Molecular Modeling, Synthesis and Biological Evaluation of N-Phenyl-4-Hydroxy-6-Methyl-2-Quinolone-3-CarboxAmides as Anticancer Agents

The emergence of phosphatidylinositol 3-kinase (PI3Kα) in cancer development has accentuated its significance as a potential target for anticancer drug design. Twenty one derivatives of N-phenyl-4-hydroxy-6-methyl-2-quinolone-3-carboxamide were synthesized and characterized using NMR (1H and 13C) and HRMS. The derivatives displayed inhibitory activity against human epithelial colorectal adenocarcinoma (Caco-2) and human colon cancer (HCT-116) cell lines: compounds 8 (IC50 Caco-2 = 98 µM, IC50 HCT-116 = 337 µM) and 16 (IC50 Caco-2 = 13 µM, IC50 HCT-116 = 240.2 µM). Results showed that compound 16 significantly affected the gene encoding AKT, BAD, and PI3K. The induced-fit docking (IFD) studies against PI3Kα demonstrated that the scaffold accommodates the kinase domains and forms H-bonds with significant binding residues.

In this context, we designed and synthesized a series of N-phenyl-4-hydroxy-6-methyl-2 -quinolone-3-carboxamides to address the effect of incorporating a methyl moiety on the biological activity and to better understand the structural activity relation (SAR) of this series. This work delineates the design and synthesis of novel N-phenyl-4-hydroxy-6-methyl-2-quinolone-3-carboxamides employing structure-based drug design and molecular modeling tactics. Biological evaluation of the prepared compounds along with pan PI3K inhibitors (LY294002) has been investigated in vitro against human epithelial colorectal adenocarcinoma (Caco-2) and human colon carcinoma (HCT-116) cell lines.

Chemistry
Target compounds  were designed to probe the influence of introducing different functionalities at the carboxamide side chain on the bioactivity of the 4-hydroxy-6-methylquinolinone scaffold. The reaction of 3 and 4 in basic medium under reflux generated the target scaffold 5 with 70% yield, as outlined in Scheme 1. Monitoring the reaction progress was applied by employing thin layer chromatography (TLC); the disappearance of the ethyl anthranilate spot indicated that the reaction was completed. Prospective compounds 7-27 have been synthesized by reacting 5 with an excess of the corresponding amine (ArNH 2 ) using DMF under reflux, as delineated in Table 1 and Scheme 1. The absence of the limiting reactant spot (5) on the TLC plate indicated that the reaction was completed. Confirmation of the identity of the chemical structures was carried out using NMR and HRMS. The detailed data in the experimental section accord with the target structures.
In this context, we designed and synthesized a series of N-phenyl-4-hydroxy-6-methyl-2quinolone-3-carboxamides to address the effect of incorporating a methyl moiety on the biological activity and to better understand the structural activity relation (SAR) of this series. This work delineates the design and synthesis of novel N-phenyl-4-hydroxy-6-methyl-2-quinolone-3carboxamides employing structure-based drug design and molecular modeling tactics. Biological evaluation of the prepared compounds along with pan PI3K inhibitors (LY294002) has been investigated in vitro against human epithelial colorectal adenocarcinoma (Caco-2) and human colon carcinoma (HCT-116) cell lines.

Chemistry
Target compounds  were designed to probe the influence of introducing different functionalities at the carboxamide side chain on the bioactivity of the 4-hydroxy-6-methylquinolinone scaffold. The reaction of 3 and 4 in basic medium under reflux generated the target scaffold 5 with 70% yield, as outlined in Scheme 1. Monitoring the reaction progress was applied by employing thin layer chromatography (TLC); the disappearance of the ethyl anthranilate spot indicated that the reaction was completed. Prospective compounds 7-27 have been synthesized by reacting 5 with an excess of the corresponding amine (ArNH2) using DMF under reflux, as delineated in Table 1 and Scheme 1. The absence of the limiting reactant spot (5) on the TLC plate indicated that the reaction was completed. Confirmation of the identity of the chemical structures was carried out using NMR and HRMS. The detailed data in the experimental section accord with the target structures.

