Journal of Molecular Modeling

, Volume 18, Issue 2, pp 569–582

Multiple receptor conformation docking and dock pose clustering as tool for CoMFA and CoMSIA analysis – a case study on HIV-1 protease inhibitors

Authors

  • Sree Kanth Sivan
    • Department of Chemistry, Nizam CollegeOsmania University
    • Department of Chemistry, Nizam CollegeOsmania University
Original Paper

DOI: 10.1007/s00894-011-1048-x

Cite this article as:
Sivan, S.K. & Manga, V. J Mol Model (2012) 18: 569. doi:10.1007/s00894-011-1048-x

Abstract

Multiple receptors conformation docking (MRCD) and clustering of dock poses allows seamless incorporation of receptor binding conformation of the molecules on wide range of ligands with varied structural scaffold. The accuracy of the approach was tested on a set of 120 cyclic urea molecules having HIV-1 protease inhibitory activity using 12 high resolution X-ray crystal structures and one NMR resolved conformation of HIV-1 protease extracted from protein data bank. A cross validation was performed on 25 non-cyclic urea HIV-1 protease inhibitor having varied structures. The comparative molecular field analysis (CoMFA) and comparative molecular similarity indices analysis (CoMSIA) models were generated using 60 molecules in the training set by applying leave one out cross validation method, rloo2 values of 0.598 and 0.674 for CoMFA and CoMSIA respectively and non-cross validated regression coefficient r2 values of 0.983 and 0.985 were obtained for CoMFA and CoMSIA respectively. The predictive ability of these models was determined using a test set of 60 cyclic urea molecules that gave predictive correlation (rpred2) of 0.684 and 0.64 respectively for CoMFA and CoMSIA indicating good internal predictive ability. Based on this information 25 non-cyclic urea molecules were taken as a test set to check the external predictive ability of these models. This gave remarkable out come with rpred2 of 0.61 and 0.53 for CoMFA and CoMSIA respectively. The results invariably show that this method is useful for performing 3D QSAR analysis on molecules having different structural motifs.

https://static-content.springer.com/image/art%3A10.1007%2Fs00894-011-1048-x/MediaObjects/894_2011_1048_Figa_HTML.gif
Figure

Schematic representation of the multiple receptor conformation docking, clustering and 3D QSAR. Comparative molecular field analysis (CoMFA) and comparative molecular similarity indices (CoMSIA) analysis is an exceptional tool for understanding the structure activity relations of molecules towards their biological activities. Receptor binding conformation of the molecule gives an added advantage to understand ligand receptor interactions required for bioactivity. There are different methods employed for obtaining the receptor based alignment of the molecules, but this method is limited to molecules having common substructure. Here we describe the new approach of multiple receptors conformation docking (MRCD) and clustering of dock poses that allows seamless incorporation of receptor binding conformation of the molecules on wide range of molecules with varied structural scaffold.

Keywords

AIDS (acquired immunodeficiency syndrome)CoMFA (Comparative molecular field analysis)CoMSIA (Comparative molecular similarity indices analysis)HIV- 1 (Human immunodeficiency virus type1)MRCD (Multiple receptor conformation docking)PLS (partial least square) analysisPR (aspartic protease)

Introduction

Human immunodeficiency virus type1 (HIV-1) is responsible for human acquired immunodeficiency syndrome (AIDS) [1, 2], one of the most urgent world health threats. With ca. 42 million HIV/AIDS patients worldwide, only 20 anti-HIV drugs are currently available for clinical use. HIV-1 genome encodes for three major enzymes protease, reverse transcriptase and integrase for HIV-1 replication. The aspartic protease (PR) of the human immunodeficiency virus type 1 (HIV-1) cleaves the viral gag-pol fusion precursor polyprotein into active viral structural proteins and replicative enzymes such as reverse transcriptase, endonuclease, and integrase, thus playing an essential role in the maturation of HIV-1 particles and virus replication [3]. Therefore, PR is an important target for the design of specific antiviral agents dedicated to treatment of HIV-1 infection and acquired immunodeficiency syndrome (AIDS) [48]. cyclic urea molecules have been reported to constitute an entirely new class of potent and perspective nonpeptidic inhibitors of PR, fundamental feature of the cyclic urea inhibitors is the carbonyl oxygen that mimics the hydrogen-bonding features of the key structural water molecule present in the active site of the PR [9]. In the present study 120 cyclic urea molecules are taken into consideration apart from that, another 25 non-cyclic urea molecules having HIV-1 protease inhibitory activity are used for comparative molecular field analysis (CoMFA) [10] and comparative similarity indices analysis (CoMSIA) [11] based 3D QSAR analysis.

3D QSAR techniques, such as the CoMFA and CoMSIA, are based on the experimental structure-activity relationship on specific bio-macromolecule and ligand pair. This method is based only on the ligand structure and thus the spatial alignment is crucial in determining the accuracy of these approaches. Another reason for the inaccuracy in traditional ligand-based 3D QSAR [12] lies in the fact that the conformation used in the alignment may not be the active conformer of that ligand. One way to surmount such inherent imperfection is to introduce three-dimensional structures of the target bio-macromolecule during alignment process this strategy is called receptor-based 3D QSAR analysis [1316]. The three-dimensional structures of the target can be obtained either from experimentally derived X-ray crystallography, NMR spectroscopy or from homology modeling.

Different methods have been employed for receptor-based alignment of the ligands. In general, the X-ray crystal structure of inhibitor complexed with the enzyme is used as a template for superimposition, assuming that this conformation represents the most probable bioactive conformation.

Receptor-docked alignment derived from the structure-based docking algorithms like GOLD, FlexX, GLIDE, and Autodock are used as such [17] or a rigid realignment of the poses from the receptor-docked alignment is performed, or the use of best docked mode of the smallest compound as template and modified for the other compounds [18]. These compounds were minimized and minimized structures at this binding mode were superimposed to get the molecular alignment for CoMFA and CoMSIA. To incorporate receptor flexibility, docking followed by molecular dynamics and minimization of protein ligand complex is performed and the ligand conformation obtained is used as template for alignment [19].

