Proteins retrievalThe Epidermal growth factor receptor protein with Uniprot id P00533, Serine/threonine-protein kinase B-raf protein with Uniport id of P15056, Tyrosine-protein kinase receptor UFO with Uniprot id P30530 and Vascular endothelial growth factor receptor 3 protein with Uniprot id P35916 were assessed from Uniprot and further their 3D structures were retrieved from Alphafold. The 3D structures were visualized by Discovery Studio is shown in Figs. 1–4.Figure 1Epidermal growth factor receptor.Figure 2Serine threonine protein kinase braf.Figure 3Tyrosine-protein kinase receptor UFO.Figure 4Vascular endothelial growth factor receptor 3.Phytochemical retrievalThe phytochemical Resveratrol with Puchem id of 445154 was attained from Pubchem and further the 3D structure was downloaded. The 3D structure was visualized by Discovery Studio is shown in (Fig. 5).Figure 53D structure of resverstrol from pubchem.FDA drug targetingThe proteins serine/threonine-protein kinase B-raf and vascular endothelial growth factor receptor 3 were given as an input for analyzing common FDA drug target against these two. The total common compounds were 4 where number of drug bank vs PDB pairs had a threshold of over 3 and common pathways were 2 including 2 Rap1 signaling pathway and focal adhesion pathway. Furthermore, the other two proteins including epidermal growth factor receptor and tyrosine-protein kinase receptor UFO were administered for common FDA drug targets. The number of total common compounds was 1 where number of drug bank vs PDB pairs had a threshold of over 7. There was no common pathway shared amongst these two proteins. The results of common FDA drugs for following proteins with MACCS similarity and ECFP4 similarity score of one is shown in the (Tables 1, 2) below. The results interpretation shows that Fostamatinib is the common FDA drug targeting all four proteins of hepatocellular carcinoma.Table 1 The proteins serine/threonine-protein kinase B-raf and vascular endothelial growth factor receptor 3 common FDA drug targets prediction.Table 2 The proteins epidermal growth factor receptor and tyrosine-protein kinase receptor UFO common FDA drug targets prediction.Fragment optimized resveratrolIn DeepFrag online server the 0:C17 fragment, which is the cysteine residue of Resveratrol, was selected as the optimization point. The results exhibited that optimization of hydroxide ion at the selected position is evidently the right fit for optimization by scoring 1 as displayed in (Fig. 6). The Fig. 7 below show the optimization outcomes and the structures of Resveratrol before and after optimization.Figure 6The different types of optimizations with varying chemical chains and their optimization score is shown.Figure 7The yellow color highlights the point 0:C17 which was optimized in resveratrol in DeepFrag webserver. The structure of resveratrol after the addition of hydroxide is shown as after optimization.ADME and toxicity analysisThe prediction results of ADME and toxicity generated by toxCSM server cover different aspects, which include organic, nuclear response, and genomics as are given in the (Tables 3, 4).Table 3 Toxicity report of Fostamatinib.Table 4 Toxicity report of fragment optimized resveratrol (FOR).General molecular properties comparisonThe molecular properties of weight, volume, and density including draggability properties of hydrogen bond donors, hydrogen bond acceptors and others are compared between Fostamatinib and fragment-optimized resveratrol in (Table 5) below.Table 5 General molecular properties comparison between fostamatinib and fragment optimized resveratrol.Molecular drug-likeness rulesIn order to evaluate the drug-likeness of Fostamatinib and Fragment Optimized Resveratrol (FOR) the molecules were assessed for the rules of Lipinski’s Rule of five, Ghose’s rules, Opera’s Notability rules, Pfizer’s rules, GSK rules and ADMETLab 2.0 Soft rules respectively. The violation of several rules are observed in case of Fostamatinib whereas, in case of Fragment Optimized Resveratrol only one rule of Oprea’s Notability rule is being violated with all other rules being fulfilled completely as shown in (Table 6) below.Table 6 The draggability of fostamatinib and fragment optimized resveratrol were evaluated for different drug-likeness rules.Molecular drug likeliness scoreFor the estimation of the drug-likeness scores of Fostamatinib and Fragment Optimized Resveratrol the molecules were assessed for the parameters of QED scores, SAS score, Fsp3 score and