so that the comparative ranking of 2-Aoeobenoxmide and Tucatinib across the five targets is computed under identical settings. Molecular interaction fingerprints Molecular interaction fingerprints (MIFs) accurately represent the key interactions between ligands and their target proteins. These fingerprints are valuable for comparing binding patterns across various compounds and protein targets, providing insights into the specific interactions contributing to binding affinity and selectivity. MIFs capture a range of interactions, including hydrogen bonds, hydrophobic contacts, π–π stacking, and salt bridges22,29. These interactions are crucial for understanding ligand-binding mechanisms and optimising drug candidates. In virtual screening and structure-based drug design, MIFs help identify key interaction motifs, which can be used to refine lead compounds and improve their binding profiles. This study generated MIFs using Schrödinger’s Maestro (v.2024-2) Interaction Fingerprints panel28. The protein-ligand complexes were aligned with the estrogen receptor alpha (PDB: 1A52) as a reference, and interaction counts were visualised in a matrix4. Residue interaction counts and other ligand–protein interactions were plotted to identify key residues involved in ligand binding. These interaction fingerprints were used to compare the binding profiles of the natural drug candidates with those of the control drug, Tucatinib, aiding in identifying the most promising candidates. Density functional theory and pharmacokinetics Density Functional Theory (DFT) is a quantum-mechanical method used to study the electronic structure of molecules, providing valuable insights into molecular stability, reactivity, and electronic interactions. DFT calculations are widely applied in drug design to optimise molecular geometries, predict the stability of drug candidates, and analyse their electronic properties. DFT calculations provide detailed information about key properties such as electron density, energy levels, and molecular orbitals50. These insights are critical for optimising the electronic structure of drug candidates and ensuring that they possess the desired properties for efficient binding to target proteins. In this study, DFT calculations were performed using Schrödinger’s Maestro (v.2024-2) Jaguar panel28,51,52. The ligand was selected within the computational environment, and the B3LYP-D3 functional (Becke, Three-Parameter, Lee-Yang-Parr with dispersion correction) was used. The 6-31G basis set with polarisation functions was used to ensure precise electronic structure calculations. In the theoretical settings, Density Functional Theory (DFT) was employed with the Self-Consistent Field (SCF) spin treatment set to automatic. Time-Dependent Density Functional Theory (TD-DFT) optimised the first excited state using full linear response28,53. One excited state was considered, with a maximum of 32 TD-DFT iterations. The energy convergence was set to a threshold of 5 × 10⁻5 Ha, while the residual convergence limit was set at 0.01 Ha. The grid density was adjusted to medium to balance accuracy with computational efficiency. For SCF calculations, the Quick setting was selected28,53. The initial guess for the calculation was based on atomic overlap, and the convergence criteria were defined with a maximum of 48 iterations, an energy change threshold of 4 × 10⁻5 Ha, and a root mean square (RMS) change in the density matrix of 5 × 10−6 Ha28,54. The maximum number of optimisation steps was limited to 100, adhering to the default convergence criteria. The initial Hessian matrix was obtained using the Schlegel method, and redundant internal coordinates were employed for geometry optimisation. In the properties calculation section, electrostatic potential (ESP), average local ionisation energy, noncovalent interactions (NCI), electron and spin density, molecular orbitals—specifically the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) within a range of 0 to 2 orbitals—were computed. Additionally, a QM-Convergence Monitor was used to analyse the results of the DFT calculations28,51. Pharmacokinetic (PK) analysis plays a crucial role in drug development as it provides insights into how a drug is absorbed, distributed, metabolised, and excreted by the body. These processes, collectively known as ADME (Absorption, Distribution, Metabolism, and Excretion), are fundamental to understanding a drug’s effectiveness, safety, and suitability for clinical use. In this study, we assessed the pharmacokinetic properties of a natural compound identified as a potential drug candidate. The evaluation was carried out using two primary methods: QikProp and Lipinski’s Rule of Five, which are widely used tools in early-stage drug development to predict key characteristics related to drug-likeness, bioavailability, and overall suitability for oral administration27,28,55,56. The QikProp tool provided predictions for various pharmacokinetic parameters, including solubility, permeability, and potential for drug-drug interactions, which are essential for assessing a compound’s likelihood of success in clinical trials28,55,56. Lipinski’s Rule of Five was applied to determine whether the compound meets key drug-like properties, such as molecular weight, hydrogen bond donors and acceptors, lipophilicity, and rotatable bonds27,28. A compound that satisfies most or all of these rules is more likely to have favourable pharmacokinetics and bioavailability, increasing its chances of success as an oral drug. In addition to assessing the natural compound, we compared its pharmacokinetic properties with those of a control drug, Tucatinib, a known targeted cancer therapy. Tucatinib was selected as a reference due to its established clinical profile and documented pharmacokinetic properties. By comparing the PK properties of the natural compound with those of Tucatinib, we aimed to evaluate its potential advantages and drawbacks in bioavailability, absorption, distribution, and elimination. This comprehensive PK analysis provided valuable insights into the compound’s pharmacokinetic profile, which is essential in determining its suitability for further clinical development29,40. The analysis helped us understand the natural compound’s drug-likeness and bioavailability, and also allowed us to gauge how it might perform in real-world therapeutic settings compared to a clinically approved drug such as Tucatinib. Ultimately, this pharmacokinetic assessment is a critical step in the preclinical evaluation of the natural compound, providing foundational data that can inform subsequent phases of drug development, including clinical trial design and optimisation of drug-delivery methods21,22,24,29,40. WaterMap studies WaterMap is a computational tool that provides insights into the role of water molecules in protein-ligand interactions. By identifying water molecules energetically unfavourable or displaced upon ligand binding, WaterMap helps design ligands that can optimise binding affinity through water displacement22,57,58. The WaterMap tool is particularly useful in drug design because it helps identify hydration sites within the protein’s binding site. When targeted for ligand optimisation, these sites can improve binding stability and overall drug efficacy59. In this study, we employed WaterMap in Schrödinger’s Maestro (v.2024-2) to perform calculations, which are supported by Desmond, and selected the ligand from the docked protein-ligand (P-L) complex28,57–61. The ligand was retained while waters within a 10 Å radius of the selected atoms were analysed. The simulation setup involved truncating the protein, and the OPLS4 force field was applied to treat the existing water as a solvent28,35. A simulation duration of 5 ns was chosen, considerably exceeding the manufacturer-recommended minimum of ~2 ns required for the convergence of hydration-site thermodynamics in protein binding pockets59, with the trajectory file set to not return after completion. Once the calculations were completed, the WaterMap-Examine Results tool in Schrödinger’s Maestro (v.2024-2) was used to analyse the data28,59. The receptor, ligand, hydrogen bonds, and markers within the complex’s WaterMap were visualised for detailed interpretation. We then assessed the enthalpy, entropy, free energy, overlap factor, hydration sites, interacting residues, and bond types, among other properties, using the project table and ligand interaction diagram for in-depth analysis28,57–59. MD simulation studies and binding free energy calculations Molecular dynamics (MD) simulations are essential for studying the dynamic behaviour of biomolecules over time. These simulations provide insights into the stability of protein-ligand interactions, conformational changes, and the flexibility of proteins and ligands. MD simulations also allow assessment of binding stability under physiological conditions, refine docking results, and provide more accurate predictions of drug efficacy. We utilised Desmond in Schrödinger’s Maestro (v.2024-2), available from D. E. Shaw Research (https://www.deshawresearch.com/) for the MD simulation28,60,61. The simulation was conducted in three key steps, each designed to ensure a comprehensive analysis of the protein-ligand complex. We built the system using the System Builder tool in the initial step. The predefined TIP3P solvent model was employed for solvation, a commonly used model that represents water molecules with three rigid point charges, offering a good balance between computational efficiency and accuracy28,62. The system was set up with orthorhombic boundary conditions, ensuring that the simulation box accommodates the complex while minimising edge effects. An orthorhombic simulation box was constructed with a 10 Å solvent buffer between the protein–ligand complex and the periodic image on each axis (which, for the systems studied here, corresponds to absolute box dimensions of 80–100 Å per side), providing sufficient space to avoid self-interaction artefacts. Following the system setup, we minimised the system’s volume to better visualise the boundary box. Na+ and Cl– counterions were added to neutralise the net charge of each system; additional ions were then placed to reach a physiological salt concentration of 0.15 M NaCl, while excluding ion placement within 20 Å of the protein–ligand complex. The previously reported numbers (7 Na+/4 Cl– for the 2-Aoeobenoxmide complexes and 6 Na+/4 Cl– for the Tucatinib complexes across PDB IDs 1A52, 3PP0, 4EJN, 4I23 and 7R9V) correspond to these combined neutralising and salt-buffering contributions. The simulation employed the OPLS4 force field, a robust choice for handling molecular interactions, including nonbonded interactions such as van der Waals and electrostatic forces28,35. In the second step, the prepared system was loaded into the Molecular Dynamics Panel of Schrödinger’s Maestro (v.2024-2) for the actual MD simulation28,60. The simulation was conducted for 100 ns, a typical duration for MD simulations, to capture the system’s dynamic behaviour. The simulation time was recorded at 100-ps intervals, yielding 1000 frames for analysis. The simulation was run under the NPT ensemble (constant Number of particles, Pressure, and Temperature), which is standard for simulating biomolecular systems28,63. The temperature was maintained at 300 K (approximately room temperature), and the pressure was set to 1.01325 bar, mimicking physiological conditions. Before initiating the production run, the system was relaxed to allow for equilibration, ensuring the protein-ligand complex reached a stable state. After equilibration, the system was ready for production, and interaction analysis was performed on the resulting trajectory. In the final step, we employed the Simulation Interaction Diagram (SID) tool to examine the results28,60. Binding free energy calculations are crucial for estimating the strength and stability of ligand-protein interactions, which directly correlate with the drug’s potency and potential efficacy. To calculate the binding free energy, we used several methods, including Molecular Mechanics/Generalised Born Surface Area (MM-GBSA), Molecular Mechanics/Poisson–Boltzmann Surface Area (MM/PBSA), and Free Energy Perturbation (FEP). These methods consider factors influencing binding affinity, such as van der Waals forces, electrostatic interactions, solvation effects, and entropy contributions64,65. These calculations are essential for refining docking results, guiding the optimisation of drug candidates, and enhancing the overall drug design process by prioritising compounds with high binding affinity and favourable