Abstract
Keywords
Introduction
The 1, 4-dihydropyridine (DHP) derivatives, known as calcium channel antagonists, are used for treatment of cardiovascular diseases such as hypertension and angina pectoris (1-4). It has been discovered that DHPs posses a wide range of other beneficial biological activities such as anticonvulsant analgesic and are also used as a chemical drug delivery systems especially in brain delivery (5-8). Recently, it has been proved that these compounds make a new class of multidrug resistance (MDR) reversals in cancer treatment. Therefore, extensive investigations were made in order to find new DHP derivatives as MDR reversal agents (9-11).
Beside the reversing activity of DHPs, there are several reports about their intrinsic Cytotoxicity. Particularly, some derivatives showed significant cytotoxicity such as dexniguldipine and some dibenzoyl derivatives (12-15). In addition, there are some reports on the effects of DHPs on potentiation of antitumoral and antimetastatic activity of some common cytotoxic drugs (16). Mechanistic investigations were proved that cytotoxicity of DHPs may not be the result of interaction with Ca2+ channels although it might be related to other pathways in which calcium is involved such as calcium/calmodulin pathway or other cellular pathways like inhibition of topoisomerase I (12, 17, 18).
In this study, we performed an investigation about the intrinsic cytotoxicity of some derivatives of DHP which were bearing a nitroimidazole group on the C4 position.
Experimental
Dihydropyridine compounds
All compounds used in this study have been synthesized in our chemistry lab and their synthesis process and their Ca+2 blocking activity were previously reported (8, 19-23). Structures of compounds and their Ca channel blocking activity are presented in Table 1.
Structure of DHP derivatives used in this study and their calcium channel antagonist activity.
Cell lines and reagents
Five cell lines including K-562 (chronic myelogenous leukemia), HeLa (carcinoma of the cervix), Raji (Human, B cell Lymphoma), and Fen (bladder carcinoma cell line) were all obtained from the Iran Pasteur Institute, Tehran, Iran. Cells were cultured in RPMI-1640 (Sigma, USA) supplemented with 10% fetal bovine serum (Gibco, USA), 100 IU/mL penicillin and 100 μg/mL streptomycin (Biosera, England). Cells were cultured in 50 cm3 flask (Nunc, Denmark) with 5 mL of culture medium in humidified 5% CO2 incubator at 37°C.
Cytotoxicity evaluation
Appropriate amount of compounds was mixed with DMSO (dimethyl sulfoxide). Then, four concentration (10, 100, 50, 1000 μM) were made with serial dilution by RPMI-1640. DMSO, as the negative control, was diluted with the same method. For positive control, doxorubicin was mixed with RPMI-1640 to make similar concentrations. MTT [3-(4, 5-dimethylthiazole-2-yl)-2, 5-diphenyl tetrazolium bromide] (Sigma, USA), was dissolved in PBS (phosphate buffered saline) at 5 mg/mL. Each well of the plate was filled with 1-5×104 cells (depending on the cell line) in 90 μL cultured medium. Then, 10 μL from the stock solutions of compounds, negative controls and positive controls, was added as triplicate to the wells to reach the final concentration of 1, 10, 50, 100 μM. Three wells containing only the same number of cells left in each plate. Plates were kept in a humidified incubator for 48 h. After the incubation period, MTT assay was carried out by the procedure described by Jabbar et al. (24).
QSAR studies
Chemical structure of the molecules was built using HyperChem software (Version 7, Hypercube Inc., http://www.hyper.com, USA) for the structural chemistry. Gaussian98 was operated to optimize the molecular structure. The structures were optimized by Ab initio calculations at the level of RHF/ STO-3G. No molecular symmetry constraint was applied; rather full optimization of all bond lengths and angles was carried out. Highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) energies, molecular polarizability (MP) and molecular dipole moment (MDP) were all calculated by Gaussian98. Local charges (LC) at each atom were calculated according to both Mulliken population and natural population analysis methods. Some theoretical QSAR descriptors including physicochemical properties and topological indices were calculated using HyperChem (Hypercube Inc.) and DRAGON (http://www.talete.mi.it/ dragon_exp.htm, Italia) software. The MLR analysis was employed to derive the QSAR models for different cytotoxic activities. In order to make better QSAR models, IC50 of compounds was transformed into the logarithmic scale (log IC50). MLR and correlation analyses were carried out by the statistics software SPSS 11.5 version. Before any MLR analysis, the correlation between the selected descriptors was examined (Table 3) and collinear descriptors (r > 0.90) were determined. Among these descriptors, the one which had higher correlation with the dependent variable was retained and the others were removed from the descriptor data matrix. The remaining descriptors were used to construct the MLR model, in accordance with the stepwise selection and elimination method. The better regression models were selected on the basis of the higher R, f-value (a statistic of assessing the overall significance) and the lower SEE. Cross-validation procedure [leave-one-out (Q2LOO)] was applied to measure the predictive capabilities of the models by using MATLAB 6.5 program (25).
