Abstract
Background/Aim: Phosphatidyl-inositol-3-kinase (PI3K), a cancer therapeutic target, has been exploited for cancer therapy. The natural compounds flavonoids have increasingly been shown to possess anticancer activity. The current study aimed to explore all known flavonoids for their ability to inhibit PI3Kγ. Materials and Methods: Virtual screening of flavonoids using molecular docking to the ATP binding site of PI3Kγ was performed. The top 10 scoring flavonoids were selected for pose analysis and binding strength scores. Results: Molecular docking revealed that the 10 selected flavonoids might inhibit PI3Kγ kinase activity. Literature search did not identify studies reporting a bioassay activity for any of these compounds. Conclusion: All 10 selected flavonoids are potential PI3Kγ kinase inhibitors and anticancer agents. Interestingly, one of the 10 least scoring flavonoids has been reported to be inactive, as expected, and thus validating the accuracy of the results.
PI3K/AKT pathway plays a pivotal role in cell survival, growth, and proliferation (1). This cellular signaling pathway is tightly regulated, however, its elevated activity is often associated with various human cancers and anti-cancer drug resistance (2-5). One of the major players and upstream targets of this pathway is phosphatidyl-inositol-3-kinase (PI3K). PI3Ks belong to a lipid kinase family, capable of phosphorylating the 3’-hydroxyl group of the inositol ring of phosphoinositides (6). Three classes of PI3K (Class I, II, III PI3K) are found in human cells; class I PI3K is the most widely studied and most involved in cancer. The class I PI3Ks phosphorylate the membrane-bound phosphatidylinositol-(4,5)-bisphosphate (PIP2) and convert it into phosphatidylinositol-(3,4,5)-trisphosphate (PIP3) (7, 8). PIP3 functions as a secondary messenger that recruits and activates the pleckstrin homology (PH) domain-containing downstream kinase AKT, to regulate growth, proliferation, and survival signaling (9). Class I PI3Ks consist of two distinct subunits, a catalytic subunit (i.e., p110) and a regulatory subunit (i.e., p85). This class is further subdivided into two groups: Class IA (PI3Kα, β, and δ), the group activated by receptor tyrosine kinases (RTKs), and Class IB (PI3Kγ), the group activated by G proteins coupled receptors (GPCRs) (9). Mutations in different subunits of PI3K are associated with various cancers (10).
Therefore, there has been increasing interest in generating inhibitors of PI3Ks and this has led to the discovery of some potentially useful compounds, many of which have reached clinical trials (11). Two well-known inhibitors of PI3Ks are Wortmannin, a fungal metabolite and LY294002, a flavonoid derivative, which cannot be administered as drugs because of their chemical properties (10, 12). Wortmanin, an irreversible potent inhibitor, binds PI3K and blocks the binding of its substrate PIP2. Unlike Wortmanin, LY294002 is a reversible ATP competitive inhibitor of PI3K. Both inhibitors inhibit all isoforms of PI3Ks. Modified versions of these two inhibitors are now tested as anti-cancer agents in clinical trials (10). Furthermore, there are other isoform specific inhibitors e.g., XL-147, specific inhibitor of class I PI3Ks and CAL-101, specific inhibitor of p110δ, that are tested in clinical trials for various cancers (10, 12, 13). Due to the increased knowledge regarding the role of different PI3K isoforms in various cancers, more specific inhibitors are required for the treatment of distinct cancer types. Given the experimental difficulties in designing a drug/inhibitor, numerous high throughput virtual screening platforms are being increasingly used for fast and cost-effective drug discovery programs. Molecular docking has been successfully used for virtual screening of drug like compounds for numerous target proteins (14-20).
Two-dimensional sketch of flavonoid scaffold (A) and selected flavonoids (B-K). The flavonoids are denoted by their PubChem CID. Heteroatoms oxygen (O), nitrogen (N), and sulfur (S) with their balancing hydrogens are shown as red, blue, and green respectively.
Flavonoids, a class of natural compounds having polyphenolic structures, belong to plant secondary metabolites. They are found in fruits, vegetables, grains, and certain beverages (21). Several epidemiological studies have indicated an association between dietary flavonoid intake and decreased cancer risk (22, 23). Various in vitro and in vivo studies have demonstrated their anticancer activities against multiple cancer types (24-34). Few studies have also indicated that flavonoids act through inhibiting PI3K kinase function (35, 36). Some flavonoids are now being tested in clinical trials for cancer therapy (37).