All these methods are helpful for the alignment of molecules having maximum common substructure, but cannot be employed for ligands having diverse structures. Here in this article we report a method to incorporate protein flexibility by applying multiple receptor conformation docking (MRCD) [2022] and clustering of docked pose for obtaining the alignment of compounds which can have a diverse substructure.

Methodology

Twelve high resolution X-ray crystal structure and one NMR resolved crystal structure of HIV-1 protease in complex with inhibitors (pdb id: 1PRO, 1BV9, 1AJX, 1AJV, 1T7K, 1QBR, 1QBS, 1QBU, 1HVR, 1HVH, 1DMP, 1G35, 1BVG) [2332] were downloaded from the protein data bank. GLIDE 5.6 [33] was used for molecular docking. The resolution of the crystal structure is given in Table 1. The proteins were prepared using protein preparation module applying the default parameters, grids were generated around the active site of the protease with receptor Van der Waals scaling for the non-polar atoms as 0.9 [34].
Table 1

PDB ids of crystal structure of HIV-1 protease, their resolution and RMSD of redocked co-crystallized ligand

S.No.

Protein PDB id

X-ray crystal structure resolution (Å)

RMSD (Å)

1

1AJV

2.00

0.942

2

1AJX

1.85

0.497

3

1BV9

2.00

2.115

4

1DMP

2.00

0.456

5

1G35

1.80

2.05

6

1HVH

1.80

2.24

7

1HVR

1.80

0.537

8

1QBR

1.80

0.421

9

1QBS

1.80

0.521

10

1QBU

1.80

0.863

11

1T7K

2.10

0.77

12

1PRO

1.80

1.115

13

1BVG

NMR resolved

0.508

A set 145 known HIV-1 protease inhibitors that includes 120 cyclic urea and 25 non cyclic urea molecules with diverse structures and varied range of inhibition constants (Ki) were selected from literature [4, 9, 27, 28, 3540], these were built using maestro build panel and prepared by LigPrep application in Schrödinger 2010 suite. Structures of cyclic urea molecules are given in Table 2. LigPrep produces the low energy conformer of the ligand using the MMFF94s force field. The lower energy RSSR conformations of the ligands were selected and docked into the grid generated from the 13 protein structures using the standard precision docking mode [34]. The crystal structure ligands were also docked and its RMSD was calculated to validate the docking process.
Table 2

Structures of cyclic urea molecules with their experimental pKi and predicted pKi

https://static-content.springer.com/image/art%3A10.1007%2Fs00894-011-1048-x/MediaObjects/894_2011_1048_Tab2r1_HTML.gifhttps://static-content.springer.com/image/art%3A10.1007%2Fs00894-011-1048-x/MediaObjects/894_2011_1048_Tab2r2_HTML.gifhttps://static-content.springer.com/image/art%3A10.1007%2Fs00894-011-1048-x/MediaObjects/894_2011_1048_Tab2r3_HTML.gifhttps://static-content.springer.com/image/art%3A10.1007%2Fs00894-011-1048-x/MediaObjects/894_2011_1048_Tab2r4_HTML.gif
The best dock pose (low binding energy conformer according to the glide dock score) of each ligand from 13 docking runs performed on 13 receptors grids were analyzed for their hydrogen bond interactions with the receptor. The pose with the required hydrogen bonding namely with carboxylate of Asp25 and amine of Ile50 were selected for further clustering. The dock poses were clustered using clustering of conformer’s script in Schrödinger 2010 suite. The clustering was performed using atomic RMSD, in this RMSD was calculated in place which does not alter the dock pose of each conformer. The lowest binding energy conformation from the most populated cluster was chosen for CoMFA and CoMSIA analysis without further alignment, i.e., super imposition of ligands based on the common substructure for a set of molecules was not done, instead the docked conformer pose obtained form clustering were taken as is for all ligands. This imparts the flexible receptor binding information of each ligand in data set. The resulting docked pose orientations is shown in Fig. 1.
https://static-content.springer.com/image/art%3A10.1007%2Fs00894-011-1048-x/MediaObjects/894_2011_1048_Fig1_HTML.gif
Fig. 1

(a) clustered pose of cyclic urea molecules from MRCD (b) clustered pose of non-cyclic urea molecules from MRCD

The molecules were imported into Sybyl 6.9 molecular modeling program package [41] Gasteiger-Hückel [42] charges were assigned. The standard Tripos force fields were employed for the CoMFA and CoMSIA analysis. A 3D cubic lattice of dimension 4Å in each direction with each lattice intersection of regularly spaced grid of 2.0 Å was created. The steric and electrostatic parameters were calculated in case of the CoMFA fields while hydrophobic, acceptor and donor parameters in addition to steric and electrostatic were calculated in case of the CoMSIA fields at each lattice. The sp3 carbon 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 electrostatic contributions were cut off at 30 kcal mol-1.

A partial least squares (PLS) 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 (pKi) for the data set of ligands. One hundred twenty cyclic urea molecules were divided into training and test set of 60 molecules each respectively, considering the set had a balanced distribution of more and less active compounds. Twenty five non cyclic inhibitors were taken as external test set. 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-1 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 determine the optimum number of components (ONC) and the coefficient rloo2. The optimum number of components obtained 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 regression coefficient (r2). Since the statistical parameters were found to be the best for the model from the LOO method, it was employed for further predictions of activity of test molecules for cross validation of the model. The schematic representation of the multiple receptor conformation docking (MRCD), clustering and 3D QSAR is given in Fig. 2.
https://static-content.springer.com/image/art%3A10.1007%2Fs00894-011-1048-x/MediaObjects/894_2011_1048_Fig2_HTML.gif
Fig. 2

Schematic representation of the multiple receptor conformation docking, clustering and 3D QSAR

Multiple receptors conformation docking is a method employed for incorporating the receptor flexibility while docking analysis. In this method different conformations of the same protein are taken and ligands are docked into the grids generated from these conformations. The accuracy of a docking procedure lies in how closely the lowest energy pose (binding conformation) predicted by the object scoring function (Glide score), resembles an experimental binding mode as determined by X-ray crystallography. In the present study, standard precision glide docking procedure was validated by removing crystal structure ligand from the binding site and redocking it to the binding site of HIV-1 protease for each of the receptor conformations.

