The three-dimensional (3D) structure of human PI3K kinase in complex with an ATP competitive inhibitor, LXX (6-(1H-pyrazolo [3,4-b]pyridin-5-yl)-4-pyridin-4-ylquinoline) was retrieved from the Protein Data Bank (PDB, https://www.rcsb.org/) with PDB ID 3L54. The 3D structures of seven flavonoid compounds (quercetin, luteolin, kaempferol, genistein, wogonin, daidzein, and flavopiridol) were obtained from PubChem (https://pubchem.ncbi.nlm.nih.gov/) with compound IDs (CID: 5280343, 5280445, 5280863, 5280961, 5281703, 5281708, and 5287969, respectively.

Molecular docking of selected compounds in the active site of PI3Kγ was performed using Dock v.6.9 with default parameters (Ewing et al., 2001; Mukerjee et al., 2022). The initial structural preparations of ligands and proteins required for docking were performed using Chimera v.1.14 (Mukerjee et al., 2022). The native ligand in complex with PI3Kγ was used as a clue for the active site, and residues within a 10 Å area around the native ligand were used for grid generation.

Chimera v.1.14 was used to generate illustrations and analyze the protein-ligand complexes (Mukerjee et al., 2022). The LigPlot v.1.4.3 program was used to generate protein-ligand interaction plots and to analyze the polar and hydrophobic interactions between the compound and amino acid residues of the binding site (Laskowski and Swindells, 2011).

MD simulations were conducted using Gromacs v.2019.6 (Abraham et al., 2015) and employed the charmm36-jul2022 force field. The ligand’s topology file was generated via the CGenFF server (https://cgenff.umaryland.edu/) utilizing the charmm force field. For solvation, the proteins were placed within a dodecahedron unit cell, maintaining a minimum edge distance of 1 nm from the protein’s surface. Solvation utilized simple point charge water molecules, specifically spc216. The solvated system was neutralized through the addition of counter ions, followed by an energy minimization step using the steepest descent method. Periodic boundary conditions were applied to mitigate surface effects. Subsequently, the system underwent equilibration under two conditions: NVT, which maintains a constant number of particles, volume, and temperature at 300K, and NPT, which keeps the number of particles, pressure, and temperature constant at 1.0 bar, each lasting for 100 ps. The simulation employed a time step of 2 fs. Once equilibrated, the systems were simulated for 200 ns, with trajectory data saved at 10 ps intervals.

The grid region is chosen as a box region around the native inhibitor covering the residues within the 10 Å vicinity of the native inhibitor and Ramachandran plot and hydrophobicity plot (Figures 1A,B). The center of the grid is located at coordinates (X = 23.871, Y = 14.967, and Z = 23.169). Whereas the dimensions of the grid along the X, Y, and Z-axes are given as 36.134, 33.499, and 26.881, respectively.

All seven flavonoids (Figure 2) were docked into the catalytic site to measure their potential for inhibiting PI3Kγ protein, and the results are summarized in Table 1.

The molecular docking results of quercetin to PI3Kγ revealed that quercetin bound deep in the catalytic site with high absolute value of dock score (−45.38) and interacted with 13 amino acid residues namely, Ser-806, Pro-810, Trp-812, Ile-831, Lys-833, Tyr-867, Ile-879, Glu-880, Val-882, Met-953, Phe-961, Ile-963 and Asp-964 with 43 non-bonded contacts (hydrophobic interactions) and 3 hydrogen bonds (Figure 4B; Table 2). Hydrogen bonds measure 2.98 Å, 3.27 Å, and 3.03 Å through Ser-806, Lys-833 and Asp-964, respectively. The significantly interacting amino acids were Met-953, Ile-963, and Asp-964, with 9, 6, and 11 non-bonded contacts, respectively. Comparing the interactions of quercetin with the native inhibitor (Figure 3), there were 10 common amino acids: Trp-812, Ile-831, Lys-833, Tyr-867, Ile-879, Glu-880, Val-882, Met-953, Ile-963, and Asp-964. This list of common amino acids includes those forming hydrogen bonds, namely, Met-953, Ile-963, and Asp-964.

