Computational Study of Quinolone Derivatives to Improve their Therapeutic Index as Anti-malaria Agents: QSAR and QSTR

authors:

avatar Maryam Iman a , avatar Asghar Davood b , * , avatar Ali Khamesipour c

Chemical Injuries Research Center, Baqiyatallah University of Medical Sciences, Tehran, Iran.
Department of Medicinal Chemistry, Faculty of Pharmacy, Pharmaceutical Sciences Branch, Islamic Azad University, Tehran, Iran (IAUPS).
Center for Research and Training in Skin Diseases and Leprosy, Tehran University of Medical Sciences, Tehran, Iran.
Iranian Journal of Pharmaceutical Research : IJPR: Vol.14, issue 3; 775-784
article type: Original Article
received: February , 2015
accepted: July , 2015
DOI: https://doi.org/10.22037/ijpr.2015.1697

how to cite: Iman M, Davood A, Khamesipour A. Computational Study of Quinolone Derivatives to Improve their Therapeutic Index as Anti-malaria Agents: QSAR and QSTR. Iran J Pharm Res. 2015;14(3):e125324. https://doi.org/10.22037/ijpr.2015.1697.

Abstract

Malaria is a parasitic disease caused by five different species of Plasmodium. More than 40% of the world’s population is at risk and malaria annual incidence is estimated to be more than two hundred million, malaria is one of the most important public health problems especially in children of the poorest parts of the world, annual mortality is about 1 million. The epidemiological status of the disease justifies to search for control measures, new therapeutic options and development of an effective vaccine. Chemotherapy options in malaria are limited, moreover, drug resistant rate is high. In spite of global efforts to develop an effective vaccine yet there is no vaccine available. In the current study, a series of quinolone derivatives were subjected to quantitative structure activity relationship (QSAR) and quantitative structure toxicity relationship (QSTR) analyses to identify the ideal physicochemical characteristics of potential anti-malaria activity and less cytotoxicity. Quinolone with desirable properties was built using HyperChem program, and conformational studies were performed through the semi-empirical method followed by the PM3 force field. Multi linear regression (MLR) was used as a chemo metric tool for quantitative structure activity relationship modeling and the developed models were shown to be statistically significant according to the validation parameters. The obtained QSAR model reveals that the descriptors PJI2, Mv, PCR, nBM, and VAR mainly affect the anti-malaria activity and descriptors MSD, MAXDP, and X1sol affect the cytotoxicity of the series of ligands.

Introduction

Malaria is a parasitic disease which is caused in human mainly by five species of Plasmodium, called P. falciparum,P. vivax, P. malariae, P. ovale spp., and P. knowlesi (1). Almost all species of human malaria parasites have a similar life cycle consists of sexual phase of multiplications, which occurs in specific species of Anopheline mosquitoes and an asexual phase of multiplications (schizogony) which take places in blood cells of the vertebrate host which might follow with invasion to the liver cells (2).

Malaria estimated annual incidence rate is about 219 million people worldwide and still considered as one of the most important public health problems especially in children of the poorest parts of the world. Currently, more than 40% of the world’s population are at risk of malaria and the annual mortality is about 1 million (3). Although more than 50 years ago the worldwide malaria eradication program was endorsed but yet high mortality and morbidity caused by malaria especially P. falciparum (WHO: World Malaria Report, 2013), studies in different aspects of the disease are highly encouraged. Every available approach should be used to design safe and effective anti-malaria drugs (4-6).

Regardless of global efforts to develop an effective vaccine against malaria and valuable publications yet no vaccine is available against malaria (7, 8).

Current epidemiological and treatment situations of malaria urge to use different approaches to develop new tools for malaria treatment (9). The rate of drug resistance in malaria is high (3), drug resistance in malaria defined as when the parasite survive or multiply, regardless of administration of enough dose of the drug which is absorbed (WHO technical report series, 1965). The speed of development of new anti-malaria drugs should be faster than rapid spread of drug resistance. In drug development it should be considered that each stage of malaria life cycle shows susceptibility to a specific anti-malaria drug so development of several drugs might be needed to control the disease (10).

Quantitative Structure Activity Relationship (QSAR) and Quantitative Structure Toxicity Relationship (QSTR) models are mathematical equations relating the chemical structure of a substance to the ‎biological activity and toxicity, respectively (11-21). The current QSAR models are developed based on the resistant malarial strains TM90-C2B (chloroquine, mefloquine, pyrimethamine, and atovaquone resistant) and for cytotoxicity against J774 mammalian of a set of ‎‎28, quinolone derivatives which have already synthesized (22). To select the set of descriptors that are more relevant to EC50 of the desired compounds, MLR models were built and QSAR and QSTR equations with stepwise selection and ‎elimination of variables were established using SPSS and Matlab software.‎

Experimental

QSAR and QSTR equations

The biological data used in this study were anti-malaria activities (EC50), and cytotoxicity of 28 quinolone derivatives against multidrug resistant p. falciparum (TM90-C2B) and J774 mammalian, respectively (21).