The current study aimed to explore all known flavonoids for their potential inhibition of PI3Kγ. A library of all known flavonoids was prepared and 10 compounds proposed as potential PI3Kγ inhibitors and anticancer agents. These proposed compounds should be further tested experimentally for their potential use as novel anti-cancer agents.
Materials and Methods
Data retrieval. The 3-dimensional-coordinate structure of PI3K p110γ (PI3Kγ) with PDB Id: 3L54 and chemical compounds were retrieved from PDB and PubChem databases, respectively. The structure of PI3Kγ with PDB Id: 3L54 was selected as it contained the ATP competitive inhibitor LXX, which was required for probing the catalytic site of the protein.
Molecular docking. The molecular docking of chemical compounds into the catalytic site of PI3Kγ was carried out using Dock v.6.5 (38). The structure preparation step for the protein and ligand required for the final docking was carried out by Chimera v.1.6.2 (39). The binding site of the protein used for docking was selected as 10 Å space around the native inhibitor LXX.
Histogram plot of dock scores of all flavonoids. The dock scores of the selected flavonoids in the left part are highlighted in yellow color.
Analyses of docked protein-ligand complex. The docked protein-ligand complexes were analyzed using Chimera v.1.6.2 (39) and PyMol v.1.3 (40). The molecular interactions were illustrated using Ligplot+ v.2.1 (41) showing non-bonding and hydrogen bonding interactions. The binding energy and dissociation constants were also predicted using XScore v.1.2.11 (42).
Results and Discussion
Flavonoids' library. The simplest flavonoid available in PubChem was ‘flavone’ with CID: 10680. Compounds similar to flavone were searched in PubChem for the presence of the flavonoid scaffold (Figure 1A) and filtered for Lipinski's rule of five (43) yielding 1,674 compounds. These 1,674 compounds were visually checked for the flavonoid scaffold and the irrelevant compounds were removed. Finally, 1173 compounds were selected, and a flavonoids' library was formed.
Virtual screening of flavonoids' library. The flavonoids' library was used for screening potential inhibitors of PI3Kγ. The dock scores of all the compounds were also plotted as histogram (Figure 2). The top 10 high dock scoring compounds were considered as potential inhibitors and evaluated through binding analyses (Figure 1B-K).
Molecular docking analyses of the top 10 screened compounds. The PubChem CID list of top the 10 high dock scoring flavonoids proposed as potential PI3Kγ inhibitors includes ‘53463223’, ‘71260095’, ‘131834212’, ‘21676336’, ‘45277410’, ‘44326949’, ‘123270861’, ‘156200’, ‘131801248, and ‘132256977’. The molecular docking analyses of these compounds are presented below.
Molecular docking of selected flavonoids to PI3Kγ. The protein is shown in cartoon representation colored light orange, while compounds are shown in stick representations in various color with the heteroatoms (H-atom colored white, O-atoms as red, N-atoms as blue, and S-atoms as green). (A) Binding of flavonoids ‘53463223’ (Pale green), ‘71260095’ (cyan-deepteal), and ‘131834212’ (magenta). (B) Binding of flavonoids ‘21676336’ (gray), ‘45277410’ (orange) and ‘44326949’ (blue-slate). (C) Binding of flavonoids ‘123270861’ (green), ‘156200’ (red-salmon), ‘131801248’ (cyan) and ‘132256977’ (yellow).
Protein-ligand interaction plots of native inhibitor (A) and selected flavonoids (B-F). The residues forming non-bonding interactions are shown as red bristles, while residues forming hydrogen bond and the bound ligand are shown as ball-and-stick representations. The carbon atoms are shown as black balls, nitrogen atoms as blue balls, oxygen atoms as red balls, and sulfur atom as yellow balls. The interacting residues common with those of the native inhibitor are shown in circle. The hydrogen bonds are shown as green dashed lines labeled with bond length (in Å).
