Molecular dockingMolecular docking is a crucial technique in the drug discovery process to do an in-depth examination of the molecular-level interaction between a small molecule and its target protein. This process plays a pivotal role in advancing the development of novel and efficacious medications43. The output files generated through molecular docking contained valuable information, including affinity scores, and docked poses for each compound screened. In the docking screening, it was found that several compounds demonstrated a robust binding affinity to the FGFR2 binding pocket that highlights them as noteworthy contenders for further investigation as potential FGFR2 inhibitors.Hits selection and drug-ability assessmentA molecular docking screening process was conducted on a pool of 2336 compounds, resulting in the identification of 10 hits that displayed a binding affinity score of − 12.4 kcal/mol to − 11.6 kcal/mol with FGFR2. Initially, these compounds showed appreciable docking values, higher than the reference inhibitors. Table 2 showcases the top 10 hits from the docking analysis and the reference inhibitors, along with their respective docking scores. These chosen compounds demonstrated incredible binding potential towards FGFR2, as evidenced by docking scores between −12.4 kcal/mol to −11.6 kcal/mol, with a ligand efficiency of >0.30 kcal/mol/non-H atom. All these selected hits exhibited a superior affinity for FGFR2 compared to the control molecules, Erdafitinib, Infigratinib, Zoligratinib, and AZD4547. This highlights the exceptional binding capabilities of the selected molecules.Table 2 List of selected compounds and their docking parameters with FGFR2.To avoid the Pan-assay interference compounds (PAINS)44 in the top 10 hits obtained from the FGFR2 molecular docking study, additional evaluations were conducted using PAINS filter. Further, the compounds’ physicochemical properties were thoroughly examined using various models to assess their bioavailability and drug-like attributes (Supplementary Table S1). Moreover, we assessed the ADMET for the top 10 hits using the SwissADME and pkCSM web tools. This assessment leads to the identification of one compound, PubChem CID:507883 (1-[7-(1H-benzimidazol-2-yl)-4-fluoro-1H-indol-3-yl]-2-(4-benzoylpiperazin-1-yl)ethane-1,2-dione) that displayed no toxic patterns, and showing better properties than the reference drugs (Supplementary Table S1). Compound CID:23646206 also exhibits similar pharmacokinetic characteristics to CID:507883; however, it was noted to function as an OCT2 substrate. Consequently, it might engage in competitive interactions with other substrates during transport, potentially modifying their pharmacokinetic profiles. Such interactions could result in fluctuations in drug concentrations and efficacy, heightening the likelihood of adverse effects. Consequently, only one compound with CID:507883 appeared as a promising candidate for further development as an FGFR2 inhibitor. The summarized results of this analysis can be found in Table 3. Here, CID:507883 was identified as a promising candidate warranting further investigation.Table 3 ADMET properties of the selected compound CID:507883 and the reference drug Zoligratinib.Visualization and evaluationThe FGFR2 structure chosen for screening purposes consists of a protein kinase domain (amino acids 481-770). Within this domain, certain residues, such as L487-V495, K417, E565-A567, and N571, hold great significance due to their ATP-binding motifs45. Our investigation revealed that these residues engage in various close interactions with CID:507883 and Zoligratinib, which can potentially impact the catalytic activity of FGFR2 (Fig. 1). The selected compound CID:507883 and the reference drug Zoligratinib bind at the same location, with similar orientations, and establish several close interactions comparable to the co-crystal ligand 1,3,5-triazine derivative, EVC in PDB ID: 6LVK. Interaction patterns of the screened compounds from the top ten hits and three remaining inhibitors, AZD4547, Erdafitinib, and Infigratinib are shown in Supplementary Figure S2. They also showed various similar interactions toward FGFR2 as CID:507883. Compound CID:507883 resides deep within the FGFR2 cavity, demonstrating close interactions with the binding residues (Fig. 1A). Notably, the ATP-binding residue Leu487 forms a hydrogen bond with