DOI 10.5012/bkcs.2011.32.7.2433
Molecular Docking, 3D QSAR and Designing of New Quinazolinone Analogues as DHFR Inhibitors
L. Yamini, K. Meena Kumari, and M. Vijjulatha*
Department of Chemistry, Nizam College, Osmania University, Hyderabad-500 001, India
*E-mail: [email protected] Received January 10, 2011, Accepted June 1, 2011
The three dimensional quantitative structure activity relationship (3D QSAR) models were developed using Comparative molecular field analysis (CoMFA), comparative molecular similarity indices analysis (CoMSIA) and docking studies. The fit of Quinazolinone antifolates inside the active site of modeled bovine dihydrofolate reductase (DHFR) was assessed. Both ligand based (LB) and receptor based (RB) QSAR models were generated, these models showed good internal and external statistical reliability that is evident from the q2loo, r2ncv and r2pred. The identified key features enabled us to design new Quinazolinone analogues as DHFR inhibitors. This study is a building bridge between docking studies of homology modeled bovine DHFR protein as well as ligand and target based 3D QSAR techniques of CoMFA and CoMSIA approaches.
Key Words : Dihydrofolate Reductase (DHFR), Three dimensional quantitative structure activity relationship (3D QSAR), Ligand Based (LB), Receptor Based (RB), Comparative molecular field analysis (CoMFA), Com- parative molecular similarity indices analysis (CoMSIA)
Introduction
Dihydrofolate reductase enzyme catalyzes the reduction of 7,8-dihydrofolate (DHF) to 5,6,7,8-tetrahydrofolate (THF) by stereo specific hydride transfer from the NADPH cofactor to the C6 atom of the Pterin ring with concomitant protonation of N5 atom. DHFR plays a central role in the maintenance of cellular pools of THF and its derivatives which are essential for Purine and Thymidine synthesis and hence for cell growth and proliferation. This enzyme has been the target of important anticancer drugs and anti- biotics.1,2 Quinazolinone skeleton is a frequently encounter- ed heterocyclic moiety in medicinal chemistry with appli- cations including antibacterial,3 analgesic,4 anti-inflam- matory,5,6 antifungal,7 anti malarial,8 anti hypersensitive,9 CNS depressant,10 anticonvulsant,11 antihistaminic and local anesthetic,12 anti Parkinson’s,13 antiviral14 and anticancer.15 S.T. Al- Rashood et al.16 synthesized and tested the bio- logical activities of Quinazolinone molecules on bovine DHFR, but the docking studies were done on hDHFR. Since the pharmacological activities of the inhibitors were deter- mined on the bovine DHFR, the enzyme should be bovine DHFR instead of human DHFR (hDHFR) for performing molecular modeling studies. Based on this statement bovine DHFR was considered for our study. The difference between the bovine and human DHFR was an incremental change in the amino acid chronological order. Docking studies were performed on modeled bovine DHFR, in order to investigate the interactions between the inhibitors and the target, bovine DHFR was modeled using mouse DHFR as a template. In this study, lower energy conformation with atom fit align- ment (ligand-based LB) and docking conformation (receptor- based RB) were used to build 3D QSAR models. Partial
least square (PLS)17,18 based statistical analysis was carried out on 44 molecules to identify the correlation. The contour maps generated enabled us to design new molecules.
Experimental Section
FASTA sequence of bovine DHFR residues with accession code of P00376 was retrieved from swiss-port database server.19 The sequence when subjected to the basic local alignment search tool (BLAST) by setting the server to search the protein data bank.20 Sequence identity, E values (a statistical measure) and secondary structure similarities were used as constraints during the selection of the template. Pair wise alignment was carried out with Clustal Xto define conserved regions, identities, similarities and differences, between the target and the template.21 The mouse DHFR shared a sequence similarity of 88% with bovine DHFR with identical sequence of amino acids except for the increase in one digit in the chronological order of the amino acids. The active site of the hDHFR represents Ile-7, Ala-9, Trp-24, Glu-30, Gln-35, Asn-64, Arg-70, Val-115, Tyr-121 and Thr- 136 and bovine DHFR active site is represented by Ile-8, Ala-10, Trp-25, Glu-31, Gln-36, Asn-65, Arg-71, Val-116, Tyr-122 and Thr-137.22 The homology models were generated using MODELLER 8v2.23,24 The low energy conformation of the protein structure of bovine DHFR was analyzed with PROCHECK,25 Verify 3D26,27 and PROSA.28,29 The built model with > 80% sequence similarity with the template is equivalent to the low resolution X-ray crystal structure.30 The active sites of both the crystal structure and modeled protein had similar binding pockets, bovine DHFR was considered for molecular docking.