The top high dock scoring compound (1st rank) with CID: 53463223 fitted well in the catalytic site (Figure 3A) and interacted with 14 residues: Met-804, Ser-806, Ile-831, Tyr-867, Ile-879, Ala-885, Thr-886, Lys-890, Asn-951, Met-953, Ile-963, Asp-964, His-967, and Leu-1090 (Figure 4B). These residues exerted non-bonding interactions with the compound and, in addition, two hydrogen bonds were formed through Asn-951 stabilizing the protein-ligand complex. The binding strength scores of the compound with PI3Kγ including the dock score (−49.50), binding energy (−6.76), and dissociation constant (pKd, 4.96) showed quality binding as required for adequate inhibition (Table I). The interacting residues and their interactions with the native inhibitor are also provided for comparison (Figure 4A). Seven of the 14 residues that interacted with the 1st rank compound were common with those of the native inhibitor (Figure 4B, Table II). This observation of binding to the same set of residues in the catalytic site also provided credence to our prediction that the proposed compound is as a good inhibitor as the native one.
The next high dock scoring compound (2nd rank) with CID: 71260095 docked well to the catalytic site (Figure 3A). The protein-ligand complex was stabilized by non-bonding interactions through 12 interacting residues including Met-804, Trp-812, Ile-831, Tyr-867, Glu-880, Val-882, Lys-883, Asp-884, Ala-885, Met-953, Ile-963, and Asp-964 (Figure 4C). The three scores for binding strength including dock score (−45.05), binding energy (−8.22), and dissociation constant (pKd, 6.03) were reasonably high, as required for good inhibition (Table I). Of the 12 interacting residues, nine residues were common with those of the native inhibitor (Figure 4C, Table II). This also indicated that the proposed compound was blocking the same set of residues as the native inhibitor and thus, would inhibit the activity of the protein.
The next dock scoring compound (3rd rank) with CID: 131834212 docked well to the catalytic site (Figure 3A), interacting with 14 residues namely Met-804, Ala-805, Pro-810, Ile-831, Lys-833, Tyr-867, Ile-879, Glu-880, Thr-887, Lys-890, Asp-950, Met-953, Ile-963, and Asp-964 (Figure 4D). These residues formed non-bonding interactions and a hydrogen bond (through Thr-887) stabilizing the protein-ligand complex. The high absolute values of the dock score (−44.37), binding energy (−7.56), and dissociation constant (pKd, 5.54) showed quality binding as required for good inhibition (Table I). Of the 14 interacting residues, 10 residues were common with those of the native inhibitor (Figure 4D, Table II). Thus, the observation of binding of both the proposed compound and the native inhibitor to common residues indicated that they inhibit the protein in a similar way.
The selected flavonoids and the binding strength scores (dock score, binding energy and pKd). The higher the absolute values of the scores, the better is the binding.
The next dock scoring compound (4th rank) with CID: 21676336 bound to the catalytic site of P3Kγ (Figure 3B) interacting with 10 residues namely Ser-806, Lys-807, Lys-808, Lys-833, Asp-950, Asn-951, Ile-963, Asp-964, His-967, and Leu-1090 (Figure 4E). The protein-ligand complex was stabilized by non-bonding interactions and six hydrogen bonds through His-967, Lys-833, Lys-808, Ser-806 (contributing 2 hydrogen bonds), and Leu-1090. The binding strength of the compound with the protein was also high as indicated by the dock score (−43.42), binding energy (−7.24), and dissociation constant (pKd, 5.31) for adequate inhibition (Table I). Of the 10 interacting residues, three residues were common among the interacting residues of the native inhibitor (Figure 4E, Table II).
The next dock scoring compound (5th rank) with CID: 45277410 bound to PI3Kγ catalytic site (Figure 3B) interacting with 11 residues namely Met-804, Ala-805, Ser-806, Ile-831, Lys-833, Ile-879, Lys-890, Asp-950, Met-953, Asp-964, and His-967 (Figure 4F). These interacting residues were forming non-bonding interactions. The dock score (−43.39), binding energy (−7.86), and dissociation constant (pKd, 5.76) were also reasonably high as required for adequate PI3Kγ kinase inhibition (Table I). Of the 11 interacting residues, six residues were common among the interacting residues of the native inhibitor, and thus, they might inhibit the PI3Kγ kinase activity similar to the native inhibitor (Figure 4F, Table II).