CID:507883, complemented by various van der Waals interactions (Fig. 1B i). Furthermore, surrounding residues, including Arg630 and Asp644, contribute significant interactions to secure both ligands within the FGFR2 cavity (Fig. 1B). Surface representations vividly depict that the compounds are positioned within the internal cavity, displaying a specific binding affinity to the catalytic pocket of FGFR2 (Fig. 1C).Figure 1Interaction plots. (A) Presentation of binding mode of CID:507883 and the reference drug Zoligratinib with FGFR2. (B) Magnified cartoon representation of FGFR2 binding pocket complex with (i) CID:507883 and (ii) Zoligratinib. (C) Surface representation of FGFR2 complex with (i) CID:507883 and (ii) Zoligratinib. CID:507883: green element, Zoligratinib: yellow element. The angle and distance cut-off for hydrogen bonds between donor and acceptor were set to 3.5 Ã… and 150-180°, respectively.Overall, this analysis revealed that CID:507883 and Zoligratinib exhibit a remarkable degree of closeness in their interactions. These interactions primarily involve the formation of hydrogen bonds and van der Waals forces, which play crucial roles in stabilizing the binding between the compounds and FGFR2. Figure 2 illustrates the spatial arrangement of these interactions, providing visual evidence of the closeness and specificity of the compound-protein binding. Notably, both CID:507883 and Zoligratinib were found to interact with several common residues of FGFR2. The analysis identified that CID:507883 and Zoligratinib establish several intimate interactions with the ATP-binding motif within the catalytic domain of FGFR2 (Fig. 2A, B). Both compounds establish several close interactions comparable to the co-crystal ligand EVC in PDB ID: 6LVK (Fig. 2C).Figure 22D interaction plots of compound (A) CID:507883. (B) Zoligratinib, and (C) Co-crystalized inhibitor EVC toward FGFR2.Previous studies on FGFR2 inhibition have identified the same key residues highlighted in our study46. A de novo design and development study aimed at developing a selective FGFR2 inhibitor found that their compounds interacted with the ATP-binding residues of the protein47. Similarly, a virtual screening study investigating Gefitinib-like compounds as potential therapeutic candidates against FGFR2 reported a similar set of interactions48. These interactions are of significant interest since the ATP-binding motif is responsible for facilitating the catalytic activity of FGFR2. By forming close associations with this motif, CID:507883 has the potential to modulate or inhibit the catalytic activity of the FGFR2 kinase domain. Therefore, the presence of compound CID:507883 may hold great promise in the context of preventing diseases associated with aberrant FGFR2 activity. Overall, these findings shed light on the molecular interactions between CID:507883 and Zoligratinib with the FGFR2 which opens up new possibilities for the drug development process.MD simulationsThe interaction between a compound and a protein’s binding pocket can trigger substantial conformational alterations in the protein’s structure49. To evaluate the stability of protein structures, one fundamental property is the root mean square deviation (RMSD). In this study, the average RMSD values were determined as 0.33 nm for FGFR2, 0.42 nm for FGFR2-CID:507883, and 0.35 nm for FGFR2-Zoligratinib (Table 4). The RMSD plot indicated that the binding of compound CID:507883 slightly increase the backbone RMSD, but overall stabilized FGFR2 structure without any major peak (Fig. 3A). The orientation of CID:507883 in the binding pocket of FGFR2 demonstrated a stable distribution throughout the 200 ns MD simulation (Fig. 3B). In the case of the FGFR2-Zoligratinib complex, the initial 50 ns of the MD trajectories exhibited somewhat unstable fluctuations in the binding of Zoligratinib within the active pocket of FGFR2. This behavior could be attributed to the specific orientation of Zoligratinib within the active pocket of FGFR2. However, as the simulation progressed, the system eventually reached a stable equilibrium, albeit with minor RMSD fluctuations in the FGFR2 binding pocket. In summary, the analysis of RMSD values revealed minimal fluctuations in all systems, suggesting negligible structural deviations after ligand binding. Table 4 The average values of systematic parameters obtained after 200 ns MD simulations.Figure 3Dynamics