A set of 44 molecules with reported IC50 values for
inhibition of bovine DHFR activity,16 were taken and con- verted to corresponding pIC50 values (Table 1). The dataset was divided into training set of 36 molecules and test set of 8 molecules. All molecular modeling calculations were per-
formed on Linux operating system. Three dimensional struc- ture building and all modeling studies were performed using Sybyl 6.9 molecular modeling program package.31
Gasteiger-Huckel32 charges were assigned and then energy minimization of each molecule was performed using conju- gate gradient method and tripos standard force field33 with a distance dependant dielectric function. The minimization was terminated when the energy gradient convergence criterion of 0.001 kcal/mol/Å was reached.34-36 The protein was minimized using conjugate gradient method and tripos standard force field was applied after adding all hydrogen atoms. The minimization was terminated when the energy gradient convergence reached to 0.05 kcal/mol/Å.
FlexX37 was used for molecular docking. The analysis of dock poses of all the molecules showed limited interactions with the active site residues with amino acids like ASN65 and ARG71. Ligand based alignment was based on the lowest energy conformation of molecule 23 in the series.
The receptor based alignment process is a standard rigid RMSD overlay of selected common structural motive, where the docked poses of the most active molecules was used as template onto which the data set was aligned.38 The align- ment was done using ALIGN DATABASE command in Table 1. Structures of Molecules used in the 3D QSAR study
Mol.No IC50 (µM) pIC50 R1 R2 R3 Core Structure-1
1 20.0 4.70 Et H 6-NO2
2 30.0 4.52 Ph H 6-NO2
3 20.0 4.70 Bn H 6-NO2
4 30.0 4.52 Bn H 7-NO2
5 25.0 4.60 Et CH3 6-NO2
6 25.0 4.60 Ph CH3 6-NO2
7 25.0 4.60 Bn CH3 6-NO2
8 30.0 4.52 Bn CH3 7-NO2
9 50.0 4.30 Ph CH3 6-NH2
10 50.0 4.30 Bn CH3 7-NH2
Core Structure-2
11 10.0 5.00 Bn 4-OCH3 -
12* 50.0 4.30 Ph 3,4-OCH3 -
13 40.0 4.40 Bn 3,4-OCH3 -
14 13.0 4.89 Ph 3,4,5-OCH3 -
15 20.0 4.70 Bn 3,4,5-OCH3 -
Core Structure -3
16 10.0 5.00 Bn H 4-OCH3
17 20.0 4.70 Ph H 3,4-OCH3
18 15.0 4.82 Bn H 3,4-OCH3
19 30.0 4.52 Ph H 3,4,5-OCH3
20 30.0 4.52 Bn H 3,4,5-OCH3
21 1.0 6.00 Ph CH3 4-OCH3
22 0.5 6.30 Bn CH3 4-OCH3
23 0.4 6.39 Ph CH3 3,4-OCH3
24 0.4 6.39 Bn CH3 3,4-OCH3
25 1.0 6.00 Ph CH3 3,4,5-OCH3
Core Structure -4
26 2.0 5.70 Ph H -
27 10.0 5.00 Bn H -
28 2.0 5.70 Ph Br -
29 5.0 5.30 Bn Br -
30 10.0 5.00 Ph CH3 -
31 7.0 5.15 Bn CH3 -
Table 1. Continued
Mol.No IC50 (µM) pIC50 R1 R2 R3 Core Structure -5
32* 50.0 4.30 Ph Et -
33* 20.0 4.70 Bn Et -
34* 10.0 5.00 Ph Bn -
35 8.0 5.10 Bn Bn -
Core Structure -6
36 20.0 4.70 Ph H -
37 15.0 4.82 Bn H -
38 50.0 4.30 Ph 4-Br -
39 8.0 5.10 Bn 4-Br -
40 10.0 5.00 Ph 4-CH3 -
41* 7.0 5.15 Bn 4-CH3 -
42* 70.0 4.15 Ph 4-OCH3 -
43* 10.0 5.00 Bn 4-OCH3 -
44* 8.0 5.10 Bn 3,4-OCH3 -
pIC50 = −log IC50. *represents molecules taken for test set, Ph = Phenyl and Bn = Benzyl
Figure 1. Common substructure used for alignment.