The next dock scoring compound (6th rank) with CID: 44326949 fitted well in the catalytic site (Figure 3B) interacting with 11 residues namely Met-804, Ala-805, Ser-806, Pro-810, Ile-831, Lys-833, Lys-890, Asp-950, Met-953, Ile-963, and Asp-964. For binding comparison, the interacting residues and their molecular interaction with the native inhibitor are provided (Figure 5A). The 11 residues interacting with the 6th rank compound formed non-bonding interactions and in addition, two hydrogen bonds were formed by Lys-833 and Ser-806 (Figure 5B). The dock score (−43.21), binding energy (−7.29), and dissociation constant (pKd, 5.34) were also reasonably good to have adequate PI3Kγ kinase inhibition (Table I). Of the 11 interacting residues, six residues were common among the interacting residues of the native inhibitor (Figure 5B, Table II) and thus it may inhibit PI3Kγ similar to the native inhibitor.
Flavonoids and their interacting residues. Each column represents a flavonoid with its name on top followed by its interacting residues. The common residues of all columns are aligned in the same row. The interacting residues common with those of the native inhibitor are shown in bold.
The next dock scoring compound (7th rank) with CID: 123270861 docked well to PI3Kγ catalytic site (Figure 3C) and interacted with 10 residues namely Met-804, Ala-805, Ile-831, Thr-887, Lys-890, Asp-950, Asn-951, Met-953, Ile-963, and Asp-964 (Figure 5C). These 10 interacting residues exerted non-bonding interactions and Lys-890 also contributed towards a hydrogen bond stabilizing the protein-ligand complex. The dock score (−43.14), binding energy (−7.57), and dissociation constant (pKd, 5.55) were also reasonably good as required for adequate inhibition of PI3Kγ kinase activity (Table I). Of the 10 interacting residues, six residues were common among the interacting residues of the native ligand (Figure 5C, Table II). This suggested that the proposed compound is a potential inhibitor like the native one.
The next dock scoring compound (8th rank) with CID: 156200 also fitted well in the catalytic site (Figure 3C) interacting with 10 residues including Met-804, Trp-812, Ile-831, Tyr-867, Ile-879, Ala-885, Asp-950, Met-953, Ile-963, and Asp-964 (Figure 5D). These 10 interacting residues formed non-bonding interactions and a hydrogen bond through Asp-950 stabilizing the protein-ligand complex. The dock score (−42.93), binding energy (−7.98), and dissociation constant (pKd, 5.85) were reasonably high as required for adequate inhibition of PI3Kγ kinase activity (Table I). Of the 10 interacting residues, eight residues were common among the interacting residues of the native inhibitor (Figure 5D, Table II). This indicates that the proposed compound was also blocking the same set of residues as that of the native inhibitor and thus may inhibit the protein in the similar way.
The next dock scoring compound (9th rank) with CID: 131801248 also fitted well in the catalytic site (Figure 3C) interacting with nine residues including Ser-806, Pro-810, Ile-831, Lys-833, Ile-879, Met-953, Phe-961, Ile-963, and Asp-964 (Figure 5E). These nine interacting residues exerted non-bonding interactions and two hydrogen bonds through residues Lys-833 and Ser-806. The dock score (−42.88), binding energy (−7.85), and dissociation constant (pKd, 5.76) were reasonably high as required for adequate inhibition of PI3Kγ kinase activity (Table I). Of the nine interacting residues, six were common among the interacting residues of the native inhibitor (Figure 5E, Table II). This indicated that the proposed compound might block PI3Kγ kinase activity similar to the native inhibitor.
Protein-ligand interaction plots of native inhibitor (A) and selected flavonoids (B–F). The residues forming non-bonding interactions are shown as red bristles, while residues forming hydrogen bond and the bound ligand are shown as ball-and-stick representations. The carbon atoms are shown as black balls, nitrogen atoms as blue balls, and oxygen atoms as red balls. The interacting residues common with those of the native inhibitor are shown in circle. The hydrogen bonds are shown as green dashed lines labeled with bond length (in Å).
