New arylated benzo[h]quinolines induce anti-cancer activity by oxidative stress-mediated DNA damage

The anti-cancer activity of the benzo[h]quinolines was evaluated on cultured human skin cancer (G361), lung cancer (H460), breast cancer (MCF7) and colon cancer (HCT116) cell lines. The inhibitory effect of these compounds on the cell growth was determined by the MTT assay. The compounds 3e, 3f, 3h and 3j showed potential cytotoxicity against these human cancer cell lines. Effect of active compounds on DNA oxidation and expression of apoptosis related gene was studied. We also developed a quantitative method to measure the activity of cyclin-dependent kinases-2 (CDK2) by western blotting in the presence of active compound. In addition, molecular docking revealed that benzo[h]quinolines can correctly dock into the hydrophobic pocket of the targets receptor protein aromatase and CDK2, while their bioavailability/drug-likeness was predicted to be acceptable but requires future optimization. These findings reveal that benzo[h]quinolines act as anti-cancer agents by inducing oxidative stress-mediated DNA damage.


Results and Discussion
Chemistry. The synthesis of benzo[h]quinolines is presented in Fig. 1. The required precursor was synthesized in two steps. In the first step, synthesis of 6-aryl-4-methylthio-2-oxo-2H-pyran-3-carbonitriles was performed by the reaction of methyl 2-cyano-3,3-bis-methylthio-acrylate with various aryl/heteroaryl methyl ketones in dimethylsulphoxide (DMSO) under basic conditions at room temperature. Further, 6-aryl-4-sec.amino-2-oxo-2H-pyran-3-carbonitriles 1 was obtained by amination 25 of 6-aryl-4-methylthio-2-oxo-2H-pyran-3-carbonitriles by various secondary amines in refluxing ethanol ( Fig. 1) [26][27][28] . Synthesis of highly functionalized benzo[h]quinolines was carried out by stirring an equimolar mixture of 6-aryl-4-sec.amino-2H-pyran-2-one-3-carbonitriles, 2-cynomethylbenzonitrile 2 and sodium amide in N,N-dimethylformamide at 100 °C for 35-50 h (Fig. 1). The presence of an electron donating or withdrawing functional group on the aryl group at the C-6 position of the pyran ring did not affect the yield of the desired product. Interestingly, the presence of thienyl and furyl rings in lieu of an aryl ring required longer duration of the reaction and afforded good yields. Synthesis of 2-amino-5-aryl-4-sec.aminobenzo[h]quinoline-6-carbonitriles 3 containing piperidine and morpholine at the C-4 position was also carried out. Conventional heating was replaced with microwave-assisted heating to reduce the reaction time ( Fig. 1). Heating of a mixture of 6-aryl-2-oxo-4-sec.amino-2H-pyran-3-carbonitriles 1, 2-cynomethylbenzonitrile and sodium amide in DMF at 100 °C for 55 minutes in a microwave reactor resulted excellent yields of 2-amino-5-aryl-4-sec.aminobenzo[h]quinoline-6-carbonitriles. All the studied compounds were synthesized by using the same procedure and characterized by spectroscopic analysis 29-31 . Structure activity Relationship. On the basis of result obtained after evaluation of anticancer activity, we have tried to understand the role of different functional groups on the reactivity. As it is clear from Table 1 that benzo[h]quinolines (3e, 3f, 3h and 3j) are most active amongst reported compound. Compound 3f has an IC 50 of 5.4, 4.7, 4.9 μ M against H460, MCF7 and HCT116 cells respectively, which is close to the reference compound doxorubicin (2.1 μ M) and similarly compound 3e exhibit IC 50 value 5.3, 6.8, 7.6 and 6.8 against G361, H460, MCF7 and HCT116 cancer cell line respectively. From this result we conclude that compound with furyl group 3e and thienyl group 3f at position 5 of benzo[h]quinolines have good activity, which proposed that five member aromatic ring plays important role in activity. From the comparison of compound 3b and 3h it is clear that only change of piperidin to morpholine of benzo[h]quinolines affect the cytotoxicity in almost all of the cell lines, as shown in Table 1. The presence of the 4-methoxyphenyl group (3j) results in IC 50 values of 4.8, 5.2 and 6.8 μ M against the H460, MCF7 and HCT116 cell lines, respectively, while the presence of the 4-bromophenyl group (3i) results in an IC 50 value of 6.9 μ M with the HCT116 cancer cell line. In summary, the presence of 4-chlorophenyl and 4-methoxyphenyl, furyl and theinyl ring in benzo[h]quinoline with suitable piperidine/morpholine moiety result in significant cytotoxicity.