Sybyl 6.9 taking the substructure that is common to all (Fig.
1). The resultant alignment is shown in Figure 2.
Standard Tripos force field was employed for the Com- parative Molecular Field Analysis (CoMFA)39 and Com- parative Molecular Similarity Indices (CoMSIA)40,41 analysis.
A 3D cubic lattice overlapping all entered molecules and extended by at least 4 Å in each direction with each lattice intersection of a regularly spaced grid of 2.0 Å was created.
The steric and electrostatic parameters were calculated in case of the CoMFA fields, while hydrophobic, H-bond acceptor and H-bond donor parameters in addition to steric and electrostatic were calculated in case of the CoMSIA fields at each lattice. A sp3 hybridized carbon atom was used as a probe atom to generate steric (Lennard-Jones potential) field energies and a charge of +1 to generate electrostatic (Coulombic potential) field energies. A distance dependent dielectric constant of 1.00 was used. The steric and electro- static fields were truncated at +30.00 kcal/mol. The similarity indices descriptors were calculated using the same lattice box employed for CoMFA calculations, using sp3 carbon as a probe atom with a +1 charge, +1hydrophobicity and +1 H- bond donor and +1 H-bond acceptor properties.
A partial least squares regression was used to generate a linear relationship that correlates changes in the computed fields with changes in the corresponding experimental values of biological activity (pIC50) for the data set of ligands.
Biological activity values of ligands were used as dependent variables in a PLS statistical analysis. The column filtering value (s) was set to 2.0 kcal/mol to improve the signal-to- noise ratio by omitting those lattice points whose energy variations were below this threshold. Cross-validations were
performed by the leave-one-out (LOO) procedure to deter- mine the optimum number of components (ONC) and the coefficient q2. The optimum number of components obtain- ed is then used to derive the final QSAR model using all of the training set compounds with non-cross validation and to obtain the conventional correlation coefficient (r2). To vali- date the CoMFA and CoMSIA derived models, the predic- tive ability for the test set of compounds (expressed as r2pred) was determined by using the following equation:
r2pred = (SD – PRESS)/SD
SD is the sum of the squared deviations between the biological activities of the test set molecules and the mean activity of the training set compounds. PRESS is the sum of the squared deviation between the observed and the pre- dicted activities of the test set compounds. Since the statistical parameters were found to be the best for the model from the LOO method, it was employed for further predic- tions of the designed molecules. The designed molecules were also constructed, minimized and docked into the protein active site, as mentioned above.
Results and Discussion
This section has been divided into discussion of the results from homology modeling, active site identification, QSAR results, contour map analysis and designing of new mole- cules.
During this study no crystallographic data for bovine DHFR was available. Sequence homologous to bovine DHFR having 187 amino acid residues was obtained from the BLAST server with corresponding E values for mouse being 3.0 × 10−94 and for human as 6.0 × 10−92. A low E- value indicates a high protein sequence identity.42 The mouse DHFR (pdb id: 2FZJ; 2.5 Å) was taken as the template protein based on E-value and maximum sequence align- ment. Alignment of bovine DHFR protein with the template protein gave a sequence identity of 89% and a sequence similarity of 93%.
The initial homology models were generated using MODELLER 8v2. A Ramachandran contour plot of phi (Φ) verses Psi (Ψ) (backbone dihedral angles) for the modeled low energy conformer of bovine DHFR protein, along with plot statistics is shown in Figure 3. Among the 187 amino acids, 128 were in the most favored region, 31 residues in additionally allowed region, 0 residues in generously allow- ed and one in the disallowed region excluding glycine and proline. This shows that the model generated was stereo chemically valid.
The PROSA II server gave pair wise energy and surface energy values based on the mean force potential (a distance based pair potential) as a function of amino acid sequence.