Results and Discussion
The data of cytotoxicity is shown in Tables 2, 3 and 4. IC15, IC30 and IC50 were calculated for each compound. The most cytotoxic effect was seen on Raji cell line while the lowest activity was observed in K562 and HeLa, since at least 7 compounds had IC30 and IC50 of higher than 100 μM.
Cell growth inhibitory activity of compounds a-e in-vitro (IC15
| IC15a(μM) | |||||
|---|---|---|---|---|---|
| HeLa | Fen | K562 | Raji | Compound | |
| 38.1 | < 1 | 41.8 | < 1 | ||
| 54.7 | < 1 | 84.0 | < 1 | A2 | |
| 38.8 | < 1 | 6.2 | < 1 | A3 | |
| 45.5 | < 1 | 20.2 | < 1 | A4 | |
| 72.4 | < 1 | < 1 | < 1 | A5 | |
| < 1 | < 1 | < 1 | < 1 | A6 | |
| < 1 | < 1 | 4.1 | < 1 | A7 | |
| 1.2 | < 1 | < 1 | < 1 | A8 | |
| < 1 | < 1 | < 1 | < 1 | A9 | |
| < 1 | < 1 | < 1 | < 1 | A10 | |
| 53.5 | 51.5 | 52.5 | > 100 | A11 | |
| < 1 | < 1 | < 1 | < 1 | A12 | |
| < 1 | < 1 | < 1 | < 1 | A13 | |
| < 1 | < 1 | < 1 | < 1 | A14 | |
| < 1 | < 1 | < 1 | < 1 | A15 | |
| < 1 | < 1 | < 1 | < 1 | A16 | |
| 35.5 | 44.7 | 44.4 | 22.4 | A17 | |
| < 1 | < 1 | < 1 | < 1 | Doxorubicin | |
| 81.1 | 71.4 | 80.4 | 92.4 | DMSO | |
Cell growth inhibitory activity of compounds a-e in-vitro (IC30).
| IC30a(μM) | Compound | |||||
|---|---|---|---|---|---|---|
| HeLa | Fen | K562 | Raji | |||
| 100 < | 33.1 | 100 < | 58.3 | A1 | ||
| 100 < | 57.8 | 100 < | 1 > | A2 | ||
| 100 < | 37.3 | 100 < | 32.5 | A3 | ||
| 100 < | 58.9 | 100 < | 1 > | A4 | ||
| 100< | 71.1 | 100 < | 100< | A5 | ||
| 1 > | 1 > | 1 > | 1 > | A6 | ||
| 12.6 | 10.5 | 100 < | 12.3 | A7 | ||
| 5.6 | 1.4 | 1.3 | 1.4 | A8 | ||
| 5.4 | 1.1 | 1.1 | 1.1 | A9 | ||
| 3.1 | 1 > | 1 > | 1 | A10 | ||
| 100 < | 100 < | 100 < | 100 < | A11 | ||
| 1.7 | 1.2 | 1.2 | 1.4 | A12 | ||
| 1 > | 1 > | 1 > | 1 > | A13 | ||
| 1 > | 1 > | 1 > | 1 > | A14 | ||
| 3.1 | 1.3 | 2.1 | 1.1 | A15 | ||
| 3.4 | 1> | 1 > | 1 | A16 | ||
| 100 < | 100 < | 1 > | 100< | A17 | ||
| 1 > | 1 > | 1 > | 1 > | Doxorubicin | ||
| 100 < | 100 < | 100 < | 100 < | DMSO | ||
Cell growth inhibitory activity of compounds a-e in-vitro (IC50).