interaction profiles66. For the MM-GBSA calculations, we executed the following command to run the MM-GBSA.py code file: $SCHRODINGER/run thermal_MM-GBSA.py desmond_md_job_NAME-out.cms. After completing the MM-GBSA calculations, we analysed the resulting poses and energies. This analysis included the binding free energy and the total complex energy, which were plotted to show the energy profile over 0 to 1000 frames. The resulting graphs provided valuable insights into the stability of the complex and the ligand’s binding strength to the protein over time. This multi-step MD simulation pipeline enabled an in-depth analysis of protein-ligand interactions, providing crucial insights into the binding mechanisms and the stability of the complexes65,66. By utilising advanced simulation tools and free energy calculation methods, we were able to obtain a comprehensive understanding of the ligand’s binding profile, paving the way for further drug optimisation and design65,66. Results Validation and analysis of protein structures The energy profile obtained from the Protein Preparation Wizard (PPW) in Schrödinger Maestro provides crucial insights into the structural stability and readiness of the proteins for subsequent computational studies, such as molecular docking and dynamics simulations. The evaluated proteins—identified by their PDB IDs 1A52, 3PP0, 4EJN, 4I23, and 7R9V—exhibit negative total and potential energies, indicative of energetically favourable and minimised conformations post-preparation. These energies, representing the cumulative effects of bonded and non-bonded interactions, range from -1002.69 kcal/mol for 1A52 to -3595.09 kcal/mol for 7R9V, with more negative values reflecting greater structural stability due to lower internal strain and more favourable intermolecular forces. The fact that the kinetic energy and temperature values are zero across all systems is expected, as the energy minimisation step does not involve dynamic simulations but purely geometry-based optimisation, affirming that these energies reflect a static, equilibrated state of the protein conformations. The bond stretch energies across the proteins, from 113.35 kcal/mol to 452.78 kcal/mol, correspond to distortions from ideal bond lengths. While absolute values vary with protein size and complexity, they fall within expected ranges, indicating that bond lengths have been reasonably constrained during minimisation. Similarly, angle bending energies—which range from 505.49 to 2004.06 kcal/mol—capture deviations from ideal bond angles. The gradual increase in these energies from smaller to larger or more complex proteins, such as 7R9V, is consistent with increased structural intricacy and the number of atoms. Torsional energies, representing resistance to rotation around bonds, span from 351.93 kcal/mol in 1A52 to 1639.13 kcal/mol in 7R9V and are also reflective of internal flexibility and conformational freedom of side chains and backbone dihedrals. These energies are within physiologically acceptable thresholds, supporting the conclusion that the protein geometries have been refined to stable, low-energy conformers. The 1,4 Lennard-Jones and 1,4 electrostatic interactions provide insights into nonbonded interactions between atoms separated by three covalent bonds. The 1,4 Lennard-Jones energies range from 1141.75 to 4386.65 kcal/mol, and the 1,4 electrostatic energies range from 343.99 to 1421.34 kcal/mol. These values demonstrate the presence of substantial intramolecular stabilising interactions that contribute to the proteins’ folded structures. The general increase in the magnitude of these terms with protein size is expected and correlates with the increasing number of such interactions. The overall van der Waals interactions, as captured by the Lennard-Jones energy term, are markedly negative across all proteins, with values such as -2298.12 kcal/mol for 1A52 and -8944.51 kcal/mol for 7R9V, indicating significant stabilisation through favourable dispersion interactions. Similarly, the electrostatic energy component is highly favourable, ranging from –1291.57 to –4720.78 kcal/mol, indicating the formation of stabilising charge–charge and dipole–dipole interactions that further anchor the structural conformation. Notably, the hydrogen-bond energy term is reported as zero for all proteins; in the Protein Preparation Wizard output, the hydrogen-bond contribution is implicitly included within the broader electrostatic energy term and is therefore not reported as a separate component. The absence of restraining energy also suggests that no artificial positional restraints were applied during minimisation, ensuring that the structures reflect natural minima in the potential energy surface. Taken together, the total energy landscape of these proteins illustrates that they have been adequately minimised to low-energy conformations with favourable intramolecular interactions and minimised steric and electrostatic repulsion. The balance and magnitude of bonded (bond, angle, torsion) and non-bonded (van der Waals, electrostatics, 1,4 interactions) energy terms all point toward well-optimised, structurally stable protein models. As such, these prepared structures are well-suited for downstream applications, including molecular docking, virtual screening, and molecular dynamics simulations. Their minimised, energetically favourable conformations provide reliable binding-site geometries and accurate energetic baselines for ligand-binding assessments. Further, in Fig. 2, we have shown the prepared structures and their respective Ramachandran Plots, indicating that they are ready for docking and other studies.Fig. 2Showing the Ramachandran Plot for prepared proteins and 3D cartoon structures of prepared proteins, highlighting the bound native ligands for breast cancer transferase proteins with PDB IDs (A) 1A52, (B) 3PP0, (C) 4EJN, (D) 4I23, and (E) 7R9V. Showing the Ramachandran Plot for prepared proteins and 3D cartoon structures of prepared proteins, highlighting the bound native ligands for breast cancer transferase proteins with PDB IDs (A) 1A52, (B) 3PP0, (C) 4EJN, (D) 4I23, and (E) 7R9V. Analysis of multitargeted molecular docking results and control comparison The molecular docking study conducted using Schrödinger’s Glide and pose processing using MM-GBSA approaches helped to identify 2-Aoeobenoxmide (2-[1-(2-amino-2-oxo-ethoxy)-6-oxo-benzo[c]chromen-3-yl]oxyacetamide, ZINC134008) across five breast cancer-related protein targets, and we compared it with the same grid and other parameters for Tucatinib (breast cancer drug). These findings are evaluated not only in terms of docking scores and MM-GBSA binding free energy (ΔG_bind), but also in terms of ligand efficiency metrics, interaction profiles, and grid parameters that reflect the accuracy of the docking environment. For the Human Estrogen Receptor (PDB ID: 1A52), Tucatinib produced a modest docking score of –4.875 kcal/mol and an MM-GBSA binding free energy of –29.68 kcal/mol. The interactions involve a hydrogen bond with ASP351, a salt bridge facilitated by this residue, aromatic interactions via π-π stacking with TRP383 and PHE404, and a π-cation interaction with LYS529. In contrast, 2-Aoeobenoxmide displayed a more negative docking score of –10.327 kcal/mol and MM-GBSA ΔG_bind of –47.18 kcal/mol. This ligand formed strong hydrogen bonding interactions with LEU346 and GLU353 via its NH₂ moiety. Ligand efficiency improved significantly for 2-Aoeobenoxmide (–11.182 kcal/mol for ligand efficiency in terms of interaction energy and –2.831 kcal/mol per surface area), relative to Tucatinib (–6.475 and –1.237 kcal/mol, respectively). The MM-GBSA Prime energy terms showed a comparable hydrogen bonding potential (–134.58 vs –133.42 kcal/mol), although the Coulomb interaction term computed from the Prime MM-GBSA decomposition was markedly higher for 2-Aoeobenoxmide (1000.74 kcal/mol). These Prime Coulomb values are long-range electrostatic terms reported in kcal/mol and should be interpreted as relative comparisons between ligands rather than as absolute interaction energies; even so, the comparison indicates a stronger polar interaction contribution for 2-Aoeobenoxmide. The docking grid for 1A52 was appropriately defined with a resolution of 2.8 Å and grid centre coordinates at x = 106.693, y = 14.7, and z = 96.263 Å with a uniform grid range of 23.845 Å in all dimensions, ensuring comprehensive coverage of the ligand binding domain (Fig. 3, Table 1). Within the kinase domain of the Human HER2 receptor (PDB ID: 3PP0), Tucatinib demonstrated a high binding affinity with a docking score of –10.948 kcal/mol and MM-GBSA ΔG_bind of –68.73 kcal/mol, mediated by hydrogen bonding with LEU726 and MET801. However, 2-Aoeobenoxmide, despite a slightly less favourable docking score of –9.668 kcal/mol, retained a comparable MM-GBSA score of –53.35 kcal/mol and established four hydrogen bonds with SER728, ASN850, ASP863, and MET801. These additional contacts suggest improved binding specificity and potentially broader interaction coverage within the kinase pocket. Ligand efficiency for 2-Aoeobenoxmide was again higher (–12.645 kcal/mol interaction and –3.201 kcal/mol per surface area) compared to Tucatinib (–14.995 and –2.864 kcal/mol, respectively). Prime energy analysis showed increased electrostatic contributions for 2-Aoeobenoxmide (1300.65 kcal/mol vs. 137.508 kcal/mol for Tucatinib), with slightly improved hydrogen bonding and comparable covalent bond stabilisation (–8748.16 vs. –8813.45 kcal/mol). The HER2 grid parameters used a high-resolution crystal structure (2.25 Å) and were centred at x = 16.642, y = 15.92, z = 27.037 Å with a grid box size of 31.913 Å in all dimensions (Fig. 3, Table 1). For AKT1 (PDB ID: 4EJN), Tucatinib showed a docking score of –7.933 kcal/mol and MM-GBSA ΔG_bind of –50.77 kcal/mol. Its key interactions were π-π stacking contacts with TRP80, suggesting limited direct polar interaction. Conversely, 2-Aoeobenoxmide achieved a stronger docking score of –10.006 kcal/mol and an improved MM-GBSA score of –57.62 kcal/mol. The ligand formed four hydrogen bonds with residues ILE290, THR291, SER205, and THR211, along with robust π-π stacking with TRP80 via three aromatic rings. These features reflect enhanced conformational fit and interaction diversity. The ligand efficiency metrics again favoured 2-Aoeobenoxmide (–13.657 and –3.457 kcal/mol) compared to Tucatinib (–11.076 and –2.115 kcal/mol). Coulombic interactions were dramatically higher in 2-Aoeobenoxmide (2541.08 kcal/mol) than in Tucatinib (209.235 kcal/mol), indicating a highly polarisable interaction environment. The docking grid, defined with a resolution of 2.19 Å and centred at x = 35.298, y = 43.719, z = 18.723 Å, and spanning 27.473 Å on all axes, accurately encompassed the functional domain of interest (Fig. 3, Table 1). In the case of EGFR (PDB ID: 4I23), Tucatinib presented a docking score of –5.782 kcal/mol and MM-GBSA binding energy of –54.39 kcal/mol. Interactions included three hydrogen bonds and a salt bridge involving ASP837, ASN842, ASP855, and MET793. 