Biological Evaluation of the Synthesized Compounds
In order to investigate the inhibitory activity of the verified compounds (5, 7-27) against PI3Kα, we probed their antiproliferative activity in human epithelial colorectal adenocarcinoma (Caco-2) and human colon cancer (HCT-116) cell lines as shown in Table 2. The epithelial colorectal adenocarcinoma (Caco-2) expresses the wild-type (WT) PI3Kα [25][26][27]. The malignant (HCT-116) carcinoma encodes both wild-type (WT) and mutant (MUT) (H1047R) PI3Kα [28]. Therefore, the difference in activity against Caco-2 and HCT-116 accounts for MUT (H1047R) PI3Kα. All examined compounds were examined against skin fibroblast cells at concentration ranges between the IC 50 and double IC 50 values for each compound. At each concentration point, cell viability was compared to the negative control of untreated cells. We found less than 10% growth reduction for all examined compounds at any concentration point.
The inhibitory activity against Caco-2 for compounds 14, 15, and 16 reveals that attaching the p-F moiety on the phenyl carboxamide side chain (13) induces activity that suggests that a hydrophobic and/or H-bond interaction mediate(s) ligand/PI3Kα complex formation. The activity of 14, 15, and 16 against HCT-116 indicates distinct inhibitory activity and highlights the activity of 16 in Caco-2; 16 exerted potent inhibitory activity in Caco-2. In contrast, the activity of 17, 18, and 19 in Caco-2 and HCT-116 shows that a steric factor might hinder their proper orientation in the binding site.
The biological data of 11, 12, and 13 in Caco-2 suggests that an H-bond interaction presides ligand/PI3Kα complex formation. Furthermore, comparing the activity of p-F (16) with that of p-OH (11) clarifies that a steric factor affects the accommodation of -OH in the binding domain. Additionally, contrasting the inhibitory activity of p-OCH 3 (12) and of p-CH 3 (13) in Caco-2 implies that an H-bond interaction might drive ligand/PI3Kα binding and/or an oxygen atom might push -CH 3 deeply into the binding cleft. Furthermore, the inhibitory activity of 11, 12, and 13 in HCT-116 declares that an H-bond interaction mediates ligand/PI3Kα interaction; 11 provides an H-bond donor and acceptor, whereas 12 offers an H-bond acceptor. Interestingly, the activity of p-OCH 3 (12) and p-CH 3 (13) affirms that an H-bond donor, exemplified by (11), dominates ligand/PI3Kα complex formation and suggests that a small hydrophobic cleft encloses the -CH 3 motif.
Comparing the activity of 7 and 8 in Caco-2 infers that a nitrogen atom in 8 might mediate the H-bond interaction with PI3Kα backbones and/or the solubility and polarity factor of pyridine might influence the compound's distribution in the cell line. Comparing the activity of 9 and 10 in Caco-2 reveals that elongation of the carboxamide side chain by the -NH motif (10) improves the activity and thus infers that the H-bond might drive ligand/PI3Kα interactions. Such a result shows the low activity of 9 and underlines the significance of the -NH motif.
The biological data of 7, 8, 9, and 10 in HCT-116 show that introducing the phenyl moiety on the carboxamide side chain (7) provokes the activity interrogating the tightness of the binding cleft, and thus, 7 accommodates the binding cleft. Contrasting the activity of 7 and 8 suggests that a hydrophobic lining encloses the phenyl motif and understates the significance of the H-bond in HCT-116. Furthermore, the activity of 9 and 10 confirms the narrowness of the binding cleft and provides a further clue to the proper orientation of 7 in the kinase domain. Furthermore, the antiproliferative activity of 7, 8, 9, and 10 in Caco-2 and HCT-116 displays comparable activity for 7 and 9 and distinct activity for 8 and 10 in both cell lines.
Tailoring the derivatives with -COOH (20, 21, and 22) confirms that the H-bond mediates ligand/PI3Kα complex formation in Caco-2 on the p-position (22). Such a finding emphasizes the importance of the H-bond on the ligand/PI3Kα interaction and further explains the activity of p-F (16) and p-OH (11) in Caco-2. Additionally, the result confirms that the steric factor impedes their proper location in the binding site. Distinctly, p-F (16) and p-OH (11) exerted higher activity than that of p-COOH (22) in Caco-2.
The activity of 20, 21, and 22 in HCT-116 demonstrates that ionic and/or H-bond(s) guide(s) ligand/PI3Kα complex formation on the m-position (21). Furthermore, the activity of m-F (15), m-COOH (16), and m-CF3 (18) in HCT-116 accentuates the significance of the H-bond donor and/or ionic bond on the ligand/PI3Kα interaction. Contrasting the activity of 20 and 23 in Caco-2 and HCT-116 infers that the ester moiety improves cell membrane permeability and consequently enhances the activity.
The inhibitory activity of 24 and 25 in Caco-2 declares that tailoring the benzoic acid with -Cl (24) and -CH 3 (25) improves the activity, suggesting that -Cl and -CH 3 might orientate the derivatives properly in the binding site and/or provide an extra binding interaction. Contrasting the activity of 20, 24, and 25 in HCT-116 shows that attaching -Cl (24) and -CH 3 (25) induces the activity, suggesting that -Cl and -CH 3 might place the ligands suitably in the binding cleft and/or furnish an extra binding interaction.
The activity of p-SH (26) in Caco-2 provides an extra clue for the H-bond interaction on the p-site. Comparing the activity of p-OH (11) and p-SH (26) confirms that the H-bond drives the ligand/PI3Kα interaction. Moreover, contrasting the activity of p-SCH 3 (27) to that of p-OCH 3 (12) in Caco-2 suggests that S and O push -CH 3 deeply into the binding site and thus in turn improves the activity. The difference in activity between 12 and 27 might be due to the H bond supplied by the O atom. It is worth noting that the series exhibited distinct antiproliferative activity in both cell lines, suggesting differences in the binding sites of WT PI3Kα and MUT (H1047R) PI3Kα, which accords with our previous data on PI3Kα [13,14,[16][17][18][19]23,24].
Comparing the biological data of p-OH (11) and p-SH (26) in HCT-116 highlights the significance of the H-bond interaction on p-site. Furthermore, the difference in activity between p-OCH 3 (12) and p-SCH 3 (27) in HCT-116 shows the significance of the H bond assigned by the O atom and/or the steric interference of the S atom, which impedes the proper conformation of 27 in the binding cleft. Finally, the activity of 5 against Caco-2 and HCT-116 cell lines highlights the significance of the tailored carboxamide motif.
The explored cell lines encode both WT and MUT PI3Kαs; therefore, isolating either gene is highly recommended to confirm the inhibitory activity against purified PI3Kα. Future validations should be performed through knocking down any gene to interpret the mechanism of inhibition against each purified enzyme.
The effect of compound 16 on the PI3K/AKT signaling pathway was investigated using real time PCR. According to our results, the relative gene expression of PI3K, AKT, and BAD was significantly affected by treatment with compound 16 (1 µM) in a manner consistent with the positive control treatment (1 uM) ( Figure 3). A significant decrease in PI3K and AKT gene expression was evident upon treatment with either compound 16 or the positive control LY-294002. On the other hand, the expression of the pro-apoptotic gene BAD was significantly increased in treated cells (for both compound 16 and the positive control) when compared to the negative control where there was no treatment applied.

