In Vitro and In Silico Biological Studies of 4-Phenyl-2-quinolone (4-PQ) Derivatives as Anticancer Agents

Our previous study found that 2-phenyl-4-quinolone (2-PQ) derivatives are antimitotic agents, and we adopted the drug design concept of scaffold hopping to replace the 2-aromatic ring of 2-PQs with a 4-aromatic ring, representing 4-phenyl-2-quinolones (4-PQs). The 4-PQ compounds, whose structural backbones also mimic analogs of podophyllotoxin (PPT), maybe a new class of anticancer drugs with simplified PPT structures. In addition, 4-PQs are a new generation of anticancer lead compounds as apoptosis stimulators. On the other hand, previous studies showed that 4-arylcoumarin derivatives with 5-, 6-, and 7-methoxy substitutions displayed remarkable anticancer activities. Therefore, we further synthesized a series of 5-, 6-, and 7-methoxy-substituted 4-PQ derivatives (19–32) by Knorr quinoline cyclization, and examined their anticancer effectiveness. Among these 4-PQs, compound 22 demonstrated excellent antiproliferative activities against the COLO205 cell line (50% inhibitory concentration (IC50) = 0.32 μM) and H460 cell line (IC50 = 0.89 μM). Furthermore, we utilized molecular docking studies to explain the possible anticancer mechanisms of these 4-PQs by the docking mode in the colchicine-binding pocket of the tubulin receptor. Consequently, we selected the candidate compounds 19, 20, 21, 22, 25, 27, and 28 to predict their absorption, distribution, metabolism, excretion, and toxicity (ADMET) profiles. Pharmacokinetics (PKs) indicated that these 4-PQs displayed good drug-likeness and bioavailability, and had no cardiotoxic side effects or carcinogenicity, but we detected risks of drug–drug interactions and AMES toxicity (mutagenic). However, structural modifications of these 4-PQs could improve their PK properties and reduce their side effects, and their promising anticancer activities attracted our attention for further studies.


Introduction
Chemotherapy is the most common remedy for localized and metastatic cancers, which uses a single drug or a combination of drugs to kill rapidly growing cancer cells [1]. Microtubules are an important chemotherapeutic target, and the anticancer activities afforded by tubulin inhibition from various small-molecular drugs were assessed [2,3]. Based on our previous studies of 4-phenyl-2-quinolone (4-PQ) analogs as microtubule inhibitors [4,5], we continued to focus on discovering novel 4-PQ derivatives in our current work.
Microtubules are composed of αand β-tubulin heterodimers [2,3,6,7], and their main functions are to form the skeleton of cells, maintain cell morphology, fix and support the position of organelles, help the transport of intracellular substances, and assist in the movement of organelles, such as the separation of chromosomes, which are moved by microtubule traction. Anticancer drugs in nature against microtubules can be divided into two categories according to their mechanism of action [2,3,6,7]. One mechanism of antimitotic agents is to inhibit microtubule polymerization (assembly), known as destabilizing binders [2,3,6,7], such as colchicine, steganacin, podophyllin-lignan analogs (podophyllotoxin; PPT), Vinca alkaloids (vinblastine and vincristine), and combretastatins (combretastatin A-4; CA-4, and CA-2). The other antimitotic agent mechanism is to promote microtubule polymerization and inhibit microtubule depolymerization, which is referred to as stabilizing binders [2,3,6,7], such as taxanes (paclitaxel and docetaxel), laulimalide, and epothilone.
Because the skeleton of 4-PQs shows remarkable anticancer activities [4,21], we attempted to further develop 4-PQ derivatives. As shown in Figure 2, previous reports demonstrated that several synthetic 4-arylcoumarins (neoflavonoids, 4-phenylcoumarin compounds A-D) could inhibit the polymerization of microtubules and induce cancer cell apoptosis [22][23][24][25]. In particular, 4-arylcoumarin (C) was isolated from Exostema species (of the family Rubiaceae) and displayed cytotoxic activity against various cancer cell lines [26]. These neoflavones are abundant components that widely exist naturally in plants and have anticancer properties [27,28]. Studies reported that microtubule inhibition by 4-arylcoumarin derivatives was similar to that of combretastatin A-4, both of which bind to the colchicine-binding site [29,30]. Previous studies showed that these 4-arylcoumarin derivatives have 5-, 6-, and 7-methoxy substitutions on the coumarin core which exhibited good anticancer activity. The 2-quinolone core is the bio-isosteric structure of coumarin ( Figure 2). Consequently, these results encouraged us to replace the oxygen atom (O) on the coumarin nucleus of 4-arylcoumarin with a nitrogen atom (N) by applying the concept of bio-isosterism drug design [31]. Since our previously synthesized 6,7-methylenedioxy-4phenylquinolin-2(1H)-one (4-PQ) derivatives used 2-quinolone as the parent nucleus and displayed excellent anticancer activities [4], we attempted to synthesize further modified molecules of 4-PQs. According to the example of these 4-arylcoumarins having methoxy groups at positions 5, 6, and 7 with potential anticancer activities, a series of 4-PQ derivatives with 5-, 6-, and 7-methoxy groups were optimized and synthesized. Collectively, we established more-complete structure-activity relationships (SARs) of 4-PQ derivatives. In addition, we conducted in-silico studies to assess these 4-PQs by molecular modeling of αβ-tubulin inhibition and pharmacokinetic (PK) predictions. methoxy groups at positions 5, 6, and 7 with potential anticancer activities, a series of 4-PQ derivatives with 5-, 6-, and 7-methoxy groups were optimized and synthesized. Collectively, we established more-complete structure-activity relationships (SARs) of 4-PQ derivatives. In addition, we conducted in-silico studies to assess these 4-PQs by molecular modeling of αβ-tubulin inhibition and pharmacokinetic (PK) predictions.

Chemistry
The synthetic route of the target compound 4-PQ derivatives 19-32 was based on our previous methods [4] and illustrated in Scheme 1. Initially, various substituted acetophenones were used as starting materials, and those acetophenones 1a-e were reacted with a diethyl carbonate (2) to obtain corresponding benzoylacetates 3a-e. Compounds 3a-e were then condensed with a substituted aniline 4a-d to generate corresponding benzoylacetanilide intermediates 5-18. Eventually, the target compound 4-phenylquinolin-2(1H)-one derivatives 19-32 were obtained through Knorr quinoline cyclization (acid catalysis) [4,32,33]. This synthesis method utilized intermediate benzoylacetanilides 5-18 and heat with polyphosphoric acid (PPA) at 100-110 °C to obtain the corresponding 4-PQ compounds 19-32 by intramolecular cyclization. Synthetic 4-PQs were characterized by infrared (IR), and 1 H nuclear magnetic resonance (NMR) and 13 C NMR, and mass spectrometry, and the results of the spectrum are shown in Supplementary Information (Figures S1-S87).

Chemistry
The synthetic route of the target compound 4-PQ derivatives 19-32 was based on our previous methods [4] and illustrated in Scheme 1. Initially, various substituted acetophenones were used as starting materials, and those acetophenones 1a-e were reacted with a diethyl carbonate (2) to obtain corresponding benzoylacetates 3a-e. Compounds 3a-e were then condensed with a substituted aniline 4a-d to generate corresponding benzoylacetanilide intermediates 5-18. Eventually, the target compound 4-phenylquinolin-2(1H)-one derivatives 19-32 were obtained through Knorr quinoline cyclization (acid catalysis) [4,32,33]. This synthesis method utilized intermediate benzoylacetanilides 5-18 and heat with polyphosphoric acid (PPA) at 100-110 • C to obtain the corresponding 4-PQ compounds 19-32 by intramolecular cyclization. Synthetic 4-PQs were characterized by infrared (IR), and 1 H nuclear magnetic resonance (NMR) and 13 C NMR, and mass spectrometry, and the results of the spectrum are shown in Supplementary Information  (Figures S1-S87). The 1 H NMR spectra of target compounds 19-32 exhibited a characteristic broad peak in the range of 11.56-11.90 ppm which can be attributed to the NH of 2-quinolone and a single peak in the region of 5.86-6.39 ppm which can be attributed to H-3 of the ethylene proton in the 2-quinolone core structure. For instance, the 1 H NMR spectrum of compound 22 was characterized by the appearance of single peaks at 3.84 and 3.86 ppm, which were attributable to the methoxy groups of the 2-quinolone moiety and 4-phenyl ring. The characteristic H-3 signal of 2-quinolone appeared at 6.38 ppm. The formation of 4-phenyl-2-quinolone was also confirmed by 13 C NMR studies. N-C=O signals in the 2-quinolone core structure of 19-32 were detected within 159.94-161.92 ppm.

Anticancer Activities and SAR Studies of New 4-PQ Derivatives
We first screened the synthesized 4-PQ 19-32 to identify the pharmacology of the anticancer activities against four human cancer cell lines, including the COLO205 (colorectal adenocarcinoma), A498 (renal cell carcinoma), H460 (non-small-cell lung cancer), and Hep 3B (liver cancer) cell lines. The cell viability of cancer cells was evaluated by a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazoliun bromide (MTT) assay to determine cell proliferation inhibition. Results are presented as the 50% inhibition of cancer cell growth based on the drug concentration (IC 50 ), and the preliminary results of antiproliferation are shown in Table 2.
The preliminary results of anti-proliferation indicated that the active compounds 19, 20, 21, 22, 25, 27, and 28 only demonstrated anticancer activities in COLO205 and H460 cells (other cell lines were shown IC 50 > 50 µM). As shown in Table 3, we further investigated the selectivity index (SI) to analyze the sensitivities of each compound against COLO205 and H460, respectively. Compound 22 exhibited the highest sensitivity (SI = 66.04) toward COLO205 cells, while compounds 27 and HPK showed moderate selected ratios (SI are 2.69 and 2.86). Other compounds (19, 20, 25, and 28) displayed no sensitivities (SI < 1) compared with 22, 27 as well as HPK in COLO205. Similar results were shown in H460 cells, compound 22 had the highest selected ratio (SI = 33.64). Notably, compound HPK also displayed high sensitivity (SI = 33.27) against H460 cells. However, compounds 19, 20, 21, 25, and 28 showed no selectivity (SI < 1) in H460 cells. Overall, compound 22 is the most active and sensitive within these 4-PQ derivatives against both COLO205 and H460 cell lines.

Anticancer Activities and SAR Studies of New 4-PQ Derivatives
We first screened the synthesized 4-PQ 19-32 to identify the pharmacology of the anticancer activities against four human cancer cell lines, including the COLO205 (colorectal adenocarcinoma), A498 (renal cell carcinoma), H460 (non-small-cell lung cancer), and Hep 3B (liver cancer) cell lines. The cell viability of cancer cells was evaluated by a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazoliun bromide (MTT) assay to determine cell proliferation inhibition. Results are presented as the 50% inhibition of cancer cell growth based on the drug concentration (IC50), and the preliminary results of antiproliferation are shown in Table 2.  [4]. e The data were downloaded from the National Cancer Institute (NCI) via the Developmental Therapeutics Program (DTP) of NCI 60 cell lines screen 5 dose-response data (https://dtp.cancer.gov/; accessed on 21 December 2022), and the data were listed in the Supplementary Materials.

Compound
Ring A Ring C IC 50

Molecular Modeling of 4-Phenyl-2-quinolone (4-PQ) Derivatives Proposes Their Docking into the Colchicine-Binding Site
As previously reported, the cytotoxic activity of 4-phenyl-2-quinolone (4-PQ) [35,36] and its 4-arylcoumarin analog [24,25,29,30] are well-known as antimitotic agents associated with their ability to inhibit tubulin formation via binding of compounds to the colchicinebinding site. In this study, molecular docking was simulated to investigate the binding modalities of 4-PQ derivatives against tubulin using AutoDock4 [37] and AutoDock Vina software [38]. In 2004, the co-crystal structure of αβ-tubulin complexed with N-deacetyl-N-(2-mercaptoacetyl)colchicine (DAMA-colchicine, an analog of colchicine) was discovered by X-ray analysis (PDB ID: 1SA0) [39]. At the same time, the X-ray structure of the tubulin/PPT complex also revealed that PPT inhibits tubulin polymerization by blocking the colchicine-binding site (PDB ID: 1SA1) [39] and presents a similar orientation as that of DAMA-colchicine. Accordingly, as shown in Figure 3A, molecular docking experiments were performed on the 4-PQ series and their virtual binding interactions with αand β-tubulin (PDB ID: 1SA1), and results were compared to PPT and colchicine as reference drugs. To validate docking protocols, the co-crystallized ligand PPT and colchicine were re-docked into the colchicine-binding site of isolated αand β-tubulin proteins. As shown in Figure 3B,C, co-crystallized poses of the reference drugs were superimposed on the re-docking pose. PPT and colchicine could be re-docked to their respective target proteins with root mean squared deviation lower bounds (RMSD/lbs) of 0.017 ( Figure 3B) and 1.879 Å ( Figure 3C) via AutoDock Vina. Additionally, the re-docked conformations of PPT and colchicine had similar orientations to their respective co-crystallized ligands. The docked poses of the 4-PQs are superimposed with PPT and colchicines in Figure 3D. Results showed that the methoxy groups on the 4-phenyl ring (C ring) of 4-PQ were located in the tri-methoxy groups' position of PPT and colchicines (as indicated by yellow arrows in Figure 3D). Additionally, the methoxy groups on the 2-quinolone core (A ring) of 4-PQ were located in the methylenedioxy moiety of PPT and the ketone group of colchicines (as indicated by red arrows).
We selected some potential 4-PQ compounds, representing 19, 20, 21, 22, 25, 27, and 28, which had IC 50 values of <50 µM by the MTT assay to predict their binding modalities into the colchicine site. Results suggested that these 4-PQ derivatives could snugly occupy the colchicine-binding site of αβ-tubulin ( Figure 3A). On the other hand, a previous report on deciding which program to use in docking studies indicated that Vina is much faster and adopts more accurate binding poses but Autodock4 forms better binding affinity [40]. Therefore, we executed both docking programs to estimate the binding energies, and the possible docking poses were predicted via the AutoDock Vina. The binding affinity of each 4-PQ attached to αβ-tubulin is discussed based on the estimated ∆G value, and results are given in Table 4. The binding energies of these compounds were in the range of −7.46 to −9.91 (kcal/mol) via AutoDock4 calculation, while the range of Vina estimation is −5. mations of PPT and colchicine had similar orientations to their respective co-crystallized ligands. The docked poses of the 4-PQs are superimposed with PPT and colchicines in Figure 3D. Results showed that the methoxy groups on the 4-phenyl ring (C ring) of 4-PQ were located in the tri-methoxy groups' position of PPT and colchicines (as indicated by yellow arrows in Figure 3D). Additionally, the methoxy groups on the 2-quinolone core (A ring) of 4-PQ were located in the methylenedioxy moiety of PPT and the ketone group of colchicines (as indicated by red arrows).   All docked molecules interacted with amino acid residues of αβ-tubulin within the colchicine-binding active site (Figures 4-6); furthermore, the results were analyzed by Discovery Studio Visualizer software and summarized in Table 5. As shown in Figure 4A, the proposed binding mode of PPT formed five carbon-hydrogen bonds with ALA250, VAL315, ALA317, LYS352, and ASN350 residues. The methylenedioxy pharmacophore formed two hydrogen bonds with VAL315 and ASN350. The carbonyl group of the lactone ring formed one hydrogen bond with ALA250, and the methoxy group of PPT's phenyl ring (C-ring) formed two hydrogen bonds with ALA317 and LYS352. The phenyl ring (ring A) showed one pi-sulfur bond with MET259. In addition, the A-ring, C-ring, methylenedioxy moiety, and tri-methoxy groups of PPT formed several hydrophobic interactions with VAL181, CYS241, LEU242, ALA250, LEU255, ALA316, VAL318, LYS352, ALA354, and ILE378. Similar results were also seen for colchicine ( Figure 4B); tri-methoxy groups, the A-ring, and C-ring of colchicines formed hydrophobic interactions with CYS241, LEU242, ALA250, LEU255, ALA316, VAL318, LYS352, ALA354, and ILE378, while the carbonyl group and methoxy group of the A-ring moiety formed four carbon-hydrogen bonds with ALA180, ASN258, VAL315, and ASN350. In addition, one methoxy group of the C-ring of colchicine showed one carbon-hydrogen bond with VAL238. Taken together, these results revealed that the methoxy groups or some hydrogen bond acceptors (containing oxygen, such as carbonyl) on the A-and C-rings of the ligand played pivotal roles with the αβ-tubulin receptor in the colchicine-binding site. Interactions with the A-ring site were delineated by the following amino acid residues: ALA180, VAL181, ASN258, MET259, VAL315, ALA316, ASN350, and LYS352. On the other hand, the ligand's moiety in the C-ring site interacted with amino acid residues of αβ-tubulin, including residues CYS241, LEU242, VAL238, ALA250, LEU255, ALA316, ALA317, VAL318, LYS352, ALA354, and ILE378.    Figure 5), which is similar to the carbonyl group in the γ-lactone ring (D-ring) of PPT displayed carbon hydrogen bond with β-tubulin. The A-and B-rings of the 2-quinolone moiety displayed several hydrophobic interactions with LEU248, ALA316, and LYS352 ( Figure 6A-D). In addition, the amide groups (NH) of compound 28 formed a conventional hydrogen bond with THR179 ( Figure 6C). These findings implied that the 2-quinolone core structure plays an important role in the colchicine-binding domain, which is consistent with our previous study of 4-benzyloxy-2-quinolone derivatives on tubulin inhibition [5]. Each of their A-and C-rings revealed similar hydrophobic interactions with αβ-tubulin, which were just like those of PPT and colchicine.

In Silico Prediction of Drug-Likeness Studies
Drug-likeness is considered an important tool for predicting whether a molecule can be a potential drug candidate. Therefore, we employed SwissADME [41] to forecast the

In Silico Prediction of Drug-Likeness Studies
Drug-likeness is considered an important tool for predicting whether a molecule can be a potential drug candidate. Therefore, we employed SwissADME [41] to forecast the bioavailability and drug-likeness properties, and the results are shown in Table 6. Several rules were used to assess the drug-likeness, the most famous of these descriptions is Lipinski's rule of five [42]. According to Lipinski's rule, a compound is labeled as drug-like when it meets the following criteria: molecular weight < 500 Da, H-bond donors (HBDs; the sum of OHs and NHs) ≤ 5, H-bond acceptors (HBAs; the sum of Ns and Os) ≤ 10, and Log P ≤ 5. The Log P values are measured as the partition coefficient between n-octanol and water and represented the molecular hydrophobicity or lipophilicity. High Log P values indicate poor absorption or low permeability, while low Log P values indicate increased absorption and permeability. Log P values of all predicted compounds were <5 (ranging from 2.25-3.05), which is within an acceptable range. The molecular polar surface area (PSA) is a very useful parameter for predicting drug transport properties. The PSA is defined as the sum of surface polar fragments (usually oxygens, nitrogens, and attached hydrogens) in a molecule. The topological PSA (TPSA) provides results of practically the same quality as the classical three-dimensional PSA; the calculations, however, are two to three orders of magnitude faster [43]. The 4-PQ compounds had TPSA values ranging from 32.86-69.78 Å 2 , which is a good range for oral bioavailability since most therapeutic molecules have TPSA values of <140 Å 2 for good passive membrane permeability (in a non-polar environment) [44].
Moreover, none of the 4-PQs were predicted as pan assay interference compounds (PAINS) [45], which suggests the absence of potential pan-assay interferences or being frequent hitters of promiscuous compounds. On the other hand, we found no violations of Lipinski's [42], Veber's [44], Ghose's [46], Egan's [47], or Mugge's [48] rules for any of the synthesized 4-PQ compounds based on the obtained data.

In Silico Analysis of Potential PK (ADMET) and Toxicological Properties (T)
It is clear that in addition to pharmacological properties, ADMET (absorption, distribution, metabolism, elimination, and toxicity) research plays a crucial role in the success of drug candidates [49,50]. Because of the impact on eventual success, such studies now occur early in the drug discovery process [51]. The PK characteristics of 4-PQ compounds (19, 20, 21, 22, 25, 27, and 28), PPT, and colchicine were assessed using online SwissADME software [41] and admetSAR [52]. As shown in Table 7, all compounds were predicted to have high gastrointestinal (GI) absorption (human intestinal absorption; HIA). In addition, all of the 4-PQ derivatives were predicted to have the ability to pass through the blood-brain barrier (BBB), except for PPT and colchicine. P-glycoprotein (P-gp), a member of ATP-binding cassette (ABC) transporters, is an efflux transporter that can pump drugs out of cells to reduce the biological effects of anticancer drugs [53], which may cause a failed treatment. If a drug under investigation is identified as a P-gp substrate, this means that its concentration levels in cells may be clinically affected by a P-gp inhibitor. On the other hand, P-gp plays a key role in multidrug resistance (MDR) in cancer [54]. Most of these were not P-gp substrates, as only HPK and colchicine were predicted to have potential interference with P-gp.
Cytochrome P450 (CYP450) enzymes are pivotal mediators in the metabolism of many medicines. The CYP450 expression level determines a drug's metabolic rate. Unanticipated drug-drug interactions (DDIs) and drug metabolism problems based on CYP450 enzymes are also common causes of adverse drug events (ADEs) [55]. In the metabolic profiles, all of these compounds served as different kinds of CYP450 enzyme inhibitors ( Table 7), suggesting that they may compete with other drugs, which means that they can block other drugs' metabolic activity to extend the half-life of therapeutic drugs in patients' serum levels. Further PKs were investigated in the future for any potential drug-drug interactions. In particular, compounds 22, 25, 27, 28, and HPK inhibited more than three types of CYP450 enzymes, including CYP1A2, CYP2C9, CYP2D6, and CYP3A4, which reveals that they would likely interfere with the metabolism of other drugs using the same CYP450 pathway. These findings implied that using these 4-PQs should be done while being aware of the risks of CYP450 enzyme-based unanticipated DDIs and metabolism problems. On the other hand, several computational approaches can help predict the potential toxicity of new compounds in the initial stages of drug discovery, primarily considering human ether-à-gogo-related gene (hERG) inhibition, AMES toxicity, and carcinogenicity. hERG potassium channels are essential for regulating electrical activity in the human heart [56]. Inhibition of hERG potassium channels may lead to long QT syndrome (LQTS), which is known as a fatal ventricular tachyarrhythmia called torsades de pointes in clinical studies [57,58]. As shown in Table 8, the toxicity profile predicted by hERG indicated that 4-PQs were non-hERG inhibitors. Moreover, the carcinogenicity assessment showed that 4-PQ compounds were non-carcinogens. However, AMES toxicity showed that most of the 4-PQ compounds were found to be AMES toxic (mutagens), including compounds 20, 21, 22, 25, and HPK. These results revealed that when using these 4-PQs, one should be aware of potential DDI side effects and risks of mutagenicity. The U.S. Environmental Protection Agency (EPA) established four toxicity categories for acute hazards: class I is high toxicity and severely irritating, class II is moderate toxicity, class III is slightly toxic and low toxicity, and class IV is non-toxic [59]. All of the predicted compounds were found to have 50% lethal dose (LD 50 ) values that ranged from 2.1994-3.0013 mol/kg and were classified as class III. These results implied that these 4-PQs have low acute toxicity. Table 7. ADMET analysis of 4-PQ compounds, podophyllotoxin (PPT), and colchicine was evaluated by SwissADME [41].
The SwissADME software briefly describes six essential properties of oral bioavailability to create a bioavailability radar and boiled egg plots for analysis. These properties and suitable values are lipophilicity (LIPO): Log P (XLOGP3) value from −0.7 to 5.0; SIZE, molecular weight in the 150-500 g/mol range; polarity (POLAR), TPSA values from 20 to 130 Å 2 ; insolubility (INSOLU), solubility with Log S (ESOL) from −6 to 0; insaturation (INSATU), saturation fraction of carbons with sp3 hybridization (fraction Csp3) from 0.25 to 1; and flexibility (FLEX), the number of the rotatable bonds from 0 to 9 [41]. Results are shown in Figure 7A−J with bioavailability radar for 4-PQ compounds, and the pink area in the radar is a range of optimal values. Bioavailability radars for all 4-PQ molecules exhibited suitable parameters in terms of lipophilicity, size, polarity, insolubility, and flexibility, but were not good at insaturation (INSATU). The purpose of the boiled egg plot is to correctly predict HIA and BBB properties of substances that are important for therapeutic research [60]. The blue dots represent P-gp substrates (PGP + ) and the red dots indicate P-gp non-substrates (PGP − ). The boiled egg forecast of 4PQs is shown in Figure 7K, the outer gray area is for compounds with lower GI absorption and limited BBB penetration. Furthermore, the white area represents the physicochemical space of the molecule with the highest probability of passive HIA, while the yellow area represents the physicochemical space of the molecule with the highest probability of penetrating the BBB. All of the 4-PQs exhibited predicted high BBB penetration, as demonstrated by their location inside the yellow ellipse (yolk) defining BBB absorption ( Figure 7K). Essentially, all of the 4-PQs that were predicted to enter the brain also displayed a predicted high HIA (Table 7). Only colchicine and PPT were predicted to display low brain permeation but high intestinal absorption.
We further employed the site of metabolism (SOM) prediction (SOMP) [61], a web service for in silico prediction of the SOM, to gain insights into the biotransformed positions of the most effective compound 22 (the original data can be found in Supplementary Information Tables S1-S5). The predicted results for compound 22 are illustrated in Figure 8A. In addition, possible metabolites of compound 22 were presented by BioTransformer 3.0 [62], a web server that accurately predicts metabolic transformation products. As shown in Figure 8B, several metabolites of 4-PQ compound 22 were formed by aromatic hydroxylation on its aromatic A-and C-rings via phase I metabolism of CYP450. Typical metabolism of the methoxy group, O-dealkylation, was also found on compound 22's 6-methoxy of the 2-quinolone core and 3 -methoxy of the 4-phenyl ring. Of note, the epoxidation of an alkene was catalyzed by CYP450 to form an epoxide on ring B of compound 22. Further phase II metabolism takes place to form OH-glucuronidation and glutathione (GSH)-conjugation of epoxide by UDP-glucuronosyltransferase and glutathione transferase, respectively. These hydrophilic conjugates and their metabolites can be excreted in phase III of their metabolism.