2-Aoeobenoxmide performed better with a docking score of –8.162 kcal/mol and MM-GBSA ΔG_bind of –51.39 kcal/mol. It formed five hydrogen bonds with ASP800, ARG841, ASN842, MET793, and CYS797, reflecting a denser interaction network. Ligand efficiency (–12.181 and –3.083 kcal/mol for 2-Aoeobenoxmide) surpassed Tucatinib’s (–11.866 and –2.266 kcal/mol). The docking grid was centred at x = –0.219, y = –52.639, z = –22.501 Å, with a grid range of 25.144 Å, and the structure resolution of 2.8 Å offered a reliable template for modelling ligand interactions (Fig. 3, Table 1). Finally, in the PIK3CA protein (PDB ID: 7R9V), Tucatinib showed the weakest performance among all proteins, with a docking score of –6.319 kcal/mol, an MM-GBSA ΔG_bind of –29.40 kcal/mol, and no specific interactions observed. In contrast, 2-Aoeobenoxmide displayed a markedly superior docking score of –8.691 kcal/mol and a highly favourable MM-GBSA score of –53.24 kcal/mol. It engaged SER919 via hydrogen bonding, contacting PHE930 and ASP933 with its O atom and a π-π stacking interaction with TYR836. Ligand efficiency values (–12.619 and –3.194 kcal/mol) significantly exceeded the control’s (–6.414 and –1.225 kcal/mol). Notably, the electrostatic component from the MM-GBSA calculation was exceptionally high for 2-Aoeobenoxmide (4731.49 kcal/mol), underscoring the strength of polar interactions in stabilising the ligand. Grid box parameters were based on a 2.69 Å resolution structure, with the box centred at x = –19.281, y = 10.026, z = 28.391 Å and a range of 37.238 Å across the axes, ensuring complete sampling of the active site (Fig. 3, Table 1). Across the five protein targets, 2-Aoeobenoxmide showed more favourable in silico binding metrics than Tucatinib, with more negative docking scores, more negative MM-GBSA binding free energies, and richer interaction profiles. It consistently achieved higher ligand efficiency and Coulombic interaction scores, supported by extensive hydrogen bonding and π- π stacking interactions. These results collectively underscore the strong potential of 2-Aoeobenoxmide as a multitarget inhibitor in breast cancer pathways and affirm its suitability for further experimental validation and lead optimisation.Fig. 3Showing the Multitargeted Docking results in 3D surface view and 2D as Ligand Interaction Diagram of 2-Aoeobenoxmide in complex with PDB IDs (Aa) 1A52, (Ba) 3PP0, (Ca) 4EJN, (Da) 4I23, and (Ea) 7R9V, and Tucatinib in complex with (Ab) 1A52, (Bb) 3PP0, (Cb) 4EJN, (Db) 4I23, and (Eb) 7R9V for comparative understanding.Table 1Showing the Grid parameters and various energies produced during the multitargeted docking studies of ZINC natural compounds and their comparative docking evaluation with FDA-approved Tucatinib against Breast Cancer Transferase proteins.PDB IDCaseLigandDocking ScoreMM-GBSA dG Bindlig efficiency Inlig efficiency saPrime HbondPrime Covalent1A52ControlTucatinib –4.875 –29.68 –6.475 –1.237 –133.42 –7640.023PP0Control –10.948 –68.73 –14.995 –2.864 –154.04 –8813.454EJNControl –7.933 –50.77 –11.076 –2.115 –199.25 –11727.044I23Control –5.782 –54.39 –11.866 –2.266 –157.41 –9506.877R9VControl –6.319 –29.4 –6.414 –1.225 –466.3 –27846.061A52New2-Aoeobe noxmide –10.327 –47.18 –11.182 –2.831 –134.58 –7629.773PP0New –9.668 –53.35 –12.645 –3.201 –155.65 –8748.164EJNNew –10.006 –57.62 –13.657 –3.457 –199.25 –11727.044I23New –8.162 –51.39 –12.181 –3.083 –158 –9527.797R9VNew –8.691 –53.24 –12.619 –3.194 –466.3 –27846.06PDB IDCaseResolutiongridbox xcentgridbox xrangegridbox ycentgridbox yrangegridbox zcentgridbox zrange1A52All2.8106.69323.84514.723.84596.26323.8453PP0All2.2516.64231.91315.9231.91327.03731.9134EJNAll2.1935.29827.47343.71927.47318.72327.4734I23All2.8 –0.21925.144 –52.63925.144 –22.50125.1447R9VAll2.69 –19.28137.23810.02637.23828.39137.238 Showing the Multitargeted Docking results in 3D surface view and 2D as Ligand Interaction Diagram of 2-Aoeobenoxmide in complex with PDB IDs (Aa) 1A52, (Ba) 3PP0, (Ca) 4EJN, (Da) 4I23, and (Ea) 7R9V, and Tucatinib in complex with (Ab) 1A52, (Bb) 3PP0, (Cb) 4EJN, (Db) 4I23, and (Eb) 7R9V for comparative understanding. Table 1Showing the Grid parameters and various energies produced during the multitargeted docking studies of ZINC natural compounds and their comparative docking evaluation with FDA-approved Tucatinib against Breast Cancer Transferase proteins.PDB IDCaseLigandDocking ScoreMM-GBSA dG Bindlig efficiency Inlig efficiency saPrime HbondPrime Covalent1A52ControlTucatinib –4.875 –29.68 –6.475 –1.237 –133.42 –7640.023PP0Control –10.948 –68.73 –14.995 –2.864 –154.04 –8813.454EJNControl –7.933 –50.77 –11.076 –2.115 –199.25 –11727.044I23Control –5.782 –54.39 –11.866 –2.266 –157.41 –9506.877R9VControl –6.319 –29.4 –6.414 –1.225 –466.3 –27846.061A52New2-Aoeobe noxmide –10.327 –47.18 –11.182 –2.831 –134.58 –7629.773PP0New –9.668 –53.35 –12.645 –3.201 –155.65 –8748.164EJNNew –10.006 –57.62 –13.657 –3.457 –199.25 –11727.044I23New –8.162 –51.39 –12.181 –3.083 –158 –9527.797R9VNew –8.691 –53.24 –12.619 –3.194 –466.3 –27846.06PDB IDCaseResolutiongridbox xcentgridbox xrangegridbox ycentgridbox yrangegridbox zcentgridbox zrange1A52All2.8106.69323.84514.723.84596.26323.8453PP0All2.2516.64231.91315.9231.91327.03731.9134EJNAll2.1935.29827.47343.71927.47318.72327.4734I23All2.8 –0.21925.144 –52.63925.144 –22.50125.1447R9VAll2.69 –19.28137.23810.02637.23828.39137.238 Showing the Grid parameters and various energies produced during the multitargeted docking studies of ZINC natural compounds and their comparative docking evaluation with FDA-approved Tucatinib against Breast Cancer Transferase proteins. Analysis of interaction patterns The residue interaction profile derived from the Interaction Fingerprint analysis in Maestro for the docked poses of the five breast cancer-related protein targets in complex with Tucatinib and 2-Aoeobenoxmide provides a comprehensive overview of the molecular interactions contributing to ligand stabilisation within the binding pockets. Leucine (LEU) appears 26 times among the most frequently involved residues, underscoring its critical role in mediating hydrophobic interactions. Leucine is a nonpolar, aliphatic amino acid commonly buried within the hydrophobic cores of proteins. It is often involved in van der Waals contacts with hydrophobic regions of ligands. Its high occurrence indicates that both Tucatinib and 2-Aoeobenoxmide engage extensively with the lipophilic environment of the binding pockets, contributing significantly to the stabilisation of the ligand via favourable entropy-driven interactions and shape complementarity. Methionine (MET) and Phenylalanine (PHE), each observed in 11 interaction events, further emphasise the dominance of hydrophobic interactions. Methionine, with its thioether side chain, is nonpolar but slightly more polarisable, which may allow it to engage in weak dipole interactions in addition to van der Waals contacts. Phenylalanine, an aromatic residue, is central to π-π stacking and π-alkyl interactions, particularly with aromatic moieties of ligands, such as the benzene rings found in both Tucatinib and 2-Aoeobenoxmide. Aspartic acid (ASP), present in 10 interactions, is a negatively charged, polar residue that plays a critical role in hydrogen bonding and salt-bridge formation. Its carboxylate group can act as a hydrogen bond acceptor or participate in electrostatic interactions with positively charged or polarisable regions of the ligand, as seen in the interaction between ASP351 in the estrogen receptor and Tucatinib. The prevalence of aspartic acid implies that electrostatic interactions are a recurring theme in ligand stabilisation, especially when ligands contain protonated amine groups or donor atoms. Asparagine (ASN), appearing in 7 interactions, is a polar but uncharged residue capable of forming strong hydrogen bonds via its amide side chain. Asparagine often contributes to ligand-binding specificity by acting as both a donor and an acceptor in hydrogen-bonding networks, thereby helping anchor the ligand in a specific orientation. This residue’s presence, along with threonine (THR), involved in 6 interactions, reflects the importance of hydrogen bonding in defining ligand orientation and increasing binding enthalpy. Threonine, a polar residue with a hydroxyl group, is often solvent-exposed and participates in directional hydrogen bonding with ligand polar atoms, further contributing to binding stability. Glutamic acid (GLU), another negatively charged, polar residue, was identified in 5 interaction cases. Similar to aspartic acid, glutamic acid’s side chain provides a longer reach, enabling electrostatic or hydrogen-bonding interactions with positively charged or hydrogen-accepting groups on the ligand. Its presence supports the role of ionic and polar interactions in the binding interface. Histidine (HIS), also appearing 5 times, is a versatile residue due to its imidazole side chain, which can exist in both protonated and neutral forms at physiological pH. Histidine often acts as a hydrogen bond donor or acceptor and occasionally participates in π-cation or π-π stacking interactions, especially in kinase domains. Lysine (LYS), present in 4 interactions, is a positively charged residue and commonly involved in forming salt bridges or hydrogen bonds with negatively charged ligand moieties (Fig. 4). This reinforces the significance of electrostatic complementarity in ligand-protein complexes. Alanine (ALA) and Isoleucine (ILE), each with 3 interaction events, are small, nonpolar residues that typically stabilise ligands through van der Waals forces, especially in the hydrophobic interior of proteins. Though less prominent, their roles are consistent with maintaining a tightly packed binding site environment. Arginine (ARG) and Tyrosine (TYR), each involved in 2 interaction events, represent a combination of electrostatic and aromatic contributions. Arginine, with its guanidinium group, is highly polar and can form multiple hydrogen bonds or salt bridges, especially with carboxylates or phosphate groups. Tyrosine, an aromatic polar residue, can engage in both π-stacking and hydrogen bonding due to its phenolic –OH group, thereby contributing to specificity and binding enthalpy. Single occurrences of Cysteine (CYS), Glycine (GLY), Serine (SER), and Tryptophan (TRP) round out the interaction profile. Cysteine, with its thiol group, may participate in weak polar interactions or covalent bonding in certain contexts, though the latter is rare in reversible docking scenarios. Glycine, the smallest residue, does not contribute much directly through side-chain interactions but allows for conformational flexibility in the binding pocket. Serine, a small polar residue, contributes to hydrogen bonding through its hydroxyl group. Tryptophan, though occurring only once, is a bulky aromatic residue capable of participating in extensive π-stacking interactions, often crucial for ligands containing large aromatic systems. Collectively, the interaction residue profile demonstrates a well-balanced contribution of hydrophobic (LEU, MET, PHE, ALA, ILE), polar (ASN, THR, GLU, SER), aromatic (PHE, TYR, TRP), and charged (ASP, GLU, LYS, ARG, HIS) residues across the protein-ligand interfaces (Fig. 4). The higher frequency and diversity of these contacts in the case of 2-Aoeobenoxmide further support its enhanced binding affinity and multitarget potential in breast cancer therapy.Fig. 4Showing the Molecular Interaction Fingerprint of the docked poses of 2-Aoeobenoxmide and Tucatinib in complex PDB IDs 1A52, 3PP0, 4EJN, 4I23, and 7R9V. The main plot shows N-to-C terminal residues highlighted from blue to red; the upper side shows the count of residue interactions, and the right side shows the count of ligand interactions. Showing the Molecular Interaction Fingerprint of the docked poses of 2-Aoeobenoxmide and Tucatinib in complex PDB IDs 1A52, 3PP0, 4EJN, 4I23, and 7R9V. The main plot shows N-to-C terminal residues highlighted from blue to red; the upper side shows the count of residue interactions, and the right side shows the count of ligand interactions. Analysis and comparison of pharmacokinetics, DFT results The pharmacokinetic profiling of 2-Aoeobenoxmide and the control drug Tucatinib reveals distinct differences in absorption, distribution and key molecular properties. These differences are pertinent to bioavailability and overall drug-like behaviour22,27,28,55,56. In terms of human oral absorption, a key determinant of oral bioavailability, Tucatinib demonstrates excellent absorption with a predicted 100% human oral absorption and a corresponding qualitative score of 1, indicating high permeability and favourable pharmacokinetics. In contrast, 2-Aoeobenoxmide exhibits significantly lower predicted oral absorption (28.36%), with a qualitative score of 2, suggesting poor absorption and potential challenges in achieving therapeutic plasma concentrations when administered orally22,27,28,55,56. This is further supported by the Caco-2 (QPPCaco) and MDCK (QPPMDCK) cell-permeability values27,28,55,56, where Tucatinib shows high permeability (496.085 nm/sec and 231.89 nm/sec, respectively), while 2-Aoeobenoxmide shows markedly lower permeability values (3.451 nm/sec and 6.092 nm/sec), indicating poor intestinal and cellular uptake. The logP values also reflect contrasting physicochemical properties; logP is a measure of lipophilicity and a key determinant of passive membrane permeability27,28,55,56. Tucatinib, with a QPlogPo/w of 4.635, is lipophilic, favouring membrane permeability, which aligns with its high absorption and distribution profile (Table 2). On the other hand, 2-Aoeobenoxmide has a significantly lower logP of –1.403, indicating it is hydrophilic, which typically results in reduced passive membrane diffusion and poor permeability unless aided by active transport mechanisms. This trend is reinforced by the polar surface area (PSA) values, where 2-Aoeobenoxmide exhibits a higher PSA of 159.341 Å2 compared to 103.512 Å2 for Tucatinib. High PSA values, especially those exceeding 140 Å2, are generally associated with poor oral absorption and reduced blood–brain barrier (BBB) penetration27,28,55,56. Despite these challenges, 2-Aoeobenoxmide