| IC50a(μM) | ||||
|---|---|---|---|---|
| HeLa | Fen | K562 | Raji | Compound |
| > 100 | > 100 | > 100 | > 100 | A1 |
| > 100 | > 100 | > 100 | > 100 | A2 |
| > 100 | > 100 | > 100 | > 100 | A3 |
| > 100 | > 100 | > 100 | > 100 | A4 |
| > 100 | > 100 | > 100 | > 100 | A5 |
| 5.4 | 4.1 | 2.8 | 4.0 | A6 |
| 68.3 | 56.3 | > 100 | 69.3 | A7 |
| 30.9 | 15.1 | 14.1 | 15.4 | A8 |
| 29.4 | 6.8 | 3.0 | 7.1 | A9 |
| 25.3 | 4.3 | 3.8 | 8.1 | A10 |
| > 100 | > 100 | > 100 | > 100 | A11 |
| 14.9 | 14.3 | 14.3 | 14.1 | A12 |
| < 1 | < 1 | < 1 | < 1 | A13 |
| 2.8 | 2.4 | 2.5 | 3.1 | A14 |
| 23.4 | 8.8 | 20.2 | 8.4 | A15 |
| 28.5 | 7.2 | 3.3 | 8.2 | A16 |
| > 100 | > 100 | > 100 | > 100 | A17 |
| < 1 | < 1 | < 1 | < 1 | Doxorubicin |
| > 100 | > 100 | > 100 | > 100 | DMSO |
However, synthesized compounds had moderate cytotoxicity. It was obvious that symmetric compounds are significantly less potent since all of them have IC50 greater than 100 μM. On the other hand, the asymmetric derivatives had good cytotoxicity. Among them, compound A13 showed the best activity since its IC50 in all cell lines were lower than 1 μM which was comparable with the reference drug (doxorubicin).
QSAR studies
MLR analysis with the stepwise selection and elimination of variables was utilized to model the structure-activity relationships with diverse set of descriptors.
In the first step, it was tried to find an appropriate model for cytotoxic activity of all compounds in Raji. The obtained equation is as follows:
1og IC50 = 5.966 - 0.551 (±0.097) nN - 0.897 (±0.242) C - 040
N = 17, R2 = 0.812, Q2LOO = 0.772, F = 30.1, SE = 0.311
In this equation, the values in the parenthesis stand for the standard deviation of the coefficients. N, R2, SE, and F are number of components, correlation coefficient, standard error of regression, and Fisher’s F-ratio, respectively. Noticeably, this equation has a good statistical feature, which can explain and predict 82% of variances in the cytotoxic activity data. Q2LOO represents acceptable correlation coefficient, which is close to that of calibration.
This explains the satisfactory prediction ability and stability of the resulted model. This two-parametric equation, holding the number of nitrogen atoms (nN) and an atom center fragment parameters (C-040), implies the importance of constitutional parameters in their activity.
The next equation was obtained for cytotoxicity of K562 cell line:
1og IC50 = 6.002 - 0.571 (±0.142) nN - 0.878 (±0.353)
C - 040
N = 17, R2 = 0.680, Q2Loo = 0.600, F = 14.88, SE = 0.453
It is obvious that cytotoxic activity K562 cell line has a great similarity with the activity in Raji cell line, though statistical quality of latter equation is much lower than the first equation. This might imply the similar mechanism of cytotoxicity in both cell lines.
The next equation was obtained for the cytotoxicity of Fen cell line:
1og IC50 = 4.147 - 0.689 (±0.117) nN - 0.367 (±0.140) MLOGP
N = 17, R2 = 0.812, Q2LOO = 0.636, F = 19.1, SE = 0.383
In this equation as well, the number of nitrogen is the most important factor while, dissimilar to the former equation, the second parameter is MLOGP which octanol-water partition coefficient. This may suggest that transversing of biological membranes is a key factor in cytotoxic activity in this cell lines, since it was proved that log P is a crucial factor in crossing the biological membrane (26).
Finally, an equation for HeLa cell line was obtained as described below:
1og IC50 = 7.23 - 1.641 (±0.290) nCN - 0.211 (±0.036) SLCDHP + 9.123 (±2.91) nHAcc
N = 17, R2 = 0.856, Q2LOO = 0.644, F = 25.8, SE = 0.257
This three-parametric equation containing the holding number of nitrile groups (nCN), sum of local charge on DHP ring (SLCDHP) and the number of acceptor atoms for hydrogen bonding (nHAcc), shows that cytotoxic activity of DHP derivatives in this cell line probably is slightly different from other cell lines, while the presence of nitrogen (here in nitrile form) is a key factor.