Molecular Docking
In order to test whether the anticancer activity of the synthesized compounds (5, 7-27) in Caco-2 and HCT-116 cell lines (both strongly upregulate PI3Kα) can be partly attributed to PI3Kα modulation, we performed molecular docking studies in PI3Kα crystal structures. For that, we employed the coordinates of WT (PDB ID: 2RD0) [4] and MUT (H1047R) (PDB ID: 3HHM) [29] PI3Kα to determine the structural basis of the PI3Kα/ligand interaction.
In order to explore the binding interaction of PI3Kα and the verified compounds (5, in the kinase domain of PI3Kα, we carried out induced-fit docking (IFD) studies [30][31][32] against the kinase domains of 2RD0 and 3HHM. IFD data illustrated that 5, 7-27 reside in PI3Kαs kinase domains and the docked pose of 14 superposes the crystal structure of X6K in the kinase domain of PI3Kα (PDB ID: 4L23) [33] (Figure 4).
The IFD approach probes the conformational changes in proteins in the following manner: ligands are docked to a protein's binding site recruiting Glide docking and the top ligand geometries are minimized along with protein binding site using the Prime module. Next, a redocking procedure is employed against the relaxed protein. Therefore, protein plasticity is considered during the docking tactic. The backbones of the synthesized molecules form H-bonds with S773, S774, A775, K776, W780, K802, D810, Y836, E849, V851, N853, S854, Q859, D915, H917, S919, N920, and D933 (Table 3, Figure 5). Furthermore, other computational [14,16,17,19,20,34] and experimental studies [4] have assigned the contribution of these key binding amino acids in the PI3Kα/ligand binding interaction. Interestingly, 5, 7-27 displayed comparable affinity toward both WT (2RD0) and MUT (H1047R) (3HHM) PI3Kα. Moreover, the prevalence of S774 in the ligand/PI3Kα interaction suggests that the series might be selective PI3Kα inhibitors [14].   In order to get further details about the tailored functionalities of 7-27, we screened them against an adopted pharmacophore model of active PI3Kα inhibitors [13].   Our analysis further revealed that our synthesized molecules differ in their lead-like and drug-like properties as shown in Figure 7B. All synthesized molecules passed the lead-like scoring filter LLS-02 [36], but only molecules 5, 7 and 8 passed the lead-like scoring filter DDL-01 [37]. Our molecules also showed diverse drug-like properties based on three drug-like scoring filters: DLS-04 [38], DLS-05 [39], and Dragon Consensus Score for drug-likeness [40]. For example, molecule 5 was the only molecule that passed the DLS-05 drug-like filter, while molecules 5, 7, 9-13, 23, 25-27 passed the DLS-04 drug-like filter. All values for all calculated drug-like scores are found in Supplementary Materilas Table S1. Drugs that have higher overall drug-like scores have better chances to succeed in further experimental testing including animal studies.