The final dock scoring compound (10th rank) with CID: 132256977 also fitted well in the catalytic site (Figure 3C) and interacted with 10 residues namely Met-804, Ala-805, Ser-806, Ile-831, Thr-887, Lys-890, Asp-950, Met-953, Ile-963, and Asp-964 (Figure 5F). These 10 residues exerted non-bonding interactions and, in addition, Thr-887 contributed towards a hydrogen bond and thus stabilized the protein-ligand complex. The dock score (−42.82), binding energy (−7.68), and dissociation constant (pKd, 5.63) were also adequate to efficiently inhibit PI3Kγ kinase (Table I). Of the 10 interacting residues, six residues were common with the interacting residues of the native inhibitor (Figure 5F, Table II). This indicated that the proposed compound was blocking the same set of residues like the native inhibitor and thus inhibits the kinase activity in a similar way.
The PubChem Bioassay database (44) was searched for any reported bioassay activity of these compounds, and only two compounds were reported to have been tested for bioassay activity. The compound ‘45277410’ was found to have nonspecific activity in antibacterial and antifungal tests (13 Bioassays for the compound with CID: 45277410), and was able to cross the membrane bilayer (PubChem Bioassay AID: 468522). The compound 44326949 was found to be active in free radical scavenging (PubChem Bioassay AID: 235284, 235285) and inhibition of yeast Alpha-glucosidase (PubChem Bioassay AID: 37406), whereas nonspecific activity was reported towards the inhibition of human xanthine dehydrogenase, XDH (PubChem Bioassay AID: 219595). None of these compounds was tested for anticancer activity.
When the 10 lowest scoring compounds were searched for bioassay activity in PubChem Bioassay database (44), only two compounds were reported to have been tested. One compound with CID: 11196546 was reported to have nonspecific activity towards the inhibition of porcine pancreatic lipase (PubChem Bioassay AID: 657039, 657041). Notably, another compound with CID: 369615 was found inactive on the National Cancer Institute (NCI) human tumor cell line growth inhibition assay for 53 cancer cell lines and unspecified activity for one cancer cell line (Bioassays results for the compound with CID: 369615). It was inactive on NCI Yeast anticancer drug screening for six strains and inactive for NCI AIDS antiviral assay (Bioassay results for the compound with CID: 369615). These interesting findings regarding the inactivity of the low scoring compound ‘369615’ on 53 cancer cell lines and six strains NCI Yeast anticancer drug screening validated our predictions. This agreed with what was expected as low scoring compounds should not possess anticancer activity.
Conclusion
The current work retrieved flavonoids from existing compounds in the PubChem database and formed a library of 1173 flavonoids. The virtual screening of the flavonoids' library yielded the top 10 dock scoring compounds as potential inhibitors of PI3Kγ kinase. The list of top 10 selected compounds with their CID ranked in a descending order includes ‘53463223’, ‘71260095’, ‘131834212’, ‘21676336’, ‘45277410’, ‘44326949’, ‘123270861’, ‘156200’, ‘131801248’, and ‘132256977’. The detailed binding analyses of these compounds revealed binding pose, interacting residues, binding strength scores, and molecular interactions. Finally, the binding of the proposed compounds was compared with the binding of the native inhibitor. The binding of the proposed compounds to residue sets common with those of the native inhibitor strengthens the prediction that the compounds may inhibit the protein similar to the native inhibitor. These compounds were not tested for anticancer activity as reported in the PubChem Bioassay database. Interestingly, among lowest 10 scoring compounds, one compound was found inactive on 53 cancer cell lines and on NCI Yeast anticancer drug screening. This was confirmed by our prediction model (low scoring should not possess anticancer activity) and thus validates our predictions.
Acknowledgements
This project was funded by the Deanship of Scientific Research (DSR) at King Abdulaziz University, Jeddah, under grant No: (G: 418-141-38). The Authors, therefore, acknowledge with thanks DSR for the technical and financial support.
Footnotes
Authors' Contributions
Conceived and designed the experiments: MR. Performed the experiments: MR. Analyzed the data: MR. Wrote the paper: MR, MMM, ST, HMAH, GMA.
Conflicts of Interest
The Authors have no conflicts of interest to disclose in relation to this study.
- Received June 3, 2020.
- Revision received July 3, 2020.
- Accepted July 6, 2020.
- Copyright© 2020, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved