Translate this page into:
3D QSAR and HQSAR analysis of protein kinase B (PKB/Akt) inhibitors using various alignment methods
⁎Corresponding author. Tel.: +91 9624931060; fax: +91 2717 241916. vicky_1744@yahoo.com (Vivek K. Vyas) vivekvyas@nirmauni.ac.in (Vivek K. Vyas)
-
Received: ,
Accepted: ,
This article was originally published by Elsevier and was migrated to Scientific Scholar after the change of Publisher.
Peer review under responsibility of King Saud University.
Abstract
Protein kinase B (PKB/Akt) regulates all aspects of cell growth, differentiation, and division. PKB/Akt has recently garnered a great deal of attention as a promising molecular target for cancer therapy due to its involvement in the development of several human cancers. In this study a diverse set of 56 Akt1 inhibitors were aligned by three different methods (pharmacophore-, docking-based and rigid body alignment) for CoMFA, CoMSIA and HQSAR analysis. The best QSAR models were obtained using rigid body alignment (Distill). CoMFA and CoMSIA models were found statistically significant with leave-one-out correlation coefficient (q2) of 0.627 and 0.598, respectively, cross validated coefficient ( ) of 0.644 and 0.563, respectively, and conventional coefficient (r2) of 0.867 and 0.865, respectively. QSAR models were validated by a test set of 9 compounds giving satisfactory predicted correlation coefficient ( ) of 0.603 and 0.613 for CoMFA and CoMSIA models, respectively. Leave-one-out correlation value (q2) of 0.687, of 0.742 and r2 of 0.868 were obtained for HQSAR analysis and found satisfactory. This study provides valuable clues to design new compounds against PKB/Akt.
Keywords
PKB/Akt inhibitors
CoMFA
CoMSIA
HQSAR
Tripos
1 Introduction
Protein kinase B (PKB) also known as Akt, is a serine/threonine kinase that belongs to the AGC family of kinases (Vivanco and Sawyers, 2002). Serine/threonine protein kinase is a key enzyme in the phosphatidylinositol-3-kinase (PI3-K) cascade with a crucial role in the balance of cellular survival and apoptosis (Vyas et al., 2013). In mammals protein kinase B is comprised of three highly homologous isoforms, namely PKBα (Akt1), PKBβ (Akt2), and PKBγ (Akt3) with a common structure made up of three domains: a PH domain at the N terminus binds phosphatidylinositol-3,4,5 triphosphate (PI(3,4,5)P3) and, is essential in the activation of the enzyme, a catalytic kinase domain contains a classical kinase ATP-binding site and a hydrophobic motif (HM) at the C terminus (Barile et al., 2010). The activation of PKB by PI3K is antagonized by the tumor suppressor phosphatase and tensin homolog (PTEN). The increased PKB activity observed in most human tumors could be the result of (a) increased PKB expression, (b) increased PI3K activity, or (c) decreased PTEN activity (Scheid and Woodgett, 2003). PKBs are over-expressed in a variety of human tumors, and at the genomic level, PKBα and PKBβ have been shown to be amplified in a number of cancer types. PKB is known to phosphorylate over 20 substrates, many of which are involved in the induction of apoptosis and in the arrest of cell proliferation (Sarbassov et al., 2005). Three main strategies can be applied to the development of novel PKB inhibitors, targeting toward pleckstrin homology (PH) domain (Mahadevan et al., 2008), ATP-binding kinase (Lindsley et al., 2007), and hinge-region domain (Lindsley et al., 2005). Enhanced PKB/Akt activity is the hallmark of several aggressive malignancies, as a result, PKB/Akt has been considered as the most promising target for the development of new anticancer therapeutics (Collins, 2009). The present study was undertaken to explore key structural requirements of different chemical scaffolds as PKB/Akt inhibitors by utilizing comparative molecular field analysis (CoMFA), comparative molecular similarity indices analysis (CoMSIA) and hologram quantitative structure activity relationship (HQSAR) analysis with three different (pharmacophore-, docking-based and rigid body) alignment methods. CoMFA is a versatile and powerful tool in rational drug design. CoMFA calculates steric and electrostatic fields surrounding the molecules and correlating the differences in these fields to inhibitory activity (Vyas and Ghate, 2012). In CoMSIA analysis, similarity indices are calculated at regularly placed grid points for aligned molecules. CoMSIA calculates other molecular descriptors like hydrophobic fields, hydrogen-bond donor and acceptor fields (Singh et al., 2017). Contour maps of CoMFA and CoMSIA fields describe the ‘favorable’ or ‘unfavorable’ contribution of a region of interest surrounding the ligands to the target property. The purpose of the HQSAR study is to explore individual atomic contributions to molecular bioactivity with visual display of active centers in a compound. Furthermore, HQSAR results can be also somehow used as control to assess our CoMFA and CoMSIA results. HQSAR technique does not require molecular superposition (Sridhar et al., 2011). QSAR models generated in this study would provide some new useful information for designing new PKB inhibitors.
2 Materials and methods
2.1 Dataset
Available data from these literatures were used for the QSAR study consisting of 56 compounds, obtained from same laboratory consisting of pyridine–pyrazolopyridine-based inhibitors (12 compounds) (Zhu et al., 2007) and trans-3,4′-bispyridinylethylenes (44 compounds) (Li et al., 2006). Akt1 IC50 values were measured against Akt1 with ATP concentration of 10 μM employed similar experimental conditions and procedures. The IC50 values (nM) were converted to pIC50 and subsequently used as a dependent variable for 3D QSAR studies (Table 1).
Compound
R1
R2
Akt1 IC50 (nM)a
bpIC50
1
3-Indole
1.5
8.824
2
3-Indole
1.0
9.000
3
3-Indole
1.8
8.745
4
3-Indole
953
6.021
5
3-Indole
51
7.292
6
3-Indole
7.1
8.149
7⁎
3-Indole
0.6
9.222
8
3-Indole
0.34
9.469
9
Ph
223
6.652
10
Ph
25
7.602
11
Ph
59
7.229
12
Ph
104
6.983
Compound
R1
R2
R3
Akt1 IC50 (nM)a
bpIC50
13
Cl
H
5290
5.277
14
H
H
5080
5.294
15
H
H
14710
4.832
16
H
H
19320
4.714
17
H
H
10870
4.964
18
H
H
23910
4.621
19
H
H
690
6.161
20
H
H
6760
5.170
21
H
H
8020
5.096
22
H
H
4040
5.394
23
H
H
875
6.058
24
H
H
320
6.495
25
H
H
324
6.489
26⁎
H
H
146
6.836
27
H
H
142
6.848
28
H
H
14
7.854
29
H
H
360
6.444
30
Cl
H
23
7.638
31
H
Cl
273
6.564
32⁎
H
H
500
6.301
33⁎
H
H
1100
5.959
34⁎
H
H
24010
4.620
Compound
Ar1
A
R4
Akt1 IC50 (nM)a
bpIC50
35
CH2
198
6.703
36
NH
19
7.721
37⁎
NH
79
7.102
38
NH
9.7
8.013
39
NH
278
6.556
40
NH
3200
5.495
41
NH
668
6.175
42
O
32250
4.491
43
Bond
284
6.547
44
Bond
760
6.119
45
O
23270
4.633
46
O
43010
4.366
Compound
X
Akt1 IC50 (nM)a
bpIC50
47⁎
MeO
2670
5.573
48
6450
5.190
49
6350
5.197
50
3110
5.507
51
–NH2
121
6.917
52
MeNH–
176
6.754
53
EtNH–
163
6.788
54⁎
69
7.161
55⁎
147
6.833
56
3340
5.476
2.2 Selection of training and test sets
The total set of 56 inhibitors was divided into the training set (47 compounds) for generating the QSAR model and a test set (9 compounds) for validating the quality of the models. Selection of the training and test set molecules was done by considering the fact that test set molecules represent a range of inhibitory activity similar to that of the training set. Thus, the test set was the true representative of the training set. This was achieved by arbitrarily setting aside 9 compounds as a test set with regularly distributed biological data.
2.3 Computational details
QSAR modeling, calculations and visualizations for CoMFA, CoMSIA and HQSAR analysis were performed using SYBYL X 1.2 software from Tripos Inc., St. Louis, MO, USA. Compound 8 was selected as the template molecule because of its highest inhibitory activity. The structures of all other compounds were constructed from the template molecule using “SKETCH” function in SYBYL and partial atomic charges were calculated by the Gasteiger–Huckel method and energy minimization was performed using the Tripos force field (Gasteiger and Marsili, 1980) with a distance-dependent dielectric and Powell conjugate gradient algorithm. The minimum gradient difference of 0.05 kcal/mol Å was set as a convergence criterion (Clark et al., 1989).
2.4 Alignment
QSAR models are often sensitive to a particular alignment scheme. Generally, the results depend upon the alignment method of molecules. The comparisons of different alignment techniques were reported (Roy et al., 2008) in the literature for CoMFA and CoMSIA analysis. Therefore, three different alignment techniques were compared carefully in this work, in order to find the most efficient one for the present study.