As reported previously structures of desired compounds 1 to 28 (Table 1) were built and optimized using HyperChem software (version 7, Hypercube Inc.) (11-15). A large number of molecular descriptors were calculated using HyperChem (Table 2) and Dragon software (23, 24). Dragon software was used to calculate different functional groups, topological, geometrical, and constitutional descriptors for each molecule. The calculated descriptors were collected in a data matrix wherein the number of rows and columns were the number of molecules and descriptors, respectively. Some of the chemical parameters including molecular volume (V), molecular surface area (SA approx), surface area (SA grid) hydrophobicity (LogP), hydration energy (HE), refractivity (Rf), molecular polarizability (MP) and different quantum chemical descriptors including dipole moment (DM) and the highest occupied molecular orbital (HOMO) energies are shown in Table 2.

Briefly, after removing the constant or near-constant descriptors to decrease the redundancy existed in the descriptor data matrix, the correlation of descriptors with each other and with the activity and cytotoxicity (pEC50) of the molecules were examined and collinear descriptors (i.e. r>0.9) were detected. Among the collinear descriptors, the one which shows the highest correlation with activity or cytotoxicity was retained and the others were removed from the data matrix. To select the set of descriptors that were most relevant to the anti-malaria activities (pEC50) and cytotoxicity, MLR models were built and the QSAR equations with stepwise selection and elimination of variables were established using MLR method (25). SPSS (version 18) and Matlab (version 7.6.0, R2008a) software were used for MLR regression method (11-16).

In the case of each regression problem, SPSS produced many models and were ranked based on standard error of calibration and coefficient of multiple determinations, wherein some models had a large number of input variables and thus they were over-fitted. To hinder obtaining over-fitted models, the generated QSAR and QSTR models were validated by the leave-one out cross-validation procedure to check the predictability and robustness. A balance between the high cross-validation correlation coefficient and low number of descriptors were used as the criterion for model selection. The overall prediction abilities of the final models were assessed by using a prediction set containing about 25% of the original molecules. To do so, the data set of activity was randomly classified to calibrate and predict the sets. The model coefficients were calculated using calibration data and then were used to calculate the anti-malaria activities of the molecules in the prediction set.

Table 1

The chemical structure of quinolone derivatives.

Table 2

Calculated properties of quinolones using the HyperChem software.

Compoundsurface area(approx)Surface area(grid)volumehydration energylog prefractivitypolarizabilitydipolemomenthomoa
1637.07685.751152.78-5.723.75109.0542.115.62-8.78
2396.06426.22680.98-4.82164.4624.525.99-8.77
3604.956611104.66-5.193.73105.2140.194.83-8.59
4655.37701.421195.12-4.024.4114.1743.864.54-8.49
5558.7617.371030.75-6.363.26100.1338.455.12-8.53
6559.69618.081038.63-5.653.21100.3738.354.50-8.46
7567.72625.521058.52-7.672.49101.7938.995.43-8.32
8467.8492.18804.34-3.951.8575.4328.796.72-8.49
9501.99522.28860.37-8.91.2580.7430.627.21-8.50
10685.09701.641184.64-3.182.75108.9641.84.75-8.53
11526.28586.18996.12-3.421.2896.6736.76.69-8.49
12566.98559.29925.33-1.811.6287.633.176.1-8.47
13565.79623.431047.37-6.243.54102.9539.145.13-8.66
14622.7724.291261.99-2.883.63129.0148.916.65-8.48
15528.2595.95989.94-7.032.7495.3336.524.65-8.56
16668.81760.731303.62-6.275.22129.6649.854.63-8.47
17656.24754.641289.86-7.814.84127.4449.146.38-8.66
18656.41753.561292.28-7.714.84127.4449.146.00-8.6
19731.13804.681381.12-5.76.1135.6451.417.57-8.72
20745.17826.251407.73-7.766.9137.2452.056.54-8.61
21782.61885.751550.89-9.316.79156.1260.065.84-8.55
22885.15988.681707.67-10.058.43168.3164.185.87-8.56
23864.51974.071697.37-10.268.8168.5264.185.51-8.59
24748.7828.231430.31-5.036.57140.6853.257.41-8.79
25731.04814.431400.7-4.356.16139.7553.524.39-8.55
26735.48815.641393.25-5.516.24135.8551.327.15-8.78
27724.51808.521391.1-5.156.21139.5153.615.39-8.64
28683.76773.381333.12-9.053.83131.649.976.73-8.67

highest occupied molecular orbital.