Amino acid residues with negative PROSA energies are more reliable and most of the amino acid residues of bovine DHFR had negative energies. The overall Z-score was −8.86 when compared with the Z-score of −8.71 for the template.
Figure 2. Alignment of data set molecules based on common substructure using compound 23 as a template. (a) Receptor based alignment. (b) Ligand based alignment.
Sequence similarity > 80% between the template and the target is equivalent to medium resolution NMR or low resolution crystal structure, that can be used directly for docking small molecules into the protein active site.30 Active site residues of the bovine DHFR were obtained from NCBI database and further confirmed by SITEID option in Sybyl 6.9. Receptor description file (RDF) was created within the area of 5.0 Å around the active site cavity. The active site residues include ILE-8, ALA-10, TRP-25, GLU-31, GLN- 36, ASN-65, ARG-71, VAL-116, TYR-122 and THR-137.21
Six different set of quinazolinone core molecules (shown in Table 1 along with IC50 and pIC50 values) were docked into the active site, they showed maximum two hydrogen bond interactions with the active site residues. The 3D QSAR CoMFA and CoMSIA analysis were carried out using quinazolinone derivatives reported as potent inhibitors for bovine DHFR.16 Molecules having precise IC50 values were selected and the molecules that did not show interactions with the protein active site (via docking) were removed from the dataset. A set of 44 molecules were used for derivation of model, these were divided into training set of 36 mole- cules and test set of 8 molecules.
The CoMFA and CoMSIA statistical analysis are sum- marized in Table 2. Statistical data shows q2loo 0.679 and 0.450 for CoMFA ligand based (LB) and receptor based (RB), 0.569 and 0.506 for CoMSIA LB and RB respectively.
The r2ncv of 0.988 and 0.902 for CoMFA ligand based (LB) and receptor based (RB), 0.985 and 0.986 for CoMSIA LB and RB respectively, that includes a good internal predictive ability of the models. To test the predictive ability of the models a test set of eight molecules excluded from the model generation were used. The predictive correlation coefficient r2pred of 0.599 and 0.833 for CoMFA LB and RB, and 0.526 and 0.538 for the CoMSIA models respectively indicates good external predictive ability of the models. The graph for the actual and predicted pIC50 values for training set and test of CoMFA LB and RB studies are shown in Figure 4(a), (b) and CoMSIA LB and RB studies shown in Figure 4(c), (d). The CoMSIA models showed better results than CoMFA models, this shows that the hydrophobic fields which were not included in the CoMFA model are important for explaining the potency of the molecules. The observed and PROCHECK
Ramachandran plot quality
Core Allowed General Disallowed
Model (a) 80.0 19.4 0.0 0.6
Template (b) 89.3 10.7 0.0 0.0
Figure 3. Ramachandran map: (a) 1FZJ template and (b) Bovine homology model.
Table 2. PLS results summary
Statistical Parameters CoMFA Model CoMSIA Model Ligand
based
Receptor based
Ligand based
Receptor based
q2loo 0.679 0.450 0.569 0.506
Molecules in
training set 36 36 36 36
Molecules in test set 8 8 8 8
ONC 8 9 10 10
r2 ncv 0.988 0.992 0.985 0.986
SEE 0.074 0.063 0.087 0.085
F 286.607 349.666 165.250 173.707
r2pred 0.559 0.833 0.526 0.538
Fraction of Field Contributions:
Steric 0.64 0.575 0.434 0.438
Electrostatic 0.36 0.425 - -
Hydrophobic - - 0.352 0.338
Acceptor - - 0.214 0.223
q2loo = cross-validated correlation coefficient by leave one out method, r2ncv = conventional corrélation coefficient, ONC = optimum number of components, SEE = standard error of estimate, F = Fisher test value, r2pred
= cross-validated correlation coefficient on test set
Figure 4. (a)-(d) Scatter Plot of experimental vs. predicted pIC50 (test set is represented as triangles). (a) Ligand based CoMFA. (b) Ligand based CoMSIA. (c) Receptor based CoMFA and (d) Receptor based CoMSIA.
Table 3. Data set used for 3D QSAR analysis with corresponding actual and predicted activities for LB and RB CoMFA and CoMSIA values
Mol.
No
Obs.
Activity
LB Predicted Activity RB Predicted Activity
CoMFA CoMSIA CoMFA CoMSIA
1 4.70 4.84 4.80 4.81 4.82
2 4.52 4.57 4.66 4.61 4.51
3 4.70 4.77 4.72 4.70 4.74
4 4.52 4.54 4.59 4.60 4.58
5 4.60 4.69 4.67 4.67 4.69
6 4.60 4.54 4.50 4.51 4.53
7 4.60 4.55 4.59 4.62 4.51
8 4.52 4.38 4.47 4.46 4.45
9 4.30 4.37 4.29 4.30 4.35
10 4.30 4.31 4.23 4.23 4.33
11 5.00 4.97 5.03 5.02 4.99
12* 4.30 5.00 4.81 4.51 4.51
13 4.40 4.34 4.34 4.37 4.44
14 4.89 4.85 4.85 4.87 4.91
15 4.70 4.74 4.74 4.70 4.76
16 5.00 5.02 5.02 4.99 4.97
17 4.70 4.65 4.65 4.63 4.63
18 4.82 4.79 4.79 4.88 4.76
19 4.52 4.56 4.56 4.63 4.51
20 4.52 4.60 4.60 4.56 4.53
21 6.00 6.03 6.03 6.08 6.00
22 6.30 6.23 6.23 6.17 6.22
Table 3. Continued Mol.
No
Obs.
Activity
LB Predicted Activity RB Predicted Activity
CoMFA CoMSIA CoMFA CoMSIA
23 6.39 6.48 6.48 6.43 6.43
24 6.39 6.42 6.42 6.37 6.43
25 6.00 5.98 5.98 5.98 6.02
26 5.70 5.70 5.70 5.74 5.71
27 5.00 5.10 5.10 5.12 5.08
28 5.70 5.50 5.50 5.58 5.58
29 5.30 5.33 5.33 5.31 5.31
30 5.00 5.07 5.07 5.11 5.06
31 5.15 5.13 5.13 5.03 5.19
32* 4.30 4.51 4.64 4.45 4.45
33* 4.70 5.01 4.87 5.03 5.03
34* 5.00 4.87 4.99 4.83 4.83
35 5.10 5.07 5.08 5.07 5.08
36 4.70 4.72 4.82 4.81 4.75
37 4.82 4.88 4.78 4.77 4.86
38 4.30 4.34 4.28 4.27 4.26
39 5.10 5.12 5.17 5.17 5.14
40 5.00 4.69 4.65 4.65 4.71
41* 5.15 4.75 4.69 4.71 4.71
42* 4.15 4.41 4.64 4.33 4.33
43* 5.00 5.16 4.65 5.06 5.06
44* 5.10 5.48 5.48 5.24 5.24
predictive ability of the molecules are provided in Table 3.
To visualize the information content of the derived 3D QSAR models, CoMFA and CoMSIA contour maps were generated. The contour plots are the representation of the lattice points and the difference in the molecular field values at lattice points, strongly connected with difference in the receptor binding affinity. Molecular fields define the favorable or unfavorable interaction energies of aligned molecules with a probe atom traversing across the lattice plots sug- gesting the modification required to design new molecules.
The contour maps of CoMFA denote the region in the space were the molecules would favorably or unfavorably interact with the receptor, while CoMSIA contour maps denote areas within the specified region where the presence of a group with a particular physicochemical property binds to the receptor. The CoMFA and CoMSIA results were graphically interpreted by field contribution maps using the ‘STDEV COEFF’ field type. Compound 23 the most potent inhibitor among the series was displayed on the maps for visuali- zation.
Figure 5(a)-(d) shows the contour maps for LB and RB CoMFA and CoMSIA steric fields with default 80 and 20%
level contributions, steric maps of these two models were similar. In the steric contours green and yellow represent favorable and unfavorable respectively.
Yellow contour near R1 substituent indicates steric bulk would decrease the activity and green contour near R2 substituent indicates steric favored region that accentuates the inhibitory activity by increasing the bulkiness. This is evident from the activity variations seen in different struc-
tures. Compounds 21-25 have maximum activity due to presence of methyl group as R2 substituent. Where as in Figure 5. (a)-(d) Steric contour maps for both ligand and receptor based CoMFA and CoMSIA. (a) & (b) Contours for ligand and receptor based CoMFA (c) & (d) Contours for ligand and receptor based CoMSIA.
Figure 6. (a) Electrostatic contour maps for ligand based and (b) receptor based CoMFA, regions favored for electropositive groups shown in blue where as regions favored for electronegative groups shown in red.
compounds 16-20 the presence of hydrogen atom a less bulky group has decreased the activity. To satisfy the steric bulkiness at this region, glutamic acid has been substituted that showed maximum hydrogen bond interactions with the active site, hence increasing the activity.
Electrostatic contour maps for both LB and RB CoMFA are shown in Figure 6(a) & (b) with field contributions of 90 and 10% for favored and disfavored respectively. The blue
and red contours depict the position where positively charged groups and negatively charged groups would be beneficial to inhibitory activity. This can be depicted in the compound 21-25 where CH3 group has been introduced on the core structure 3 in the positive favorable contour at region B are better than, respectively compound 16-20.
There is a disparity between the LB and RB CoMFA in the red contour region. There is a large red contour found Figure 7. Hydrophobic contour maps of (a) ligand based and (b) receptor based. Yellow contour represents hydrophobic favored regions and white contour represents hydrophobic disfavored. (c) ligand based and (d) receptor based CoMSIA. Purple contour for acceptor favored and red contour for acceptor disfavored.
Figure 8. Structures of newly designed molecules by change in substitutions at R1 Region named Q1 and Q2. Change in substitution at R1 and R2 region named Q3-Q8.
around the region B in the LB CoMFA suggesting an electronegative group will increase the activity. In contrary to the LB CoMFA the red contour has been shifted away from this region in RB CoMFA, suggesting a flipping in methylene group for better interaction with the receptor binding site.
The hydrophobic fields were represented in Figure 7(a) &
(b) with field contributions of 90 and 10% for favored and disfavored respectively. Yellow and white contours highlighted areas where hydrophobic and hydrophilic pro- perties were preferred respectively. A large yellow patch at R2 substituent in core 3, suggesting a hydrophobic group like a methyl increases the activity. The phenyl group at R1 in core 3 is protruding into the hydrophilic region that would decrease the activity of the molecules. Suggesting an hydro- philic group at this region would still increase the activity.
This was taken care while designing the new molecules where small hydrophilic groups like C2H5OH, CH3NH2 and CH3COOH were used to replace phenyl ring, inorder to increase the activity.
Hydrogen bond acceptor contour maps for both ligand and receptor based CoMSIA models are shown in Figure 7(c) &
(d) with filed contributions of 90 and 10% for favored and disfavored respectively. The red contour indicates dis- favored for hydrogen bond acceptor and purple indicates favored for hydrogen bond acceptor hence the presence of CH3 group in region B of core-3 at red contour showed highest activity.
The detailed contour map analysis of both CoMFA and CoMSIA models empowered us to identify structural require- ments for the observed inhibitory activity. The analogues were designed to improve the inhibitory activity. The best Table 4. Dock score and predicted activity and calculated IC50
values of newly designed molecules Molecule
No.
Ligand Based
Receptor
Based Calculated IC50 (µM)
CoMFA CoMSIA CoMFA CoMSIA
23 6.39 6.48 6.48 6.43 0.4
Q1 6.14 6.61 6.40 7.04 0.09
Q2 6.26 7.24 6.24 7.22 0.06
Q3 6.07 6.55 6.24 7.15 0.07
Q4 6.06 6.61 6.28 7.15 0.07
Q5 6.11 6.66 6.31 7.25 0.05
Q6 6.14 7.29 6.08 7.38 0.04
Q7 6.21 7.24 6.12 7.34 0.04
Q8 6.26 7.30 6.17 7.44 0.03
Figure 9. (a) Modeled protein overlapped with the template (Ribbon form) incorporated with the active site amino acids (Balls and sticks) for template (red) and modeled (peach) protein respectively. (b) Illustrating the field contributions from 3D QSAR depicting steric (green and yellow) and electrostatic (blue and red) for the designed (Q6) molecule embedded in fig a. (c) Hydrogen bonding interactions of the newly designed molecule (Q6) with the active site.
active molecule has been taken as a reference structure to design new molecules (Fig. 8) and to obtain new potent inhibitors.