NP score respectively. The Fragment Optimized Resveratrol scores higher for all aspects of its molecular properties against Fostamatinib proving it to be a better candidate for drug (Table 7).Table 7 The molecular drug likeliness score of Fostamatinib and fragment optimized resveratrol were assessed for different properties.ADMET radar chartThe ADMET radar provided with the visual representation of Lipinki’s rule’s characteristics, such as polar surface area, logP, mass, range of atoms, range of OH, range of N or O, and range of rotatable bonds for Fostamatinib and Fragment Optimized Resveratrol. The results show that Fragment Optimized Resveratrol passed the Lipinski’s rule of five as present within the radar boundaries whereas, in case of Fostamatinib violation of Lipinki’s rule of 5 is apparent as the green lines surpassed the boundaries of optimal drug properties as shown in (Fig. 8).Figure 8The ADMET radar with the representation of Lipinki’s rule’s characteristics showing fostamatinib and fragment optimized resveratrol results for following the Lipinski’s rule of 5.Docking analysisThe four main proteins of hepatocellular carcinoma were docked with Resveratrol and Fragment fragment-optimized resveratrol as shown in (Figs. 9–12). The energies were optimized after docking of all four proteins with Fragment Optimized Resveratrol proving its better efficiency as shown in (Table 8) below.Figure 9Docking complex of EGFR_HUMAN with FOR.Figure 10Docking complex of VGFR3_HUMAN with FOR.Figure 11Docking complex of UFO_HUMAN with FOR.Figure 12Docking complex of BRAF_HUMAN with FOR.Table 8 The binding affinities of resveratrol and fragment optimized resveratrol with respective proteins.Molecular dynamic simulation evaluationMolecular dynamics simulations were performed on the top hits containing high binding energies. Over the simulation period, the projected conformational changes from the initial structure were presented in terms of root mean square deviation (RMSD). Moreover, structural stability, atomic mobility, and residue flexibility at times of interaction of protein-hit were expressed with root mean square fluctuation (RMSF) values. The peaks of RMSF graph represent the fluctuation portion of the protein through the simulation. The N- and C-terminal show greater fluctuations overall. Alpha helices and beta strands show high fluctuation as well. All protein frames are first aligned on the reference frame backbone, and then the RMSD is calculated based on the atom selection. Monitoring the RMSD of the protein can give insights into its structural conformation throughout the simulation. The RMSD of FOR and VGFR3_HUMAN complex showed small deviation from almost 2 to 8.5 Å till 10 ns and then it was stable at almost 9 to 10 Å from 10 to 30 ns. After a peak at 16 Å on 40 ns the system remained equilibrated till 90 ns with one peak at 16 Å at almost 95 Å towards the 100 ns frame (Fig. 13). After looking the trajectories, it was found that the both systems were stable and ligands remained inside the binding pockets and made important interactions and the backbones were consistent. Similarly, estimated RMSF values of up to 4 Å indicated high stability of the complex whereas fluctuation changes at almost 5 Å is the highest where ligand made interaction with the receptor shown in green lines (Fig. 14).Figure 13Root mean square deviation plot of disentanglement docked complex of FOR and VGFR3_HUMAN.Figure 14Root mean square fluctuation plot of disentanglement docked complex of FOR and VGFR3_HUMAN.To determine the ligand’s effect on whole protein, the Disentangle docked complex of FOR and VGFR3_HUMAN was investigated. During the 100 ns simulation, six properties were taken into consideration at in order to demonstrate the stabilities of the chosen ligands in the binding pocket. Ligand RMSD, or root mean square deviation, is the measurement of a ligand’s deviation from the reference conformation, which is usually the first frame, which is considered to be time t = 0. (2) Radius of gyration (rGyr): This quantity is equal to a ligand’s primary moment of inertia and is used to gauge how “extended” a ligand is; The quantity of internal hydrogen bonds (HB) in a ligand molecule is known as the intramolecular hydrogen bond (intraHB). (4) Molecular surface area (MolSA): A probe radius of 1.4 Å was used to compute the molecular surface; (5) Polar surface area (PSA): Solvent accessible surface area in a molecule given only by oxygen and nitrogen atoms; (6) Solvent accessible surface area (SASA): Molecular surface area of accessible by a water molecule (Fig. 15).Figure 15Variation in the ligand’s properties in disentanglement docked complex of FOR and VGFR3_HUMAN with respect to time during the course of 100 ns simulation.Ligand RMSDAs shown in Fig. 15 the RMSD of the ligand was 0.5 Å till 10 ns and after that it was stabilized at 0.75 Å till 100 ns.Radius of gyration (rGyr)The radius of gyration measures the ‘extendedness’ of a ligand, and is equivalent to its principal moment of inertia. The rGyr value of ligand was from 4.16 to 4.3 Å at the start and remained constant at almost 4.20 Å till the end. The constant values indicate the steady behaviour in (Fig. 15).Intramolecular hydrogen bonds (intraHB)The number of internal hydrogen bonds (HB) within a ligand molecule showed by intraHB are not found in this particular simulation of the FOR and VGFR3_HUMAN complex as shown in (Fig. 15).Molecular surface area (MolSA)Molecular surface is the calculation with 1.4 Å value of probe radius. This value is equivalent to a van der Waals surface area. The MolSA value was 242 to 244 Ã…2 throughout remining stabilized till the time of 100 ns as shown in (Fig. 15)Solvent accessible surface area (SASA)SASA is the Surface area of a molecule accessible by a water molecule. The value of SASA was almost 600 Ã…2 till the 100 ns of simulation as shown in (Fig. 15). No unusual fluctuations in the graph were observed throughout the simulation representing the steady behavior of ligand.Polar surface area (PSA)The PSA is Solvent accessible surface area in a molecule contributed only by oxygen and nitrogen atoms. The PSA value was 192 to 196 Ã…2 till 100 ns timeframe depicting complete stability throughout as shown in (Fig. 15). Several energy components inside a system can be analyzed to get insight into prevailing forces and reach simulated equilibrium. An easy-to-use interface is provided by the Trajectory Plot tool for configuring and performing these kinds of analysis. It is necessary to have access to a Linux-based host computer with a compatible GPU card in order to perform energy analysis. To investigate the ligand’s internal energy, a Linux system with two GPU cards—the RTX 3070 and 3080—was used in this work. Correlating energy fluctuations with conformational changes and examining the energy exchange between a protein and receptor over the course of a 100 ns simulation were the objectives of this investigation. With an indicated value of −21514 cal/mol, the energy breakdown covering Coulomb (electrostatic) and van der Waals (vdW) interactions demonstrated a significant binding strength (Fig. 16).Figure 16Receptor-ligand complex energy analysis throughout 100 ns simulation.MM/GBSA binding free energy of complexThe Desmond’s thermal_mmgbsa.py script and Molecular Mechanics Generalised Born Surface Area (MM/GBSA) were used to calculate the binding free energies. The following formula is used to compute the MM/GBSA free energy of binding (ΔGbind) for the docked postures and Desmond trajectories:$$\Delta {\text{GBind }} = \, \Delta {\text{GComplex }} – \, \Delta {\text{GLigand }}{-} \, \Delta {\text{GReceptor}}$$where the variables ΔGComplex, ΔGLigand, and ΔGReceptor in MM/GBSA reflect the energy estimations of the optimised complex (complex), optimised free ligand (ligand), and optimised free receptor (receptor).MM/GBSA is unquestionably one of the most popular methods for estimating binding free energy because it is more accurate than the majority of molecular docking score systems. MM/GBSA is widely used in bimolecular research describing, among other things, protein folding, protein-ligand binding, and protein-protein interaction since it uses less processing power than other chemical free energy scoring techniques. One of the most crucial problems in bimolecular investigations is the precise calculation of the free energy, which drives molecular motions. The solvation free energy difference upon binding and the interaction energy between the ligand and receptor complex can be used to compute the binding free energy using the MM/GBSA method.Throughout the post-MM-GBSA computation of 100 ns MD data of both hit ligands identified by the dynamic’s investigations, a total of 1000 frames were processed and examined. Every 10 ns, we performed binding free energy estimations using 11 recorded images. The mean ΔG was discovered to be −70.80 kcal/mol. The energies were displayed in (Fig. 17) every 10 ns.Figure 17MM/GBSA approach to calculate Binding free energy of complex.