Results and Discussion

QSAR equation

Based on the procedure explained in the Materials and Methods section, using a stepwise multiple linear regression method, the following five-parametric equation 1 (Equation 1) was derived for quinolones 1 to 28.

In the QSAR and QSTR equations, n is the number of data points, R2 is the correlation coefficient, S is the standard deviation, F is the Fisher’s F-value and q2 is the leave-one-out (LOO) cross validated coefficient that was obtained by a multiple linear regression.

The anti-malaria activities of the quinolone derivatives were tested against multidrug resistant P. falciparum (TM90-C2B). Herewith, we explain QSAR model for anti-malaria activity against TM90-C2B and cytotoxicity against the J774 mammalian in equations 1 and 2, respectively.

Equation (1):

pEC50= (-18.1 ± 9.02) – (8.8 ± 1.7) PJI2 – (12.99 ± 8.18) Mv + (29.6 ± 3.58) PCR – (0.68 ± 0.11) nBM + (0.03 ± 0.006) VAR

n=28, F=33.157, R2=0.92, S=0.89, p<0.000, q2=0.71

Equation 1 explains 92% of the variance in pEC50 data wherein the relative error prediction (REP) of the equation is shown in Table 5. This equation describes the effect of PJI2, Mv, PCR, nBM and VAR indices on the anti TM90-c2B activity. The PJI2 and VAR are among topological descriptors and Mv and nBM are among constitutional descriptors. PCR is among walk and path counts descriptor that show ratio of multiple path count over path count.

Equation 1 indicates that PCR and VAR demonstrate positive contribution and PJI2, Mv, and nBM show a negative contribution towards the anti-malaria activities. Comparison of the coefficient of descriptors reveals that anti-malaria (TM90-c2B) activities might be affected mainly by PCR, Mv, and PJI2 with 29.6, 12.99 and 8.8 values, respectively. The calculated pEC50 using equation 1 is presented in Table 3 and the graphical representation of cross validated calculated activity and the experimental values using equation 1 are presented in Figure 1. The correlation coefficient matrix for the descriptors that were used in the MLR equation 1 is shown in Table 4. Based on the equation 1, in order to design a new ligand with high potency, substitutions with high amount of PCR and low amounts of Mv and PJI2 should be considered.

Table 3

Anti-malaria activity of quinolones.

NumberapEC50 Exp.bpEC50 Calcc|REP|
13.92083.860370.0157
21.11631.419860.2138
32.35564.941600.5233
44.82394.593900.0500
52.14212.199510.0261
62.55752.459310.0399
72.709963.332580.1868
82.09852.309130.0912
91.11011.361550.1846
100.21250.438810.5156
111.13911.112760.0236
121.218960.838080.4545
132.36152.069610.1410
142.03761.44040.4146
152.14691.567890.3693
161.11691.192090.0631
170.99311.192090.1669
181.23061.192090.0323
190.61080.63730.0415
200.60030.38490.5597
210.99911.239540.1939
220.39251.165110.6631
231.32331.482330.1073
241.85391.955890.0522
250.78781.271790.3805
262.51001.825990.3746
271.26601.141890.1087
281.89280.5282.5848

The experimentally activity (pEC50) in TM90-C2B.

The calculated pEC50 using multi linear regression equation 1.

The absolute value of Percent of the relative error of prediction.

Table 4

Pearson correlation coefficient matrix for the descriptors of quinolone used in the MLR activity equation 1.

Correlations
pEC50PJI2MvPCRnBMVAR
Pearson CorrelationpEC501.000-0.514-0.2090.183-0.226-0.290
PJI2-0.5141.000-0.377-0.1350.1400.305
Mv-0.209-0.3771.0000.3670.4050.236
PCR0.183-0.1350.3671.0000.8330.649
nBM-0.2260.1400.4050.8331.0000.935
VAR-0.2900.3050.2360.6490.9351.000
Sig. (1-tailed)pEC50.0.0030.1430.1760.1230.067
PJI20.003.0.0240.2470.2390.057
Mv0.1430.024.0.0270.0160.113
PCR0.1760.2470.027.0.0000.000
nBM0.1230.2390.0160.000.0.000
VAR0.0670.0570.1130.0000.000.
NpEC50282828282828
PJI2282828282828
Mv282828282828
PCR282828282828
nBM282828282828
VAR282828282828
Plot of cross-validated calculated activity of TM90-C2B obtained by QSAR equation 1.