General Procedure for the Synthesis of Benzoylacetates (3a-e)
Compound ethyl 3-oxo-3-phenyl propanoate (3a) was purchased commercially and used without further purification. Compounds 3b-e were prepared according to our previously reported procedures [4]. A mixture of diethyl carbonate (2; 1 equivalent) and appropriate acetophenone (1b-e; 1 equivalent) was dissolved into 50-100 mL toluene, and then sodium hydride (NaH) (2 equivalents, 60% dispersed in mineral oil) was added. The mixture was stirred and heated to reflux for 30 min; the reaction was monitored by thin layer chromatography (TLC; with n-hexane/ethyl acetate = 10/1 as the eluent solution). After the reaction was complete, the mixture was poured into 200 mL of ice water and acidified with glacial acetic acid to pH 4-5. The mixture was extracted with ethyl acetate, dried, filtered with anhydrous magnesium sulfate (MgSO 4 ), concentrated under reduced pressure, and separated and purified by column chromatography (n-hexane/ethyl acetate = 10/1) to obtain the corresponding benzoylacetates (3b-e). All synthetic compounds were in agreement with 1 H NMR and 13 C NMR spectroscopic data.

General Procedure for the Synthesis of Benzoylacetanilides (5-18)
A mixture of the benzoylacetates (3a-e, 1 equivalent) and substituted aniline (4a-d, 1 equivalent) was stirred and heated to reflux in 50-100 mL toluene overnight. Reactions were monitored by TLC. After the reaction was complete, the mixture was cooled down to room temperature, and if a solid had formed, it was filtered by suction. If no solid had formed, then it was extracted with 10% NaOH, the aqueous layer with glacial acetic acid was acidified to pH 4-5, and the solid was separated out. The resulting precipitate was isolated by suction filtration. If no solid was found after acidification, the residue was extracted with ethyl acetate, and the organic layer was dried with anhydrous magnesium sulfate and filtered. The filtrate was concentrated under reduced pressure and purified by column chromatography to obtain the corresponding benzoylacetanilides (5-18).
The tested compounds were initially prepared in the stock solution with a concentration of 100 µM and diluted to evaluate until found the IC 50 values. The IC 50 values of more than 50 µM are regarded as having no anti-proliferation effect. After 48 h of treatment, the cell growth rate was measured on an enzyme-linked immunosorbent assay (ELISA) reader at a wavelength of 570 nm, and IC 50 values of the test compounds were calculated.