2.4.1 Distill (align 1)
Rigid body alignment of molecules in a Mol2 database was performed using maximum common substructure (MCS) defined by Distill. Compound 8 was used as a template and all other compounds were aligned on the basis of the common structure. MCS represents a common core of all the structures used for the alignment. Distill generates MCS on the basic of group of connected atoms common to a set of structures used for the alignment. A rigid alignment attempts to align molecules in a database to a template molecule on a common backbone or core (MCS). This core will typically have been produced by Distill. The minimum atom count in MCS fragments can be as small as 3. For the alignment of the molecules using Distill, first the core is looked for in all the molecules, if the core may be found more than once, or there may be more than one mapping of the core atoms to the molecule atoms. In this case a single mapping is chosen. Finally, all the molecule are fit to the template using the best mapping of the core to the molecules and the database is updated with the molecule’s new orientation. Alignment of training and test set compounds using the Distill module is shown in Fig. 1A.
Alignment of training and test set compounds (A) rigid body alignment using Distill, (B) pharmacophore-based (C) docking-based.
2.4.2 Pharmacophore-based alignment (align 2)
All the compounds (training and test sets) were selected to generate the pharmacophore model using DISCOtech. All the compounds were aligned on some common features depending upon the position rotation and conformation. Generated pharmacophore model contains 1 donor site, 1 acceptor atom and 2 hydrophobic regions, which were then used for the alignment of the compounds in QSAR studies (Fig. 1B).
2.4.3 Docking-based alignment (align 3)
The active conformation of all the compounds was achieved by molecular docking studies. Docking experiments were performed using Surflex–Dock module of SYBYL X 1.3. The X-ray crystallographic structure of PKB (PDB ID: 2ZDO) (Davies et al., 2007) solved at 1.8 Å resolution was retrieved from the PDB databank. Each inhibitor was docked into the ALK using the flexible docking module implemented in Surflex–Dock (Jain, 1996). Active conformation was selected from the binding orientation in the active site of ALK and evaluated by consideration of binding free energy scores (Surflex–Dock score). Since for all compounds the best-docked geometries (active conformation) were in agreement with the crystallographic data of the PKB complex (and thus already aligned) (Fig. 1C), they were directly submitted to QSAR studies.
These three different alignment methods of molecules yielded very good statistical results, but rigid body alignment of molecules by Distill (align 1) gave us the best results (Table 2) with a significant statistical value of q2 and
as compared to align 2 and 3, so further QSAR studies was carried out using align 1 (Distill). N is the optimal number of components (PLS components); q2 is the leave-one-out (LOO) validation coefficient;
is cross-validation coefficient;
is the non-cross-validation coefficient;
is the predictive correlation coefficient; SEE is the standard error of estimation; F is the F-test value;
is mean r2 of bootstrapping analysis (100 runs).
Statistical parameters
Align 1 (rigid body)
Align 2 (pharmacophore-based)
Align 3 (docking-based)
CoMFA
CoMSIA
HQSAR
CoMFA
CoMSIA
HQSAR
CoMFA
CoMSIA
HQSAR
q2
0.627
0.598
0.687
0.364
0.208
0.451
0.303
0.415
0.446
0.867
0.865
0.868
0.880
0.847
0.537
0.752
0.806
0.732
0.644
0.563
0.210
0.288
0.268
0.206
0.945
0.932
0.915
0.918
0.861
0.913
N
6
6
3
6
4
3
3
2
2
F
1511
1258
352
281
46
147
SEE
0.067
0.074
0.048
0.195
0.304
0.238
0.665
0.410
0.318
0.603
0.613
0.742
0.414
0.511
0.655
0.232
0.423
0.623
Probability of
0.0
0.0
0.0
0.0
0.0
0.0
Field contribution
Steric
0.544
0.154
0.551
0.148
0.391
0.105
Electrostatic
0.456
0.183
0.449
0.176
0.609
0.219
Hydrophobic
0.236
0.198
0.202
H-bond donor
0.235
0.202
0.237
H-bond acceptor
0.193
0.276
0.237
2.5 CoMFA model
In CoMFA analysis, steric and electrostatic potential energies were calculated using the Tripos force field with a probe atom having a van der waals radius of sp3-hybridized carbon and a + 1 charge to generate steric (Lennard–Jones 6–12 potential) field energies and electrostatic (Coulombic potential) fields with a distance-dependent dielectric at each lattice point. A lattice with 2 Å grid spacing extending at least 4 Å in each direction beyond the aligned molecules was used. The steric and electrostatic energy values were truncated at 30.0 kcal/mol. In order to reduce noise and improve efficiency, column filtering (minimum sigma) was set to 2.0 kcal/mol.