Cytotoxicity and apoptosis studies. Compounds 3a-3j was evaluated for in vitro cytotoxic activity by MTT assay against four human cancer cell lines: G361, H460, MCF7 and HCT116. Doxorubicin was used as a positive control (Table 1). Results obtained reveals that 3f is cytotoxic against the HCT116 (IC 50  cell lines but not against the HCT116 cell line. Apoptotic effect of active compound 3e was studied at its IC 50 concentration (see Table 1) on G361, H460, MCF7 and HCT 116 cancer cells. In this assay, cells were harvested and annex in V-FITC/PI staining was performed to determine apoptotic activity. Compound 3e induced apoptosis in the range ~25-30% of cells in G361, H460, MCF7 and HCT116 cancer cell lines (Fig. 3). This result strongly  suggests that activation of intracellular ROS and oxidative DNA damage might be responsible for induced apoptosis in these cancer cells.
Intracellular ROS activation, mRNA expression and DNA oxidation analysis. Intracellular ROS levels were estimated using fluorescent probes. The ROS level detected by H2DCFDA staining was dramatically increased upon addition of compounds 3e, 3f, 3h and 3j in all the four cancer cell lines, and was significantly higher in comparison with the positive control ( Fig. 4a,b). The increase in intracellular ROS may cause oxidative stress. Hence in order to investigate the effect of the above active compounds on DNA oxidation, formation of 8-oxoguanine as a typical oxidative base lesion 7,19,29,36 was measured through the increased formation of 8-OHdG, as shown in Fig. 4c. Exposure to all active compounds lead to a substantial increase in the amount of 8-OHdG in all cancer cells. This reveals that these compounds can activate oxidative stress signaling leading to DNA oxidation. Thus, this result suggested that 3e and 3f treatment markedly induced DNA damage in cancer cells. In Fig. 4d shows that compound 3e induce the cells death might be depends on inhibition of cyclin-dependent CDK2, a serine-threonine protein kinase associated with cell cycle progression and DNA damage. In order to establish the role of increased intracellular ROS in cancer cell death induced by the above active compounds, mRNA expression was evaluated for three oxidative stress-related genes, ATM, H2AX, and BAX ( Fig. 5a-e). The ataxia telangiectasia mutated (ATM) is a serine/threonine protein kinase and responsible for activating cellular responses to DNA damage, which may be activated by H2AX and CDK2 in the DNA damage checkpoint 32-34 but a mechanistic link between these two pathways has not been clearly elucidated. H2AX plays a key role in DNA damage response and is required for the assembly of DNA repair proteins at sites containing damaged chromatin as well as for activation of checkpoint proteins, which arrest the cell cycle progression 35 . This study also demonstrates that exposure to active arylated benzo[h]quinolines increases levels of intracellular ROS, which may be partially mediated through changes in mitochondrial potential leading to cell death induced by H2AX and ATM over-expression. As shown in Fig. 5a-e reveals that cancer cells exposed to active compounds showed over expression of H2AX and ATM genes leading to DNA damage in cancer cells. Therefore, over expression of these genes plays a major role in initiating cell death in cancer cells through the DNA damage signalling pathway. Molecular modeling. The synthesized benzo[h]quinolines were evaluated for their inhibitory effects in different cancer cell lines. Molecular docking studies were carried out to ascertain the mode of action towards the molecular targets, that is known human breast cancer target receptor aromatase (PDB ID: 3EQM) 37,38 and colon cancer target protein CDK2 (PDB ID: 2R3J) 39,40 . Binding affinities were predicted by the Sybyl docking total score upon docking with the Surflex-Dock program (Sybyl X 2.0). Docking studies were carried out to evaluate the binding affinity and interactions with their target proteins. Hydrogen bonds (H-bonds, with a donor-receptor distance of 3 Å) between the ligand and amino acids in the binding site of the protein were used for the ranking of compounds. The mode of interaction of the co-crystallized ligand (3-bromo-5-phenyl-N-(pyridin-3-ylmethyl) pyrazolo [1,5-a]pyrimidin-7-amine) (SCJ) within the crystal structure of CDK2 was used as a reference binding model. The binding site of CDK2 contains the important residue Leu-83 is important for hydrophobic interactions with the ligand.