of FGFR2 upon ligands binding. (A) RMSD plot as a function of time. (B) Backbone RMS fluctuations in FGFR2 upon ligands binding.Additionally, we assessed the root mean square fluctuation (RMSF) of FGFR2 to gauge the average fluctuation of all residues throughout the simulation period following the binding of compounds (Fig. 3B). The RMSF plot unveiled residual fluctuations occurring in different regions of the FGFR2 protein structure. However, these fluctuations were observed to be slightly increase upon the binding of CID:507883 and Zoligratinib. Notably, it was observed that the binding of Zoligratinib resulted in higher residual fluctuations compared to CID:507883. In summary, the examination of RMSF values indicated that there were no notable fluctuations seen in the residues, except in the loop regions, especially R580-T599. This finding implies that the binding of the compounds had a minimal impact on the overall protein structure, indicating little to no conformational changes upon their interaction.The stability of the protein within the biological system was evaluated through the computation of the radius of gyration (rGyr). The average rGyr values were determined for three different states: free FGFR2, FGFR2 bound to CID:507883, and FGFR2 bound to Zoligratinib, resulting in values of 1.94 nm, 1.97 nm, and 1.98 nm, respectively (Table 4). The rGyr plot provided valuable insights into the conformational dynamics of FGFR2 under these conditions. In its free form, FGFR2 exhibited a tightly packed and well-defined conformation. However, when bound to Zoligratinib, FGFR2 experienced higher structural deviations compared to both its free state and when bound to CID:507883 (Fig. 4A). This increase in rGyr values can be attributed to the occupancy of intramolecular space within FGFR2 by the ligands, particularly Zoligratinib. Ovearll, the analysis of rGyr values indicated that while both CID:507883 and Zoligratinib binding had an impact on the conformation of FGFR2. Zoligratinib exhibited a more pronounced effect that caused slightly higher structural deviations and reduced compactness compared to the other conditions.Figure 4(A) Time evolution of radius of gyration (rGyr) values during 200 ns of MD simulation. (B) Solvent Accessible Surface Area (SASA) as a function of time. Black, red, and aqua represent values obtained from FGFR2, FGFR2-CID:507883 and FGFR2-Zoligratinib complexes respectively.The solvent accessible surface area (SASA) is the surface area of a protein that is accessible to a solvent50. During the simulations, the average SASA values were calculated for the FGFR2, FGFR2-CID:507883, and FGFR2-Zoligratinib complexes. The average SASA values were determined to be 152.02 nm2, 151.74 nm2, and 156.78 nm2 for FGFR2, FGFR2-CID:507883, FGFR2-Zoligratinib, respectively (Table 3). Notably, no significant changes were observed in the SASA values due to the binding of the compounds (Figure 4B). This analysis indicates that the folding state of the protein, FGFR2, remains stable upon compounds binding with minimal conformational changes.Hydrogen bondingHydrogen bonds are pivotal for assessing the stability and directionality of protein and protein-ligand interactions. To further evaluate the structural integrity and stability of protein folding, the presence of intramolecular hydrogen bonding within a 0.35 nm distance was investigated in three different systems: free FGFR2, FGFR2 bound to CID:507883, and FGFR2 bound to Zoligratinib, during the simulations. This analysis played a critical role in validating the stability of the docked complexes and aimed to assess the persistence of hydrogen bonds between FGFR2 and the ligands over time (Fig. 5). Remarkably, throughout the simulations, no significant changes were observed in the average hydrogen bonding patterns resulting from the binding of the compounds (Fig. 5A). The findings suggest that the docked complexes maintained their stability, and the intramolecular hydrogen bonds between FGFR2 and both CID:507883 and Zoligratinib remained relatively constant during the simulation period. This consistency in hydrogen bonding indicates that the interactions between the protein and the ligands were robust and enduring (Fig. 5B). This supports the overall stability of the protein-ligand complexes throughout the MD simulation.Figure 5Dynamics of hydrogen bonding. (A) Formation of intramolecular hydrogen