ZHANG Lei ZHANG Qing-Qing TANG Feng ZHANG Ji WANG Jing YAO Qi-Zheng

?

3-QSAR Studies of the Pteridine Analogues as iNOS Inhibitors①

ZHANG Leia②ZHANG Qing-Qingb②TANG FengbZHANG JicWANG Jinga③YAO Qi-Zhengb③

a(563003)b(210009)c(211198)

Inducible nitric oxide synthase (iNOS), which can produce nitric oxide (NO) in the induction of cytokines and other factors, has an important impact on the physiological functions of the body for the transmission of information. However, continuous generation of NO will produce a lot of great damages to organisms. Therefore, iNOS inhibitors with good inhibitory activity and selectivity have been important means of treating a variety of diseases. Based on the public-alignment of pteridine, 3-QSAR (Three-Dimensional Quantitative Structure-Activity Relationship) models of pteridine analogues as iNOS inhibitors were established by the 3-QSAR protocol of Discovery Studio 3.0. Pteridine molecules divided in different groups obtained four approximate models, indicating good stability of such models, in which A3 is preferable (2= 0.672,2= 0.996,2pred.= 0.888,2denotes the cross-validation coefficient,2denotes the non-cross-validation coefficient). This study should be significant for the future structure design and modification of pteridine analogues as iNOS inhibitors.

3-QSAR, pteridine analogue, iNOS inhibitor;

1 INTRODUCTION

Nitric oxide (NO), an endogenous molecule, is an important signaling mediator for various physiolo- gical functions such as neurotransmission, synaptic plasticity, vasorelaxation, immune responses and inflammatory[1-3].NO is generated through the oxi- dation of L-arginine to L-citrulline catalyzed by nitric oxide synthase (NOS) family, nicotinamide adenine dinu-cleotide phosphate (NADPH) and oxygen (O2)[4].There are three subtypes of NOS isoforms including neuronal NOS (nNOS), endo- thelial NOS (eNOS) and inducible NOS (iNOS, Fig. 1). Generally, under physiological conditions, NO is produced by nNOS and eNOS at a low level. And iNOS is not expressed in healthy cells, however, when iNOS is stimulated by proinflammatory cytokines or lipopolysaccharide, aberrant NO biosynthesis has been proved to be toxic and can cause many pathological diseases including hyper- tension, cancer, septic stroke, ischemia, inflamma- tion and Alzheimer’s disease[5, 6]. Thus, iNOS has drawn considerable attention and it may be con- sidered as a potential target for the treatment of various diseases[7, 8].

Considerable studies have reported that the L-arginine, heme and 5,6,7,8-tetrahydrobiopterin (BH4, Fig. 1) were the three crucial targets for the iNOS inhibitors[9, 10]. In our previous studies, we mainly chose the BH4target of iNOS, and prepared lots of pteridine derivatives, which were identified as potential iNOS inhibitors[11-17]. However, the interaction between iNOS and our inhibitors is not understood completely, and the related mechanism is not clear. In this paper, we report a 3-QSAR analysis of these pteridine derivatives as iNOS inhibitors using 35 known molecules from our previous work for the first time.

Fig. 1. Structures of human iNOS and BH4molecule

2 MATERIALS AND METHODS

2.1 Data set

The 35 pteridine analogues were selected for our study. The structures and biological activities of molecules 1～5 were cited from our previous researches. The structural optimization and energy minimization of these compounds were performed using the AM1 method in Hyperchem 7.0. The experimental bioactivities were expressed inIC0values, which were converted into the p50values by the formula p50= –log50(Table 1). The data set was divided into a training set of 28 molecules and a test set of 7 molecules.

Table 1. Experimental Bioactivities of Peridine Analogues as iNOS Inhibitors

No.RR5R6IC50(μmol/L)pIC50 27HCH3CH337.904.42 28HC6H4CH3C2H537.214.43 29HC6H4OCH3C2H533.774.47 30CH3C6H4OCH3C2H527.624.56

No.RR1IC50(μmol/L)pIC50 31C2H5C2H552.354.28 32C6H11H114.553.94 33CHCH2CH2H54.204.27 34C4H9H53.094.28 35CH2CH2OHH66.914.18

2.2 Molecular modeling and alignment

A 2.55 ? crystal structure of iNOS (PDB code 1 NSI) combined with BH4 was selected from PDB database[18]. Compound 1 with the highest activity was used as a template, which was docked into the active site of 1 NSI using the Libdock and CDOCKER docking programs in Discovery Studio 3.0[19]. Public template was selected for building the 3-QSAR models based on the molecular alignment and docking results.

2.3 3D-QSAR models

The 3-QSAR models were generated using Discovery Studio 3.0. Two probe types are used to compute the energy grids that indicate electrostatic and steric effects. The CHARMm force field was used considering the electrostatic potential. Full cross validated Partial Least-Squares (PLS) method of LOO (leave-one-out) was used to perform the regression analysis.

In order to examine the stability of 3-QSAR model, we exchanged a few groups of training set and test set compounds which were selected ran- domly by software for debugging, and established four models, A1～A4, respectively, in order to compare their effects on statistical results.