Overall, it can be concluded that presence of nitrogen has a crucial importance in the cytotoxicity DHP derivatives in both total number of nitrogen and also groups containing nitrogen (i.e. nitrile group).
On the other hand, modeling the Ca+2 channel blocking activity of these compounds gives the following equation:
1og IC50 = 9.74 + 3.60 (±0.60) SLCNI - 1.562 (±0.352)
H - 046 - 1.459 (±0.569)Mv
N = 15, R2 = 0.852, Q2LOO = 0.701, F = 19.1, SE = 0.49
Where SLCNI is the sum of local charge on nitroimidazole ring, H-046 is an atom-centered fragment (H attached to C0 (sp3)) and Mv is the mean atomic van der Waals volume. This equation shows the importance of charges in C4 position while volume of molecules has negative effect on Ca+2 channel blocking activity.
Conclusion
DHPs are one of the major classes of Ca2+ channel blockers. Recently, their MDR reversing activity attracted many researchers. In this way, a large set of new derivatives of DHP were synthesized as MDR reversals, some of which have reached to the upper stage of studies.
In addition, some derivatives showed intrinsic cytotoxicity on cancer cell lines. As a remarkable example, dexniguldipine, a potent MDR reversal, exhibited strong cytotoxicity on various cell lines. On the other hand, it was shown that nitroimidazole and its derivatives possessed appropriate cytotoxic properties (27-29). Therefore, it seems that DHP compounds including nitroimidazole moiety at C4 position may have strong cytotoxicity due to the simultaneous presence of DHP base structure and nitroimidazole substituent. In the present study, cytotoxic activity of some DHP derivatives was evaluated on four different cancer cell lines. Asymmetric derivatives showed good cytotoxicity while symmetric ones were less potent.
In addition, QSAR studies revealed the significance of N atom and group containing it in cytotoxic activity. This is compatible with previous findings about MDR reversing activity of DHPs which demonstrated the good association of amine groups with MDR reversing effect. For example, Ford’s findings showed that tertiary and cyclic amine groups are decisive for MDR reversing activity (30) and in recent years, in most synthesized DHP derivatives as MDR modulators, this rule has been applied (10, 31). This correlation between MDR reversing activity and cytotoxic effect might be the sign of possible similar mechanisms interfering in both phenomena.
On the other hand, QSAR model of Ca+2 channel blocking activity confirms that key factors in this activity are to some extent different from cytotoxic activity. For example, while local charges of DHP is an important feature in cytotoxic effect, the significant part for Ca+2 channel blocking activity is nitroimidazole moiety. Besides, the most significant factor in cytotoxic effect (number of nitrogen atoms) does not have any considerable role in Ca+2 channel blocking activity. Another remarkable point about Ca+2 channel blocking activity is that the existence of local charge of nitroimidazole moiety in QSAR model is in agreement with previous works which highlighted the importance of the substituent of C4 position (32).
Acknowledgements
References
-
1.
Coburn RA, Wierzba M, Suto MJ, Solo AJ, Triggle AM, Triggle DJ. 1, 4-Dihydropyridine antagonist activities at the calcium channel: a quantitative structure-activity relationship approach. J. Med. Chem. 1988;31:2103-2107. [PubMed ID: 2846838].
-
2.
Shekarchi M, Binesh Marvasti M, Sharifzadeh M, Shafiee A. Anticonvulsant activities of 7-phenyl-5H-thiazolo[5, 4-e][1, 2, 3, 4]tetrazolo[5, 1-c]pyrrolo[1, 2-a][1, 4]diazepine and 7-phenyl-5H-thiazolo[5, 4-e][1, 3, 4]triazolo[5, 1-c] pyrrolo[1, 2-a][1, 4]diazepines. Iranian J. Pharm. Res. 2005;4:33-36.
-
3.
Miri R, Javidnia K, Sarkarzadeh H, Hemmateenejad B. Synthesis, study of 3D structures, and pharmacological activities of lipophilic nitroimidazolyl-1, 4-dihydropyridines as calcium channel antagonist. Bioorg. Med. Chem. 2006;14:4842-4849. [PubMed ID: 16603367].