possesses several pharmacologically desirable properties. It has a lower molecular weight (342.307 Da) than Tucatinib (480.528 Da) (Table 2), which falls within the molecular-weight range commonly associated with drug-like compounds. However, oral suitability cannot be inferred from this descriptor alone and should be interpreted alongside intestinal absorption, Caco-2 permeability, PSA and logP. Both compounds comply with Lipinski’s Rule of Five23, indicating no violations of major drug-likeness parameters, and both also satisfy the Rule of Three conditions23, supporting their qualification as drug-like candidates for early-stage screening. In terms of hydrogen bonding, 2-Aoeobenoxmide shows a higher number of both hydrogen bond donors (4) and acceptors (9) than Tucatinib (2 and 8, respectively), contributing to increased polarity and reduced membrane permeability, but potentially enhancing target binding affinity and specificity through polar interactions. The solubility (QPlogS) of 2-Aoeobenoxmide (–0.111) is considerably better than that of Tucatinib (–7.373), indicating improved aqueous solubility, which is advantageous for formulation and overall bioavailability, especially for intravenous administration or when oral routes are bypassed22,27,28,55,56. Additional descriptors such as polarisability (QPpolrz) and dipole moment further differentiate the two molecules. Tucatinib exhibits a higher dipole moment (13.266 Debye) and polarisability (53.452 Å3) compared to 2-Aoeobenoxmide (8.609 Debye and 32.06 Å3) (Table 2), suggesting greater electronic flexibility and a more favourable profile for dynamic interactions within protein binding sites. However, 2-Aoeobenoxmide has a more compact and globular structure, with a globularity index of 0.8346 versus 0.7564 for Tucatinib, which may enhance fit within confined binding pockets. Metabolic stability, another critical pharmacokinetic aspect, also differs notably between the compounds. The number of predicted metabolic reactions (#metab)51 for 2-Aoeobenoxmide is 6, compared with 2 for Tucatinib; a higher #metab count is generally associated with a shorter predicted half-life and may indicate that the compound would require structural modification or formulation strategies to improve its metabolic stability. The absence of toxicophore alerts (#stars = 0) in 2-Aoeobenoxmide, as opposed to 2 alerts in Tucatinib, suggests a potentially safer in silico safety profile51,52, further supporting its candidacy for drug repurposing despite its absorption limitations67. Regarding CNS penetration, both compounds are predicted to be non-penetrant, with CNS scores of –2 and QPlogBB values of –1.396 (Tucatinib) and –2.586 (2-Aoeobenoxmide), indicating poor brain penetration, which is desirable for reducing off-target neurological side effects in the treatment of peripheral cancers (Table 2). The hERG channel inhibition potential (QPlogHERG), a widely used in silico predictor of cardiac toxicity22,55,68, shows a more favourable score for 2-Aoeobenoxmide (–2.499) compared with Tucatinib (–7.499), implying a lower predicted risk of cardiotoxicity associated with hERG potassium-channel blockade, a common liability among kinase inhibitors68. Lastly, the surface area descriptors such as FISA (hydrophilic surface area), FOSA (hydrophobic surface area), and SASA (solvent accessible surface area) suggest that 2-Aoeobenoxmide has a significantly higher hydrophilic character (FISA = 291.322 Å2) and lower hydrophobic contact area (FOSA = 69.913 Å2) compared to Tucatinib (FISA = 137.124 Å2, FOSA = 269.985 Å2), correlating with its better solubility but reduced permeability (Table 2). These attributes suggest that while 2-Aoeobenoxmide holds promise as a low-toxicity anticancer agent, its oral bioavailability may be a limiting factor requiring optimisation through prodrug strategies, formulation enhancements, or alternative delivery routes to realise its full therapeutic potential. Taken together, the ADME profile suggests that the most relevant clinical liability of 2-Aoeobenoxmide is its low predicted intestinal absorption and permeability; these limitations could be mitigated in future work through prodrug design, lipid-based or nanoparticle formulations, or parenteral routes of administration21,67.Table 2Showing the Pharmacokinetics results of 2-Aoeobenoxmide computed with QikProp and compared with Tucatinib.DescriptorsTucatinib2-AoeobenoxmideDescriptorsTucatinib2-Aoeobenoxmide#acid00HumanOralAbsorption12#amide02IP(eV)8.1569.176#amidine00Jm00.755#amine00mol MW480.528342.307#in3400Moleculesmallsmall#in563014PISA424.801217.894#metab26PSA103.512159.341#NandO108QPlogBB –1.396 –2.586#noncon20QPlogHERG –7.499 –2.499#nonHatm3625QPlogKhsa0.752 –1.178#ringatoms3014QPlogKp –1.974 –5.545#rotor66QPlogPC1616.72211.952#rtvFG01QPlogPo/w4.635 –1.403#stars20QPlogPoct26.79622.783% HumanOralAbsorption10028.36QPlogPw15.02924.441accptHB89QPlogS –7.373 –0.111ACxDN^.5/SA0.01359970.0310812QPPCaco496.0853.451CIQPlogS –7.401 –2.191QPPMDCK231.896.092CNS –2 –2QPpolrz53.45232.06dip^2/V0.11856030.0741785RuleOfFive00dipole13.2668.609RuleOfThree11donorHB24SAamideO087.361EA(eV)1.0060.94SAfluorine00FISA137.124291.322SASA831.91579.129FOSA269.98569.913volume1484.293999.139glob0.75642990.8345912WPSA00 Table 2Showing the Pharmacokinetics results of 2-Aoeobenoxmide computed with QikProp and compared with Tucatinib.DescriptorsTucatinib2-AoeobenoxmideDescriptorsTucatinib2-Aoeobenoxmide#acid00HumanOralAbsorption12#amide02IP(eV)8.1569.176#amidine00Jm00.755#amine00mol MW480.528342.307#in3400Moleculesmallsmall#in563014PISA424.801217.894#metab26PSA103.512159.341#NandO108QPlogBB –1.396 –2.586#noncon20QPlogHERG –7.499 –2.499#nonHatm3625QPlogKhsa0.752 –1.178#ringatoms3014QPlogKp –1.974 –5.545#rotor66QPlogPC1616.72211.952#rtvFG01QPlogPo/w4.635 –1.403#stars20QPlogPoct26.79622.783% HumanOralAbsorption10028.36QPlogPw15.02924.441accptHB89QPlogS –7.373 –0.111ACxDN^.5/SA0.01359970.0310812QPPCaco496.0853.451CIQPlogS –7.401 –2.191QPPMDCK231.896.092CNS –2 –2QPpolrz53.45232.06dip^2/V0.11856030.0741785RuleOfFive00dipole13.2668.609RuleOfThree11donorHB24SAamideO087.361EA(eV)1.0060.94SAfluorine00FISA137.124291.322SASA831.91579.129FOSA269.98569.913volume1484.293999.139glob0.75642990.8345912WPSA00 Showing the Pharmacokinetics results of 2-Aoeobenoxmide computed with QikProp and compared with Tucatinib. Density functional theory (DFT) and time-dependent DFT (TDDFT) analyses were performed using Jaguar (B3LYP-D3 functional with the 6-31G basis set augmented with polarisation functions, and an implicit solvation treatment)28,51. The calculations provide insights into the quantum-chemical and electronic properties of 2-Aoeobenoxmide compared with Tucatinib, highlighting differences in reactivity, electrostatic characteristics and molecular stability in both gas and solvated phases. Both compounds achieved a geometry-convergence category of 4, corresponding to the default tight-convergence criterion in the Jaguar geometry optimiser and indicating complete and stable convergence to local minima during optimisation47,48. The spin multiplicity was 1 for both molecules, confirming their singlet ground states and indicating closed-shell electronic configurations, typical of neutral organic drug-like compounds. The number of canonical orbitals, a reflection of molecular size and electronic complexity, was higher for Tucatinib (663) than for 2-Aoeobenoxmide (443), in line with their respective molecular weights and atom counts. This suggests that Tucatinib has a more electronically complex structure, which may translate to a broader electron density distribution across its framework. The Average Local Ionisation Energy (ALIE) descriptors, which characterise the energetic demand for electron removal at every point on the molecular surface and are widely used in chemical-reactivity analysis47,48, reveal comparable ionisation profiles for both molecules, with 2-Aoeobenoxmide showing a mean ALIE of 270.06 kcal/mol and Tucatinib slightly higher at 272.94 kcal/mol (Fig. 5). These values indicate similar energetic demands for electron removal from the molecule, a parameter relevant for assessing chemical hardness and oxidative stability. The maximum ALIE value, which represents the most difficult-to-ionise region, is slightly lower for 2-Aoeobenoxmide (353.83 kcal/mol) than Tucatinib (383.15 kcal/mol), suggesting a marginally more reactive or polarisable electronic surface in the latter. The minimum ALIE values are also close (201.58 kcal/mol for 2-Aoeobenoxmide and 203.53 kcal/mol for Tucatinib), suggesting that both have comparable lowest-energy ionisable sites, which are often associated with reactive hot spots in molecular interactions. Electrostatic potential (ESP) analysis50,51 further delineates the molecular electrostatic environments at the binding interface. The ESP balance, a measure of the spatial distribution of positive and negative charges, shows significantly more asymmetry in 2-Aoeobenoxmide (0.248) than in Tucatinib (0.019), implying a more pronounced dipole and a less homogeneous charge distribution. This is consistent with the ESP mean value of –0.48 kcal/mol for 2-Aoeobenoxmide, indicating an overall slightly negative potential, compared to Tucatinib’s 58.27 kcal/mol, which is substantially more positive and suggests strong electron-withdrawing characteristics or polarised functional groups. The ESP maximum and minimum values further confirm this polarity contrast, with 2-Aoeobenoxmide exhibiting a lower ESP max (77.5 kcal/mol) and a significantly more negative ESP min (–59.9 kcal/mol) than Tucatinib (141.51 and –16.69 kcal/mol, respectively) (Fig. 5). These characteristics suggest that 2-Aoeobenoxmide has stronger localised negative regions, which may promote specific hydrogen bonding or ionic interactions in a biological context. Moreover, the ESP positive and negative variances, representing the heterogeneity of potential across the molecular surface, are notably higher for Tucatinib (1044.85 and 20.91 (kcal/mol)^2, respectively) than for 2-Aoeobenoxmide (331.01 and 272.78 (kcal/mol)^2), supporting the interpretation that Tucatinib has more extreme and spatially segregated electrostatic features. These variances are consistent with Tucatinib’s more rigid and bulky aromatic framework, whereas the more balanced profile of 2-Aoeobenoxmide may confer favourable binding interactions through more uniformly distributed polar contacts. In terms of energetics, 2-Aoeobenoxmide demonstrated a final energy of –1217.088 kcal/mol, with a gas-phase ground-state energy of –1217.169 kcal/mol and a solution-phase energy of –1217.207 kcal/mol, indicating a solvation energy of –23.91 kcal/mol. These values reflect a relatively stable molecule in isolation and a solvated environment with moderate solvation favourability. For Tucatinib, although the ground-state energy in the gas phase was reported as –1594.1869 kcal/mol, the solution-phase energy was –1594.2922 kcal/mol, indicating a solvation energy of –66.07 kcal/mol, which is significantly more favourable than that of 2-Aoeobenoxmide. This higher solvation energy reflects Tucatinib’s greater polarity and likely stronger interaction with solvent molecules, especially due to its fluorinated aromatic systems and polar functionalities. Frontier molecular orbital analysis reveals close HOMO–LUMO gaps for both molecules; the HOMO–LUMO gap is widely used as a measure of kinetic stability and chemical reactivity, with smaller gaps generally associated with higher reactivity50,51,69,70. 