Yuhuuu
Yuhuuu
Yuhuuu
Cairo University
ORIGINAL ARTICLE
G R A P H I C A L A B S T R A C T
A R T I C L E I N F O A B S T R A C T
Article history: Quantitative structure-activity relationship and molecular docking studies were carried out on a
Received 4 July 2016 series of quinazolinonyl analogues as anticonvulsant inhibitors. Density Functional Theory
Received in revised form 11 October (DFT) quantum chemical calculation method was used to nd the optimized geometry of the
2016 anticonvulsants inhibitors. Four types of molecular descriptors were used to derive a quantita-
tive relation between anticonvulsant activity and structural properties. The relevant molecular
http://dx.doi.org/10.1016/j.jare.2016.10.004
2090-1232 2016 Production and hosting by Elsevier B.V. on behalf of Cairo University.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
34 U. Abdulfatai et al.
Accepted 15 October 2016 descriptors were selected by Genetic Function Algorithm (GFA). The best model was validated
Available online 16 November 2016 and found to be statistically signicant with squared correlation coefcient (R2) of 0.934,
adjusted squared correlation coefcient (R2adj) value of 0.912, Leave one out (LOO) cross valida-
Keywords: tion coefcient (Q2) value of 0.8695 and the external validation (R2pred) of 0.72. Docking analysis
QSAR method revealed that the best compound with the docking scores of 9.5 kcal/mol formed hydrophobic
Gamma aminobutyric acid interaction and H-bonding with amino acid residues of gamma aminobutyric acid aminotrans-
aminotransferase ferase (GABAAT). This research has shown that the binding afnity generated was found to be
Molecular docking better than the commercially sold anti-epilepsy drug, vigabatrin. Also, it was found to be better
Density functional theory than the one reported by other researcher. Our QSAR model and molecular docking results cor-
Anticonvulsant roborate with each other and propose the directions for the design of new inhibitors with better
Genetic function algorithm activity against GABAAT. The present study will help in rational drug design and synthesis of
new selective GABAAT inhibitors with predetermined afnity and activity and provides valuable
information for the understanding of interactions between GABAAT and the anticonvulsants
inhibitors.
2016 Production and hosting by Elsevier B.V. on behalf of Cairo University. This is an open
access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/
4.0/).
Table 1 (continued)
Comp. number Compound pED50 Pred.Pred.pED50ED50 Residual
Table 1 (continued)
Comp. number Compound pED50 Pred.Pred.pED50ED50 Residual
a
Training set.
b
Test set.
Computational method where SSE is the sum of squares of errors, c is the number of
terms in the model, other than the constant term, d is an user-
In order to obtain validated QSAR models, the descriptors dened smoothing parameter, p is the total number of descrip-
(1D-3D) generated from the PaDEL version 2.18 tool kits tors contained in all model terms (ignoring the constant term)
[17] were divided into training and test sets. The training set and M is the number of samples in the training set. Unlike the
was used to generate the model, while the test set was used commonly used least squares measure, the LOF measure can-
for the external validation of the model [18]. The correlation not always be reduced by adding more terms to the regression
between activity values of the molecules against GABAAT model. While the new term may reduce the SSE, it also
and the calculated descriptors was obtained through correla- increases the values of c and p, which tend to increase the
tion analysis using the material studio software version 8. LOF score. Thus, adding a new term may reduce the SSE,
Pearsons correlation matrix was used as a qualitative model, but actually increases the LOF score. By limiting the tendency
in order to select the suitable descriptors for regression analy- to simply add more terms, the LOF measure resists over tting
sis. The generated descriptors from the PaDEL version 2.18 better than the SSE measure (Materials Studio 8.0 Manual).
tool kits [17] were subjected to regression analysis with the
experimentally determined activities as the dependent variable Quality assurance of the model
and the selected descriptors as the independent variables using
Genetic Function Algorithm (GFA) method in material studio The reliability and predictive ability of the developed QSAR
software version. The number of descriptors in the regression models were evaluated by internal and external validation
equation was 4, and Population and Generation were set to parameters.