-
4.
Velazquez C, Knaus EE. Synthesis and biological evaluation of 1, 4-dihydropyridine calcium channel modulators having a diazen-1-ium-1, 2-diolate nitric oxide donor moiety for the potential treatment of congestive heart failure. Bioorg. Med. Chem. 2004;12:3831-3840. [PubMed ID: 15210150].
-
5.
Bodor N. Redox drug delivery systems for targeting drugs to the brain. Ann. NY. Acad. Sci. 1987;507:289-306. [PubMed ID: 2894793].
-
6.
Mortazavi SM. Advances in targeted drug delivery. Iranian J. Pharm. Res. 2007:6 149-150.
-
7.
Kaymaz AA, Tan H, Altug T, Buyukdevrim AS. The effects of calcium channel blockers, verapamil, nifedipine and diltiazem, on metabolic control in diabetic rats. Diabetes Res. Clin. Pract. 1995;28:201-205. [PubMed ID: 8529499].
-
8.
Miri R, Javidnia K, Kebriaie-Zadeh A, Niknahad H, Shaygani N, Semnanian S, Shafiee A. Synthesis and evaluation of pharmacological activities of 3, 5-dialkyl 1, 4-dihydro-2, 6-dimethyl-4-nitroimidazole-3, 5-pyridine dicarboxylates. Arch. Pharm. 2003;336:422-428. Weinheim.
-
9.
Kawase M, Motohashi N. New multidrug resistance reversal agents. Curr. Drug Targets. 2003;4:31-43. [PubMed ID: 12528988].
-
10.
Mehdipour AR, Javidnia K, Hemmateenejad B, Amirghofran Z, Miri R. Dihydropyridine derivatives to overcome atypical multidrug resistance: design, synthesis, QSAR studies, and evaluation of their cytotoxic and pharmacological activities. Chem. Biol. Drug Des. 2007;70:337-346. [PubMed ID: 17937779].
-
11.
Zhou XF, Yang X, Wang Q, Coburn RA, Morris ME. Effects of dihydropyridines and pyridines on multidrug resistance mediated by breast cancer resistance protein: in-vitro and in-vivo studies. Drug Metab. Dispos. 2005;33:1220-1228. [PubMed ID: 15908473].
-
12.
Gietzen K, Abdallah F, Bai G. Inhibition of tumor cell growth by a novel dihydropyridine derivative. Eur. J. Cancer (1990;26:922-923.
-
13.
Hahn KA, Legendre AM, Schuller HM. Amputation and dexniguldipine as treatment for canine appendicular osteosarcoma. J. Cancer Res. Clin. Oncol. 1997;123:34-38. [PubMed ID: 8996538].
-
14.
Jorjani M, Rastegar H, Roshanzamir F, Varmazyari M, Abdollahi M, Zarghi A. Synthesis and biological evaluation of new 1, 4-dihydropyridines as antihypertensives agents in rats. Iranian J. Pharm. Res. 2003;2:43-46.
-
15.
Morshed SR, Hashimoto K, Murotani Y, Kawase M, Shah A, Satoh K, Kikuchi H, Nishikawa H, Maki J, Sakagami H. Tumor-specific cytotoxicity of 3,5-dibenzoyl-1,4-dihydropyridines. Anticancer Res. 2005;25:2033-2038. [PubMed ID: 16158941].
-
16.
Fedeli L, Colozza M, Boschetti E, Sabalich I, Aristei C, Guerciolini R, Del Favero A, Rossetti R, Tonato M, Rambotti P, Davis S. Pharmacokinetics of vincristine in cancer patients treated with nifedipine. Cancer. 1989;64:1805-1811. [PubMed ID: 2790693].
-
17.
Orth SR, Nobiling R, Bonisch S, Ritz E. Inhibitory effect of calcium channel blockers on human mesangial cell growth: evidence for actions independent of L-type Ca2+ channels. Kidney Int. 1996;49:868-879. [PubMed ID: 8648932].
-
18.
Straub T, Boesenberg C, Gekeler V, Boege F. The dihydropyridine dexniguldipine hydrochloride inhibits cleavage and relegation reactions of eukaryotic DNA topoisomerase I. Biochemistry. 1997;36:10777-10783. [PubMed ID: 9271509].
-
19.