Molecular docking of luteolin to PI3Kγ binding site showed that it bound deep in the catalytic site with a high absolute value of the docking score (−42.70) and interacted with 12 residues, including Ser-806, Pro-810, Trp-812, Ile-831, Lys-833, Tyr-867, Ile-879, Val-882, Met-953, Phe-961, Ile-963, and Asp-964, with 37 non-bonded contacts (hydrophobic interactions) and two hydrogen bonds (Figure 4C; Table 3). The 12 residues interacted with PI3Kγ through a high binding affinity of −8.11 Kcal/mol and a dissociation constant of 5.95 p K d . The protein-ligand interactions are shown in Table 3. Hydrogen bonds measure 2.93 Å and 3.29 Å through Ser-806 and Lys-833. The significantly interacting amino acids were Met-953 and Asp-964, with 7 and 10 non-bonded contacts, respectively. A comparison of the interactions of luteolin with the native inhibitor (Figure 4C) revealed nine common amino acids: Trp-812, Ile-831, Lys-833, Tyr-867, Ile-879, Val-882, Met-953, Ile-963, and Asp-964. This list of common residues includes the important amino acids Met-953, Ile-963, and Asp-964, which have a high number of non-bonded contacts.

The molecular docking results of kaempferol to PI3Kγ revealed that kaempferol bound deep in the catalytic site with a high absolute docking score (−42.86) and interacted with nine residues, namely, Ser-806, Trp-812, Ile-831, Tyr-867, Val-882, Met-953, Phe-961, Ile-963, and Asp-964, with 34 non-bonded contacts (hydrophobic interactions) and two hydrogen bonds (Figure 4D; Table 4). The nine residues interacted with PI3Kγ through a high binding affinity of −8.11 kcal/mol and dissociation constant (5.95, p K d ). The protein-ligand interactions are shown in Table 5. Hydrogen bonds measure 2.93 Å and 3.19 Å through Ser-806 and Asp-964 respectively. The most important interacting amino acids were Met-953 and Asp-964, with a high number of non-bonded contacts (6 and 10, respectively). Comparing the interaction between kaempferol and the native inhibitor (Figure 4D), there were seven common amino acids: Trp-812, Ile-831, Tyr-867, Val-882, Met-953, Ile-963, and Asp-964, including the important amino acids Met-953, Ile-963, and Asp-964.

Molecular docking results showed that genistein bound to PI3Kγ deep in the cavity (high absolute value of dock score, −38.45) and formed interactions with nine residues, including Ile-831, Asp-841, Tyr-867, Ile-879, Glu-880, Val-882, Met-953, Ile-963, and Asp-964, with 37 non-bonded contacts (hydrophobic interactions) and one hydrogen bond (Figure 4E; Table 5). Genistein interacts with the protein through nine amino acids with a binding affinity of −8.67 Kcal/mol and dissociation constant of 6.35 p K d . The protein-ligand interactions are shown in Table 6. The hydrogen bond measures 3.14 Å through Asp-841. The most important interacting amino acids were Ile-831, Met-953, and Asp-964, with a high number of non-bonded contacts (6, 7, and 10, respectively). By comparing the interaction of genistein with the native inhibitor (Figure 4E), eight common amino acids were identified: Ile-831, Tyr-867, Ile-879, Glu-880, Val-882, Met-953, Ile-963, and Asp-964. The list of common interacting residues includes the important interacting amino acids Met-953, Ile-963, and Asp-964.

The molecular docking results of wogonin to PI3Kγ revealed that wogonin bound deep in the catalytic site with high absolute value of dock score −38.23 and interact with 11 residues namely, Met-804, Ile-831, Tyr-867, Ile-879, Glu-880, Val-882, Ala-885, Met-953, Phe-961, Ile-963, Asp-964 with 41 non-bonded contact (hydrophobic interactions) and one hydrogen bond (Table 6). The hydrogen bond measures 3.13 Å through Val-882. The most important interacting amino acids were Tyr-867 and Met-953, with 11 and 8 non-bonded contacts, respectively. Comparing the interaction of wogonin with the native inhibitor (Figure 4F), there were 10 common amino acids: Met-804, Ile-831, Tyr-867, Ile-879, Glu-880, Val-882, Met-953, Ile-963, and Asp-964, including the important residues, Met-953, Ile-963 and Asp-964. Wogonin interacted with the protein through 11 amino acids with a binding affinity of −8.26 Kcal/mol and dissociation constant of 6.06 p K d . The protein-ligand interactions are shown in Table 6.

Daidzein bound deep in the catalytic site of PI3Kγ kinase domain with high absolute value of dock score (−36.35) and interacted with 9 residues including Trp-812, Ile-831, Asp-841, Tyr-867, Ile-879, Glu-880, Val-882, Met-953 and Asp-964 with 33 non-bonded contact (hydrophobic interactions) and one hydrogen bond (Figure 5B; Table 7). The ligand interacted with the protein through nine amino acids with a binding affinity of −8.56 Kcal/mol and dissociation constant of 6.27 p K d . The hydrogen bond measures 3.07 Å through Asp-841. The important interacting amino acids were Ile-831, Ile-879, and Asp-964, with high numbers of 6, 6, and 10 non-bonded contacts, respectively. Comparing the interaction of daidzein with the native inhibitor (Figure 5B), there were eight common amino acids: Trp-812, Ile-831, Tyr-867, Ile-879, Glu-880, Val-882, Met-953 and Asp-964 including the important amino acids Met-953 and Asp-964.