2.6 CoMSIA model
The CoMSIA similarity index descriptors were calculated using a dummy sp3-hybridized carbon with +1 charge. The same lattice box used in CoMFA calculations was also applied to CoMSIA calculations with a grid spacing of 2 Å with a radius of 1.0 Å as implemented in SYBYL. Similarity indices were calculated between a probe and each atom of the molecules based on a Gaussian distance function. CoMSIA not only computes steric and electrostatic fields, but also calculates hydrophobic, hydrogen-bond donor (HBD), and hydrogen-bond acceptor (HBA) fields. For the distance dependence between the probe atom and the molecule atoms, a Gaussian function was used. Because of the different shape of the Gaussian function, the similarity indices calculated at all grid points, both inside and outside the molecular surface.
2.7 Hologram QSAR (HQSAR)
HQSAR is a technique based on the concept of using molecular substructures expressed in a binary pattern (molecular hologram) as descriptors in QSAR models. The premise of HQSAR is that the two-dimensional (2D) fingerprint encodes the structure of a molecule, which is a key determinant of all molecular properties (Kulkarni et al., 2008). 2D chemical database storage and searching technologies rely on linear notations that define chemical structures (WLN–Wiswesser line-formula notation; SMILES–simplified molecular input line entry system; SLN–SYBYL line notation). The process involves the generation of fragments that are hashed into array bins in the range of 1 to L (length) wherein the array is called molecular hologram and the bin occupancies are the descriptor variables. In this study, fingerprints were generated for all substructures between four and seven atoms in size for all molecules. The substructure fingerprints were then hashed into hologram bins with lengths of 97, 151, 199, 257, 307 and 353. Leave-one-out (LOO) and cross-validation (CV) methods were applied to determine the number of components that yield a good predictive model.
2.8 Partial least square (PLS) analysis
CoMFA, CoMSIA and HQSAR models were derived using PLS regression analysis. Calculated CoMFA and CoMSIA descriptors were used as independent variables and PKB inhibitory activity (pIC50) as the dependent variables in the PLS analysis. In HQSAR, PLS analysis yields a mathematical equation that related the molecular hologram bin values to the inhibition activity of the compounds in the database (Honorio et al., 2005). PLS analysis was performed using the leave-one-out (LOO) and cross-validation (CV) methods for 3D and HQSAR analysis, which gives q2 and , respectively as a statistical index of predictive power. The non cross-validated models were assessed by the conventional correlation coefficient (r2), standard error of estimation (SEE) and F values. A 100-cycle bootstrap analysis was performed to assess the statistical confidence of the derived models. The mean correlation coefficient is represented as bootstrap r2 ( ). The PLS analysis was then repeated with no validation using the optimal number of components to generate CoMFA, CoMSIA and HQSAR models (Cramer III et al., 1988).
2.9 Predictive r2 value
The predictive r2 ( ) was based only on the molecules (9 compounds) not included in the training set and is defined as = SD-PRESS/SD where, SD is the sum of the squared deviations between the inhibitory activity of molecules in a test set and the mean inhibitory activity of a training set molecules, and PRESS is the sum of squared deviations between predicted and actual activity values for every molecule in a test set.
3 Results and discussion
3.1 Results of the CoMFA analysis
The statistical parameters of standard CoMFA models constructed with steric and electrostatic fields are given in Table 2. The q2,
,
,
, F and SEE values were computed as defined in SYBYL. PLS analysis showed a q2 value of 0.627 and
of 0.644. A non-cross-validated PLS analysis results in a conventional r2 of 0.867, F = 1511 and a standard error of estimation (SEE) of 0.067. In both steric and electrostatic field contributions, former accounts for 0.544, while the latter contributes 0.456, indicating that the steric field contributed slightly higher to the binding affinity. A high bootstrapped r2 (0.945) value and low standard deviation (0.004) suggest a high degree of confidence in the analysis. The predicted and experimental pIC50 and residual values are listed in Table 3, and the correlation between the predicted and the experimental pIC50 of training and test set is depicted in Fig. 2A.