Chemistry
All chemicals, reagents, and solvents were of analytical grade and used directly without further purification. Chemicals were purchased from the corresponding companies: SD Fine-Chem Limited Nuclear magnetic resonance (NMR) 1 H-and 13 C-NMR spectra were measured on a Bruker (Mundelein, IL, USA), Avance DPX-500 MHz spectrophotometer (The University of Jordan) and Bruker NanoBay 400 MHz spectrophotometer (the Hashemite University). Chemical shifts are given in δ (ppm) using TMS as an internal reference; the samples are dissolved in DMSO-d 6 . High-resolution mass spectra (HRMS) were recorded using a Bruker APEX-IV (7 Tesla) instrument. External calibration was executed using an arginine cluster at a mass range of m/z 175-871, and the samples were dissolved in methanol and drops of formic acid.

MTT Assay
All cells were plated at density of 8 × 10 3 cells per well in 96-well plates and incubated to allow attachment for 24 h. The in vitro evaluation of the antiproliferative activities of the examined series was accomplished using the 3-(4,5-dimethylthiazol-2yl)-2,5-diphenyl tetrazolium bromide (MTT) colorimetric assay, as previously described [41]. In brief, compounds were diluted in culture media to yield the required concentration and applied to test wells for 48 h at 37 • C in a 5% CO 2 incubator. Three triplicates of each concentration for all tested compounds were evaluated in three independent assays (n = 9). DMEM samples were employed as negative controls, and doxorubicin as a positive control. At the end of the exposure period, 20 µL of 0.5 mg/mL of MTT was added to each well and incubated for 4 h, afterword its reduction to formazan by metabolically active cells was calculated via measuring the absorbance at 570 nm. Cell viability was calculated based on the measured absorbance relative to the absorbance of the cells exposed to the negative control, which represented 100% cell viability.

Statistical Analysis
Data analysis was performed using GraphPad Prism software version 7 (San Diego, CA, USA). The differences between treatment groups were determined by t-test or one-way analysis of variance (ANOVA) followed by Tukey post hoc T-test as appropriate. Data were expressed as mean ± SD and p < 0.05 was considered a statistically significant difference. A non-linear regression analysis was used to calculate IC 50 values.