Miri R, Dehpour AR, Azimi M, Shafiee A. Synthesis and smooth muscle calcium channel antagonist effect of alkyl, aminoalkyl 1, 4-dihydro-2, 6-dimethyl-4-nitroimidazole-3,5 pyridine dicarboxylates. Daru. 2001;9:40-45.
-
20.
Miri R, Javidnia K, Mirkhani H, Hemmateenejad H, Kazemi Z, Edraki N, Mehdipour AR. Synthesis and in-vitro dual calcium channel antagonist-agonist activity of some 1, 4-dihydo-2, 6-dimethyl-3-nitro and cyano-4-(1-methyl-5-nitro-1H-imidazol-2-yl)-5-pyridinecarboxylates. Daru. 2008;16:263-270.
-
21.
Miri R, Niknahad H, Vazin A, Azarpira A, Shafiee A. Synthesis and smooth muscle calcium channel antagonist effects of new derivatives of 1,4-dihydropyridine containing nitroimidazolyl substituent. Daru. 2002;10:130-136.
-
22.
Miri R, Niknahad H, Vesal G, Shafiee A. Synthesis and calcium channel antagonist activities of 3-nitrooxyalkyl, 5-alkyl 1,4-dihydro-2,6-dimethyl-4-(1-methyl-5-nitro-2-imidazolyl)-3, 5-pyridinedicarboxylates. Farmaco. 2002;57:123-128. [PubMed ID: 11902654].
-
23.
Shafiee A, Miri R, Dehpour AR, Soleymani F. Synthesis and calcium-channel antagonist activity of nifedipine analogues containing nitroimidazolyl substituent in guinea-pig ileal smooth muscle. Pharm. Sci. 1996;2:541-543.
-
24.
Jabbar SA, Twentyman PR, Watson JV. The MTT assay underestimates the growth inhibitory effects of interferons. Br. J. Cancer. 1989;60:523-528. [PubMed ID: 2529890].
-
25.
Mehdipour AR, Hemmateenejad B, Miri R. QSAR studies on the anesthetic action of some polyhalogenated ethers. Chem. Biol. Drug Des. 2007;69:362-368. [PubMed ID: 17539829].
-
26.
Hughes JD, Blagg J, Price DA, Bailey S, Decresceno GA, Devraj RV, Ellsworth E, Fobian YM, Gibbs ME, Gills RW, Greene N, Huang E, Krieger-Burke T, Loesel J, Wager T, Whiteley L, Zhang Y. Physiochemical drug properties associated with in-vivo toxicological outcomes. Bioorg. Med. Chem. Lett. 2008;18:4872-4875. [PubMed ID: 18691886].
-
27.
Brezden CB, Horn L, McClelland RA, Rauth AM. Oxidative stress and 1-methyl-2-nitroimidazole cytotoxicity. Biochem. Pharmacol. 1998;56:335-344. [PubMed ID: 9744571].
-
28.
Lee HH, Palmer BD, Wilson WR, Denny WA. Synthesis and hypoxia-selective cytotoxicity of 2-nitroimidazole mustard. Bioorg. Med. Chem. Lett. 1998;8:1741-1744. [PubMed ID: 9873426].
-
29.
Moselen JW, Hay MP, Denny WA, Wilson WR. N-[2-(2-methyl-5-nitroimidazolyl) ethyl]-4-(2-nitroimidazolyl)butanamide (NSC 639862), a bisnitroimidazole with enhanced selectivity as a bioreductive drug. Cancer Res. 1995;55:574-580. [PubMed ID: 7834627].
-
30.
Ford JM, Hait WN. Pharmacology of drugs that alter multidrug resistance in cancer. Pharmacol. Rev. 1990;42:155-199. [PubMed ID: 2217530].
-
31.
Tasaka S, Ohmori H, Gomi N, Lino M, Machida T, Kiue A, Natio S, Kuwano M. Synthesis and structure-activity analysis of novel dihydropyridine derivatives to overcome multidrug resistance. Bioorg. Med. Chem. Lett. 2001;11:275-257. [PubMed ID: 11206476].
-
32.
Mohajeri A, Hemmateenejad B, Mehdipour A, Miri R. Modeling calcium channel antagonistic activity of dihydropyridine derivatives using QTMS indices analyzed by GA-PLS and PC-GA-PLS. J. Mol. Graph. Model. 2008;26:1057-1065. [PubMed ID: 17959402].