600 and 600, respectively. The number of top equations
returned was 4. Mutation probability was 0.1, and the smooth- Internal and external validations
ing parameter was 0.5. The models were scored based on
Friedmans Lack of Fit (LOF). In GFA algorithm, an individ- The internal and external validation parameters were com-
ual or model was represented as one-dimensional string of bits. pared with the minimum recommended value for the evalua-
It was a distinctive characteristic of GFA that it could create a tion of the quantitative QSAR model [20] as shown in
population of models rather than a single model. GFA algo- Table 2. The square of the correlation coefcient (R2) describes
rithm, selecting the basic functions genetically, developed bet- the fraction of the total variation attributed to the model. The
ter models than those made using stepwise regression methods. closer the value of R2 is to 1.0, the better the regression equa-
And then, the models were estimated using the LOF, which tion explains the Y variable. R2 is the most commonly used
was measured using a slight variation of the original Friedman internal validation indicator and is expressed as follows:
formula, so that the best tness score can be received. The P
Yobs Ypred2
revised formula of LOF [19] is as follows: R2 1 P 2
, 2 Yobs Ytraining2
C dp
LOF SSE 1 1 where Yobs, Ypred, and Ytraining are the experimental prop-
M
erty, the predicted property and the mean experimental prop-
38 U. Abdulfatai et al.
Table 2 General minimum recommended value for the evaluation of the quantitative QSAR model.
Symbol Name Value
R2 Coecient of determination P0.6
P(95%) Condence interval at 95% condence level <0.05
Q2 Cross validation coecient P0.5
R2 - Q2 Dierence between R2 and Q2 60.3
Next. test set Minimum number of external test set P5
R2ext Coecient of determination for external test set P0.6
erty of the samples in the training set, respectively [20]. of Pyrx by describing the target GABAAT protein. The energy
Adjusted R2 (R2adj) value varies directly with the increase in grid was performed based on Lamarckian genetic algorithm
number of repressors i.e. descriptors; thus, R2 cannot be an [24]. Ligplot, discovery studio 3.5 and PyMol visualization
useful measure for the goodness of model tness. Therefore, software were used to perform the virtual analysis of docking
R2 is adjusted for the number of explanatory variables in the site.
model. The adjusted R2 is dened as follows:
Preparation of the target receptor
n1 n 1R2 P
R2adj 1 1 R2 3 The 3D structure of GABAAT receptor (1OHV) was obtained
np1 np1
from the protein data bank in PDB format. All Heteroatomic
where molecules were excluded from the le using Discovery Studio
n is the number of training compounds. 3.5 software. GABAAT receptor structure was minimized, pro-
p = number of independent variables in the model [21]. tonated and saved in PDBQT le format in all polar residues.
Fig. 1(a and b) shows the prepared three dimensional structure
The leave one out cross validation coefcient (Q2) is given of GABAAT (10HV).
by the following:
P Preparation of the ligands
Yp Y2
Q2 1 P 4 The 24 synthesized compounds of quinazolinonyl derivatives
Y Ym2
(Table 1) were selected from the literature and used as ligands
where Yp and Y represent the predicted and observed activity [15]. Chemdraw software was used to draw the 2D structures
respectively of the training set and Ym the mean activity value of these compounds and was then converted to 3D structures,
of the training set [22]. optimized and saved in pdb le format by Spartan14 version
1.1.2 [16]. The compounds were converted to PDBQT format
Applicability domain by Autodock 4.2 software. The 3D structures of the prepared
ligands are shown in Fig. 2.
The applicability domain (AD) of the generated models was
assessed in order to specify the scope of their proposed models Structure validation
by dening the mathematical model limitations with respect to Native ligands present in the protein structure were removed.
its structural domain and response space. In order to check the conrmation, root mean square devia-
tion (RMSD) value was calculated between the original struc-
Docking study ture and the ligand deleted structure [25,26].
Fig. 1 (a) Structure of GABAAT (10HV), (b) Structure of GABAAT(10HV) Preparation of compounds for docking.
Molecular modeling and docking of some anticonvulsant agents 39
(GABAAT) was kept as rigid while ligand molecules were kept Model 2
as exible throughout the docking studies. pED50 = 0.279901890 * VP-6 + 0.188955711 * XLogP +
0.033018384 * PPSA-3 + 3.694884401 * RNCG 0.404755657,
Results and discussion N = 17, R2ext = 0.62704, R2 = 0.932637, R2a = 0.910182,
Q2cv = 0.832929, LOF 0:002876, Min expt. error for non-
QSAR studies signicant LOF (95%) = 0.018704.