Plot of cross-validated calculated activity of TM90-C2B obtained by QSAR equation 1.

Figure 1

QSTR equation

Based on the procedure explained in the Materials and Methods section, using a stepwise multiple linear regression method, the following three-parametric equation 2 was derived for quinolones 1 to 28.

Equation (2):

pEC50= (-7.35± 0.657) + (22.5 ± 1.46) MSD + (0.815 ± 0.085) MAXDP – (0.097 ± 0.014) X1sol

n=18, F=81.14, R2=0.96, S=0.049, p<0.000, q2=0.812

Equation 2 explains 96% of the variance in pEC50 data which describes the effect of MSD, MAXDP and X1sol indices on cytotoxicity. MSD and MAXDP are among topological descriptors and X1sol is among connectivity indices. The relative error prediction (REP) of the equation 2 is shown in Table 5.

Equation 2 indicates that MSD and MAXDP demonstrate positive contribution and X1sol show a negative contribution towards the cytotoxicity which all descriptors have positive concept, so comparison of the coefficient of descriptors reveals that which cytotoxicity might be affected mainly by MSD. The calculated pEC50 using equation 2 is presented in Table 5 and the graphical representation of cross validated calculated activities and the experimental values using equation 2 are presented in Figure 2. The correlation coefficient matrix for the descriptors that were used in the MLR equation 2 is shown in Table 6. Based on this equation in order to have a new ligand with low cytotoxicity, some substitution with low amounts of MSD properties should be considered.

Table 5

Cytotoxicity of quinolone in term of pEC50.

NumberapEC50 Exp.bpEC50 Calc.c|REP|
11.3372421.7293470.226736
21.4317981.4902770.039240
31.5528421.7417330.108450
431.7368740.727241
51.7212461.6372860.051280
62.3979402.2376760.071621
71.6382721.6800070.024842
81.6382721.6751070.021989
91.6382721.6733920.020987
101.6989701.6456390.032407
111.6989701.7052420.003678
121.7212461.8128650.050538
131.7958801.9755040.090926
141.7958801.8113270.008528
151.6989701.5847430.072079
161.6575771.6401810.010606
171.7212461.6121830.067649
181.6575771.59810.037217

The experimentally cytotoxicity pEC50 in J774.

The calculated pEC50 using multi linear regression equation 2.

The absolute value of Percent of the relative error of prediction.

Table 6

Pearson correlation coefficient matrix for the descriptors of quinolones used in the MLR cytotoxicity equation 2.

Correlations
pEC50MSDMAXDPX1sol
Pearson CorrelationpEC501.0000.379-0.068-0.043
MSD0.3791.000-0.686-0.484
MAXDP-0.068-0.6861.0000.931
X1sol-0.043-0.4840.9311.000
Sig. (1-tailed)pEC50.0.0600.3940.432
MSD0.060.0.0010.021
MAXDP0.3940.001.0.000
X1sol0.4320.0210.000.
NpEC5018181818
MSD18181818
MAXDP18181818
X1sol18181818
Plot of cross-validated calculated activity of Cytotoxicity obtained by QSTR equation 2.

Plot of cross-validated calculated activity of Cytotoxicity obtained by QSTR equation 2.

Figure 2

Conclusion

Twenty eight analogs of quinolone with anti-malaria activities were subjected to QSAR and QSTR studies in order to design new ligands with an improved therapeutic index. Based on our QSAR and QSTR equations (Equations 1 and 2) PJI2, Mv, PCR, nBM, and VAR can effect on anti-malaria activity and MSD, MAXDP, and X1sol can effect on cytotoxicity. Our QSAR and QSTR calculations reveals in order to increase the therapeutic index of the series of anti-malaria compounds, some moieties with high amount of PCR and low amounts of Mv, PJI2, and MSD should be considered in the lead compounds.

Acknowledgements

References

  • 1.

    Birkett AJ, Moorthy VS, Loucq C, Chitnis CE, Kaslow DC. Malaria vaccine R&D in the Decade of Vaccines: Breakthroughs, challenges and opportunities. Vaccine. 2013;31:233-243.

  • 2.

    Krotoski WA, Collins WE, Bray RS, Garnham PC, Cogswell FB, Gwadz RW, Killick-Kendrick R, Wolf R, Sinden R, Koontz LC, Stanfill PS. Demonstation of hypnozoites in sporozoite-transmitted plasmodium vivax infection. Am. J. Trop. Med. Hyg. 1982;31:1291-1293. [PubMed ID: 6816080].