bonds within FGFR2 as a function of time. (B) The average number of hydrogen bonds distribution and their probability. Formation of intermolecular hydrogen bonds between FGFR2 and (C) CID:507883 and (D) Zoligratinib. Black, red, and aqua represent the time-evolution of hydrogen bonds formed within FGFR2, FGFR2-CID:507883, and FGFR2-Zoligratinib complexes, respectively.Furthermore, we assessed the stability of ligand-FGFR2 interactions by quantifying the hydrogen bonds formed between ligands CID:507883 and Zoligratinib with FGFR2 throughout the simulation. Over the entire simulation duration, both ligands were consistently engaged with FGFR2 through 1-2 enduring hydrogen bonds, highlighting the durability and strength of the ligand-protein interactions (Fig. 5C, D). These bonds formed up to three at some time during the simulation. These results validate the interaction patterns identified during molecular docking, reinforcing the accuracy and reliability of the docking predictions. The intermolecular interactions observed after 200 ns of MD simulations were analyzed. The interactions from the representative pose of the final snapshot at 200 ns revealed that the hits exhibited similar interactions with only slight variations compared to those obtained from docking studies (Supplementary Figure S3). The IUPAC names of CID:507883 and Zoligratinib, along with their structural features are shown in Table 5.Table 5 The IUPAC names and structural features of CID:507883 and Zoligratinib.Principal component analysisPCA is a method utilized to analyze the overall expansion of a protein under different simulation conditions51. It provides insights into the flexibility and dynamics of the protein. In this study, PCA was performed using the gmx covar module to calculate the dynamics of FGFR2 in relation to the protein backbone. Notably, the eigenvalues for the free FGFR2 protein were relatively lower compared to the complexes, indicating an increase in random fluctuations upon binding of the compounds. Elevated eigenvalues indicate a more pronounced expansion of FGFR2, signifying reduced compactness of the complexes. Figure 6 illustrates the multidimensional atomic covariance matrix, which captures the covariances between each pair of atoms. Additionally, we utilized the gmx anaeig module to project the MD trajectory onto specific eigenvector values. The 2D projections of the trajectories on these eigenvectors revealed overlap between FGFR2, FGFR2-CID:507883, and FGFR2-Zoligratinib complexes.Figure 6Two-dimensional projections of PCA trajectories on both eigenvectors showed conformational landscapes of FGFR2 and its complexes with CID:507883 and Zoligratinib.Gibbs free energy landscapeThe FELs were computed based on the projections of the first (PC1) and second (PC2) eigenvectors. The resulting color-coded FELs are depicted in Fig. 7. These landscapes examine the fluctuation direction of the two systems for all Cα atoms in the structures of FGFR2, FGFR2-CID:507883, and FGFR2-Zoligratinib complexes. The corresponding free energy contour map indicates lower energy with deeper blue, as observed in FGFR2, FGFR2-CID:507883, and FGFR2-Zoligratinib. Notably, FGFR2, FGFR2-CID:507883, and FGFR2-Zoligratinib complexes displayed 1-2 stable conformational states. Comparing the full views of FELs among these complexes, it was observed that FGFR2-CID:507883 covered larger ranges of PC1 and PC2, indicating a more rugged free energy surface than FGFR2-Zoligratinib (Fig. 7B, C). Overall, FGFR2 showed 2 free energy wells in 2-3 basins, while FGFR2-CID:507883, and FGFR2-Zoligratinib exhibited 2-3 stable conformational states in 3-4 basins. The structural snapshots of FGFR2 and its docked complexes were extracted from the global minima following the simulations. Analysis of these structures revealed that FGFR2 did not undergo significant structural alterations compared to its initial state (Fig. 7, lower panels). Overall, the analysis of essential dynamics indicated no significant changes in the structural stability of FGFR2 upon compounds binding. This suggests strong stability of the FGFR2-compounds complexes.Figure 7The free energy landscape plots obtained during 200 ns MD simulation for (A) FGFR2, (B) FGFR2-CID:507883, (C) FGFR2-Zoligratinib. Lower panels showed the structural snapshots of FGFR2, and its docked complexes fetched from the global minima after the simulations.