Compound
Experimental pIC50
Predicted pIC50
Residual
CoMFA
CoMSIA
HQSAR
CoMFA
CoMSIA
HQSAR
1
8.824
8.804
8.884
9.114
−0.02
0.06
0.29
2
9.000
9.140
8.980
9.080
0.14
−0.02
0.08
3
8.745
8.735
8.755
9.025
−0.01
0.01
0.28
4
6.021
6.011
5.981
5.311
−0.01
−0.04
−0.71
5
7.292
7.192
7.332
6.682
−0.10
0.04
−0.61
6
8.149
8.149
8.079
8.409
0.00
−0.07
0.26
7⁎
9.222
11.232
11.012
9.462
2.01
1.79
0.24
8
9.469
9.499
9.459
9.649
0.03
−0.01
0.18
9
6.652
6.572
6.582
5.742
−0.08
−0.07
−0.91
10
7.602
7.682
7.692
7.332
0.08
0.09
−0.27
11
7.229
7.129
7.199
6.719
−0.10
−0.03
−0.51
12
6.983
7.013
6.983
7.473
0.03
0.00
0.49
13
5.277
5.097
5.117
5.437
−0.18
−0.16
0.16
14
5.294
5.444
5.464
5.774
0.15
0.17
0.48
15
4.832
4.852
4.872
4.622
0.02
0.04
−0.21
16
4.714
4.714
4.714
4.704
0.00
0.00
−0.01
17
4.964
4.944
4.974
4.674
−0.02
0.01
−0.29
18
4.621
4.701
4.631
4.491
0.08
0.01
−0.13
19
6.161
6.241
6.131
5.491
0.08
−0.03
−0.67
20
5.170
5.180
5.160
4.370
0.01
−0.01
−0.80
21
5.096
5.046
5.116
4.526
−0.05
0.02
−0.57
22
5.394
5.434
5.374
4.494
0.04
−0.02
−0.90
23
6.058
6.078
6.078
6.628
0.02
0.02
0.57
24
6.495
6.495
6.445
5.645
0.00
−0.05
−0.85
25
6.489
6.519
6.519
5.919
0.03
0.03
−0.57
26⁎
6.836
7.666
7.706
6.646
0.83
0.87
−0.19
27
6.848
6.768
6.758
7.188
−0.08
−0.09
0.34
28
7.854
7.784
7.794
8.764
−0.07
−0.06
0.91
29
6.444
6.434
6.424
7.054
−0.01
−0.02
0.61
30
7.638
7.678
7.688
7.778
0.04
0.05
0.14
31
6.564
6.574
6.594
6.674
0.01
0.03
0.11
32⁎
6.301
7.781
8.381
6.641
1.48
2.08
0.34
33⁎
5.959
8.009
7.469
6.469
2.05
1.51
0.51
34⁎
4.620
4.580
5.260
5.050
−0.04
0.64
0.43
35
6.703
6.663
6.723
7.253
−0.04
0.02
0.55
36
7.721
7.741
7.761
7.671
0.02
0.04
−0.05
37⁎
7.102
5.712
5.632
6.702
−1.39
−1.47
−0.40
38
8.013
8.043
7.983
8.533
0.03
−0.03
0.52
39
6.556
6.496
6.526
7.006
−0.06
−0.03
0.45
40
5.495
5.535
5.505
6.325
0.04
0.01
0.83
41
6.175
6.145
6.145
7.065
−0.03
−0.03
0.89
42
4.491
4.501
4.491
4.921
0.01
0.00
0.43
43
6.547
6.587
6.577
6.357
0.04
0.03
−0.19
44
6.119
6.159
6.129
6.639
0.04
0.01
0.52
45
4.633
4.603
4.623
4.143
−0.03
−0.01
−0.49
46
4.366
4.336
4.406
4.476
−0.03
0.04
0.11
47⁎
5.573
6.393
6.133
5.973
0.82
0.56
0.40
48
5.190
5.140
5.030
4.810
−0.05
−0.16
−0.38
49
5.197
5.177
5.197
4.947
−0.02
0.00
−0.25
50
5.507
5.507
5.737
5.067
0.00
0.23
−0.44
51
6.917
6.857
6.857
6.017
−0.06
−0.06
−0.90
52
6.754
6.814
6.804
5.904
0.06
0.05
−0.85
53
6.788
6.788
6.798
6.128
0.00
0.01
−0.66
54⁎
7.161
8.721
8.591
7.941
1.56
1.43
0.78
55⁎
6.833
7.493
7.603
7.423
0.66
0.77
0.59
56
5.476
5.506
5.476
6.096
0.03
0.00
0.62
Plot of experimental versus predicted activity of training and test set compounds based on (A) CoMFA model, (B) CoMSIA model, (C) HQSAR model.