Model 3
All the four developed QSAR models were recorded out of pED50 = 0.148446854 * VP-4 + 0.190534973 * XLogP +
which the best model (model 1) was identied and reported 0.032884549 * PPSA-3 + 4.028075797 * RNCG 0.595730073,
due to the statistical signicance. The name and symbol of N = 17, R2ext 0:703963, R2 0:931777, R2a = 0.909036,
the descriptors used in the QSAR optimization model are Q2cv 0:806221, LOF 0:002912, Min expt. error for non-
shown in Table 3 below. Table 4 gives the result of Validation signicant LOF (95%) = 0.018823.
of the Genetic Function Algorithm (GFA) of model 1 that was
generated from material studio. Minimum recommended value
of validation Parameters for a generally acceptable QSAR
model [20] was in agreement with the model 1 parameters. Table 5 Pearsons correlation matrix for descriptors used in
Based on this analysis, Model 1 was selected and reported as QSAR model for the activities of anticonvulsant molecules.
the best QSAR model. ETA_Eta_L XLogP PPSA-3 RNCG
Model 1
pED 50 = 0.114383001 * ETA_Eta_L + 0.190098515 * ETA_Eta_L 1
XLogP 0.17959 1
XLogP + 0.028759587 * PPSA-3 + 4.201924750 * RNCG
PPSA-3 0.1924 0.25267 1
0.690224604, N = 17, R2ext 0.72028, R2 0:934053, RNCG 0.57017 0.35028 0.54108 1
R2a = 0.912071, Q2cv = 0.869587, LOF 0:002815, Min expt.
error for non-signicant LOF (95%) = 0.018506.
Table 3 List of some physiochemical descriptors used for the best model.
S/NO Symbol Names of descriptors Class
1 ETA_Eta_L Local index Eta_local 2D
2 XLogP XLOgP 2D
3 PPSA-3 Charge weighted partial positive surface area 3D
4 RNCG Relative negative charge most negative charge/total negative charge 3D
U. Abdulfatai et al.
Molecular modeling and docking of some anticonvulsant agents 41
Fig. 3 Three-dimensional docked GABAAT - Ligands Complex. (A) Interactions between GABAAT and Ligand 13a. (B) Interactions
between GABAAT and Ligand 15b. (C) Interactions between GABAAT and Ligand 24b. Ligand:H-bond interactions, green dashed lines:
Hydrophobic interactions, red dashed line.
Model 4 Q2cv 0:87158, LOF 0:003107, Min expt. error for non-
pED50 = 0.109267006 * SP-6 + 0.197509169 * XLogP + signicant LOF (95%) = 0.01944.
0.029112087 * PPSA-3 + 4.163767660 * RNCG 0.562852003, The result from the Correlation matrix (Table 5) shows
N = 17, R2ext = 0.69, R2 = 0.92723, R2a = 0.902973, clearly that the correlation coefcients between each pair of
42 U. Abdulfatai et al.
descriptors are very low, and this means that there exist no sig- Asn423, Arg422, His44, Arg430, Leu436, Ile426, Tyr438,
nicant inter correlation among the descriptors used in devel- Ile72, Tyr69, His206, Gly438, Lys203, and Glu270.