Doxorubicin (standard compound) was docked into CDK2 with a predicted total score of 5.1647. The hydrophobic cavity made by residues Asp-86, Lys-89 and Leu-83 interacted with the ligand through formation of H-bonds. Docking of compound 3e with CDK2 resulted in a predicted high binding affinity with a total score of 5.8273 and the formation of a strong H-bond (of length 2.0 Å) to the hydrophobic residue Leu-83 (Fig. 6). The predicted binding mode of compound 3e with CDK2 involved a number of binding site residues within a radius of 3 Å with diverse properties ( Table 2):basic (polar, hydrophobic, positively charged), for example Lys-33, Lys-89, His-84; aromatic (hydrophobic), for example, Phe-82; acidic (polar, negatively charged), for example, Glu-12, Asp-145 and Asp-86; polar amide, for example, Gln-85, Gln-131 and Asn-132; hydrophobic, for example, Ala-177, Ala-131, Ala-144, Gly-13, Leu-83, Leu-134, Ile-10, Val-18 and Gly-11. As a consequence, compound 3e was predicted to have strong hydrophobic interactions with CDK2, which provides a molecular rational for the activity in this compound (Fig. 6). Docking of compound 3f also predicted high binding affinity with a total score of 5.5206 and the formation of a strong H-bond (length 2.1 Å) to the hydrophobic residue Leu-83, with interactions with almost the same amino acid residues in the binding site (Fig. 7).
Docking of compounds 3f, 3g, 3h and 3j with the breast cancer target protein aromatase (PDB ID:3EQM) was also carried out. For comparison, the docking score of the control anti-cancer drug doxorubicin was predicted to be 2.2253. Docking of compound 3 f resulted in a predicted high binding affinity with a total score of 5.4971 and the formation of an H-bond with Ala-438 ( Table 3). The predicted binding mode of compound 3f with aromatase involved a number of binding site residues within a radius of 3 Å with diverse properties: hydrophobic, for example Met-374, Ala-438, Ala-306, Ile-133, Ile-132, Leu-372, Val-373, and Val-370; aromatic (hydrophobic), for example Phe-134; nucleophilic (polar, hydrophobic), for example Cys-437; nucleophilic (polar, hydrophobic), for example Thr-310, Cys-437; polar amide, for example Gln-86, Asn-105; hydrophobic positively charged, for example Arg-115; acidic (polar, negatively charged), for example, Asp-309. As a consequence, compound 3f was predicted to have strong hydrophobic interactions with aromatase, which provides a molecular rationale for its observed activity (Fig. 8).
Similarly, docking of compound 3h resulted in a predicted high binding affinity with a total score of 5.5392. The predicted binding mode of compound 3h with aromatase involved a number of binding site residues within a radius of 4 Å that were acidic (polar, negative charged), for example Asp-309. Compound 3 h was also predicted to have strong hydrophobic interactions with aromatase, which explain its observed activity (Fig. 9). Likewise, docking of compound 3j against aromatase gave a predicted high binding affinity with a total score of 5.0609 ( Table 3). The predicted binding mode of compound 3j involved several binding site residues within a radius of 4 Å that were acidic (polar, negative charged), for example Asp-309, and hydrophobic, for example Met-374. The strong hydrophobic interactions involved in the binding of compound 3j explain its observed activity (Fig. 10).