3.2 Results of the CoMSIA analysis
CoMSIA offered steric, electrostatic, hydrophobic, HBD and HBA fields’ information. CoMSIA models were generated using steric, electrostatic, hydrophobic, HBD and HBA fields. CoMSIA models showed higher correlation and high predictive properties. We found that the CoMSIA descriptors such as steric, electrostatic, hydrophobic, and HBD fields played a significant role in the prediction of inhibitory activity. These factors result in best CoMSIA models. Statistically significant CoMSIA model gives q2 of 0.598, of 0.563, r2 of 0.865, F of 1258 and SEE of 0.074 values with 6 components. The corresponding field contributions are 0.154 (steric), 0.183 (electroststic), 0.236 (hydrophobic), 0.235 (HBD) and 0.193 (HBA). Plot of experimental and predicted pIC50 of training and test sets is depicted in Fig. 2B.
3.3 Results of the HQSAR analysis
As per HQSAR calculations, the lowest SEE of 0.048 occurred at a cross-validated q2 of 0.687 with 3 components. The hologram result with the lowest SEE has a hologram length of 353. The PLS analysis gave a conventional r2 of 0.868 for all the studied compounds.
3.4 Predictive power of CoMFA, CoMSIA and HQSAR models
The predictive abilities of QSAR models were further validated using a test set of 9 compounds, not included in the model generation study. The predicted r2 ( ) values of CoMFA, CoMSIA and HQSAR models are 0.603, 0.613 and 0.742, respectively (Table 2). By comparison of experimental and predicted pIC50 values of a test set of compounds, it is observed that CoMFA, CoMSIA and HQSAR models performed well in the predication of inhibitory activity.
3.5 CoMFA contour maps
The CoMFA and CoMSIA contour maps are more or less similar and both models established good predictive ability. Compound 8 is shown inside the field for demonstrating contours generated from CoMFA and CoMSIA analysis. favored and disfavored levels fixed at 80% and 20%, respectively. In the CoMFA steric contour map (Fig. 3A) reference molecule’s (8) 3-methyl group substituted on the pyrazolo[3,4-c]pyridine ring was observed in the region of green colored (80% contribution) contour, which showed favorable bulky substitution. Differences in the activity of 8 and 7 (8, IC50 = 0.34 nM, 7 = 0.6 nM) were due to the presence of the more sterically favored 3-methyl group in 8. Similar could be the reason for activity diversity in molecules 10 (IC50 = 25 nM) and 11 (IC50 = 59 nM). In the case of inactive molecules, the extension of various groups led to the loss of inhibitory activity as they were oriented away from the green contour map as in the case of molecules 9 (IC50 = 223 nM) and 4 (IC50 = 953 nM). A second favorable steric contour was found at the methylene spacer in between pyridine and 3-indole ring system indicating a favorable effect of steric bulk. The bulky/steric unfavorable yellow colored (20% contribution) region flanked near the pyrazolo[3,4-c]pyridine ring system showed the sterically disfavored point. Unsubstituted pyrazolo[3,4-c]pyridine ring in compound 1 showed better activity (IC50 = 1.5 nM) as compared to sterically favored (–CF3) substituted pyrazolo[3,4-c]pyridine in compound 3 (IC50 = 1.8 nM). A second unfavorable yellow region flanked near the propan-2-amine bridge suggested that there is a definite structural requirement of a substructure with appropriate shape to exhibit good inhibitory activity. The CoMFA electrostatic contour map is shown in Fig. 3B. A large blue colored (80% contribution) contour was observed around the indole ring and indicated that electropositive groups in this region would increase the activity. For example, compound 10 (IC50 = 25 nM) had a phenyl ring in this region and exhibits good activity as compared to compound 4 (IC50 = 953 nM), which has an indole ring in this region. The electrostatic contour map displays three red colored polyhedrons (20% contribution). A large red colored contour close to the 2-amine group suggested that electronegative groups at this position would increase the activity. A small red colored contour was very close to negatively charged oxygen atom of propan-2-amine linker, which is necessary for a red colored favorable isopleth in proximity to this area. Oxygen atom in such a position usually can form H-bonds with amino acid residues of the binding site of PKB/Akt. Similar could be the reason for diverse activity in molecules 8 (IC50 = 0.34 nM) and 35 (IC50 = 198 nM). A small red colored contour at the methylene spacer indicated that the presence of electronegative substituents is for good activity.
CoMFA (std∗coeff) contour maps. Compound 8 is shown inside the field, (A) contour maps of CoMFA steric map shown in green (80% contribution) refer to sterically favored regions; yellow (20% contribution) indicates disfavored areas, (B) contour maps of CoMFA electrostatic field. Electrostatic contour map shown in red (20% contribution) indicates regions where negatively charged substituents are favored and blue contours (80% contribution) refer to regions where negatively charged substituents are disfavored.