Results and discussion

We found a very good agreement between the localization of the inhibitor upon docking and from the crystal structure, i.e., having similar hydrogen bonding interactions with Asp 25 and Ile 50. The root mean square deviations between the predicted conformation and the observed X-ray crystallographic conformation for the ligands ranged from 0.4Åto 2.24Å, the values are provided in Table 1, these values suggests the reliability of Glide docking in reproducing the experimentally observed binding mode for HIV-1 protease inhibitor and the parameter set for the Glide docking is reasonable to reproduce the X-ray structure.

The advantage of MRCD is clearly understood when each molecules dock pose was analyzed to confirm the presence of required hydrogen bond interaction with the active site amino acids namely with carboxylate of Asp25 and amine of Ile50. In some receptor confirmation the interactions were missing or inverse, in this manner the redundant poses of the molecules can be screened out and the biologically active conformation of the ligand can be obtained. To obtain the most preferred biologically active conformer of the ligand, a cluster analysis of the dock poses of each ligand was performed using clustering of conformer’s script. Pose from the most frequent cluster having lowest binding energy was selected. This pose of each molecule were taken as the basis for the CoMFA and CoMSIA analysis, these conformation provide the most vital information for the binding of ligand into the protein active site.

3D QSAR analysis was done by dividing the molecules into training and test set having 60 molecules each, keeping in view that the activity range is at least 5 log units different in both the sets. The CoMFA and CoMSIA statistical analysis is summarized in Table 3. Statistical data shows rloo2 0.598 for CoMFA and 0.674 for the CoMSIA models, respectively, which indicates a good internal predictive ability of both models. The models developed also exhibited r2 of 0.983 and 0.985 for CoMFA and CoMSIA, respectively. To test the predictive ability of the models, a test set of 60 molecules excluded from the model derivation was used. The predictive correlation coefficient rpred2 of 0.684 for CoMFA and 0.640 for CoMSIA models indicate good external predictive ability of the model. The experimental and predicted activity from CoMFA and CoMSIA model is given in Table 2.
Table 3

Summary of CoMFA and CoMSIA statistical analysis 

Statistical parameters

CoMFA

CoMSIA

PLS result summary for model derived from cyclic urea molecules

r2looa

0.598

0.674

Number of cyclic urea molecules in training set

60

60

Number of cyclic urea molecules in test set

60

60

Number of non-cyclic urea molecules in test set for external validation

25

25

ONCb

8

10

SEEc

0.199

0.189

r2 d

0.983

0.985

Fratioe

153.166

311.772

r2predf

0.684

0.640

rpred2 on non-cyclic urea molecules

0.61

0.53

Fraction of field contributions

Steric

56.2

14.2

Electrostatic

43.8

28.8

Hydrophobic

--

23.3

Acceptor

--

18.6

Donor

--

14.8

PLS result summary for model derived from non cyclic urea molecules

r2looa

0.595

0.568

Number of non-cyclic urea molecules in training set

25

25

Number of cyclic urea molecules in test set

60

60

ONCb

10

10

SEEc

0.048

0.076

r2 d

0.997

0.998

Fratioe

225.814

895.159

r2predf

0.42

0.41

Fraction of field contributions

Steric

57.8

14.9

Electrostatic

42.2

21.7

Hydrophobic

--

17.7

Acceptor

--

20.2

Donor

--

25.5

a correlation coefficient from leave one out method

b optimum number of components

c standard error of estimate

d conventional regression coefficient

e Fisher test value

f predictive r2 on test set using equation (r2 = (SD-PRESS)/SD) where SD is the sum of the squared deviations between the biological activities of the test molecules and the mean of training set. PRESS is the sum of the squared deviation between the observed and the predicted activities of the test set

Based on this information 25 non-cyclic urea molecules were taken as a test set to check the external predictive ability of these models on diverse set of molecules (structure are given in Fig. 3). This gave remarkable outcome with rpred2 of 0.61 and 0.53 for CoMFA and CoMSIA respectively. The results invariably show that this method is useful for performing 3D QSAR analysis on molecules having different structural motifs. The experimental and predicted activity is given in Table 4. Further, these diverse structures were used to generate a QSAR model, that gave statistical data of rloo2 0.595 for CoMFA and 0.568 for the CoMSIA models respectively, gave an r2 of 0.997 and 0.998 for CoMFA and CoMSIA, respectively. The graph for the experimental and predicted pKi values for training set, test set and external data set of non-cyclic urea molecules are shown in Fig. 4. The 60 cyclic urea molecule of the training set was used to check the predictive ability of this model; it showed an acceptable but not very enthusiastic result of 0.42 and 0.41 for CoMFA and CoMSIA respectively. The probable reason for this result could be the smaller training set of 25 non cyclic urea molecules and larger test set of 60 cyclic urea molecules. By using the QSAR ANALYSIS LIST command the intercept for the PLS equations were obtained, the command does not provide the regression equation, hence a method followed by Wheelock and Nakagawa et al. [43] was used to obtain the QSAR equation where the intercept was used. The equations derived for the CoMFA and CoMSIA analysis is provided below (Eqs. 1, 2, 3, 4) where n is number of molecules involved in the model generation and r is the correlation coefficient for the equations obtained.
  • PLS derived from cyclic urea molecules
    $$ \begin{array}{*{20}{c}} {{\text{pKi}} = \left[ {\text{CoMFA terms}} \right] + {5}.{118}} \hfill \\ {{\text{n}} = {6}0,\;{\text{r}} = 0.{991}} \hfill \\ \end{array}, $$
    (1)
    $$ \begin{array}{*{20}{c}} {{\text{pKi}} = \left[ {\text{CoMSIA terms}} \right] + {6}.{214}} \hfill \\ {{\text{n}} = {6}0,\;{\text{r}} = 0.{992}} \hfill \\ \end{array}. $$
    (2)
  • PLS derived from non-cyclic urea molecules
    $$ \begin{array}{*{20}{c}} {{\text{pKi}} = \left[ {{\text{CoMFA}}\;{\text{terms}}} \right] + {5}.{193}} \hfill \\ {{\text{n}} = {25},\;{\text{r}} = 0.{965}} \hfill \\ \end{array}, $$
    (3)
    $$ \begin{array}{*{20}{c}} {{\text{pKi}} = \left[ {{\text{CoMSIA}}\;{\text{terms}}} \right] + {6}.{192}} \hfill \\ {{\text{n}} = {25},\;{\text{r}} = 0.{936}} \hfill \\ \end{array}. $$
    (4)
https://static-content.springer.com/image/art%3A10.1007%2Fs00894-011-1048-x/MediaObjects/894_2011_1048_Fig3_HTML.gif
Fig. 3