2-Aoeobenoxmide exhibits a HOMO of –0.214 eV and a LUMO of –0.074 eV, resulting in a gap of ~0.1395 eV, while Tucatinib has a HOMO of –0.225 eV and a LUMO of –0.072 eV, with a slightly narrower gap of ~0.1526 eV (Fig. 5). These gaps are indicative of electronic excitation energies, chemical reactivity, and kinetic stability, with the slightly smaller gap in Tucatinib suggesting marginally higher reactivity under photoinduced or redox conditions. Importantly, the TDDFT calculations for 2-Aoeobenoxmide provide additional insight into its optical properties. A lowest singlet excitation energy of 3.2341 eV and an oscillator strength of 0.0975 were observed; the combination of excitation energy and oscillator strength characterises the absorption probability and photophysical behaviour of the molecule50,51, pointing to the molecule’s ability to absorb in the UV-visible region with modest intensity, potentially influencing its photophysical behaviour in biological environments or contributing to noncovalent π–π* stacking interactions in the protein binding pocket. The DFT- and TDDFT-based Jaguar computations indicate that Tucatinib exhibits greater polarity, stronger solvation energetics, and sharper ESP features, contributing to its high membrane permeability and oral bioavailability (Fig. 5)69,70. In contrast, 2-Aoeobenoxmide, while electronically more uniform and less polar, demonstrates good structural stability and lower electronic reactivity, with sufficient polar sites for hydrogen bonding and a UV-active profile suitable for biochemical applications. These quantum mechanical characteristics reinforce the idea that 2-Aoeobenoxmide may be a pharmacologically safer compound with selective binding capabilities, although its pharmacokinetic limitations, such as solubility and membrane permeability, may require optimisation for clinical translation. The detailed Jaguar/DFT and TDDFT parameter settings (SCF treatment, convergence thresholds, iteration limits, and Hessian initialisation) are in the Supplementary sheets.Fig. 5The Density Functional Theory (DFT) results for 2-Aoeobenoxmide and Tucatinib, along with the legend, are provided for better understanding and comparison. The Density Functional Theory (DFT) results for 2-Aoeobenoxmide and Tucatinib, along with the legend, are provided for better understanding and comparison. Analysis of WaterMap results and control comparison The WaterMap results59 provide a comprehensive comparative analysis of hydration-site displacement and protein–ligand interactions for the identified candidate compound, 2-Aoeobenoxmide, and the reference drug, Tucatinib. Figure 6 displays the 3D and 2D binding poses, highlighting displaced hydration sites as spheres of varying colours and sizes mapped onto the binding cleft. These spheres represent individual water molecules predicted to be energetically unfavourable and displaced upon ligand binding59; the displacement of such high-energy hydration sites favours the ligand’s binding free energy. The accompanying 2D interaction diagrams beneath each 3D structure delineate the specific residues involved in ligand interaction, annotated with detailed legends that denote interaction types, including hydrogen bonds, hydrophobic contacts, polar contacts, water bridges, pi-stacking, and salt bridges, as well as the displaced hydration sites marked with red Xs. In the 2Aoeobenoxmide–1A52 complex (Aa), the binding pocket is densely occupied by displaced hydration sites, particularly within a hydrophobic cleft, with key interactions involving residues such as Val108, Leu103, Phe101, and Thr98. The 2D interaction map shows that the ligand forms multiple hydrogen bonds with Thr98 and polar contacts with Gln104 and His105. Pi-stacking is evident with Phe101. In contrast, the Tucatinib–1A52 complex (Ab) shows a relatively sparser displacement of hydration sites. While it also forms hydrogen bonds with Thr98 and polar contacts with Gln104, the extent of interaction and the disruption of the hydration site are visibly lower, indicating weaker water-entropy-driven binding energetics (Fig. 6). The 3PP0 complexes further exemplify this contrast. In Ba, 2-Aoeobenoxmide forms tight interactions with residues, including Tyr200, Ser202, and Arg195, displacing numerous hydration spheres. A salt bridge is formed with Arg195, and hydrogen bonding with Ser202 stabilises the ligand orientation. The 2D diagram maps these clearly and shows a distinct cluster of red X-marked hydration sites expelled from the active site. Conversely, Bb (Tucatinib–3PP0) reveals weaker displacement and fewer electrostatic interactions, predominantly hydrogen bonds with Tyr200 and a single polar interaction with Asp196, which may limit its binding efficiency from a thermodynamic standpoint (Fig. 6). In the 4EJN complexes, Ca shows that 2-Aoeobenoxmide occupies the central cavity of the binding pocket, displacing several hydration sites and forming multiple hydrogen bonds with residues Ser135, Glu138, and Lys140. The 2D diagram illustrates a well-coordinated water bridge and π-π interactions with Phe132. Tucatinib, in Cb, binds similarly but shows less extensive hydrogen bonding and no salt bridges. The hydration site profile in Cb is less disrupted, indicating a less favourable entropic gain upon binding than Ca (Fig. 6). The interaction of 2-Aoeobenoxmide with 4I23 (Da) shows extensive hydration displacement, especially in the lower cleft region near Arg87, Lys90, and Glu91 residues. The ligand forms strong electrostatic interactions, including a salt bridge with Glu91 and hydrogen bonds with Arg87. The 2D visualisation captures several water molecules displaced from this electrostatically charged pocket. In contrast, Db reveals that Tucatinib fails to occupy this cleft effectively, forming fewer hydrogen bonds and only weak hydrophobic contacts, with significantly fewer displaced hydration sites, suggesting suboptimal binding thermodynamics (Fig. 6). Finally, in the 7R9V complexes, Ea shows that 2-Aoeobenoxmide forms a stable complex with multiple hydration sites expelled from the binding pocket near key residues such as Lys245, Glu249, and Tyr251. Hydrogen bonds and a well-defined pi-cation interaction with Lys245 are observed, contributing to ligand stabilisation. The 2D interaction plot shows a broad region of hydration disruption, suggesting a favourable entropic contribution. Tucatinib, in Eb, binds more superficially and displaces fewer hydration spheres. Though hydrogen bonding with Glu249 and Tyr251 is present, the interaction profile and the hydration site disruption are not as pronounced as with 2-Aoeobenoxmide (Fig. 6). The colour-coded legend in the 2D plots provides detailed insight into the interaction types: charged residues (positive in deep blue, negative in red), polar residues (light purple), glycine (grey), and hydrophobic residues (tan). Displaced hydration sites are denoted with red Xs, and the types of interactions are represented with dotted lines—green for hydrogen bonds, pink for salt bridges, and orange for pi-stacking. The black contour represents the ligand’s orientation, and the surrounding circle of residues emphasises the pocket’s physicochemical environment. All results show that 2-Aoeobenoxmide consistently demonstrates superior hydration-site displacement across all five protein–ligand complexes and more diverse, robust binding interactions than Tucatinib; quantitatively, 2-Aoeobenoxmide displaces, on average, two- to three-fold more unfavourable hydration sites per complex than Tucatinib, with associated ΔG contributions of the order of 2–5 kcal/mol per displaced site (full per-site data are provided in the Supplementary Sheets). This suggests that the newly identified compound may have a thermodynamically more favourable binding profile, driven by both enthalpic interactions and significant entropic gains from water displacement. These WaterMap results strongly support the candidate’s potential as a multitargeted therapeutic agent with high binding affinity and specificity. Furthermore, details are provided in Supplementary Sheets for Site, Entry ID, Occupancy, Overlap, dH, -TdS, dG, #HB(WW), #HB(PW), and #HB(LW).Fig. 6Showing the WaterMap Results in 3D cartoon view and 2D as Ligand Interaction Diagram of 2-Aoeobenoxmide PDB IDs in complex with (Aa) 1A52, (Ba) 3PP0, (Ca) 4EJN, (Da) 4I23, and (Ea) 7R9V, and Tucatinib in complex with (Ab) 1A52, (Bb) 3PP0, (Cb) 4EJN, (Db) 4I23, and (Eb) 7R9V for comparative understanding. Showing the WaterMap Results in 3D cartoon view and 2D as Ligand Interaction Diagram of 2-Aoeobenoxmide PDB IDs in complex with (Aa) 1A52, (Ba) 3PP0, (Ca) 4EJN, (Da) 4I23, and (Ea) 7R9V, and Tucatinib in complex with (Ab) 1A52, (Bb) 3PP0, (Cb) 4EJN, (Db) 4I23, and (Eb) 7R9V for comparative understanding. Analysis of MD simulation trajectories and control comparison We performed 100 ns MD simulations for each protein–ligand complex (5 targets × 2 ligands) using the TIP3P water model62, which is widely adopted in computational drug-discovery studies for trajectories of this length29,39,60. For each of the ten complexes (five targets × two ligands), the mean ± standard deviation of the protein backbone RMSD, ligand RMSD, RMSF at ligand-contacting residues and MM/GBSA ΔG_bind were computed across the 1000 frames sampled uniformly every 100 ps from the 100 ns trajectories. These summary statistics are provided in full in the Supplementary Sheets and are also referenced in the corresponding result subsections, so that the comparative interpretation of the trajectories is supported by explicit quantitative measures rather than by descriptive statements alone. Analysis of root mean square deviation The Root Mean Square Deviation (RMSD) results for the 100 ns MD simulations reveal a comparative evaluation of the structural stability and conformational integrity of the five protein–ligand complexes involving the identified candidate compound 2-Aoeobenoxmide and the reference drug Tucatinib; RMSD is a standard quantitative metric for monitoring structural stability across MD trajectories60,62. Each frame in the plots corresponds to 100 ps, and the trajectories were analysed to monitor deviations in the protein backbone (orange), alpha-carbon atoms (Prot_CA, red), side chains (Prot_sidechain, green), heavy atoms of the protein (Prot_allHeavy, blue), and the ligand in terms of both its all-heavy atoms (Lig_allHeavy, purple) and its RMS fitting to the initial conformation (Lig_RMSFit, brown). For the 2-Aoeobenoxmide–1A52 complex (Aa), the protein backbone, alpha-carbons, side chains, and heavy atoms maintain relatively stable RMSD values between 2.5 and 3.5 Å throughout the majority of the simulation period, with no erratic deviations post-initial equilibration. The ligand heavy-atom deviation stabilises around 2.5–3 Å after 100 frames (10 ns)—a range typically considered indicative of a stable docked pose for protein–ligand complexes60,66—and the Lig_RMSFit line remains consistently below 1 Å for the rest of the trajectory, indicating tight fitting and minimal conformational drift from the original docked orientation. In contrast, in the Tucatinib–1A52 complex (Ab), while the protein RMSD traces behave similarly to Aa, the ligand heavy-atom deviation reaches 6–7 Å, indicating significant movement within the binding site. The Lig_RMSFit, although relatively stable at around 1.5 Å, is consistently higher than that observed for 2-Aoeobenoxmide, implying a less stable ligand conformation throughout the simulation. For the 2-Aoeobenoxmide–3PP0 system (Ba), protein-related deviations hover between 2.5 and 4.0 Å with acceptable fluctuation margins. However, several spikes in the Lig_allHeavy line above 5 Å are observed, possibly indicating transient conformational shifts or rotation within the binding pocket. Despite this, the Lig_RMSFit remains below 2 Å throughout, suggesting that the ligand retains its core orientation and is not fully displaced. Conversely, in the Tucatinib–3PP0 complex (Bb), all protein RMSD trajectories are noticeably higher, climbing steadily above 4 Å and approaching 5 Å in the side chain and heavy atom deviations. The ligand also demonstrates a consistent deviation exceeding 6 Å for Lig_allHeavy, and the Lig_RMSFit line remains closer to 2 Å, indicative of a much less stable protein-ligand interaction than the candidate compound. In the 4EJN complex with 2-Aoeobenoxmide (Ca), the protein trajectories are tightly grouped with Prot_CA, Backbone, and side chains ranging between 2.5 and 3.5 Å. The ligand heavy atom deviation remains stable at 2.5 Å, and the Lig_RMSFit remains well below 1.5 Å, indicating excellent conformational retention and binding pose consistency. The Tucatinib–4EJN system (Cb) shows moderate protein deviations up to 3.5 Å, but the Lig_allHeavy trace is unstable, peaking at over 4 Å in several regions. Although Lig_RMSFit stays under 2 Å, it is not as consistent or low as observed for 2-Aoeobenoxmide, indicating a comparatively weaker or less stable binding orientation. The results from the 4I23 complex simulations present a stark contrast. In Da, 2-Aoeobenoxmide’s protein RMSD values fluctuate mildly around 3.5 Å with well-contained deviations post-equilibration. Lig_allHeavy values oscillate between 2.5 and 3.5 Å, but the Lig_RMSFit is consistently below 1.5 Å, signifying that the ligand remains firmly positioned in the binding site. In comparison, the Tucatinib–4I23 system (Db) shows comparable protein deviation patterns, but the ligand all-heavy atom deviation fluctuates significantly between 3.5 and 5.5 Å, with Lig_RMSFit frequently approaching or exceeding 2 Å, denoting weaker stability and possible minor reorientations or exits from the optimal binding pocket. Finally, the RMSD plots for 2-Aoeobenoxmide