3.6 CoMSIA contour maps
CoMSIA contribution maps denote those areas within the specified region where the presence of a group with a particular physicochemical property will be favored or disfavoured for good inhibitory activity. CoMSIA calculates both steric and electrostatic fields, as in CoMFA, but additionally uses hydrophobic, HBD and HBA fields. favored and disfavored levels fixed at 80% and 20%, respectively The CoMSIA hydrophobic contour map is shown in Fig. 4A, represented by yellow (80% contribution) and gray (20% contribution) colored contours. Yellow colored contours indicated the regions where hydrophobic groups on ligands are favored and gray colored contours represent those areas where hydrophobic groups are unfavored (or favorable for hydrophilic groups on ligands). One large yellow colored polyhedral is observed near the indole ring indicating that the hydrophobic property in this region would increase the activity. It can be observed by comparing the structures of 10 (25 nM) and 5 (51 nM), 5 contains the indole ring which is less hydrophobic than the phenyl ring of 10 in this region. Second small yellow colored region was observed near the 3-methyl group substituted on the pyrazolo[3,4-c]pyridine ring system suggesting that hydrophobic groups in this region favor Akt1 inhibitory activity. Difference in the activity of 3 and 4 (3, IC50 = 1.8 nM, 4 = 953 nM) was due to the presence of a more sterically favored –CF3 group in 3. A yellow colored contour in this region of the hydrophobic map was in agreement with the green colored contour of the steric map. A large gray colored contour is observed near the pyrazolo[3,4-c]pyridine ring, indicating that the hydrophilic group in this region would increase the activity. This can be explained be comparing structure and activity of compound 1 (IC50 = 1.5 nM) and 6 (IC50 = 7.1 nM). A second small gray colored contour was observed near the propan-2-amine bridge, illustrated that adding the hydrophobic group at this position would decrease the activity. The graphical interpretation of the HBD interactions in the CoMSIA model is represented in Fig. 4B. Cyan colored contours (80% contribution) indicated the regions where HBD substituents on ligands are favored and purple colored contours (20% contribution) represent those areas where HBD properties on compounds are disfavored. In the HBD contour map, three cyan colored areas were observed; one was observed away from the molecules area and indicated that there is no significance of HBD in this region, and the other two were observed near the pyrazolo[3,4-c]pyridine ring, especially one at the protonated –N atom of the pyrazolo ring, suggesting a favorable H-bonding between the active site amino acid and –N atom. It is indeed in case of compound 2 (IC50 = 1.0 nM) and 40 (IC50 = 3200 nM). A large purple colored region was observed near the propan-2-amine bridge and a small purple colored contour observed at –N atom of pyridine ring indicated that HBD properties in this region would disfavoured the activity. The graphical interpretation of the HBA interactions in the CoMSIA model is shown in Fig. 4C. Magenta colored (80% contribution) and red colored (20% contribution) contours represented the area where HBA are favored and disfavored, respectively. Two magenta colored contours and red colored contours were in complex with each other and an overlap of contours occurred near the indole ring system which is very complex to interpret. Analysis of CoMFA and CoMSIA contour plots offered enough information to understand the importance of substituents at a particular position for better activity.
CoMSIA (std∗coeff) contour maps. Compound 8 is shown inside the field. Hydrophobic field (A), HBD field (B), and HBA field (C). Yellow and gray contours indicate regions where hydrophobic groups favored and disfavored the activity, respectively. Cyan contours represent areas where HBD is favored. Magenta and red contours represent areas where HBA is favored and disfavored, respectively. Favored and disfavored levels are fixed at 80% and 20%, respectively.
4 Conclusions
PKB/Akt has emerged as an attractive target for the development of novel anticancer therapeutics. In this study, we described 3D QSAR and HQSAR analysis as a rational strategy for the design of novel PKB inhibitors, using three different alignment methods. All the models (CoMFA, CoMSIA and HQSAR) were found satisfactory according to the statistical parameters. CoMFA and CoMSIA models are satisfactory according to the statistical results as well as the contour maps analysis. The present QSAR approach provides useful information to design novel derivatives with higher selectivity and efficacy for PKB/Akt inhibition.
Acknowledgments
The authors would like to thank the Nirma University, Ahmedabad, India for providing the necessary facilities.
References
- Design, synthesis, and structure–activity relationships of 3-ethynyl-1H-indazoles as inhibitors of the phosphatidylinositol 3-kinase signaling pathway. J. Med. Chem.. 2010;53:8368-8375.
- [Google Scholar]
- Validation of the general purpose Tripos 5.2 force field. J. Comput. Chem.. 1989;10:982-1012.
- [Google Scholar]
- Targeted small molecule inhibitors of protein kinase B as anticancer agents. Anti-Cancer Agents Med. Chem.. 2009;9:32-50.
- [Google Scholar]
- Cross validation, bootstrapping, and partial least squares compared with multiple regression in conventional QSAR studies. Quant. Struct. Act. Relat.. 1988;7:18-25.
- [Google Scholar]
- A structural comparison of inhibitor binding to PKB, PKA and PKA–PKB chimera. J. Mol. Biol.. 2007;367:882-894.
- [Google Scholar]
- Iterative partial equalization of orbital electronegativity- a rapid access to atomic charges. Tetrahedron. 1980;36:3219-3228.
- [Google Scholar]
- Hologram quantitative structure-activity relationships for a series of farnesoid X receptor activators. Bioorg. Med. Chem. Lett.. 2005;15:3119-3125.
- [Google Scholar]
- Scoring noncovalent protein–ligand interactions: a continuous differentiable function tuned to compute binding affinities. J. Comput.-Aided Mol. Des.. 1996;10:427-440.
- [Google Scholar]
- CoMFA and HQSAR studies on 6,7-dimethoxy-4-pyrrolidylquinazoline derivatives as phosphodiesterase10A inhibitors. Bioorg. Med. Chem.. 2008;16:3675-3686.
- [Google Scholar]
- Li, Q., Li, T., Zhu, G.D., Gong, J., Claibone, A., Dalton, C., Luo, Y., Johnson, E.F., Shi, Y., Liu, X., Klinghofer, V., Bauch, J.L., Marsh, K.C., Bouska, J.J., Arries, S., Jong, R.D., Oltersdorf, T., Stoll, V.S., Jakob, C.G., Rosenberg, S.H., Giranda, V.L., 2006. Discovery of trans-3,4’-bispyridinylethylenes as potent and novel inhibitors of protein kinase B (PKB/Akt) for the treatment of cancer: Synthesis and biological evaluation. Bioorg. Med. Chem. Lett. 1679–1685.
- Allosteric Akt (PKB) inhibitors: discovery and SAR of isoenzyme specificity. Bioorg. Med. Chem. Lett.. 2005;15:761-764.
- [Google Scholar]
- Recent progress in the development of ATP-competitive and allosteric Akt kinase inhibitors. Curr. Top. Med. Chem.. 2007;7:1349-1363.
- [Google Scholar]
- Discovery of a novel class of AKT pleckstrin homology domain inhibitors. Mol. Cancer Ther.. 2008;7:2621-2632.
- [Google Scholar]
- An investigation of structurally diverse carbamates for acetylcholinesterase (AChE) inhibition using 3D-QSAR analysis. J. Mol. Graphics Modell.. 2008;27:197-208.
- [Google Scholar]
- Phosphorylation and regulation of Akt/PKB by the rictor-mTOR complex. Science. 2005;307:1098-1101.
- [Google Scholar]
- Unravelling the activation mechanisms of protein kinase B/Akt. FEBS Lett.. 2003;546:108-112.
- [Google Scholar]
- 3D QSAR and molecular docking analysis of (4-piperidinyl)-piperazines as acetyl-CoA carboxylases inhibitors. Arabian J. Chem.. 2017;10:S617-S626.
- [Google Scholar]
- QSAR models of cytochrome P450 enzyme 1A2 inhibitors using CoMFA, CoMSIA and HQSAR. SAR QSAR Environ. Res.. 2011;22:681-697.
- [Google Scholar]
- The phosphatidylinositol 3-kinase AKT pathway in human cancer. Nat. Rev. Cancer. 2002;2:489-501.
- [Google Scholar]
- CoMFA and CoMSIA studies on aryl carboxylic acid amide derivatives as dihydroorotate dehydrogenase (DHODH) inhibitors. Cur. Comp.-Aided Drug Des.. 2012;8:271-282.
- [Google Scholar]
- Pharmacophore modeling, virtual screening, docking and in silico ADMET analysis of protein kinase B (PKB β) inhibitors. J. Mol. Graphics Modell.. 2013;42:17-25.
- [Google Scholar]
- Design and synthesis of pyridine–pyrazolopyridine-based inhibitors of protein kinase B/Akt. Bioorg. Med. Chem. Lett.. 2007;15:2441-2452.
- [Google Scholar]