Structure of non-cyclic urea molecule

https://static-content.springer.com/image/art%3A10.1007%2Fs00894-011-1048-x/MediaObjects/894_2011_1048_Fig4_HTML.gif
Fig. 4

(a) and (b) Scatter plot of experimental vs predicted pKi values for cyclic urea molecules using (PLS) derived from cyclic urea set (test set is represented in triangles), (c) and (d) Scatter plot of experimental vs predicted pKi values for non-cyclic urea molecules using (PLS) derived from cyclic urea set, (e) and (f) Scatter plot of experimental vs predicted pKi values for non-cyclic urea molecules using PLS derived from non-cyclic urea set

Table 4

Experimental and predicted pKi values of non-cyclic urea molecules

Mol.

Experimental pKi

Predicted pKi CoMFA cyclic urea PLS

Predicted pKi CoMSIA cyclic urea PLS

Predicted pKi CoMFA non cyclic urea PLS

Predicted pKi CoMSIA non cyclic urea PLS

NCU 1

6

5.648

6.972

5.993

5.958

NCU 2

5.44

7.685

8.186

5.425

5.462

NCU 3

6.24

6.353

7.199

6.323

6.459

NCU 4

6.77

6.267

7.194

6.683

6.583

NCU 5

7.55

8.972

8.501

7.505

7.542

NCU 6

5.47

7.145

8.282

5.486

5.465

NCU 7

6.4

7.484

9.042

6.405

6.375

NCU 8

7.36

7.021

7.216

7.381

7.346

NCU 9

7.42

7.915

7.699

7.748

7.386

NCU 10

5.6

7.797

8.116

5.614

5.599

NCU 11

8.52

8.217

8.015

8.512

8.506

NCU 12

7.72

8.691

8.183

7.722

7.73

NCU 13

9.31

7.614

9.062

9.305

9.3

NCU 14

8.7

7.995

8.306

8.66

8.716

NCU 15

9.66

9.081

9.216

9.713

9.629

NCU 16

11.3

9.129

10.062

11.3

11.331

NCU 17

8.92

8.616

8.908

8.002

7.085

NCU 18

7.89

7.9

8.668

7.941

7.803

NCU 19

8.51

8.531

8.861

8.141

8.315

NCU 20

8.52

9.042

9.716

8.528

8.523

NCU 21

9

10.257

8.634

9.002

8.994

NCU 22

7.77

7.62

7.393

8.148

7.617

NCU 23

8

7.373

8.005

8.07

8.241

NCU 24

9.7

8.418

9.447

8.315

7.861

NCU 25

7.82

7.829

8.336

7.01

7.383

The contour generated from the above two QSAR models for cyclic and non-cyclic urea molecules using docking and clustering as a prerequisite gave results that invariably reveal that this method is useful for performing 3D QSAR analysis on molecules having different structural motifs.

The contour maps of CoMFA (electrostatic and steric) and CoMSIA (electrostatic, steric, hydrophobic, donor and acceptor) are represented by color codes. The contour maps of CoMFA denote the region in the space where the aligned molecules would favorably or unfavorably interact with the receptor while the CoMSIA contour maps denote those areas within the specified region where the presence of a group with a particular physicochemical activity binds to the receptor. The CoMFA/CoMSIA results were graphically interpreted by field contribution maps using the ‘STDEV * COEFF’ field type.

Steric and electrostatic contour maps

To visualize the information content of the derived 3D QSAR models, CoMFA contours maps were generated to rationalize the regions of 3D space around the cyclic urea and non-cyclic urea molecules, where changes in the steric and electrostatic fields would influence the increase or decrease in inhibitory activity. All of the contours represented the default 80 and 20% level contributions for favored and disfavored regions, respectively. The CoMFA steric and electrostatic contour maps are shown in Figs. 5 and 6 respectively.
https://static-content.springer.com/image/art%3A10.1007%2Fs00894-011-1048-x/MediaObjects/894_2011_1048_Fig5_HTML.gif
Fig. 5

CoMFA steric standard deviation (S.D.* coefficient) contour maps illustrating steric features in combination with compound (a) 52 and (b) NCU16. Green contours show favorable bulky group substitution at that point while yellow regions show disfavorable bulky group for activity

https://static-content.springer.com/image/art%3A10.1007%2Fs00894-011-1048-x/MediaObjects/894_2011_1048_Fig6_HTML.gif
Fig. 6