RNA Extraction
Total RNA was extracted from the cultured cells using Quick-RNA MiniPrep according to the manufacturer's instructions as follows: harvested cells were re-suspended in RNA lysis buffer and centrifuged, the supernatant then was transferred into a Spin-Away Filter in a collection tube and centrifuged, and then 95% ethanol was added and mixed well. The mixture was then transferred to a Zymo-Spin IIICG Column in a collection tube and centrifuged, followed by adding RNA Prep Buffer to the column, followed by two washing steps by adding RNA wash buffer to the column and then centrifuged. Finally, the RNA was eluted by adding DNase/RNase-free water to the column and it was centrifuged. Samples were quantified using a spectrophotometer via absorbance at 260/280 nm.

Complementary DNA (cDNA) Synthesis
Complementary cDNA synthesis was performed using ProtoScript First Strand cDNA Synthesis Kit following the manufacturer's instructions. In brief: 1000 ng from total RNA was added to a total of 20 µL reaction volume that included 60 µM random primer mix, reaction mix, 10× enzyme mix, and nuclease-free water. The tubes were then incubated at 25 • C for 5 min followed by incubation at 42 • C for 1 h. The cDNA was then then stored at −20 • C until further analysis.

Real-Time PCR
The sequences of the primers that were used in real-time PCR assay are shown in Table 4. The reaction mixtures consisted of 200 ng cDNA template, 10 µM of each primer, 10 µL of 2× SYBER Premix Ex Taq II (Takara BIO INC, Shiga, Japan), and the total reaction volume was 20 µL. The reaction was carried out using a BIO RAD iQ5 Multicolor real-time PCR Detection System thermal cycler under the following reaction conditions: 1 cycle of 2 min at 95.0 • C, followed by 45 cycles of 10 s at 95.0 • C, 25 s at 57.0 • C, 25 s at 60.0 • C, and a final cycle of 30 s at 55.0 • C. To confirm that only one PCR product was amplified, dissociation curve analysis of amplification products was performed at the end of the last amplification cycle.

Preparation of PI3Kα Structure
The X-ray structures of WT PI3Kα (PDB ID: 2RD0) [4] and (PDB ID: 4L23) [33] as well as MUT (H1047R) PI3Kα (PDB ID: 3HHM) [29] were retrieved from the RCSB Protein Data Bank. The coordinates of X6K in 4L23 [33] were transferred to 2RD0 and assigned as a ligand to derive the grid file. The homology modeled structures of 2RD0 and 3HHM were adopted for this study [14]. Energy minimization was employed to reduce steric clash.
Further treatment of the minimized structures was performed using Protein Preparation module in Schrödinger enterprise [32] to optimize H-bond interactions between backbones.

Preparation of Ligand Structures
The synthesized compounds (ligands) were built using wortmannin's coordinates in 3HHM. The ligands were modeled using MAESTRO [32] Build wizard and energetically minimized by the MacroModel module using OPLS2005 force field.

Induced-Fit Docking (IFD)
The co-crystallized ligand (wortmannin) was assigned as a centroid in the kinase binding site of 2RD0 [4] and 3HMM [29]. The Vander Waals scaling factors for receptor and ligand were adjusted to 0.5 to furnish flexibility for the best docked ligand conformation. Other parameters were set as default. The ligand pose with the highest XP Glide score was identified.

Molecular Descriptors
All molecular structures, sketched in ChemDraw [42] and saved in SDF file format, were standardized according to the methods described by Hajjo et al. [43]. Next, two groups of molecular descriptors, comprising 'Drug-like Indices' and 'Molecular Properties' calculated using alvaDesc software from Kode Cheminformatics [40], were generated for compounds 5, 7-27.

Principal Component Analysis (PCA)
A principal component analysis was performed on structures 5 and 7-27 using drug-like indices and molecular properties. All calculations and PCA analysis were performed using alvaDesc software from Kode Cheminformatics [40].

Conclusions
Phosphatidylinositol 3-kinase (PI3Kα) has been underlined as a potential target for anticancer drug design. We identified a series of N-phenyl-4-hydroxy-6-methyl-2-quinolone-3-carboxamide as possible PI3Ká inhibitors. Biological evaluation showed that the series exhibited high inhibitory activity against Caco-2 and HCT-116 cell lines. Results revealed that compound 16 has a substantial effect on AKT, BAD, and PI3K gene expression. Docking studies against PI3Kαs displayed that the scaffold orientates in PI3Kαs kinase domains and form H-bonds with key binding residues. We are looking to optimize the core structure of this series to induce its anticancer activity and selectivity against a panel of kinases.