opment of the model. Suppl. Fig. 1 gives the plot of predicted
activities of both training and test sets against observed activ- Conclusions
ities; the reliability of the model (best QSAR model) was fur-
ther conrmed as the GFA derived R2 value was in It has been clearly demonstrated that the approach utilized in
agreement with R2 value of 0.93 recorded in this graph. this study was successful in nding novel GABAAT inhibitors
The Williams plot, the plot of the standardized residuals from the data set developed by computational methods. The
against the leverage (suppl. Fig. 2), was used to visualize the model generated from various physicochemical descriptors
applicability domain (AD) [27]. Leverage indicates a com- corresponds to the essential structural features of quinazoli-
pounds distance from the centroid of X. The leverage of a nonyl analogues and found to have signicant correlation
compound in the original variable space is dened as follows: coefcient of determination (R2) of 0.934 with GABAAT
1 inhibiting activity. Substituted quinazolinonyl analogues
hi XTi XT X Xi 5 showed good interactions with GABAAT protein. Compound
The danger leverage (h*) is dened as follows: (13a), in particular, showed high binding afnity with docking
score of -9.5 kcal/mol against GABAAT in docking analysis
3P 1
hi 6 and predicted pED50 value of 1.77 in QSAR analysis. The
N ligand was docked deeply within the binding pocket region
where N is the number of training compounds, and p is the forming a hydrogen bond with Gly440 (3.04 A), and
number of predictor variables. Where Xi is the descriptor vec- hydrophobic interactions with Cys439, Asn423, Arg422,
tor of the considered compound and X is the descriptor matrix His44, Arg430, Leu436, Ile426, Tyr438, Ile72, Tyr69, His206,
derived from the training set descriptor values. In suppl. Fig. 3, Gly438, Lys203, and Glu270. From the docking analysis, we
it is obvious that all compounds in the test set fall inside the realized that the binding scores generated were found to be
domain of the model (the danger leverage limit is 0.88). All better than the one proposed by other researcher [28].
the training and test sets are good leverages since none of Furthermore, all the quinazolinonyl analogues were found
the chemical compounds go beyond the danger hi value, so to be docked to GABAAT better than the standard anti-
they can be regarded as good prediction for the model. epilepsy drug (vigabatrin). The physicochemical descriptors
used in QSAR analysis (model 1) in this study were important
Molecular docking studies parameters to consider in improving the potency of these sub-
stituted quinazolinonyl analogues as inhibitors of GABAAT.
Molecular docking studies were carried out between the targets Our QSAR model (high correlation coefcient of determina-
(GABAAT) and the inhibitors. All the compounds were found tion R2 of 0.934) and molecular docking results (high binding
to strongly inhibit by completely occupying the active sites in afnity with docking score of 9.5 kcal/mol) corroborate with
the target protein (GABAAT). All inhibitors showed low each other and propose the directions for the design of new
energy values (high docking scores) than the binding energies inhibitors with better activity toward GABAAT. This study will
of vigabatrin (-4.4 kcal/mol), the standard antiepileptic drug. help in rational drug design and synthesis of new selective
For target protein, binding energy values range from -6.0 to GABAAT inhibitors with predetermined afnity and activity
-9.5 kcal/mol. In Table 6, most of the inhibitors were found and provides valuable information for the understanding of
to involve in both the hydrophobic interactions and hydrogen interactions between GABAAT and the novel compounds
bonding with the receptor (GABAAT). In addition, ligand and might pave the way toward discovery of novel GABAAT
number 13a with binding energies of -9.5 kcal/mol showed bet- inhibitors.
ter binding energies than other co-ligands.
Conflict of Interest
Binding mode of inhibitors
No conflict of interest.
Table 6 shows the docking scores, hydrogen bond length (in
angstrom) and interacting residues involved in the docking Funding
of inhibitors (ligands) at the active site of GABAAT. Fig. 3
shows the best rst-three docking results. Ligand number The authors received no direct funding for this research.
24a shows that Arg422, Tyr69, Ile105, Ile72, Phe351, Tyr348,
and Glu270 residues of target are involved in hydrophobic Compliance with Ethics Requirements
interactions. In addition, it also forms hydrogen bonds
(3.06 A) with Gly440. Strong inhibitor binding is also reected
by the frequency of hydrogen bonds as shown in Table 4. This article does not contain any studies with human or animal
Compound 15b made two hydrogen bonds (3.04 A and subjects.
3.05 A) with two residues Tyr69 and Gly440, while hydropho-
bic interactions are observed with Ile426, Arg430, Arg422, Appendix A. Supplementary material
Tyr348, His44, Ile72, Ile105, Glu270, Act500, His206,
Lys203, and Cys439. Compound 13a (compound with the best Supplementary data associated with this article can be found,
binding score of -9.5 kcal/mol) forms a hydrogen bond with in the online version, at http://dx.doi.org/10.1016/j.jare.2016.
Gly440 (3.04 A), and hydrophobic interactions with Cys439, 10.004.
Molecular modeling and docking of some anticonvulsant agents 43