CoMFA electrostatic standard deviation (S.D.* coefficient) contour maps illustrating electrostatic features in combination with compound (a) 52 and (b) NCU16. Red contours indicate negative charge favoring activity, whereas blue contours indicate positive charge favoring activity

The steric field is characterized by green and yellow contours, in which green contours indicate the region where bulkier group would be favorable, while the yellow represents region were bulkier group would decrease the activity. The most potent cyclic urea analogue from the series, compound 52 was embedded in the map (Fig. 5a), and compound NCU16 (Fig. 5b) was embedded for non-cyclic urea molecule, to demonstrate their affinity for the steric regions of inhibitors. The steric contours for both cyclic and non-cyclic urea molecules showed same regions, having a large green contour at the R2 and R1 positions of the cyclic urea molecule suggesting an increase in the bulkiness would increase the activity of the molecules, similar kind of contour was obtained for non-cyclic urea molecule where the green contour was envisaged at the same position as in cyclic urea. Substitution on the phenyl ring at R1 would decrease the activity in cyclic urea molecules, a similar kind of contour was observed for non-cyclic urea where the yellow contours is observed all over the molecule expect at R2 position showing that the two contours for cyclic and non-cyclic urea molecules are similar. This depicts that in cyclic urea molecule substitution on the benzyl ring at R2 position will increase the activity, in non-cyclic urea molecules increase in bulkiness at R2 position for triazalinones, diaza sulfoxide, substitution at 3rd and 7th position of coumarin moiety and 5,6 position of pyranones will increase the enzyme inhibitory activity.

Figure 6a, b shows the CoMFA electrostatic contour maps for cyclic urea and non-cyclic urea molecules respectively. The blue and red contours depict the positions where positively charged groups and negatively charged groups would be beneficial for inhibitory activity. In both the contours a red region is seen near the carbonyl and hydroxyl groups of the cyclic urea scaffold and non-cyclic urea molecules, suggesting an electron withdrawing group will be preferred at this position.

The huge blue region around the two amino groups suggests, electron donating group will be beneficial. The red contour region around the side chain atoms of R1 nearer to the carbonyl group also suggests that an electro negative group will be a potential substituent. In non-cyclic urea molecule there is a disparity in the contour’s suggesting a flip in the molecules toward the right where the contours are more concentrated. This suggests that carbonyl group and the hydroxyl group are the pivotal substituents for the increase of inhibitory activity. The CoMSIA steric and electrostatic field contour maps were almost similar to the corresponding CoMFA contour maps

Hydrophobic contour maps

The hydrophobic fields are presented in Fig. 7, yellow and white contours highlight areas where hydrophobic and hydrophilic groups are preferred respectively. The hydrophobic contour shows the presence of large yellow region near benzyl ring of R2 substitution in cyclic urea scaffold for the compound 52, yellow contour are scattered over the benzyl rings of R2 substituents on the non-cyclic urea molecule NCU16, indicates that the groups with hydrophobic characters are preferred at these positions. White hydrophilic favored contour is observed on the amide group of the R1 position of the compound 52, suggesting group having hydrogen bond forming ability at these positions will be beneficial for protein binding, which is evident form the docking studies as shown in Fig. 8 where the amide group is interacting with Asp 30 and Gly 48 of protein active site. In non-cyclic urea molecules the hydrophilic contour is observed on hydroxyl and methoxy groups of the template NCU16 at R1 position, this indicates similarity in contours obtained from the two different QSAR model generated from a different set of molecules with diverse scaffold.
https://static-content.springer.com/image/art%3A10.1007%2Fs00894-011-1048-x/MediaObjects/894_2011_1048_Fig7_HTML.gif
Fig. 7

CoMSIA hydrophobic standard deviation (S.D.* coefficient) contour maps illustrating hydrophobic features in combination with compound (a) 52 and (b) NCU16. Yellow contours indicate hydrophobic group favored region , white contours indicate hydrophilic group favored region

https://static-content.springer.com/image/art%3A10.1007%2Fs00894-011-1048-x/MediaObjects/894_2011_1048_Fig8_HTML.gif
Fig. 8

Docked pose of molecules in the protein active site showing hydrogen bond interaction with the active site amino acids (a) compound 52, (b) compound NCU16 and (c) compound NCU 8

Acceptor and donor contour maps

The hydrogen bond acceptor and donor field contour maps of CoMSIA is shown in Fig. 9a-d using the same templates of cyclic and non-cyclic urea. The magenta and red contours represent favorable and unfavorable hydrogen bond acceptor groups respectively, cyan and purple contours represent favorable and disfavorable hydrogen bond donor groups respectively.
https://static-content.springer.com/image/art%3A10.1007%2Fs00894-011-1048-x/MediaObjects/894_2011_1048_Fig9_HTML.gif
Fig. 9

CoMSIA H-bond acceptor standard deviation (S.D.* coefficient) contour maps illustrating H-bond acceptor features in combination with compound (a) 52 and (b) NCU16. Magenta contours indicate H-bond acceptor group favored region, red contours indicate H-bond acceptor group disfavored region. CoMSIA H-bond donor standard deviation (S.D.* coefficient) contour maps illustrating H-bond donor features in combination with compound (c) 52 and (d) NCU16. Cyan contours indicate H-bond donor group favored region, purple contours indicate H-bond donor group disfavored region

The magenta contour near the carbonyl of the amide group on substituent at R1 position reveals that hydrogen bond acceptor group may increase the inhibitory activity. This is in agreement with the donor contour where a small purple region for hydrogen bond donor disfavored is seen at the same position. This is in accordance with the docking results as shown in Fig. 8 where the carbonyl of the amide group on substituent at R1 position in compound 52 shows hydrogen bond interaction with Asp 30 in protein active site. There is a big acceptor disfavored red contour near NH of amide group and surrounding the carbonyl of the cyclic scaffold, this is similar to the hydrogen bond donor contour were a cyan contour is seen at the same region which is favored for hydrogen bond donor groups. The NH group at this position shows a hydrogen bond interaction with Gly 48. Hydrogen bond acceptor and donor contours for non-cyclic urea obtained are similar to the cyclic urea showing same contour regions. The contour map analysis indeed show that the derived 3D QSAR model from the two set of molecules having diverse structural motifs which were aligned based on multiple receptor conformation docking and clustering as a prerequisite have consensus.