Molecular Docking
The crystal structure of αβ-tubulin with the co-crystallized inhibitors, N-deacetyl-N-(2-mercaptoacetyl)colchicine (DAMA-colchicine) and podophyllotoxin (PPT), was obtained from the Protein Data Bank under respective PDB codes of 1SA0 and 1SA1 [39]. The chemical structures of 4-phenyl-2-quinolone (4-PQ) derivatives were drawn, and the energy of the 3D structure was minimized using the MM2 force field [63] in ChemBio3D ultra 12.0. Molecules were docked into the colchicine-binding pocket using AutoDock4 [37] and AutoDock Vina [38]. The binding abilities were calculated as the binding free energy (∆G, kcal/mol) in both AutoDock4 and Vina. The grid map with autogrid (for grid calculation) within MGLTools (version 1.5.6) was carried out to define interactions of the protein and ligand in the colchicine-binding site. The grid box size was built of 40 × 40 × 40 points in the x, y and z directions, and the centers of x, y and z dimensions are 117, 89, and 7.5 in AutoDock4. The Vian's parameters were set as follows: a grid box size was 20 × 20 × 20, and the grid center was located at x = 117, y = 89, and z = 7.5. The molecular 3D-docking results were prepared and visualized using PyMOL [64] (https://pymol.org/2/; accessed on 6 May 2022). 2D-interaction maps of docked complexes were visualized with the BIOVIA Discovery Studio Visualizer (http://accelrys.com; accessed on 6 May 2022) to show interactions of compounds with amino acid residues of the target protein.

Conclusions
Previously, 2-quinolone derivatives were identified as anti-tubulin agents [5,36]. To discover novel 2-quinolone-bearing molecules of 4-phenyl-2-quinolone (4-PQ) derivatives with anticancer activities, we attempted to design and synthesize a series of methoxysubstituted 4-PQ compounds. In the current study, we investigated the structure-activity relationships of the methoxy groups at positions 5, 6, and 7 on the 2-quinolone core, and evaluated the anticancer effects of the methoxy substitutions on their 4-phenyl ring. Preliminary antiproliferative effects of compounds 19-32 indicated that most of the 4-PQ derivatives had IC 50 values of >50 µM. However, some of the 4-PQs, such as compounds 19, 20, 21, 22, 25, 27, and 28, showed potential anticancer effects against COLO205 and H460 cells (with IC 50 values of <50 µM). In particular, the compound 6-methoxy-4-(3-methoxyphenyl)quinolin-2(1H)-one (22) exhibited excellent anticancer activities against COLO205 and H460 cancer cells with respective IC 50 values of 0.32 and 0.89 µM. Computer modeling indicated that the colchicine-binding site of αβ-tubulin could deeply dock to these 4-PQs, which displayed similar orientations as those with podophyllotoxin and colchicine. The A-and C-rings of 4-PQ overlapped with those of colchicine and podophyllotoxin. In the drug discovery process, weak bioavailability and pharmacokinetics always lead to drug development failure. Of note, these 4-PQs showed suitable drug-likeness, adequate bioavailability, and non-cardiotoxicity; however, toxicity predictions indicated the risk of drug-drug interactions and carcinogenic side effects, so these 4-PQs must be used with caution. Overall, LD 50 values of these 4-PQs were classified in class III, indicating that these 4-PQs have low acute toxicity. Therefore, further structure optimization of 4-PQs is worth investigating to convert them into clinically useful anticancer agents in the future. Altogether, our findings suggest that 4-PQ derivatives with methoxy substituents may provide new insights into these compounds in antimitotic drug development, particularly through αβ-tubulin inhibition.