The newly designed analogues when docked into the protein active site showed increased (6) hydrogen bond interactions (GLN36 (2), VAL116, TYR122, ASN65 (2)) with the active site and showed better dock score and predicted activity with respect to the reference compound (Table 4).
In Figure 9, Q6 molecule was embedded into the QSAR contours in the modeled protein overlapped with the template, showing hydrogen bond interactions with the active site amino acids. The predicted IC50 values were calculated using the reverse formula of pIC50.
IC50 = e−2.303 × pIC50
It was found that the IC50 values calculated had an increased activity ranging from 3 fold to 13 fold with respect to the reference molecule.
Conclusion
Bovine DHFR was modeled using mouse DHFR template having 89% sequence identity, this model has been validated using PROCHECK and PROSA II. The docking methodo- logy has been used as a tool to explain the binding affinity of the newly designed molecules, both LB and RB methods are appropriate to build 3D QSAR models of quinazolinone derivatives. The established models showed good q2 and r2pred values. Factors affecting the inhibitory activities are investigated by combining the contour maps and the results are in good accordance and complementary to each other.
Bulky and hydrophobic groups at R2 and small hydrophilic group at R1 are preferred. The designed molecules based on these parameters showed better activity than the reference molecules, which indicates the 3D QSAR model generated has a good predictive ability and can be used to design potent inhibitors. All these results yield reliable and precious information for further structure based drug design optimi- zation.
Acknowledgments. We gratefully acknowledge support for this research from University Grants Commission, India, Department of Science and Technology, India and Depart- ment of chemistry, Nizam College, Hyderabad. We are greatly thankful to Dr. G. N. Sastry, Indian Institute of Chemical technology for Sybyl 6.9 software and his useful suggestions. We also thank Andrew Sali for the academic free license of MODELLER software.
References
1. Trimble, J. J.; Murthy, S. C. S.; Bakker, A.; Grassmann, R.;
Desrosiers, R. C. Science 1988, 239, 1145.
2. Collin, J. Suckling, Enzyme Chemistry; Chapman and Hall Ltd.:
733 Third Avenue, NY, 1984; p 121.
3. Suresha, G. P.; Prakasha, K. C.; Kapfo, W.; Gowda, D. C. E-J Chem. 2010, 7(2), 449.
4. Alagarsamy, V.; Muthukumar, V.; Pavalarani, N.; Vasanthanathan, P.; Revathi, R. Biol. Pharm. Bull. 2003, 26(4), 557.
5. Mani, C. P.; Yakaiah, T.; Raghu, R. R. A.; Narsaiah, B.;
ChakraReddy, N.; Sridhar, V.; Venkateshwara, R. J. Eur. J. Med.
Chem. 2007, 42, 147.
6. Ravishankar, C. H.; Devender, R. A.; Bhaskar, R. A.; Malla, R. V.;
Sattur, P. B. Curr. Sci. 1984, 53, 1069.
7. Ouyang, G.; Zhang, P.; Xu, G.; Song, B.; Yang, S.; Jin, L.; Xue, W.; Hu, D.; Lu, P.; Chen, Z. Molecules 2006, 11, 383.
8. Martin, T. A.; Wheller, A. G.; Majewski, R. F.; Corrigan, J. R. J.
Med. Chem. 1964, 7, 812.
9. Dienei, J. B.; Dowalo, F.; Hoeven, H. V.; Bender, P.; Loev, B. J.
Med. Chem. 1973, 16, 633.
10. Jatav, V.; Mishra, P.; Kashaw, S.; Stables, J. P. Eur. J. Med. Chem.
2008, 4, 135.
11. Ilangovan, P.; Ganguly, S.; Pandi, V. J. Pharm. Res. 2010, 3, 703.
12. Chandrasekhar, V.; Raghurama, R. A.; Malla, R. V. Indian Drugs 1986, 3, 24.
13. Naithani, P. K.; Gautam, P.; Srivastava, V. K.; Shankar, K. Indian J. Chem. 1989, 28B, 745.
14. Magnus, N. A.; Confalone, P. N.; Storace, L.; Patel, M.; Wood, C. C.; Davis, W. P.; Parsons, R. L. J. Org. Chem. 2003, 68, 754.