The present reported approach of multiple receptor conformation docking and clustering for obtaining receptor based conformation of ligands for CoMFA and CoMSIA analysis, is applicable for ligands having specific protein targets with at least eight to ten experimental three dimensional structures, but not for the targets having less or none. For such cases a homology modeling can be performed and different receptor conformations can be obtained for docking.

Conclusions

A new approach of multiple receptor conformation docking and clustering was employed to obtain a flexible receptor based alignment of molecules having diverse motifs, which could not be aligned based on common substructure that is a prerequisite for CoMFA and CoMSIA analysis. The statistical results and contours obtained invariably show that the method employed is having an adequate accuracy, and provides valuable information for the structural requirements for improving inhibitory activity of the molecules. This advanced and novel approach is appropriate for receptor based alignment for molecules having varied structural motifs, recommending an increase in accuracy of 3D QSAR predictions for considering diverse scaffolds while screening and designing molecules for specific targets.

Acknowledgments

We gratefully acknowledge support for this research from Department of Science and Technology, New Delhi, India, University Grants Commission, New Delhi, India and Department of chemistry, Nizam College, Hyderabad, India. We are greatly thankful to Dr. G. N. Shastry, Scientist, Indian Institute of Chemical Technology for Sybyl 6.9 software and his useful suggestions. We also acknowledge Schrödinger Inc. for GLIDE software.

Supplementary material

894_2011_1048_MOESM1_ESM.txt (6 kb)
1.pdb(TXT 5 kb)
894_2011_1048_MOESM2_ESM.txt (6 kb)
2.pdb(TXT 6 kb)
894_2011_1048_MOESM3_ESM.txt (7 kb)
3.pdb(TXT 6 kb)
894_2011_1048_MOESM4_ESM.txt (8 kb)
4.pdb(TXT 7 kb)
894_2011_1048_MOESM5_ESM.txt (8 kb)
5.pdb(TXT 8 kb)
894_2011_1048_MOESM6_ESM.txt (9 kb)
6.pdb(TXT 8 kb)
894_2011_1048_MOESM7_ESM.txt (7 kb)
7.pdb(TXT 7 kb)
894_2011_1048_MOESM8_ESM.txt (8 kb)
8.pdb(TXT 7 kb)
894_2011_1048_MOESM9_ESM.txt (8 kb)
9.pdb(TXT 7 kb)
894_2011_1048_MOESM10_ESM.txt (8 kb)
10.pdb(TXT 8 kb)
894_2011_1048_MOESM11_ESM.txt (9 kb)
11.pdb(TXT 8 kb)
894_2011_1048_MOESM12_ESM.txt (9 kb)
12.pdb(TXT 9 kb)
894_2011_1048_MOESM13_ESM.txt (7 kb)
13.pdb(TXT 6 kb)
894_2011_1048_MOESM14_ESM.txt (7 kb)
14.pdb(TXT 7 kb)
894_2011_1048_MOESM15_ESM.txt (7 kb)
15.pdb(TXT 6 kb)
894_2011_1048_MOESM16_ESM.txt (7 kb)
16.pdb(TXT 7 kb)
894_2011_1048_MOESM17_ESM.txt (7 kb)
17.pdb(TXT 7 kb)
894_2011_1048_MOESM18_ESM.txt (8 kb)
18.pdb(TXT 7 kb)
894_2011_1048_MOESM19_ESM.txt (8 kb)
19.pdb(TXT 8 kb)
894_2011_1048_MOESM20_ESM.txt (8 kb)
20.pdb(TXT 7 kb)
894_2011_1048_MOESM21_ESM.txt (8 kb)
21.pdb(TXT 7 kb)
894_2011_1048_MOESM22_ESM.txt (8 kb)
22.pdb(TXT 7 kb)
894_2011_1048_MOESM23_ESM.txt (8 kb)
23.pdb(TXT 7 kb)
894_2011_1048_MOESM24_ESM.txt (9 kb)
24.pdb(TXT 9 kb)
894_2011_1048_MOESM25_ESM.txt (8 kb)
25.pdb(TXT 7 kb)
894_2011_1048_MOESM26_ESM.txt (8 kb)
26.pdb(TXT 7 kb)
894_2011_1048_MOESM27_ESM.txt (8 kb)
27.pdb(TXT 7 kb)
894_2011_1048_MOESM28_ESM.txt (8 kb)
28.pdb(TXT 7 kb)
894_2011_1048_MOESM29_ESM.txt (8 kb)
29.pdb(TXT 7 kb)
894_2011_1048_MOESM30_ESM.txt (8 kb)
30.pdb(TXT 7 kb)
894_2011_1048_MOESM31_ESM.txt (8 kb)
31.pdb(TXT 7 kb)
894_2011_1048_MOESM32_ESM.txt (9 kb)
32.pdb(TXT 8 kb)
894_2011_1048_MOESM33_ESM.txt (9 kb)
33.pdb(TXT 8 kb)
894_2011_1048_MOESM34_ESM.txt (9 kb)
34.pdb(TXT 8 kb)
894_2011_1048_MOESM35_ESM.txt (9 kb)
35.pdb(TXT 8 kb)
894_2011_1048_MOESM36_ESM.txt (9 kb)
36.pdb(TXT 8 kb)
894_2011_1048_MOESM37_ESM.txt (9 kb)
37.pdb(TXT 8 kb)
894_2011_1048_MOESM38_ESM.txt (8 kb)
38.pdb(TXT 8 kb)
894_2011_1048_MOESM39_ESM.txt (9 kb)
39.pdb(TXT 8 kb)
894_2011_1048_MOESM40_ESM.txt (9 kb)
40.pdb(TXT 9 kb)
894_2011_1048_MOESM41_ESM.txt (10 kb)
41.pdb(TXT 10 kb)
894_2011_1048_MOESM42_ESM.txt (11 kb)
42.pdb(TXT 10 kb)
894_2011_1048_MOESM43_ESM.txt (11 kb)
43.pdb(TXT 10 kb)
894_2011_1048_MOESM44_ESM.txt (11 kb)
44.pdb(TXT 11 kb)
894_2011_1048_MOESM45_ESM.txt (11 kb)
45.pdb(TXT 10 kb)
894_2011_1048_MOESM46_ESM.txt (10 kb)
46.pdb(TXT 10 kb)
894_2011_1048_MOESM47_ESM.txt (9 kb)
47.pdb(TXT 9 kb)
894_2011_1048_MOESM48_ESM.txt (11 kb)
48.pdb(TXT 11 kb)
894_2011_1048_MOESM49_ESM.txt (11 kb)
49.pdb(TXT 10 kb)
894_2011_1048_MOESM50_ESM.txt (11 kb)
50.pdb(TXT 10 kb)
894_2011_1048_MOESM51_ESM.txt (11 kb)
51.pdb(TXT 10 kb)
894_2011_1048_MOESM52_ESM.txt (11 kb)
52.pdb(TXT 11 kb)
894_2011_1048_MOESM53_ESM.txt (11 kb)
53.pdb(TXT 11 kb)
894_2011_1048_MOESM54_ESM.txt (11 kb)
54.pdb(TXT 10 kb)
894_2011_1048_MOESM55_ESM.txt (11 kb)
55.pdb(TXT 10 kb)
894_2011_1048_MOESM56_ESM.txt (11 kb)
56.pdb(TXT 10 kb)
894_2011_1048_MOESM57_ESM.txt (11 kb)
57.pdb(TXT 11 kb)
894_2011_1048_MOESM58_ESM.txt (11 kb)
58.pdb(TXT 10 kb)
894_2011_1048_MOESM59_ESM.txt (11 kb)
59.pdb(TXT 10 kb)
894_2011_1048_MOESM60_ESM.txt (10 kb)
60.pdb(TXT 10 kb)
894_2011_1048_MOESM61_ESM.txt (11 kb)
61.pdb(TXT 10 kb)
894_2011_1048_MOESM62_ESM.txt (11 kb)
62.pdb(TXT 10 kb)
894_2011_1048_MOESM63_ESM.txt (12 kb)
63.pdb(TXT 11 kb)
894_2011_1048_MOESM64_ESM.txt (11 kb)
64.pdb(TXT 10 kb)
894_2011_1048_MOESM65_ESM.txt (10 kb)
65.pdb(TXT 9 kb)
894_2011_1048_MOESM66_ESM.txt (10 kb)
66.pdb(TXT 9 kb)
894_2011_1048_MOESM67_ESM.txt (10 kb)
67.pdb(TXT 10 kb)
894_2011_1048_MOESM68_ESM.txt (11 kb)
68.pdb(TXT 10 kb)
894_2011_1048_MOESM69_ESM.txt (11 kb)
69.pdb(TXT 11 kb)
894_2011_1048_MOESM70_ESM.txt (11 kb)
70.pdb(TXT 10 kb)
894_2011_1048_MOESM71_ESM.txt (10 kb)
71.pdb(TXT 9 kb)
894_2011_1048_MOESM72_ESM.txt (9 kb)
72.pdb(TXT 8 kb)
894_2011_1048_MOESM73_ESM.txt (9 kb)
73.pdb(TXT 9 kb)
894_2011_1048_MOESM74_ESM.txt (10 kb)