in complex with 7R9V (Ea) show reasonably stable protein trajectories between 2.5 and 3.5 Å, though some minor fluctuation peaks are observed between frames 300–600. The ligand’s heavy atom deviation is slightly noisy but remains under 3.5 Å, and Lig_RMSFit remains below 1.5 Å throughout. In contrast, the Tucatinib–7R9V complex (Eb) shows elevated protein fluctuations with side chains and heavy atoms often exceeding 4 Å, and Lig_allHeavy experiences persistent high deviations nearing 5 Å. The Lig_RMSFit values hover around 2 Å or higher, reflecting reduced binding stability and poorer pose retention than the candidate compound. In all results, 2-Aoeobenoxmide shows consistent, moderate RMSD values for both protein and ligand components after the initial equilibration phase. Protein deviations typically remain under 4 Å, with ligands mostly stable below 3.5 Å and Lig_RMSFit values comfortably under 1.5 Å, confirming its favourable stability and binding retention. In contrast, Tucatinib often shows greater deviations in both ligand and protein RMSD profiles, with Lig_allHeavy often exceeding 5 Å and Lig_RMSFit values approaching or exceeding 2 Å, suggesting weaker or unstable interactions within the binding pockets. Further, in Fig. 7, we show a detailed view of the RMSD results for all proteins with both compounds.Fig. 7Showing the Root Mean Square Deviation (RMSD) results from MD Simulation for 2-Aoeobenoxmide in complex with PDB IDs (Aa) 1A52, (Ba) 3PP0, (Ca) 4EJN, (Da) 4I23, and (Ea) 7R9V, and Tucatinib in complex with (Ab) 1A52, (Bb) 3PP0, (Cb) 4EJN, (Db) 4I23, and (Eb) 7R9V for comparative understanding. The figure shows the various components of the protein and ligand, plotted in different colours. Showing the Root Mean Square Deviation (RMSD) results from MD Simulation for 2-Aoeobenoxmide in complex with PDB IDs (Aa) 1A52, (Ba) 3PP0, (Ca) 4EJN, (Da) 4I23, and (Ea) 7R9V, and Tucatinib in complex with (Ab) 1A52, (Bb) 3PP0, (Cb) 4EJN, (Db) 4I23, and (Eb) 7R9V for comparative understanding. The figure shows the various components of the protein and ligand, plotted in different colours. Analysis of root mean square fluctuations The Root Mean Square Fluctuation (RMSF) plots provide a residue-level understanding of the structural flexibility of protein–ligand complexes over the 100 ns MD simulations; RMSF is a widely used metric for interpreting protein flexibility and the local rigidifying effect of bound ligands60,62. RMSF is a quantitative measure of the deviation of each residue from its average position, helping to distinguish flexible loop regions from more stable structural cores such as α-helices and β-sheets. The plots incorporate colour-coded RMSF profiles for different atomic types: orange for backbone atoms, red for Cα atoms, green for side chains, and blue for total heavy atoms. Superimposed vertical red bands indicate residues that interact directly with the ligand throughout the simulation. In the case of the 1A52 complex, the RMSF profile of the 2-Aoeobenoxmide-bound structure (top-left panel) reveals low to moderate flexibility across most residues, generally remaining below 3 Å. The terminal residues—especially those at the C-terminus—exhibit greater fluctuations, reaching approximately 10 Å, a common trait due to the absence of secondary-structure constraints in terminal regions. Most notably, the residues marked in red directly interact with the ligand and display low RMSF values, indicating that these regions remain conformationally stable throughout the simulation. This is a strong indication of stable ligand engagement. Conversely, the Tucatinib-bound structure (top-right panel) shows slightly increased RMSF values across the binding region. Several interacting residues display fluctuations between 3 and 6 Å, suggesting a less consistent or weaker binding affinity, which could contribute to reduced stabilisation of the binding site. For the 3PP0 complex, the RMSF profile with 2-Aoeobenoxmide (second row, left) demonstrates overall fluctuations below 4 Å. Ligand-contacting residues in this complex are again well constrained, reflecting their limited movement due to persistent interactions with the ligand. The observed peaks are generally associated with surface-exposed loops or linker regions that are not directly involved in the ligand-binding interface. In contrast, the Tucatinib-bound complex (second row, right) exhibits broader peaks across several segments in the 3–5 Å range, with some ligand-interacting residues experiencing noticeable mobility. This variation implies that Tucatinib may induce more dynamic behaviour in the protein or fail to adequately rigidify its binding site, leading to potential instability in complex formation. The 4EJN complex, shown in the third row, follows a similar trend. In the presence of 2-Aoeobenoxmide (left panel), the RMSF curve remains tightly regulated, with fluctuations rarely exceeding 2.5 Å for the interacting residues. Most of the dynamic regions appear to lie outside the ligand-interaction zone, indicating that ligand binding stabilises the core functional residues. On the other hand, the Tucatinib-bound complex (right panel) features multiple spikes above 4 Å, with ligand-contacting residues in several cases falling within these peaks. This variability suggests that Tucatinib’s binding may not be as strong or allow for greater conformational breathing at the interaction interface, potentially compromising the structural integrity of the binding site. The fourth-row panels display the 4I23 complex. With 2-Aoeobenoxmide (left), the RMSF values show a stable trend across the binding domain, remaining below 2.5 Å for most residues, with the highest fluctuations (~12 Å) restricted to the far C-terminal tail. This high flexibility at the terminus is expected and does not impact the binding site. Importantly, ligand-binding residues are well-conserved in motion, indicating consistent and strong interaction with the ligand. However, the Tucatinib-bound structure (right) exhibits higher RMSF values across the sequence, including regions of direct ligand contact, which fluctuate significantly from 4 to 7 Å. These findings indicate a lack of rigid binding and potential instability in the protein-ligand association. Finally, in the 7R9V complex (bottom row), the RMSF curve for the 2-Aoeobenoxmide-bound form (left) maintains low fluctuation values (<3.5 Å) throughout, with minimal spikes. Crucially, the ligand-binding residues, as marked by the vertical red lines, show remarkably low flexibility, underlining the stabilising influence of the ligand on the protein’s structural dynamics. This stabilisation is indicative of high-affinity, persistent binding. By contrast, the Tucatinib-bound form (right) shows sharp, erratic RMSF peaks across many regions, including the binding residues. Several of these exceed 5 Å, suggesting instability and possible conformational disruptions induced by the ligand. Across all five protein-ligand complexes analysed, 2-Aoeobenoxmide demonstrates superior binding stability as evidenced by low RMSF values at ligand-interacting residues. Tucatinib, on the other hand, fails to maintain such stability, with elevated RMSF values in interacting regions suggesting more transient or less effective binding. Further, in Fig. 8, we show a detailed view of the RMSF results for both compound complexes.Fig. 8Showing the Root Mean Square Fluctuation (RMSF) results from MD Simulation for 2-Aoeobenoxmide in complex with PDB IDs (Aa) 1A52, (Ba) 3PP0, (Ca) 4EJN, (Da) 4I23, and (Ea) 7R9V, and Tucatinib in complex with (Ab) 1A52, (Bb) 3PP0, (Cb) 4EJN, (Db) 4I23, and (Eb) 7R9V for comparative understanding. The figure shows the various components of the protein and ligand in different colours, with red lines indicating ligand interactions with protein residues. Showing the Root Mean Square Fluctuation (RMSF) results from MD Simulation for 2-Aoeobenoxmide in complex with PDB IDs (Aa) 1A52, (Ba) 3PP0, (Ca) 4EJN, (Da) 4I23, and (Ea) 7R9V, and Tucatinib in complex with (Ab) 1A52, (Bb) 3PP0, (Cb) 4EJN, (Db) 4I23, and (Eb) 7R9V for comparative understanding. The figure shows the various components of the protein and ligand in different colours, with red lines indicating ligand interactions with protein residues. Analysis of simulation interaction diagram To evaluate the binding efficiency and interaction dynamics of the repurposed drug candidate 2-Aoeobenoxmide (ZINC000000134008) compared to the FDA-approved control drug Tucatinib, MD simulation interaction diagrams were generated for five key breast cancer targets. The Simulation Interaction Diagram (SID)60 results reveal intricate hydrogen-bonding networks, water-mediated interactions and non-covalent contacts such as π–π stacking and π-cation interactions. The Human Estrogen Receptor Alpha (PDB ID: 1A52) in complex with Tucatinib establishes multiple hydrogen bonds. Key residues involved include GLU380, which forms hydrogen bonds via water bridges to N+H atoms, and THR347 and ASP351, which interact with NH atoms. Additional water-mediated contacts were observed with GLU353, LEU387, and ARG394. Notably, π-π stacking interactions were established between TRP383 and PHE404 with two aromatic rings of Tucatinib. In contrast, 2-Aoeobenoxmide exhibited a broader interaction profile. Hydrogen bonding was prominent with residues LEU346, LEU349, GLU353, PHE404, and THR347, including water-mediated bridges. Residues such as LYS529, LEU403, ARG394, and LEU387 were also involved in water interactions, while TRP383 formed hydrogen bonds with two oxygen atoms of the ligand, suggesting a tighter and more distributed binding conformation. HER2 Kinase Domain (PDB ID: 3PP0) in complex with Tucatinib revealed hydrogen bonding primarily with LEU726 and ASP808, mediated by water, to the NH atoms. Additional contacts included ASP863, PHE864, LYS753, ILE767, SER783, THR862, CYS805, and MET801, with the latter interacting with four nitrogen atoms of the ligand, indicating robust anchoring. On the other hand, 2-Aoeobenoxmide interacted through hydrogen bonds involving ASN850, SER728, ARG849, ASP863, and LEU796, supported by two NH2 groups. Four oxygen atoms of the ligand interacted with MET801, LYS753, and ASP863. With LEU726, SER783, and ASP808, water-mediated bonds were also observed, demonstrating a distributed and stabilised interaction profile similar to in-depth but more diverse than Tucatinib. AKT1 Kinase (PDB ID: 4EJN) in complex with Tucatinib engaged with LYS268 through hydrogen bonding via an NH atom. Residues TYR272, SER205, THR211, ASP292, ARG273, TYR326, and ASP331 participated in water-mediated contacts. Moreover, five nitrogen atoms of the ligand were involved in hydrogen bonding with ASN54 and GLN79, and the oxygen atom was involved in hydrogen bonding with ASP292. Importantly, ARG273 participated in a π-cation interaction, while TRP80 and TYR272 engaged in π-π stacking. The AKT1–2-Aoeobenoxmide complex showed extensive hydrogen bonding involving THR82, ILE290, ASP292, and TYR272, mediated by NH2 groups. Residues such as TRP80, ARG273, GLN79, VAL271, and THR211 contributed to water-mediated stabilisation, while three ligand oxygen atoms engaged LYS268. A π-π stacking interaction was also observed between TRP80 and the aromatic ring of 2-Aoeobenoxmide, reinforcing binding affinity. EGFR Tyrosine Kinase (PDB ID: 4I23) in complex with Tucatinib, hydrogen bonding was mainly mediated by ASP837 via N+H, with ASN842, ASP855, SER720, ARG841, and CYS797 involved through water bridges. Four nitrogen atoms of Tucatinib contacted MET793, and two oxygen atoms interacted with THR854 and GLY721. Additionally, PHE723 and LYS745 formed pi-cation interactions with the ligand’s aromatic ring and N+H group, respectively. 2-Aoeobenoxmide exhibited enhanced hydrogen bonding with SER720, ASP800, LEU718, and PRO794 through NH2 groups. Its four oxygen atoms were engaged with LYS716, GLN791, ARG841, ASP800, and ASP855 via water-mediated interactions. Other key contacts included LYS728 and MET793, which together support a stable and multifaceted binding mode. PIK3CA (PDB ID: 7R9V) in complex with Tucatinib’s binding to PIK3CA involved hydrogen bonds with VAL851, SER854, and ASP933, and water-mediated interactions with TYR836, SER773, LYS802, ASP810, and ALA775 via nitrogen atoms. 