15. Jantova, S.; Urbancikova, M.; Maliar, T.; Mikuldsova, M.; Rauko, P.; Cipak, L.; Kubikova, J.; Stankovsky, S.; Spirkova, K.
Neoplasm. 2001, 48, 52.
16. Sarah, T. R.; Ihsan, A.; Aboldahab et al. Bio. Org. Med. Chem.
2006, 14, 8608.
17. Klebe, G.; Abraham, U.; Mietzner, T. J. Med. Chem. 1994, 37, 4130.
18. Wold, S.; Ruhe, A.; Wold, H.; Dunn, W. J. I. SIAM J. Sci. Stat.
Comput. 1984, 5, 735.
19. Cavalli, A.; Greco, G.; Novelliono, E.; Recanatini, M. Boiorg.
Med. Chem. 2000, 8, 2771.
20. Boeckmann, B.; Bairoch, A.; Apwweiler, R.; Blatter, M. C.;
Estreicher, A.; Gasteiger, E.; Martin, M. J.; Michoud, K.;
O’Doonovan, C.; Phan, I.; Pilbout, S.; Schneider, M. Nucl. Acids Res. 2003, 31, 365.
21. Altschul, S. F.; Thomas, L. M.; Alejandro, A. S.; Jinghui, Z.;
Zheng, Z.; Miller, W.; Lipman, D. J. Nucleic Acids Res. 1997, 25, 3389.
22. Balazs, J.; Arpad, M. J. Mol. Grap. and Model. 2007, 25, 711.
23. Sali, A.; Blundell, T. L. J. Mol. Biol. 1993, 234, 779.
24. Marti-Renom, M. A.; Stuart, A. C.; Fiser, A.; Sanchez, R.;
Melo, F.; Sali, A. Annu. Rev. Biophys. Biomol. Struct. 2000, 29, 291.
25. Laskowski, R. A.; Mac, A. M. W.; Moss, D. S.; Thornton, J. M. J.
Appl. Cryst. 1993, 26, 283.
26. Bowie, J. U.; Lüthy, R.; Eisenberg, D. Science 1991, 253, 164.
27. Lüthy, R.; Bowie, J. U.; Eisenberg, D. Nature 1992, 356, 83.
28. Wiederstein & Sippl Nucleic Acid Res. 2007, 35, W407.
29. Sippl, M. J. Proteins 1993, 17, 355.
30. Sali, A.; Kurian, J. Trends Biochem. Sci. 1999, 22, M20.
31. SYBYL Molecular Modeling System, version 6.9, Tripos Inc., St.
Louis, MO, 63144.
32. Gasteiger, J.; Marsili, M. Tetrahedron 1980, 363, 3219.
33. Stitch, I.; Car, R.; Parrinello, M.; Baroni, S. Phys. Rev. B 1989, 39, 4997.
34. Leach, A. Molecular Modelling, Principles and Applications;
Longman: Harlow, Essex, England, 1996.
35. Forcefield-Based Simulations; Accelerys, Corp.: San Diego, CA.
Chapter 4, Minimization.
36. Jensen, F. Introduction to Computational Chemistry; John Wiley:
Chichester, England, 1999; p 322.
37. Rarey, M.; Kramer, B.; Lengauer, T.; Kleb, G. A. J. Mol. Biol.
1996, 261, 470.
38. Balazs, J.; Arpad, M. J. Mol. Grap. And Model 2007, 25, 711.
39. Cramer, R. D., III.; Patterson, D. E.; Bunce, J. D. J. Am. Chem Soc. 1988, 110, 5959.
40. Cramer, R. D., III.; Bunce, J. D.; Patterson, D. E. Quant. Struct.
Act. Relat. 1988, 7, 18.
41. Klebe, G.; Abraham, U.; Mietzner, T. J. Med. Chem 1994, 37, 4130.
42. Altschul, S. F.; Gish, W.; Miller, W.; Myers, E. W.; Lipman, D. J.
J. Mol. Biol. 1990, 215, 403.