Pharmacokinetic and toxicity properties. Pharmacokinetic (PK) ADME properties are important descriptors for human therapeutic use of any drug molecule. These ADME descriptors were calculated for all active molecules and compared with standard ranges. All the derivatives possessed a good number of hydrogen bond donors and acceptors 39,40 . These derivatives were designed to increase the binding of the drug with the receptor through hydrogen bonding. These derivatives were found to follow Lipinski's rule of 5, affording drug likeness to the designed compounds. Polar surface area was calculated to estimate the ability of the compounds  to permeate cell membranes. Lipophilicity (ratio of octanol solubility to water solubility) measured through logP, which has been implicated in blood brain barrier penetration and permeability prediction. Excretion of drugs depends on MW and logP 41 .
While the benzo[h]quinolines showed significant anti-cancer activity, they also exhibited some PK limitations. The calculated TPSA values of these compounds were within acceptable limits. The distribution of compounds in the human body was described by the predicted blood-brain barrier coefficient (logBB), apparent Caco-2 permeability, log Kp for skin permeability, volume of distribution and plasma protein binding (log Khsa for serum protein binding) [42][43][44] . All compounds show poor aqueous solubility ( Table 1 in Supplementary Information). All compounds except 3g were predicted to be CYP2D6 non-inhibitors, and all derivatives were predicted to be hepatotoxic and highly-bound to plasma binding protein . Compounds 3c, 3d, 3i and 3j show medium apparent MDCK permeability. The calculated BBB values of the analogues were within the acceptable interval. The calculated values for these ADME parameters showed close similarity between the analogues and that of the reference drug doxorubicin and lie within the standard range of values exhibited by 95% of all known drugs shown in Table 4 and supplementary Figure S1.
Toxicity risks assessment. The reproductive/developmental toxicity, Ames mutagenicity, skin irritant effects, Ames score, rat oral LD 50 (mg/kg), and rat carcinogenic potency TD 50 (mg/kg) were predicted for all benzo[h]quinoline derivatives. These compounds were estimated to possess no risk of skin irritation. Compounds 3c, 3d, 3i and 3j may show developmental toxicity. All compounds were predicted to possess high mutagenicity potential at high doses or long term therapeutic use in human, except for compound 3c, which was predicted to be non-mutagenic 45,46 . Other predicted toxicity parameters are summarized in Table 5. The toxicity risks for benzo[h]quinoline derivative scanned to find the moderate to good compared with doxorubicin. Similarly, toxicity screening for USFDA rodent carcinogenicity, Ames mutagenicity, developmental toxicity potential, and carcinogenic effect irritancy are predicted to have a positive response for these benzo[h]quinoline derivatives.   were purchased from Sigma Aldrich and Alfa Aesar. All compounds were synthesized by the method described by Singh et. al. 20 (Detail data available in S1). Detailed structural characterization is mentioned therein. These derivatives were defined as benzo[h]quinoline anti-cancer derivatives. In this study, we have further explored their cytotoxic effect on G361, H460, MCF7 and HCT116 cancer cell lines using in-vitro, and in-silico bioassays.