2-Aoeobenoxmide formed hydrogen bonds with SER854, SER919, SER774, ASP933, and GLU793, coordinated through two NH2 atoms. Water-mediated interactions were observed with ARG770, VAL851, THR856, LYS802, and GLU798 via three oxygen atoms. Additionally, two π- π stacking interactions were observed between TRP780 and TYR836, indicating aromatic stabilisation. The SID analysis highlights that 2-Aoeobenoxmide exhibits comparable or enhanced interaction profiles relative to Tucatinib across all five protein targets. These findings suggest that 2-Aoeobenoxmide may offer comparable or potentially superior stability and binding specificity, warranting further investigation as a potential multitargeted therapeutic candidate for breast cancer. Further, in Fig. 9, we have shown a detailed view of the Simulation Interaction Diagram and its histogram representations.Fig. 9Showing the Simulation Interaction Diagram (SID) results from MD Simulation highlighting the interactions and count of interaction as histogram for 2-Aoeobenoxmide in complex with PDB IDs (A) 1A52, (B) 3PP0, (C) 4EJN, (D) 4I23, and (E) 7R9V, and Tucatinib in complex with (F) 1A52, (G) 3PP0, (H) 4EJN, (I) 4I23, and (J) 7R9V for comparative understanding. Showing the Simulation Interaction Diagram (SID) results from MD Simulation highlighting the interactions and count of interaction as histogram for 2-Aoeobenoxmide in complex with PDB IDs (A) 1A52, (B) 3PP0, (C) 4EJN, (D) 4I23, and (E) 7R9V, and Tucatinib in complex with (F) 1A52, (G) 3PP0, (H) 4EJN, (I) 4I23, and (J) 7R9V for comparative understanding. Analysis of binding free energy computation and control comparison The molecular mechanics/generalised Born surface area (MM-GBSA) calculations64–66 performed on the MD simulation trajectories offer comprehensive insights into the binding energetics and complex stability of the control drug Tucatinib and the identified drug candidate 2-Aoeobenoxmide (ZINC000000134008). It is noted, however, that MM-GBSA is an end-point approximation: it neglects explicit conformational entropy and depends on the chosen solvation model and pose set; the reported energies should therefore be interpreted comparatively rather than as absolute affinities64,65. across five breast cancer-related target proteins: Human Estrogen Receptor Alpha (PDB ID: 1A52), the kinase domain of HER2 (PDB ID: 3PP0), AKT1 (PDB ID: 4EJN), Epidermal Growth Factor Receptor (EGFR, PDB ID: 4I23), and PIK3CA (PDB ID: 7R9V). The MM-GBSA results are visualised in two subplots: Panel A shows the binding free energy across the MD trajectory for all protein-ligand (P-L) complexes, while Panel B depicts the total complex energy for each complex throughout the simulation. In Panel A, the binding free energy trajectories show the dynamic interaction strength of each ligand–protein complex over 1000 uniformly sampled frames, every 100 ps, from the 100 ns trajectory; MM-GBSA energies were computed on the full 1000-frame ensemble and averaged to obtain the reported values. For all targets, both Tucatinib and 2-Aoeobenoxmide exhibit consistently negative binding free energy values, indicating thermodynamically favourable binding events65,66. However, notable differences in magnitude and stability exist between the two ligands. Among all complexes, 2-Aoeobenoxmide bound to AKT1 (4EJN_ZINC000000134008) demonstrated the most negative and stable binding free energy profile, with values often reaching or exceeding -100 kcal/mol. This strong, energetic performance suggests a highly stable and favourable interaction between the ligand and the active site of AKT1, surpassing the control drug’s corresponding complex, which showed a less negative (less stabilising) and more fluctuating energy profile. Similarly, the Estrogen Receptor complex with 2-Aoeobenoxmide (1A52_ZINC000000134008) also demonstrated stronger binding energy than its Tucatinib counterpart. Its trajectory was more tightly clustered around −85 to −90 kcal/mol, highlighting stability and strong affinity. In most frames, the HER2 kinase domain (3PP0) complexed with 2-Aoeobenoxmide showed moderately favourable binding energies comparable to or slightly better than the Tucatinib complex. Although both exhibited fluctuations, the repurposed ligand maintained consistent trends with fewer abrupt spikes in binding energy, indicating a better accommodation within the HER2 binding pocket. In the case of the EGFR (4I23) complexes, 2-Aoeobenoxmide again showed slightly superior performance in lower average binding energy and reduced variability throughout the simulation. The difference in binding profiles suggests that the repurposed compound engages more favourably with the EGFR binding site and maintains stable contacts throughout the simulation. Likewise, for the PIK3CA catalytic domain (7R9V), the ZINC compound exhibited lower energy values, typically ranging from -65 to -75 kcal/mol, compared to Tucatinib, which hovered around -50 to -60 kcal/mol. The greater negative energy values observed for 2-Aoeobenoxmide suggest enhanced electrostatic and hydrophobic contributions stabilising the ligand within the PIK3CA binding pocket. Panel B supports these findings by showing the total complex energy of all simulated systems over time. Across all protein targets, the total complex energies for 2-Aoeobenoxmide were consistently lower than those of Tucatinib, indicating better overall system stability when bound to the identified ligand. The AKT1 complex (4EJN_ZINC000000134008) again stood out with the lowest total energy, hovering well below -25,000 kcal/mol and remaining tightly grouped over the 1000 frames, signifying a highly stable interaction landscape. Tucatinib’s complex with the same target showed a higher total energy and slightly more variability, reinforcing the superior stability profile of the repurposed ligand. For the Estrogen Receptor (1A52), EGFR (4I23), and PIK3CA (7R9V), 2-Aoeobenoxmide-bound complexes maintained lower energy baselines than those of Tucatinib, suggesting that the presence of this ligand contributed to a more energetically favourable structural environment for the protein-ligand interaction. The HER2 (3PP0) complexes demonstrated similar trends, though the difference in total energy was slightly narrower than in the other cases. Further details are available in the Supplementary Sheets for all energies. Table 3 provides a better representation of the key energies generated, and Fig. 10 shows the binding-free and total complex energies from MM-GBSA computations. These MM-GBSA results derived from MD trajectories indicate that 2-Aoeobenoxmide forms energetically more favourable and structurally more stable complexes with all five breast cancer-related targets than the FDA-approved control drug, Tucatinib. The sustained negative binding free energies, lower total complex energies, and reduced fluctuation patterns highlight the repurposed ligand’s potential as a promising multitargeted candidate. Its particularly strong performance with AKT1 and the Estrogen Receptor suggests it may significantly modulate key signalling pathways in breast cancer progression. These findings lay the groundwork for further validation through experimental assays or extended in vivo studies.Table 3Molecular Mechanics/Generalised Born Surface Area (MM-GBSA) calculations result for all protein-ligand complexes.PDB ID_LiganddG AveragedG Std DevdG RangedG(NS) AveragedG(NS) Std DevdG(NS) Range1A52_Tucatinib –77.839.24 –102.8153 to –54.7751 –81.129.09 –105.6683 to –58.06493PP0_Tucatinib –80.363.96 –92.0752 to –67.4905 –82.813.92 –94.8367 to –70.32464EJN_Tucatinib –67.154.99 –91.3444 to –49.5038 –69.605.06 –93.6836 to –52.44174I23_Tucatinib –80.234.27 –91.4356 to –58.3183 –84.824.38 –96.1106 to –63.40207R9V_Tucatinib –45.684.65 –65.7577 to –30.5572 –48.874.99 –68.6182 to –32.54081A52_2-Aoeobenoxmide –68.505.25 –81.7815 to –50.9230 –71.085.28 –84.9365 to –54.07743PP0_2-Aoeobenoxmide –61.814.15 –74.5897 to –48.8599 –64.734.36 –77.3668 to –53.22064EJN_2-Aoeobenoxmide –64.355.71 –82.0637 to –49.3007 –66.835.43 –84.0397 to –52.56544I23_2-Aoeobenoxmide –52.494.28 –72.7083 to –39.3807 –54.764.47 –74.4158 to –41.08337R9V_2-Aoeobenoxmide –58.315.62 –76.4789 to –40.4065 –60.535.71 –79.3097 to –43.1854Fig. 10Showing the Molecular Mechanics/Generalised Born Surface Area (MM-GBSA) calculations on 100ns MD Simulation results for (A) Binding Free Energy and (B) total complex energy of the Protein-ligand complex of 2-Aoeobenoxmide and Tucatinib with PDB IDs 1A52, 3PP0, 4EJN, 4I23, and 7R9V. Table 3Molecular Mechanics/Generalised Born Surface Area (MM-GBSA) calculations result for all protein-ligand complexes.PDB ID_LiganddG AveragedG Std DevdG RangedG(NS) AveragedG(NS) Std DevdG(NS) Range1A52_Tucatinib –77.839.24 –102.8153 to –54.7751 –81.129.09 –105.6683 to –58.06493PP0_Tucatinib –80.363.96 –92.0752 to –67.4905 –82.813.92 –94.8367 to –70.32464EJN_Tucatinib –67.154.99 –91.3444 to –49.5038 –69.605.06 –93.6836 to –52.44174I23_Tucatinib –80.234.27 –91.4356 to –58.3183 –84.824.38 –96.1106 to –63.40207R9V_Tucatinib –45.684.65 –65.7577 to –30.5572 –48.874.99 –68.6182 to –32.54081A52_2-Aoeobenoxmide –68.505.25 –81.7815 to –50.9230 –71.085.28 –84.9365 to –54.07743PP0_2-Aoeobenoxmide –61.814.15 –74.5897 to –48.8599 –64.734.36 –77.3668 to –53.22064EJN_2-Aoeobenoxmide –64.355.71 –82.0637 to –49.3007 –66.835.43 –84.0397 to –52.56544I23_2-Aoeobenoxmide –52.494.28 –72.7083 to –39.3807 –54.764.47 –74.4158 to –41.08337R9V_2-Aoeobenoxmide –58.315.62 –76.4789 to –40.4065 –60.535.71 –79.3097 to –43.1854 Molecular Mechanics/Generalised Born Surface Area (MM-GBSA) calculations result for all protein-ligand complexes. Showing the Molecular Mechanics/Generalised Born Surface Area (MM-GBSA) calculations on 100ns MD Simulation results for (A) Binding Free Energy and (B) total complex energy of the Protein-ligand complex of 2-Aoeobenoxmide and Tucatinib with PDB IDs 1A52, 3PP0, 4EJN, 4I23, and 7R9V. Discussion Breast cancer continues to represent a global health burden, ranking as the most frequently diagnosed cancer among women and a leading cause of cancer-related mortality. As reported by the World Health Organisation and the most recent GLOBOCAN 2022 estimates1,2, the disease accounts for an estimated 685,000 deaths annually, and its incidence is rising due to factors such as genetic susceptibility, lifestyle modifications and environmental exposures. Despite the development of sophisticated diagnostic tools and targeted treatments—particularly those aimed at HER2-positive subtypes—therapeutic resistance, drug-induced toxicity, and limited efficacy in heterogeneous disease settings remain pressing challenges. In this context, drug repurposing emerges as a practical and time-efficient strategy to uncover novel applications for known compounds, reducing the cost and duration of the drug development pipeline while minimising safety concerns10,39. This study leveraged a comprehensive in silico pipeline to evaluate the anti-breast cancer potential of the novel compound 2-Aoeobenoxmide (ZINC000000134008) compared with the FDA-approved drug Tucatinib, which is currently used to treat HER2-positive metastatic breast cancer. Tucatinib, a tyrosine kinase inhibitor (TKI), selectively targets the HER2 receptor and has demonstrated clinical benefit when combined with trastuzumab and capecitabine. However, resistance to Tucatinib and other HER2-targeted therapies has been frequently reported, particularly due to mutations in downstream effectors such as PIK3CA and AKT1 or upregulation of alternative signalling pathways such as EGFR9,11,13. The computational evaluation incorporated five key breast cancer-associated protein targets—ERα (1A52), HER2 (3PP0), AKT1 (4EJN), EGFR (4I23), and PIK3CA (7R9V)—with energy-minimised structures confirming their suitability for docking4. While Tucatinib displayed potent binding to HER2 (docking score: –10.94 kcal/mol; MM-GBSA: –68.73 kcal/mol), its affinity toward other targets, such as ERα and PIK3CA, was modest. In contrast, 2-Aoeobenoxmide demonstrated improved or comparable in silico binding scores across multiple targets in this study, particularly at ERα (–10.32 vs –4.87 kcal/mol) and AKT1 (–10.00 vs –7.93 kcal/mol), supported by more negative MM-GBSA values; given the single-replicate nature of the docking and MM-GBSA calculations, these differences are reported as relative trends rather than as statistically validated comparisons. These findings suggest that 2-Aoeobenoxmide could serve as a broad-spectrum multitarget inhibitor, addressing the therapeutic limitations of Tucatinib’s HER2-selective mechanism. The interaction fingerprint analysis reinforced this conclusion, revealing an enriched spectrum of stabilising interactions for 2-Aoeobenoxmide, including extensive hydrophobic contacts with leucine residues (26 occurrences), π–π stacking with aromatic residues like PHE and TRP, and strong electrostatic interactions via charged and polar residues such as ASP, GLU, LYS, and ASN. These detailed interaction networks