3 RESULTS AND DISCUSSION

3.1 Molecular modeling and alignment

Compared with BH4, Fig. 2 shows that, in Libdock docking method, the conformation of molecule 1 was similar to that of BH4. However, it was not similar in the CDOCKER docking method, indicating that Libdock was the better method for docking.

Fig. 2. Comparison of molecule 1 with BH4location in two docking methods (The green denotes BH4, pink denotes the conformation of molecule 1 in Libdock docking method, and yellow denotes the conformation of molecule 1 in the CDOCKER docking method)

Moreover, Fig. 3 shows the docking mode of molecule 1 with human iNOS (PDB code 1NSI) using Libdock method (Green denotes the experi- mental conformation of BH4, and pink denotes molecule 1). Perspicuously, in Fig. 3, the docking conformation of molecules 1 in binding pocket of iNOS matched with the BH4experimental confor- mation pretty well. The 4-carbonyl group of BH4formed hydrogen bond with the guanidinate of Arg381. However, molecules 1 could not form the same hydrogen bond because that N was an electron donating atom, which formed another indirect hydrogen bond with the guanidinate of Arg381 by water. In addition, BH4generated hydrogen bonds with Phe476, Trp463, Ile462 and Ser118.

Fig. 3. Docking mode of molecule 1 with 1NSI (The green denotes the experimental conformation of BH4, and pink denotes molecule 1)

Subsequently, public-template was selected for building the 3-QSAR model based on the mole- cular alignment and docking results, and four models, A1～A4, are presented in Figs.4～7, respectively.

Fig. 4. Public-template alignment of training (a) and test (b) molecules in model A1

Fig. 5. Public-template alignment of training (a) and test (b) molecules in model A2

Fig. 6. Public-template alignment of training (a) and test (b) molecules in model A3

Fig. 7. Public-template alignment of training (a) and test (b) molecules in model A4

3.2 3D-QSAR statistical results

The statistical parameters associated with A1～A4 models are listed in Table 2. According to Table 2, the non-cross-validated correlation coefficients (2) of A1～A4 are 0.977, 0.992, 0.996, 0.954, cross- validated correlation coefficient (2) values are 0.660, 0.692, 0.672, 0.713, and predicted2are 0.869, 0.870, 0.888, 0.870, respectively, which proved that A1～A4 models owned favorable external predictive ability. Moreover, A3 model was selected in the present study for its better property. The predicted inhibitory activities and residues obtained from A3 model are shown in Table 3, and linear relationship between the experimental and predicted p50is shown in Fig. 8.

Table 2. Results of 3D-QSAR Models Obtained from Discovery Studio 3.0

n: optimal components;2: non-cross-validated correlation coefficient;2: cross-validated correlation coefficient;

2.: predicted correlation coefficient for the test set;: standard estimated error

Table 3. Predicted Inhibitory Activities (Pred.) and Residues (Res.) Obtained from Model A3

* means the test molecules

Fig. 8. Linear relationship between the experimental and predicted p50for the model A3

3.3 Contour maps analysis

The electrostatic potential contour map is displayed in Figs. 9a and 9b, where blue color showed positive coefficients and red color exhibited negative coefficients. Meanwhile, steric interaction is shown in Figs. 9c and 9d, where green color revealed positive coefficients and yellow color displayed negative coefficients. In order to aid the visualization, molecule 13 (pink) is shown in Fig. 9.

Fig. 9. Electrostatic (a, b) and steric (c, d) contour maps of model A3

Firstly, at the 4 position of pteridine, small green contour was closer than peripheral yellow contour, suggesting that bulky group would increase the activity, while bulky group with long linker chain was unfavorable. Simultaneously, blue contour was closer than red, however, blue contour was much smaller, which indicated that small positive group would increase the activity, while if the substituent had a long chain group, negatively charged substi- tuent would be favorable. For example, the repla- cement of isopropyl in N-4 position with bulky benzyl group should increase the activity, but increasing the electronegativity would decrease the activity. Generally, two opposite coefficients resulted in the reduction of activity (13,50= 24.17 μmol/L and 2050= 49.21 μmol/L), thereby intimating that the influence of electrostatic interaction was greater than the steric interaction. The same phenomenon was observed when com- pared to compounds 1 and 4.

As shown in Fig. 9, green and red contours were near to the 2 position of pteridine, which indicated that bulky benzyl and negative group in this region will increase the activity. This was a possible reason why compounds 14, 16 and 19 with contained higher electron-donating groups -OCH3and -CH3as substituents on the pyridine ring had higher activity than compound 20 with halogen group. Moreover, the introduction of larger substitutions at the 2-position was unfavorable to enhance the activity, such as compounds 12 and 21.

At the 6-position, green and blue contours were found around the substituent, while yellow and red contours were found outside, suggesting that bulky group with positive potential or small group with long chain and negative potential should strengthen the activity. For example, increasing the size of the substituent from methyl (650= 27.60 μmol/L) to bulky group-methylphenyl (550= 33.62 μmol/L) on pteridine aromatic ring at the 6-position decreased the activity, which showed that the effect of the electrostatic interaction was greater than the steric interaction.