The molecular docking results of flavopiridol to PI3Kγ revealed that flavopiridol bound deep in the catalytic site with a high absolute docking score of −44.40 and interacted with 11 residues, namely, Met-804, Ser-806, Ile-831, Lys-833, Tyr-867, Ile-879, Lys-890, Asp-950, Met-953, Ile-963, and Asp-964, with 41 non-bonded contacts (hydrophobic interactions) and one hydrogen bond (Figure 5C; Table 8). The hydrogen bonds measure 3.32 Å through Asp-950. The significantly interacting amino acids were Ile-831, Met-953, and Asp-964, with high numbers of 5, 5, and 13 non-bonded contacts, respectively. Comparison of the interaction of flavopiridol with the native inhibitor (Figure 5C) revealed 8 common amino acids: Met-804, Ile-831, Lys-833, Tyr-867, Ile-879, Met-953, Ile-963, and Asp-964, including the significant amino acids Met-953, Ile-963, and Asp-964. The ligand interacted with the protein through 11 amino acids with a binding affinity of −8.97 Kcal/mol and dissociation constant of 6.58 p K d ), as shown in Table 8.

The seven selected flavonoids were observed to engage with PI3Kγ by interacting with a range of 9–13 specific residues. In contrast, the native inhibitor exhibited interactions with 13 distinct residues. Notably, there was an overlap of 8–10 interacting residues between the natural compounds and the native inhibitor, as illustrated in Table 9 and Figures 4, 5. Consequently, a significant proportion of the interacting residues involved in binding these natural compounds coincided with those of the native inhibitor. Among these, Ile-831, Tyr-867, Met-953, and Asp-964 consistently appeared as common interacting residues shared by the native inhibitor and all five proposed compounds. Additionally, the natural compounds occupied the catalytic site in an overlapping manner and bound to the same location as the native inhibitor, as depicted in Figure 3. It is important to note that our study explored a vast search space, encompassing residues within a 10 Å radius of the native inhibitor. Despite this extensive search space, the natural compounds consistently targeted and blocked a similar set of residues, underscoring the accuracy of our docking approach. This finding reinforces the notion that these compounds inhibit PI3Kγ in a manner akin to the native inhibitor.

In our investigation of binding energy trends concerning structural variations among compounds, we made intriguing observations. Based on the structure of the compounds, we classified these compounds into three groups and examined whether the binding energy trends were in agreement with the structural change in the compounds. The first group contains a set of compounds, namely, quercetin, luteolin, kaempferol, and wogonin, which share a common structural backbone characterized by the presence of a benzene ring attached to a double ring of benzopyrone. While these compounds maintained a common structure, they vary in the number and positions of hydroxyl groups and a methyl substituent. Consequently, their binding energy values demonstrated comparability, displaying minimal variation within this group.

The second group contains genistein and daidzein, which share a similar structural backbone but differ in the orientation of a single benzene ring compared to the first group. This distinction was reflected in slightly higher binding energy values compared to the first group.

Lastly, we examined the compound flavopiridol, which stood alone in a distinct third group. Notably, despite sharing a three-ring structural backbone identical to that of compounds in the first group, it featured an additional ring and the presence of a chlorine atom. As a result, it exhibited the highest binding energy among all the compounds examined. In summary, our investigation revealed distinctive binding energy trends that aligned seamlessly with the structural variations present in the compounds.

To gain a better understanding of the protein-ligand interaction between PI3Kγ and flavopiridol a molecular dynamic simulation was performed for 200 ns. The trajectories were used to analyze root-mean-square deviation (RMSD), root-mean-square fluctuation (RMSF), the radius of gyration, and the number of hydrogen bonds. The ligand conformation RMSD after super-posing on the ligand was plotted as a function of simulation time. Conformational fluctuations of the ligand stabilized at 70 ns. After 80 ns, the aforementioned RMSD values were observed to remain confined within an acceptable narrow range of 0.25 nm–0.3 nm during the 200 ns (Figure 6A). The RMSF plot shows protein fluctuation between 0.1 nm and 0.5 nm (Figure 6B) and the radius of gyration around 3 nm during whole 200 ns simulation (Figure 6C). In (Figure 6D) shows number of hydrogen bonds and pairs within 0.35 nm where five is the maximum number of interactions and three of hydrogen bonds during the simulation. These results indicate the tight binding and stability of the studied complex. However, the scope remains to crystallize the complex.