  • 3.

    Bloland PB. Drug Resistance in Malaria. Switzerland: World Health Organization; 2001.

  • 4.

    Bartoloni A, Zammarchi L. Clinical aspects of uncomplicated and severe malaria. Mediterr. J. Hematol. Infect. Dis. 2012;4:2012-2026.

  • 5.

    Kwiatkowski DP. How malaria has affected the human genome and what human genetics can teach us about malaria. Am. J. Hum. Genet. 2005;77:171-192. [PubMed ID: 16001361].

  • 6.

    Hedrick PW. Population genetics of malaria resistance in humans. Heredity. 2011;107:283-304. [PubMed ID: 21427751].

  • 7.

    Dubovsky F, Rabinovich NR. Malaria vaccines. In: Plotkin SA, Orenstein WA, editors. Vaccines. 4nd ed. Philadelphia: Saunders; 2004. p. 1283-1289.

  • 8.

    Girard MP, Reed ZH, Friede M, Kieny MP. A review of human vaccine research and development: malaria. Vaccine. 2007;25:1567-1580. [PubMed ID: 17045367].

  • 9.

    Okenu DMN. An integrated approach for malaria control in africa. Malar. Infect. Dis. Afr. 1999;10:4-13.

  • 10.

    Delves M, Plouffe D, Scheurer C, Meister S, Wittlin S, Winzeler EA, Sinden RE, Leroy D. The activities of current antimalarial drugs on the life cycle stages of plasmodium: a comparative study with human and rodent parasites. PLoS Med. 2012;9:1001-1169.

  • 11.

    Iman M, Davood A. QSAR and QSTR study of pyrimidine derivatives to improve their therapeutic index as antileishmanial agents. Med. Chem. Res. 2013;22:5029-5035.

  • 12.

    Iman M, Davood A. QSAR and QSTR study of selenocyanate derivatives to improve their therapeutic index as antileishmanial agents. Med. Chem. Res. 2014;23:818-826.

  • 13.

    Davood A, Nematollahi A, Iman M, Shafiee A. Computational studies of new 1,4-dihydropyridines containing 4-(5)-chloro-2-ethyl-5-(4)-imidazolyl substituent: QSAR and docking. Med. Chem. Res. 2010;19:58-70.

  • 14.

    Davood A, Iman M. Docking and QSAR studies of β-phenylethylidenehydrazine derivatives as a Gamma-aminobutyric acid aminotransferase inhibitor. Med. Chem. Res. 2011;20:955-961.

  • 15.

    Davood A, Iman M. Molecular docking and QSAR study on imidazole derivatives as 14-demethylase inhibitors. Turk. J. Chem. 2013;37:119-133.

  • 16.

    Hansch C. Quantitative approach to biochemical structure-activity relationships. Accounts Chem. Res. 1969;2:232-239.

  • 17.

    Hansch C, Fujita T. A method for the correlation of biological activity and chemical structure. J. Am. Chem. Soc. 1964;86:1616-1626.

  • 18.

    Kubinyi H. QSAR and 3D QSAR in drug design Part 1: Methodology. Drug Discov. Today. 1997a;2:457-467.

  • 19.

    Kubinyi H. QSAR and 3D QSAR in drug design Part 2: Applications and problems. Drug Discov.Today. 1997b;2:538-546.

  • 20.

    Hansch C, Leo A. Exploring QSAR Fundamentals and applications in chemistry and biology. Washington, DC: American Chemical Society; 1995.

  • 21.

    Cross RM, Namelikonda NK, Mutka TS, Luong L, Kyle DE, Manetsch R. Synthesis, antimalarial activity and structure−activity relationship of 7-(2-phenoxyethoxy)-4(1h)-quinolones. J. Med. Chem. 2011;54:8321-8327. [PubMed ID: 22111907].

  • 22.

    Todeschini R, Consonni V. Mannhold R, Kubinyi H, Folkers G, editors. Molecular Descriptors for Chemoinformatics: 2nd ed. Volume I: Alphabetical Listing/Volume II: Appendices. 2nd ed. Weinheim: Wiley-VCH; 2009.

  • 23.

  • 24.

    Hansch C, Kurup A, Garg R, Gao H. Chem-Bioinformatics and QSAR: A Review of QSAR Lacking Positive Hydrophobic Terms. Chem. Rev. 2001;101:619-672. [PubMed ID: 11712499].