74.pdb(TXT 10 kb)
894_2011_1048_MOESM75_ESM.txt (6 kb)
75.pdb(TXT 5 kb)
894_2011_1048_MOESM76_ESM.txt (10 kb)
76.pdb(TXT 9 kb)
894_2011_1048_MOESM77_ESM.txt (9 kb)
77.pdb(TXT 8 kb)
894_2011_1048_MOESM78_ESM.txt (8 kb)
78.pdb(TXT 7 kb)
894_2011_1048_MOESM79_ESM.txt (7 kb)
79.pdb(TXT 7 kb)
894_2011_1048_MOESM80_ESM.txt (9 kb)
80.pdb(TXT 8 kb)
894_2011_1048_MOESM81_ESM.txt (9 kb)
81.pdb(TXT 8 kb)
894_2011_1048_MOESM82_ESM.txt (9 kb)
82.pdb(TXT 9 kb)
894_2011_1048_MOESM83_ESM.txt (9 kb)
83.pdb(TXT 8 kb)
894_2011_1048_MOESM84_ESM.txt (9 kb)
84.pdb(TXT 9 kb)
894_2011_1048_MOESM85_ESM.txt (9 kb)
85.pdb(TXT 8 kb)
894_2011_1048_MOESM86_ESM.txt (8 kb)
86.pdb(TXT 8 kb)
894_2011_1048_MOESM87_ESM.txt (9 kb)
87.pdb(TXT 8 kb)
894_2011_1048_MOESM88_ESM.txt (9 kb)
88.pdb(TXT 8 kb)
894_2011_1048_MOESM89_ESM.txt (8 kb)
89.pdb(TXT 8 kb)
894_2011_1048_MOESM90_ESM.txt (8 kb)
90.pdb(TXT 8 kb)
894_2011_1048_MOESM91_ESM.txt (10 kb)
91.pdb(TXT 9 kb)
894_2011_1048_MOESM92_ESM.txt (8 kb)
92.pdb(TXT 8 kb)
894_2011_1048_MOESM93_ESM.txt (8 kb)
93.pdb(TXT 7 kb)
894_2011_1048_MOESM94_ESM.txt (9 kb)
94.pdb(TXT 8 kb)
894_2011_1048_MOESM95_ESM.txt (10 kb)
95.pdb(TXT 9 kb)
894_2011_1048_MOESM96_ESM.txt (11 kb)
96.pdb(TXT 10 kb)
894_2011_1048_MOESM97_ESM.txt (7 kb)
97.pdb(TXT 6 kb)
894_2011_1048_MOESM98_ESM.txt (6 kb)
98.pdb(TXT 6 kb)
894_2011_1048_MOESM99_ESM.txt (6 kb)
99.pdb(TXT 6 kb)
894_2011_1048_MOESM100_ESM.txt (7 kb)
100.pdb(TXT 7 kb)
894_2011_1048_MOESM101_ESM.txt (8 kb)
101.pdb(TXT 7 kb)
894_2011_1048_MOESM102_ESM.txt (7 kb)
102.pdb(TXT 7 kb)
894_2011_1048_MOESM103_ESM.txt (8 kb)
103.pdb(TXT 7 kb)
894_2011_1048_MOESM104_ESM.txt (9 kb)
104.pdb(TXT 8 kb)
894_2011_1048_MOESM105_ESM.txt (7 kb)
105.pdb(TXT 6 kb)
894_2011_1048_MOESM106_ESM.txt (7 kb)
106.pdb(TXT 7 kb)
894_2011_1048_MOESM107_ESM.txt (9 kb)
107.pdb(TXT 9 kb)
894_2011_1048_MOESM108_ESM.txt (8 kb)
108.pdb(TXT 8 kb)
894_2011_1048_MOESM109_ESM.txt (9 kb)
109.pdb(TXT 8 kb)
894_2011_1048_MOESM110_ESM.txt (8 kb)
110.pdb(TXT 8 kb)
894_2011_1048_MOESM111_ESM.txt (8 kb)
111.pdb(TXT 7 kb)
894_2011_1048_MOESM112_ESM.txt (9 kb)
112.pdb(TXT 8 kb)
894_2011_1048_MOESM113_ESM.txt (8 kb)
113.pdb(TXT 7 kb)
894_2011_1048_MOESM114_ESM.txt (10 kb)
114.pdb(TXT 9 kb)
894_2011_1048_MOESM115_ESM.txt (10 kb)
115.pdb(TXT 10 kb)
894_2011_1048_MOESM116_ESM.txt (10 kb)
116.pdb(TXT 9 kb)
894_2011_1048_MOESM117_ESM.txt (10 kb)
117.pdb(TXT 10 kb)
894_2011_1048_MOESM118_ESM.txt (12 kb)
118.pdb(TXT 12 kb)
894_2011_1048_MOESM119_ESM.txt (10 kb)
119.pdb(TXT 9 kb)
894_2011_1048_MOESM120_ESM.txt (10 kb)
120.pdb(TXT 10 kb)
894_2011_1048_MOESM121_ESM.txt (4 kb)
ncu1.pdb(TXT 4 kb)
894_2011_1048_MOESM122_ESM.txt (7 kb)
ncu2.pdb(TXT 7 kb)
894_2011_1048_MOESM123_ESM.txt (5 kb)
ncu3.pdb(TXT 4 kb)
894_2011_1048_MOESM124_ESM.txt (5 kb)
ncu4.pdb(TXT 4 kb)
894_2011_1048_MOESM125_ESM.txt (8 kb)
ncu5.pdb(TXT 7 kb)
894_2011_1048_MOESM126_ESM.txt (7 kb)
ncu6.pdb(TXT 6 kb)
894_2011_1048_MOESM127_ESM.txt (8 kb)
ncu7.pdb(TXT 7 kb)
894_2011_1048_MOESM128_ESM.txt (8 kb)
ncu8.pdb(TXT 7 kb)
894_2011_1048_MOESM129_ESM.txt (6 kb)
ncu9.pdb(TXT 5 kb)
894_2011_1048_MOESM130_ESM.txt (8 kb)
ncu10.pdb(TXT 7 kb)
894_2011_1048_MOESM131_ESM.txt (9 kb)
ncu11.pdb(TXT 8 kb)
894_2011_1048_MOESM132_ESM.txt (8 kb)
ncu12.pdb(TXT 8 kb)
894_2011_1048_MOESM133_ESM.txt (8 kb)
ncu13.pdb(TXT 8 kb)
894_2011_1048_MOESM134_ESM.txt (8 kb)
ncu14.pdb(TXT 7 kb)
894_2011_1048_MOESM135_ESM.txt (9 kb)
ncu15.pdb(TXT 8 kb)
894_2011_1048_MOESM136_ESM.txt (9 kb)
ncu16.pdb(TXT 8 kb)
894_2011_1048_MOESM137_ESM.txt (9 kb)
ncu17.pdb(TXT 8 kb)
894_2011_1048_MOESM138_ESM.txt (8 kb)
ncu18.pdb(TXT 7 kb)
894_2011_1048_MOESM139_ESM.txt (9 kb)
ncu19.pdb(TXT 8 kb)
894_2011_1048_MOESM140_ESM.txt (8 kb)
ncu20.pdb(TXT 8 kb)
894_2011_1048_MOESM141_ESM.txt (9 kb)
ncu21.pdb(TXT 8 kb)
894_2011_1048_MOESM142_ESM.txt (8 kb)
ncu22.pdb(TXT 8 kb)
894_2011_1048_MOESM143_ESM.txt (8 kb)
ncu23.pdb(TXT 8 kb)
894_2011_1048_MOESM144_ESM.txt (9 kb)
ncu24.pdb(TXT 8 kb)
894_2011_1048_MOESM145_ESM.txt (6 kb)
ncu25.pdb(TXT 5 kb)

Copyright information

© Springer-Verlag 2011