Cytotoxic Activity and Three-Dimensional Quantitative Structure Activity Relationship of 2-Aryl-1,8-naphthyridin-4-ones

Abstract

A series of substituted 2-arylnaphthyridin-4-one analogues, which were previously synthesized in our laboratory, were evaluated for their in vitro cytotoxic activity against human lung cancer A549 and human renal cancer Caki-2 cells using MTT assay. Some compounds (11, 12, and 13) showed stronger cytotoxicity than colchicine against both tumor cell lines, and compound 13 exhibited the most potent activity with IC₅₀ values of 2.3 and 13.4 µM, respectively. Three-dimensional quantitative structure activity relationship (3D-QSAR) studies of comparative molecular field analysis (CoMFA) and comparative molecular similarity indices analysis (CoMSIA) were performed. Predictive 3D-QSAR models were obtained with q² values of 0.869 and 0.872 and r²_ncv values of 0.983 and 0.993 for CoMFA and CoMSIA, respectively. These results demonstrate that CoMFA and CoMSIA models could be reliably used in the design of novel cytotoxic agents.

INTRODUCTION

The microtubules play an important role in mitosis. Compounds that attack microtubules can inhibit or promote microtubule assembly, resulting in cell arrest in mitosis (Jordan and Wilson, 2004). Therefore, these antimitotic compounds have been used as anticancer drugs. For example, vinca alkaloids inhibit tubulin polymerization into microtubules and taxoids promote microtubule assembly (Rowinsky and Donehower, 1992; Verwij et al., 1994). Colchicine is another antimitotic agent, but it is not used clinically due to its high toxicity (Hastie, 1991; Chen et al., 1997).

Some flavonoid derivatives interfere with tubulin polymerization to produce potent cytotoxic activity against several cancer cell lines (Shi et al., 1995; Manthey et al., 2001). To find more potent antimitotic compounds, some researchers have synthesized new flavonoid analogues, such as 2-arylquinolin-4-ones (Xia et al., 1998; Lai et al., 2005; Nakamura et al., 2005), 2-arylnaphthyridin-4-ones (Chen et al., 1997; Zhang et al., 1999), and 2-arylquinazolin-4-ones (Hamel et al., 1996; Hour et al., 2000; Xia et al., 2001). Some of these compounds, especially 2-arylnaphthyridin-4-ones, exhibit potent cytotoxicity; thus, we synthesized a series of substituted 2-arylnaphthyridin-4-one analogues and conducted cytotoxicity assays in human tumor cell lines (Chung, 2004).

The 3D-QSAR model yields some information about the structural requirements for more potent activity of ligands when the structure of the receptor is not known (Rieger et al., 2001). In this paper, we report the cytotoxic activity and 3D-QSAR analysis of sixteen synthesized compounds to further the development of more potent cytotoxic agents.

METHODS

Materials

Phosphate-buffered saline (PBS) was purchased from Boehringer Mannheim (Indianapolis, IN). 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT), colchicine, trypsin, dimethyl sulfoxide (DMSO), and other reagents were purchased from Sigma (St. Louis, MO). Human lung cancer A549 and renal cancer Caki-2 tumor cell lines were obtained from Chong Kun Dang Pharmaceutical Corp., and 2-aryl-1,8-naphthyridin-4-one analogues were synthesized in our laboratory.

In vitro cytotoxic assay

The prepared compounds were tested for their cytotoxicity against the human lung cancer A549 and human renal cancer Caki-2 cell lines. The cytotoxic activity was determined using a MTT-based colorimetric assay. Briefly, cell lines were seeded at a density of approximately 5×10³ cells/well in 96-well plates. Each well contained 180 µl of medium to which 20 µl of 10×concentration of the prepared compounds or PBS were added. After 4 days of culture, 0.1 mg of MTT was added to each well and incubated at 37℃ for 4 h. Plates were centrifuged at 450×g to precipitate the formazan crystals. The medium was removed and 150 µl of DMSO was added to each well to dissolve formazan. In this assay, MTT was converted to blue formazan by mitochondrial enzymes. The intensity of the blue color formed was measured with an ELISA reader at 540 nm. The IC₅₀ value was defined as the concentration of the compound that reduced the absorbance at 540 nm by 50% (Park et al., 1987; Manthey and Guthrie, 2002).

Molecular modeling and alignment

A total of sixteen compounds exhibiting cytotoxic activity with IC₅₀ values ranging from 2.3~276.5 µM were used to perform the 3D-QSAR analysis. The IC₅₀ values were transformed into pIC₅₀ (-log IC₅₀) values and used as the dependent variable in CoMFA and CoMSIA studies. The structures of the training set and the test set are shown in Table 1. The data set was divided into two groups. Thirteen compounds (1~13) were used for the training set. Three compounds (T1~T3) were selected randomly as a test set and were used for external validation of the 3D-QSAR models.

Table 1.

In vitro cytotoxic activity of 2-aryl-1,8-naphthyridin-4-ones compared with colchicines

Molecular modeling was performed using the Sybyl 8.1.1 software of Tripos (SYBYL, 2009). All molecules were sketched by the sketch module in SYBYL. Each structure was optimized using a conjugate gradient method based on the TRIPOS force field with Gasteiger Huckel charges and gradient convergence criteria of 0.05 kcal/mol. Simulated annealing on the energy optimized structures was carried out with 50 cycles. They were heated at 1000 K for 1000 fs to reach equilibrium and then annealed at 200 K for 1000 fs. The 50 conformations were then minimized to obtain low energy conformations for each compound. The structures were aligned by using the align database. Compound (13), which showed the most potent activity, was selected as a template molecule. The 1,8-naphthyridin fragment of compound (13) in its optimized conformation was used as a common substructure in the alignment.

CoMFA and CoMSIA

CoMFA and CoMSIA analysis are based on the relationships between biological activity and structural properties of the compounds when the receptor structure is not known. In the CoMFA study, the aligned molecules were put in a three-dimensional grid of 2A°. The steric and electrostatic properties of the compounds were calculated at each grid point using Lennard-Jones potential and Coulombic potential, respectively. The sp³ carbon probe atom with +1 charge and Van der Waals radius of 1.52A° were used to calculate the CoMFA steric and electrostatic fields. The default value of 30 kcal/mol was used as the maximum steric and electrostatic energy cut-off. In CoMSIA analysis, five fields are calculated: steric, electrostatic, hydrophobic, hydrogen bond acceptor, and hydrogen bond donor fields. The probe atoms with radius 1A°, charge +1, hydrophobicity +1, hydrogen bond donating +1, and hydrogen bond accepting +1 were used to calculate the five fields. The attenuation factor of 0.3 was used as the default value.

The partial least squares (PLS) were carried out using the cross-validated leave one out (LOO) method to evaluate q² values. The model of highest q² value and lowest standard error of prediction suggested the optimum number of components. Leave one out (LOO) cross-validation was used to estimate the predictive ability of the models (Jackson et al., 2009).

RESULTS

In vitro cytotoxic activity

The sixteen synthesized compounds and colchicine were evaluated for their cytotoxicity in vitro against lung cancer A549 and renal cancer Caki-2 cells. The cytotoxic activities of each compound were obtained as IC₅₀ values, and the data are presented in Table 1. Colchicine showed moderate cytotoxicity with IC₅₀ values of 56 and 21.2 µM against A549 and Caki-2 tumor cell lines in our assay conditions, respectively. Among the synthesized compounds, Compounds 11, 12, and 13 showed stronger cytotoxic activity than colchicine against both tumor cell lines. They have a naphthyl group at the C₂-position in 1,8-naphthyridine and showed IC₅₀ values of 9.4, 10.9, and 2.3 µM against lung cancer A549 cells and 19.1, 17.5, and 13.4 µM against renal cancer Caki-2 cells, respectively. The compound 13 gave the most potent activity and its IC₅₀ values were 2.3 and 13.4 µM, respectively. The IC₅₀ values of the other compounds ranged from 25.7 to 194.6 µM in A549 cells and 50.7 to 276.5 µM in Caki-2 cells.

CoMFA and CoMSIA analysis

The statistical data from CoMFA and CoMSIA models are shown in Table 2.

Table 2.

Statistical data of CoMFA and CoMSIA for two tumor cell lines

^aFields used, S, steric; E, electrostatic; D, H-bond donor; A, H-bond acceptor; ^bq², cross-validated correlation coefficient from leave-one-out(LOO); ^cN, optimum number of components; ^dSEP, standard error of prediction; ^er²_ncv, non-crossvalidated correlation coefficient; ^fSEE, standard error of estimate; ^gF, F-test value.

Lung cancer A549: By use of CoMFA default settings with steric and electrostatic fields, a cross-validated coefficient (q²)=0.820 with 3 optimum number of components (N) was observed. This model had a non-cross-validated coefficient (r²_ncv)=0.983, standard error of estimate (SEE)=0.092, and F value=115.575. CoMFA with only steric field showed q²=0.869, N=3, r²_ncv=0.983, SEE=0.091, and F value=118.709. Six CoMSIA models were calculated from the same training and test sets. CoMSIA model with only steric field gave q² value=0.863, N=3, r²_ncv=0.977, SEE=0.107, and F value=83.623. These statistical parameters indicate the accuracy of model.

CoMFA model with steric field showed higher q² value (0.869) than any other CoMFA and CoMSIA models. Therefore, using this model, the predicted pIC₅₀ values for each training set and the residual were calculated (Table 3). The predicted pIC₅₀ values of the test set molecules and their residual values are shown in Table 4. These test set compounds were used to validate CoMFA model with steric field. A graph of the predicted pIC₅₀ values versus actual pIC₅₀ values for the training set and test set is shown in Fig. 1A.

Table 3.

Actual and predicted activity (pIC₅₀) of training set

Table 4.

Actual and predicted activity (pIC₅₀) of test set

Fig. 1.

Scatter plot of actual activities versus predicted activities against A549 (A) and Caki-2 (B) tumor cell lines (●: training set molecules, ▲: test set molecules).

Renal cancer Caki-2 cells: Six CoMSIA models with steric, electrostatic, hydrogen bond acceptor, and hydrogen bond donor fields were considered. In the CoMSIA model with steric and electrostatic fields, a cross-validated coefficient (q²)=0.872, optimum number of components (N)=3, non-cross-validated coefficient (r²_ncv)=0.993, standard error of estimate (SEE)=0.042, and F value=305.023 were observed. CoMFA model with electrostatic field gave q²=0.838, N=4, r²_ncv=0.984, SEE=0.064, and F value=126.912.

CoMSIA model with steric and electrostatic fields showed better q² value (0.872) than any other CoMFA and CoMSIA models. With this model, the predicted pIC₅₀ values for the training set and the residuals were calculated (Table 3). The predicted pIC₅₀ values of the test set and their residual values are shown in Table 4. The graph of predicted pIC₅₀ values versus actual pIC₅₀ values for the training set and test set is shown in Fig. 1B.

DISCUSSION

2-Phenyl-4-quinolones reportedly act as antimitotic agents that interact with tubulin at the colchicine site (Kuo et al., 1993; Li et al., 1994). The 2-phenyl-1,8-naphthyridin-4-ones can be considered isosteres of 2-phenyl-4-quinolones. Since the CH moiety in 4-quinolone is isosterically substituted to nitrogen atom at the C₈-position in 1,8-naphthyridin-4-one, the 2-phenyl-1,8-naphthyridin-4-one can be assumed to have similar biological activity to 2-phenyl-4-quinolones.

In terms of average cytotoxicity in the A549 tumor cell line, compounds with methyl substituted at the C₆-position were less active than those substituted at the C₅ or C₇-position in the 1,8-naphthyridin ring, while compounds with two methyl groups at both the C₅ and C₇-positions or no substituted group at the C₅, C₆ and C₇-positions in 1,8-naphthyridin were substantially less active. In renal cancer Caki-2 cells, methyl-substituted compounds at the C₇-position were more potent than those substituted at the C₅ or C₆-position in 1,8-naphthyridin, while methyl substituted compounds at both C₅ and C₇-positions or no substituted group at C₅, C₆ and C₇-positions in 1,8-naphthyridin were much less active.

The effects of substitutions at the C₅, C₆ and C₇-positions depend on the substitution at the C₂-position in 1,8-naphthyridin, which is consistent with the previously reported trend (Chen et al., 1997). The introduction of a naphthyl group at C₂-position increased cytotoxicity and methoxy substituted compounds at both the C₂' and C₄'-positions were more active than those substituted at the C₃' and C₄'-positions in 1,8-naphthyridin.

For lung cancer A549 cells, the statistical results from the CoMFA model with steric field demonstrated high cross-validated value q² (0.869) and non-cross-validated coefficient value r²_ncv (0.983), and these values suggest that this model was predictive. In general, a q² value higher than 0.5 is considered an indication that the model is internally predictive. This model showed optimum component number=3 and suggested that steric interaction strongly influenced the cytotoxic activity.

For renal cancer Caki-2 cells, the CoMSIA model with steric and electrostatic fields showed higher q² value than CoMFA models. Good predictivity of the model was indicated by high q² (0.872) and r²_ncv (0.993) coefficient values. The relative contributions of steric field (31.9%) and electrostatic field (68.1%) indicated that strong electrostatic interaction was involved in the activity.

The small average residual values (0.07 and 0.03) of actual and predicted activities in Table 3 indicated that the predicted activities from CoMFA and CoMSIA models correlated well with actual activities. The test set was used to validate the predictivity of these CoMFA and CoMSIA models. The residual values of the test set were calculated and their small average residual values (0.14 and 0.16) demonstrated that the predicted values of the test set also correlated well with the actual values.

Contour maps indicate the regions around molecules where increased or decreased activity was predicted by physicochemical property changes in the molecules. The CoMFA contour map of steric field in lung cancer A549 cells and the CoMSIA contour maps of steric and electrostatic fields in renal cancer Caki-2 cells are shown in Fig. 2. In the steric contour map, the green contours indicate sterically favored regions, whereas the yellow contours designate sterically unfavorable regions. In the electrostatic contour map, blue and red regions are favorable for positive charge and negative charge, respectively. The molecule in the contour maps is compound (13), which showed the most potent activity.

Fig. 2.

Contour maps of steric and electrostatic fields in lung cancer A549 (A) and renal cancer Caki-2 (B) cells. Compound 13 is shown inside the field. Color-coding is as follows: Green indicates regions where bulky groups increase activity, whereas yellow indicates regions where bulky groups decrease activity. Blue indicates that positive charge favors high activity, whereas red indicates that negative charge favors high activity.

The CoMFA steric contour map in lung cancer A549 cells (Fig. 2A) illustrated two small yellow contours above the C₂-naphthyl ring and one large yellow contour below the C₂-naphthyl ring of 1,8-naphthyridin. These steric contours indicate that bulky substituents at these sites are unfavorable for cytotoxic activity. This yellow contour map suggests that a flat phenyl or naphthyl ring at the C₂-position increases cytotoxic activity, but non-flat and bulky substituents at the C₂-position decrease cytotoxicity. The presence of green contours at the C₄' and C₈'-positions of the naphthyl ring and near the methyl group at the C₇-position in 1,8-naphthyridin indicates that bulky groups around these regions increase the cytotoxic activity. This green contour map proposes that atoms bigger than hydrogen at the C₄' and C₈'-position of the naphthyl ring and that groups bigger than methyl at the C₇-position of 1,8-naphthyridin might increase the activity.

The CoMSIA steric contour map in renal cancer Caki-2 cells (Fig. 2B) displayed sterically favorable green regions at the C₃', C₄', C₇', and C₈'-positions of the C₂-naphthyl ring and the methyl group at the C₇-position in 1,8-naphthyridin. This green contour map suggests that the in troduction of bulky groups in these areas might increase the cytotoxic activity against renal cancer Caki-2 cells. The presence of a sterically unfavorable yellow region above the C₂-naphthyl ring indicates that bulky groups around these regions decrease the cytotoxic activity, but a flat naphthyl ring at the C₂-position is acceptable for the activity. The CoMSIA electrostatic contour map in renal cancer Caki-2 cells (Fig. 2B) showed the favorable regions for positive charge (blue) near the C₃ and C₄-position of 1,8-naphthyridin and the C₄'-position of the C₂-naphthyl ring. These blue contours indicate that electron-deficient groups at these sites enhance the cytotoxic activity, and the carbonyl group in the C₄-position of 1,8-naphthyridin might be important for activity. The presence of a red contour in the C₂-naphthyl ring suggests that negative electrostatic potential around this region increases the cytotoxic activity, and the electron-rich C₂-naphthyl ring increases the activity.

In summary, compounds 11, 12, and 13 produced stronger cytotoxicity than colchicine against both tumor cell lines. The good predictivity of the CoMFA and CoMSIA models was shown by high q² (0.869 and 0.872) and r² _ncv (0.983 and 0.993) coefficient values. The small average residual values of the training (0.07 and 0.03) and test (0.14 and 0.16) sets indicated that the predicted activities from the CoMFA and CoMSIA models correlated well with the actual activity. The CoMFA contour map in lung cancer A549 cells suggested that atoms bigger than hydrogen at the C₄' and C₈'-positions of the naphthyl ring and groups bigger than methyl at the C₇-position of 1,8-naphthyridin might increase the cytotoxicity. The CoMSIA contour map in renal cancer Caki-2 cells suggested that the introduction of bulky groups at the C₃', C₄', C₇', and C₈'-positions of the C₂-naphthyl ring and the C₇-position in 1,8-naphthyridin and that positively charged groups near the C₃ and C₄-positions of 1,8-naphthyridin and the C₄'-position of the C₂-naphthyl ring might enhance the cytotoxic activity. In addition, the carbonyl group in the C₄-position and the C₂-naphthyl ring of 1,8-naphthyridin might be important for biological activity.

ABBREVIATIONS

3D-QSAR

three dimensional quantitative structure activity relationship

PBS

phosphate-buffered saline

MTT

3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide

CoMFA

comparative molecular field analysis

CoMSIA

comparative molecular similarity indices analysis