4 CONCLUSION

In our research, 3-QSAR model was applied for a series of pteridine analogues as iNOS inhibitors. The QSAR models showed good statistical results, and among them, A3 model possessed excellent credible and strong predictive ability, as indicated by high2and2and small SEE values. Further- more, external validation further verified the relia- bility of the constructed 3-QSAR model. These observations in the present study will be useful for the design of new pteridine analogues as potent iNOS inhibitors.

(1) Carpenter, A. W.; Schoenfisch, M. H. Nitric oxide release: part II. Therapeutic applications.. 2012, 41, 3742-3752.

(2) Mishra, B. B.; Rathinam, V. A. K.; Martens, G. W.; Martinot, A. J.; Kornfeld, H.; Fitzgerald, K. A.; Sassetti, C. M. Nitric oxide controls the immunopathology of tuberculosis by inhibiting NLRP3 inflammasome-dependent processing of IL-1.. 2013, 14, 52-60.

(3) Bogdan, C. Nitric oxide synthase in innate and adaptive immunity: an update.. 2015, 36, 161-178.

(4) Hobbs, A. J.; Higgs, A.; Moncada, S. Inhibition of nitric oxide synthase as a potential therapeutic target.. 1999, 39, 191-220.

(5) Nagy, G.; Clark, J. M.; Buzás, E. I.; Gorman, C. L.; Cope, A. P. Nitric oxide, chronic inflammation and autoimmunity.. 2007, 111, 1-5.

(6) Corrêa, T. A.; Alves, C. C. S.; Castro, S. B. R.; Oliveira, E. E.; Franco, L. S.; Ferreira, A. P.; de Almeida, M. V. Synthesis of 1,4-anthracene-9,10-dione derivatives and their regulation of nitric oxide, IL-1and TNF-in activated RAW264.7 cells.. 2013, 82, 463-467.

(7) Janakiram, N. B.; Rao, C. V. iNOS-selective inhibitors for cancer prevention: promise and progress.. 2012, 4, 2193-2204.

(8) Hoffmann, J.; Goadsby, P. J. New agents for acute treatment of migraine: CGRP receptor antagonists, iNOS inhibitors.. 2012, 14, 50-59.

(9) Nishimura, J. S.; Martasek, P.; Mcmillan, K.; Salerno, J. C.; Liu, Q.; Gross, S. S.; Masters, B. S. S. Modular structure of neuronal nitric oxide synthase: localization of the arginine binding site and modulation by pterin.. 1995, 210, 288-294.

(10) Saleron, L.; Sorrenti, V.; Giacomo, C.; Romeo, G.; Siracusa, M. A. Progress in the development of selective nitric oxide synthase (NOS) inhibitors.. 2002, 8, 177-200.

(11) Ma, F.; Guo, P.; Jiang, L.; Feng, M.; Shi, J.; Yao, Q. Design and synthesis of pterin derivares as nitric oxide synthase inhibitors.. 2009, 6, 575-578.

(12) Ma, F.; Lü, G.; Zhou, W. F.; Wang, Q. J.; Zhang, Y. W.; Yao, Q. Z. Synthesis and biological evaluation of 2,4-diaminopteridine derivatives as nitric oxide synthase inhibitor.. 2009, 20, 420-422.

(13) Ma, F.; Lu, G.; Zhou, W. F.; Wang, Q. J.; Zhang, Y. W.; Yao, Q. Z. Synthesis and biological activities of 2,4-diaminopteridine derivatives.. 2009, 342, 274-280.

(14) Tang, F.; Yao, Q. Z.; Hua, W. Y. Synthesis and NOS regulative effect of 2-arylacylated amino pteridines.. 2006, 37, 18-22.

(15) Jiang, L. X. Study on biopterin inhibitor of nitric oxide synthase targating tetrahydrobiopterin.. China Pharmaceutical University 2004.

(16) Ma, F. Study on design, synthesis and bioactivity of pteridine compounds as nitric oxide synthase regulator.. China Pharmaceutical University 2009.

(17) Tang, F. The Synthesis and bioactivity study of pteridines compounds as regulators of nitric oxide synthase, Ph. D. Dissertation, China Pharmaceutical University 20005.

(18) Li, H.; Raman, C. S.; Glaser, C. B.; Blasko, E.; Young, T. A.; Parkinson, J. F.; Whitlow, M.; Poulos, T. L. Crystal structures of zinc-free and-bound heme domain of human inducible nitric-oxide synthase. Implications for dimer stability and comparison with endothelial nitric-oxide synthase.. 1999, 274, 21276-21284.

(19) Ji, Y.; Shu, M.; Lin, Y.; Wang, Y.; Wang, R.; Hu, Y.; Lin, Z. Combined 3D-QSAR modeling and molecular docking study on azacycles CCR5 antagonists.. 2013, 1045, 35-41.

1 December 2017;

12 March 2018

① This project was supported by the National Natural Science Foundation of China (39870882)

② These authors contributed equally to this work

Wang Jing. E-mail: wangjing0642320@126.com; Yao Qi-Zheng. E-mail: qz_yao@163.com

10.14102/j.cnki.0254-5861.2011-1909