Cell viability assays by MTT. Cell viability was measured by the MTT assay 34,36,48 . Briefly, 2 × 10 5 cells/well were seeded in 96-well plates and incubated for 12 h at 37 °C in a 5% CO2 atmosphere to allow cell adhesion. Stock solutions of the compounds made in DMSO were sterilized through 0.45 micro size filter (Sartorus Stedim, biotech, USA), then further diluted to 0.2 mg/ml in incomplete medium for treatment against G361, H460, MCF7 and HCT116 cancer cells lines. A 100 μ L solution of compound (3e, 3f, 3h and 3j) was added respectively to a 100-μ L of fresh medium in wells to give final concentrations of 5-0.312 μ M/mL after serial dilution. A negative control (without drug), a positive control with anticancer drug doxorubicin was included in each assay 34,36 .    RNA extraction for quantitative real time PCR. We measured mRNA expression of three apoptosis related genes, including H2AX, ATM, and BAX. The total cellular RNA was isolated from all four cancer cells line after 24 h incubated at the IC 50 concentration of 3e, 3f, 3h, 3j and Doxorubicin respectively. Primer design, cDNA synthesis and quantification of every gene was carried out as previously reported 34,36 . Quantitative real-time PCR was done by using following forward and reverse primer sequences, as listed below. 18sRNA was amplified to ensure cDNA integrity and to normalize expression. Primer sequences used for mRNA expression.   Apoptosis analysis by annexin V-FITC/PI staining. Cancer cells were exposed with IC 50 concentration of active compound 3e to detect apoptosis using the annexin V-FITC apoptosis detection kit. Treated/untreated cells were trypsinized after incubation for 12 h, after first washing with 1 ml of cold 1× biding buffer. Annexin V-FITC (0.5 mg ml −1 ) was added to each sample. After incubation for 15 min at room temperature, the cells were again washed with PBS and stained with 0.3 mg ml −1 of PI (Propidium Iodide) and analyzed by flow cytometry (BD FACS Verse, BD Biosciences) 34,36,47 .
DNA damage analysis. Genomic DNA was extracted from all four cell lines after treatment with IC 50 concentrations of 3e, 3f, 3h, 3j and Doxorubicin respectively, for OH-dG detection. Genomic DNA was extracted from treated/untreated cells following a standard molecular biology protocol and re-suspended in 50 μ l water 34,36 . The same amount of genomic DNA (2 mg) extracted from cells was used for the detection of 8-OH-dG level following the standard protocol of anoxidative DNA damage ELISA kit (Cell Biolabs, Inc. USA) 34,36 .
Western blot analysis. All cells were exposed with IC 50 concentrations of 3e compounds. After exposure, protein extraction of the cells was performed. The whole cell protein extracts from treated/untreated cells were lysed in a RIPA buffer (Cell Signaling Technology, USA) and the extracted proteins were subjected to electrophoresis in 12% SDS-PAGE and blotted onto nitrocellulose membranes 36,[47][48][49] . The membrane was probed with the CDK2 antibodies for protein expression, (Cell Signaling Technology, USA) [50][51][52] . The bands were detected using the Super   Signal West Pico Chemiluminescent substrate (Pierce, Rockford, IL, USA) and images was taken using a Vilver imaging system (Vilver, Upland, CA, USA).  39,40 . The Surflexdoc module in Sybyl was used to construct a 3D model of the structures 53,54 . Details data point available in S1 of supplementary material.
Screening through in silico pharmacokinetic parameters. Pharmacokinetic (PK) properties depend on chemical descriptors of drugs, which determine their absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties, which are the key descriptors for the human therapeutic use of any compound. These PK parameters were calculated using the ADMET modules in Discovery Studio v3.5 software (Accelrys, USA), and are reported in supplementary Table S1 of the Supplementary Information. Predictive mathematical ADMET models were derived with different PK parameters, namely aqueous solubility, blood-brain barrier penetration, cytochrome P-450 2D6 inhibition, hepatotoxicity, human intestinal absorption and plasma protein binding. Predictions from these models were used to quantitatively predict properties of a set contrasted with of known rules that specify for appropriate ADMET characteristics of the chemical structure of all benzo[h]quinoline derivatives.
Statistical analysis. All values are presented as the mean ± SD of the indicated number of replicates. Statistical analyses of the data were performed using Student's t-test, and significant differences were based on P < 0.05 or P < 0.01 34,36,48 .