suggest a more deeply embedded and persistent binding conformation, especially in functionally critical regions like the catalytic and regulatory domains of AKT1 and PIK3CA. In contrast, Tucatinib’s interactions, while effective in HER2, were comparatively less diverse and less persistent across other targets. Pharmacokinetic profiling further highlighted the divergent properties of the two compounds. Tucatinib demonstrated excellent oral bioavailability (100%), high membrane permeability, and favourable lipophilicity, consistent with its clinical use as an oral agent. Conversely, 2-Aoeobenoxmide exhibited reduced oral absorption (28.36%) and low permeability, likely due to its higher topological polar surface area (TPSA) and hydrophilic nature. However, it offered significant advantages in aqueous solubility (QPlogS–0.111 vs –7.373 for Tucatinib) and predicted safety, with zero toxicophoric alerts and a lower HERG liability, suggesting potentially reduced cardiotoxicity. While suggesting formulation challenges, these trade-offs open avenues for alternative delivery methods, such as nanoparticle encapsulation or parenteral administration. Quantum-mechanical analyses using DFT and TDDFT calculations provided supportive computational evidence for the in silico characterisation, rather than independent confirmation of biological activity. While both compounds exhibited stable singlet electronic states, 2-Aoeobenoxmide had a slightly wider HOMO-LUMO gap (0.1395 eV vs. 0.1526 eV for Tucatinib), indicative of greater kinetic stability and potentially lower chemical reactivity—a desirable property for reducing off-target effects. ESP mapping and ALIE descriptors further illustrated differential charge distributions and ionisation tendencies, pointing to distinct interaction mechanisms that could be exploited for target-specific modulation. From a pathway-level perspective, the five chosen targets collectively cover the principal Tucatinib-resistance escape routes: HER2 and EGFR represent the receptor tyrosine-kinase entry point; AKT1 and PIK3CA cover the PI3K/AKT/mTOR axis frequently activated upon HER2 inhibition; and ERα represents the parallel hormone-receptor signalling axis. Engaging this combination of targets, even at the in silico level, addresses the multitarget drug-design hypothesis that motivated this study. However, pathway-level validation through phospho-signalling and gene-expression assays in HER2+, ER+ and PI3K-mutant breast cancer cell lines remains an essential next step. WaterMap analysis provided critical insight into the thermodynamic contributions of solvent displacement at protein-binding sites, a feature often overlooked in classical docking studies. Across all five protein-ligand complexes, the repurposed compound exhibited stronger hydrophobic enclosure and displaced more energetically unfavourable water molecules from high-energy hydration sites than Tucatinib. Particularly in the AKT1 (4EJN) and EGFR (4I23) complexes, 2-Aoeobenoxmide replaced several tightly bound, high-entropy water molecules, yielding an energetically favourable binding cavity. This thermodynamic optimisation contributed to the higher binding efficiency observed in downstream MM-GBSA calculations. Moreover, replacing water molecules in buried pockets enhanced the entropy component of ligand binding, thereby stabilising the complexes more effectively than Tucatinib. The MD simulations provided a time-resolved understanding of the protein-ligand dynamics and helped validate the initial binding hypotheses. RMSD trajectories demonstrated that 2-Aoeobenoxmide maintained a more consistent and lower average RMSD across all complexes, indicating better structural stability throughout the 100 ns simulations. Notably, the AKT1 and Estrogen Receptor complexes with 2-Aoeobenoxmide exhibited RMSD stabilisation as early as 10–15 ns and remained stable for the simulation. In contrast, the Tucatinib-bound complexes, while showing acceptable RMSD trends, exhibited more frequent fluctuations and occasional conformational shifts, especially in the HER2 and EGFR systems, suggesting comparatively weaker structural retention during binding. RMSF analysis further supported the stability trends by highlighting residue-wise flexibility. For most complexes, 2-Aoeobenoxmide induced lower fluctuations across key binding site residues. This was especially evident in the PIK3CA (7R9V) complex, where the ligand-bound form exhibited dampened fluctuations in the catalytic loop and αC-helix, indicating a stabilising effect on critical regulatory regions. Tucatinib, in contrast, permitted greater mobility in these regions, potentially limiting its long-term inhibitory effectiveness. The SID further reinforced this trend, which captured the residue-level interaction consistency across time. 2-Aoeobenoxmide established more stable hydrogen bonds, hydrophobic contacts, and π–π stacking interactions, particularly in the AKT1 and EGFR binding sites. The frequency of these interactions remained high throughout the simulation frames, demonstrating robust anchoring within the binding cavities. Tucatinib, while showing decent interaction profiles, lacked the same temporal persistence and showed higher dependence on transient water bridges, particularly in the PIK3CA and HER2 systems. The MM-GBSA binding free energy calculations based on MD trajectories provide a final energetic validation of the superior performance of 2-Aoeobenoxmide. The binding free energy plots reveal that all complexes formed with the repurposed ligand demonstrated more negative ΔG_bind values than those with Tucatinib. The most pronounced difference was observed in the AKT1 (4EJN) complex, where 2-Aoeobenoxmide exhibited binding energies consistently around –100 kcal/mol, far surpassing Tucatinib, which fluctuated between –70 and –80 kcal/mol. Similarly, the Estrogen Receptor complex showed more stable, negative ΔG_bind values for the identified compound, indicating strong electrostatic and van der Waals interactions. Notably, the total complex energy plots further reinforced these results, with the 2-Aoeobenoxmide complexes maintaining significantly lower total system energies across the MD trajectory in all five systems. This energy advantage translates to greater thermodynamic favourability and structural coherence. The multi-faceted computational evidence robustly supports the conclusion that 2-Aoeobenoxmide exhibits more favourable binding behaviour than Tucatinib across key breast cancer targets. This compound effectively displaces high-energy water molecules, as revealed through WaterMap analysis, and contributes to the enhanced structural stability of the target proteins over time, demonstrated by consistent RMSD and RMSF profiles. Additionally, 2-Aoeobenoxmide maintains strong and persistent interactions with critical binding-site residues, as confirmed by SID data, and exhibits a higher thermodynamic binding affinity, as supported by MM-GBSA calculations. These characteristics position 2-Aoeobenoxmide as a highly promising multitarget lead worthy of further in vitro and in vivo validation in breast cancer therapy. Moreover, the convergence of structural, energetic, and dynamic parameters highlights the critical value of integrating diverse computational methodologies in contemporary drug discovery, especially when addressing complex, multifactorial diseases such as breast cancer. This comprehensive in silico investigation underscores the potential of 2-Aoeobenoxmide as a repurposed therapeutic candidate, complementing and potentially surpassing the efficacy of Tucatinib. While Tucatinib remains effective with well-characterised pharmacokinetics and binding profiles, 2-Aoeobenoxmide offers superior docking scores, enriched interaction patterns with key residues, improved aqueous solubility, and a promising safety profile. The integration of quantum mechanical analyses further validates the compound’s molecular stability and binding readiness, bolstering the rationale for its progression toward experimental validation. However, there are notable limitations to consider. Firstly, despite favourable computational predictions, the compound’s predicted oral absorption and membrane permeability are suboptimal, which could limit its bioavailability and therapeutic efficacy in vivo. Secondly, the current study’s reliance on computational models necessitates cautious interpretation, as these predictions require extensive experimental corroboration to account for complex biological variables such as metabolism, off-target effects, and pharmacodynamics. Addressing these limitations through medicinal chemistry optimisation and rigorous biological testing will be essential in advancing 2-Aoeobenoxmide toward clinical applicability. This study exemplifies the power of integrated in silico repurposing strategies to identify novel therapeutic opportunities within existing chemical space, providing a strong foundation for future translational oncology research. Limitations and future work. The present study is computational in nature, and several limitations should be highlighted. First, no experimental in vitro (kinase activity, breast-cancer cell-line viability and downstream phospho-signalling assays) or in vivo (xenograft pharmacokinetics) validation has been performed, and the comparative ranking against Tucatinib should therefore be interpreted as an in silico binding hypothesis rather than as a pharmacological superiority claim. Second, MD simulations were performed as single 100 ns trajectories per complex due to the substantial computational cost of running ten independent 100 ns simulations on the available hardware; the absence of replicate trajectories and of formal statistical testing is acknowledged. Third, docking and MM-GBSA are end-point approximations, and the scoring functions used here have well-known limitations, including pose-dependence and approximate treatment of entropy and solvation71. Fourth, a single positive control (Tucatinib) was retained for clinical relevance, but additional target-specific positive controls and experimental ADME-Tox profiling would further strengthen the comparison. Finally, the predicted low intestinal absorption and permeability of 2-Aoeobenoxmide indicate that, even if its activity were confirmed experimentally, formulation strategies such as prodrug design, lipid-based or nanoparticle delivery, or parenteral administration would likely be required to translate the candidate towards clinical evaluation72. Future work will focus on replicating MD simulations, biochemical and cell-based assays against HER2-, ER- and PI3K-mutant breast cancer models, and experimental ADME-Tox characterisation. Conclusion This computational investigation suggests that 2-Aoeobenoxmide may serve as a promising multi-target therapeutic candidate against breast cancer, with more favourable in silico binding metrics than the FDA-approved drug Tucatinib across multiple molecular and pharmacological parameters. Docking and MM-GBSA binding free energy analyses demonstrated superior affinity of 2-Aoeobenoxmide toward the kinase domains of HER2, AKT1, EGFR, and PIK3CA, with docking scores ranging from −8.162 to −10.327 kcal/mol and MM-GBSA values ranging from −47.18 to −57.62 kcal/mol, significantly better than Tucatinib. Pharmacokinetic predictions via QikProp and electronic-structure descriptors from DFT analyses provided supportive in silico evidence regarding drug-likeness and molecular stability. WaterMap simulations revealed effective displacement of high-energy water molecules, while 100 ns molecular dynamics simulations showed reduced structural deviations and fluctuations alongside sustained key residue interactions, confirming enhanced complex stability. These integrated in silico results highlight 2-Aoeobenoxmide’s potential to overcome resistance by targeting multiple critical proteins in breast cancer progression. Despite these promising findings, limitations such as suboptimal predicted oral absorption and permeability, single-replicate trajectories and the absence of statistical validation must be acknowledged. Importantly, experimental validation through in vitro (kinase activity, breast cancer cell line viability, and downstream phospho-signalling) and in vivo (xenograft pharmacokinetics) studies is essential before any clinical translation of 2-Aoeobenoxmide can be considered. This study underscores the value of integrated computational pipelines in nominating promising candidates for subsequent experimental investigation in challenging oncological contexts. Supplementary Information Supplementary Information. Supplementary Information.